UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl) UvA-DARE (Digital Academic Repository) Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes de Wolf, G. Link to publication Citation for published version (APA): de Wolf, G. (2010). Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes. General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. Download date: 29 Jun 2020
Transcript
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes
de Wolf, G.
Link to publication
Citation for published version (APA):de Wolf, G. (2010). Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes.
General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.
The research described in this thesis was conducted at the Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, and the Academic Medical Center (University of Amsterdam), Department of Social Medicine, as part of the Amsterdam Born Children and their Development (ABCD) study. Financial support for the ABCD study was granted by the Netherlands Organisation for Health Research and Development (ZonMw), the Public Health Service of Amsterdam, the Municipal Council of Amsterdam, and the Academic Medical Center. This thesis was printed with financial support of the Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, and the Academic Medical Center.
Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctoraan de Universiteit van Amsterdamop gezag van de Rector Magnificus
prof.dr. D.C. van den Boomten overstaan van een door het college voor promoties
ingestelde commissie,in het openbaar te verdedigen in de Agnietenkapel
op donderdag 30 september 2010, te 14:00 uur
door
Geertje de Wolf
geboren te Zeist
Promotiecommissie
Promotores: Prof.dr. G.J. Bonsel Prof.dr. P. Cuijpers
Copromotor: Dr. M.F. van der Wal
Overige leden: Prof.dr. B.R.H.M. van den Bergh Prof.dr. J.A. van der Post Dr. T.J. Roseboom Prof.dr. R.P.M. Steegers-Theunissen Prof.dr. K. Stronks Prof.dr. A.P. Verhoeff Dr. C. de Weerth
Faculteit der Geneeskunde
“HEER, U kent mij, U doorgrondt mij,U was het die mijn nieren vormde,
die mij weefde in de buik van mijn moeder. Ik loof U voor het ontzaglijke wonder van mijn bestaan,
wonderbaarlijk is wat U gemaakt hebt.“
(Psalm 139:1,13-14, Nieuwe Bijbelvertaling)
Contents
Chapter 1 General introduction 9
Chapter 2 Ethnic differences in term birthweight: the role of constitutional 21 and environmental factors. Paediatric and Perinatal Epidemiology 2008;22(4):360-8.
Chapter 3 Ethnic differences in preterm birth and its subtypes: the effect of 35 a cumulative risk profile. BJOG 2008;115(6):710-9.
Chapter 4 Psychosocial problems and continued smoking during pregnancy. 53 Addictive Behaviors 2009;34(4):403-6.
Chapter 5 Maternal depressive symptoms in relation to perinatal mortality 63 and morbidity: Results from a large multi-ethnic cohort study. Psychosomatic Medicine; in press.
Chapter 6 Maternal cortisol and offspring birthweight: Results from a large 81 prospective cohort study. Psychoneuroendocrinology 2010;35(5):644-52.
Chapter 7 Maternal vitamin B-12 and folate status during pregnancy and 97 excessive infant crying. Submitted.
A healthy mother and child is the most precious outcome of human reproduction. Although medical focus and public’s concern is on unhealthy outcomes, the startling fact is that at a superficial level human reproduction rarely goes wrong despite its complexity. The development of a fetus is a mysterious but astonishing process. Growing evidence points towards a strong interaction with the maternal environment, preparing the fetus for the living conditions outside the womb.1-3 Many believe this to be an extremely ‘clever’, ‘efficient’, ultimately ‘selfish’ process, which supports the survival of human race.
Perinatal health epidemiology studies the health status of a mother and her baby during pregnancy, birth and the early postpartum period, and identifies causes and consequences of perinatal health problems. Main conventional indicators of perinatal health problems include on the one hand maternal mortality and morbidity (e.g. eclampsia), and on the other hand perinatal (fetal and neonatal) mortality and morbidity. Perinatal morbidity is mostly indicated by the following outcome measures: congenital anomalies, low birthweight or small-for-gestational-age (SGA), preterm birth (PTB) and a low Apgar score.4 Although the overall perinatal mortality rate decreased over the last decades due to improved living conditions and perinatal health care, the mortality rate in the Netherlands, and especially in its four largest cities (including Amsterdam), is high (10.0 per 1000 births) compared to other European countries.5,6 Partly as a consequence of the higher survival rate, perinatal morbidity rates increased over the last decades.4,7-9 Perinatal morbidity can be seen as a short-term marker of a poor intrauterine
Figure 1.1 Conceptual model for the emergence and persistence of health problems over the life course and across
_________________________________________________________________ General introduction
1
environment, which is not without long-term consequences. Pioneering research by Barker and colleagues1,3 strongly suggest a connection between perinatal morbidity and adult diseases like type 2 diabetes or cardiovascular disease; this is now known as the fetal programming hypothesis. Following this hypothesis, health problems may become a reinforcing circle over the life course and over generations (Figure 1.1).10,11 Increasing awareness on the multigenerational dimension of the reproductive cycle, and on the underlying genetic and epigenetic pathways, has made the field of perinatal health epidemiology more important than ever.
Ethnicity and perinatal health epidemiologyWorldwide, large differences in perinatal health outcomes exist between ethnic groups.12,13 In the United States (USA), for example, black infants are more than twice as likely to die as white infants, and in the United Kingdom (UK), Asian infants are more than twice as likely to have a low birthweight (<2500 g) as white infants.12,14 Also in the Netherlands ethnic differences in perinatal health outcomes exist, with the ethnic minority groups (e.g. Surinamese, Turkish or Moroccan groups) often having the worst outcomes.15-17 After years of research, the etiology of ethnic differences in perinatal health outcomes is still hardly understood. Ethnic inequalities in socioeconomic position or genetic differences between races are often held responsible, yet they lack explanatory exclusiveness and moreover, they lack preventive possibilities.10,12 Evidence is growing for an explanatory model in which ‘distal’ factors like socioeconomic position and genetic environment place pregnant women at greater susceptibility to ‘proximal’ risk factors, i.e. behavioral and physiological factors that directly influence maternal and fetal health.18 Potential proximal risk factors include stress, nutrition, infections, working conditions, living conditions, substance use, parity, maternal age, medical conditions and obstetric problems.18-23 In case of multiple (distal and/or proximal) risk factors, potential interaction effects could substantially increase the risk for perinatal health problems.18,19 These risk factors could, in theory, be mediators in the association between ethnic background and perinatal health outcomes.12 If confirmed, this will have huge implications for reducing ethnic differences in perinatal health outcomes. Below, we will expand on one of the potential mediating factors: maternal psychosocial stress.
Psychobiological processes in perinatal health epidemiologyAs the overall prevalence of mental illnesses and stress-related problems is rising, the psychosocial health status of mother and child becomes more and more important in the field of perinatal health epidemiology. Women are two to three times more likely to experience mental health problems than men, particularly in the reproductive age.24 The prevalence of antenatal depression is, for example, estimated to be as high as 20%,24 although some claim the prevalence is not much unlike the prevalence in a comparable female population. While psychosocial problems in general already have a large impact on a woman’s life and that of her environment, psychosocial problems during pregnancy even have a much larger impact as it also affects the health of the fetus.2,24 The commonly used term ‘psychosocial problems’ or ‘psychosocial stress’ covers a broad range of psychosocial indicators including emotional problems and disorders (e.g. depression,
anxiety), social stress, life events, work and household stress, domestic violence, racism and, in the case of pregnancy, pregnancy-related anxiety.25 Although it is not really appropriate to combine mental health problems with stress-related problems, for several reasons it is not uncommon in the field of perinatal health research. First, most studies measure mental health through self-report scales like the Center for Epidemiologic Studies Depression scale (CES-D)26 or the State-Trait Anxiety Inventory Questionnaire (STAI),27 which are not sufficiently valid to diagnose mental illness; they describe a mood state rather than a clinical entity or an etiologically coherent disease. Second, both mental health problems and stress-related problems share part of their symptoms and often appear as comorbid disorders.28 Third and most important, they probably share the same psychopathological mechanisms into perinatal health outcomes.25,28
Numerous studies have examined the potentially negative effects of maternal psychosocial problems during pregnancy on perinatal health outcomes. Higher prevalences of PTB, low birthweight, SGA, certain congenital anomalies and a low Apgar score have been observed among infants born to mothers with psychosocial problems.8,25,28-39 Nevertheless, evidence on a causal contribution is far from consistent; to a large part this can be attributed to methodological differences between studies. Psychosocial problems, for instance, are operationalized by a wide variety of self-report scales; furthermore, they are measured during different time windows of pregnancy. In addition, studies differ in their consideration of potential confounding or mediating factors, such as maternal smoking behavior or nutritional intake.24,28,35,40-42 Besides the search for distinct evidence on the negative effects of maternal psychosocial problems on perinatal health outcomes, it is even more important to explore the psychopathological framework through which psychosocial stress affects fetal growth and development. Several psychopathological mechanisms have been proposed, with on top the involvement of the hypothalamic-pituitary-adrenal (HPA) axis. When a pregnant woman is exposed to a stressor, the HPA-axis releases, among others, cortisol, also called the ‘stress-hormone’, in order to help the body to recover. Normally, a placental barrier (including the enzyme 11β-hydroxysteroid-dehydrogenase type 2) protects the fetus from large amounts of maternal cortisol, however, in case of exposure to excess levels of cortisol the placental barrier may be too weak to protect the fetus any longer.2,25 This may result in reduced fetal growth, preterm parturition or teratogenic effects.28,37,38,43 Yet evidence for such a psychopathological mechanism is, so far, mainly provided by animal rather than human research.44-48
The prevalence of maternal psychosocial problems during pregnancy is higher among ethnic minority groups. Turkish and Moroccan groups in Amsterdam had, for example, a higher prevalence of depressive and anxiety disorders (respectively 18.7% and 9.8%) compared to the Dutch group (6.6%).49 This higher prevalence of maternal psychosocial problems could, in theory, be responsible for the higher prevalence of adverse perinatal health outcomes among the ethnic minority groups.50,51
Psychobiological processes involved in fetal and infant health outcomes also include fetal origins of infant psychosocial health. In the first months of life, excessive crying behavior is a reasonable indicator of infant psychosocial stress. While many theories exist on the etiology of excessive infant crying, there is no general agreement.52 As a result, preventive measures are scarce and treatment of infants with excessive crying behavior is, so far, limited to advice parents to reduce stimulation or to
13
_________________________________________________________________ General introduction
1
try hypoallergenic formula milk.53 Recently, the role of maternal nutrients in fetal neurodevelopment and, ultimately, infant behavior has gained interest.24,54 An early nutritional origin in infant crying behavior has, however, not yet been explored. Vitamin B-12 and folate seem to be good candidates because of their involvement in essential neuroendocrinological processes.55,56
Aims and objectives of this thesis
The main aims of this thesis are (I) to explore perinatal health epidemiology in the multi-ethnic city of Amsterdam, the Netherlands, by examining (a) the perinatal health outcomes among the main ethnic groups in Amsterdam, and (b) to what extent ethnic disparities in perinatal health outcomes can be explained by ethnic differences in maternal risk factors during pregnancy; and (II) to elucidate psychobiological processes involved in (ethnic disparities in) perinatal health outcomes, by examining (c) the association between maternal psychosocial well-being during pregnancy (as indicated by self-report scales and the biomarker cortisol) and perinatal health outcomes, and vice versa, (d) the association between maternal risk factors during pregnancy, in particular maternal nutritional status, and infant psychosocial well-being (as indicated by infant crying behavior).
General Methods: ABCD study
To explore the association between (ethnic differences in) maternal risk factors during pregnancy and (ethnic disparities in) perinatal and infant health outcomes, a large prospective multi-ethnic cohort study among pregnant women in Amsterdam, the Netherlands, was initiated: the Amsterdam Born Children and their Development (ABCD) study. Initiators were the Public Health Service and Academic Medical Center in Amsterdam. Approval of the study was obtained from the Central Committee on Research involving Human Subjects in the Netherlands, the Medical Ethical Committees of participating hospitals and the Registration Committee of Amsterdam.
In the period from January 2003 until March 2004, all pregnant women living in Amsterdam were invited to enroll in the ABCD study at their first prenatal visit (median 13 pregnancy weeks, interquartile range 3 weeks) to participating obstetric care providers (general practitioners, midwives and gynecologists; overall participation rate 96%). For all approached women, the care provider completed a registration form with personal data such as name, address and country of birth. Based on this information, a pregnancy-questionnaire was sent to the pregnant woman’s home address within two weeks, to be returned by prepaid mail. Reminders were sent two weeks after the initial mailing. Questionnaires were in Dutch and accompanied by an English, Turkish or Arabic copy depending on the woman’s country of birth. Furthermore, Turkish and Moroccan women had the opportunity to contact Turkish- or Arabic-speaking trained female interviewers for oral administration of the questionnaire. This approach enhanced the inclusion of pregnant women from the main ethnic groups in Amsterdam: the Dutch, Surinamese, Antillean/Aruban, Turkish, Moroccan and Ghanaian group.
The pregnancy-questionnaire covered sociodemographic characteristics, (psychosocial) health status, lifestyle and obstetric history. Maternal psychosocial well-being during pregnancy was operationalized with several self-report scales. Depressive symptoms were measured by the Dutch version of the Center for Epidemiologic Studies Depression scale (CES-D);26,57 the 20-item CES-D scale assesses the self-reported frequency of depressive symptoms experienced over the past week. Anxiety was measured by the Dutch version of the state-scale of the State-Trait Anxiety Inventory (STAI);27,58 the 20-item state-scale asks subjects how they felt in the past week. State anxiety is conceptualized as a transient emotional condition. Pregnancy-related anxiety was measured by three scales (‘fear of giving birth’, ‘fear of bearing a handicapped child’, ‘concern about one’s appearance’) based on the Pregnancy Related Anxiety Questionnaire – Revised version (PRAQ-R).59,60 Parenting stress was measured by the Dutch version of the frequency scale of the Parenting Daily Hassles (PDH);61,62 the PDH scale lists 20 typical everyday events in parenting young children. Job strain was measured by four scales (‘pace of work’, ‘mental workload’, ‘physical workload’, ‘job control’) of the Dutch Work Experience and Appreciation Questionnaire (VBBA), based on the Job Content Questionnaire.63,64
Pregnant women were also asked to participate in the ABCD biomarker study. For this purpose, an extra blood sample was drawn during routine blood collection at the first prenatal visit. Blood samples were analyzed for nutrients (including vitamin B-12 and folate) and hormones (including cortisol) at the National Institute for Public Health and the Environment (RIVM), Bilthoven, and the Medical Laboratory Dr. Stein & Collegae, Maastricht, the Netherlands.
Three months after delivery (median 13 weeks postpartum, interquartile range 1 week), a baby-questionnaire was sent to the women who had given consent for follow-up. The baby-questionnaire covered the delivery, maternal lifestyle, and (psychosocial) health of the mother and her baby, including infant crying behavior. Women were also retrospectively asked whether they were exposed to physical and/or sexual violence during pregnancy.
Perinatal health outcomes were obtained from the Youth Health Care registration at the Public Health Service in Amsterdam. According to Dutch law, all children born in Amsterdam after 24 weeks’ gestation (either stillborn or liveborn) must be registered at the municipality’s Registry Office, after which a report is sent to the Youth Health Care Department. Between the 4th and 7th day after delivery, nurses from this department visit the liveborn infants for screening purposes. At this visit, they also record the date of delivery, infant gender, birthweight and gestational age at birth as provided by the obstetric care provider. These data are close to complete except for migrating families. Additional perinatal outcome data was obtained from the Dutch Perinatal Registration (PRN), which includes the national obstetric database for midwives (LVR-1), the national obstetric database for gynecologists (LVR-2), and the national neonatal database for pediatricians and neonatologists (LNR). These national data cover 95% of pregnancies, in the larger cities close to 100%.
Of all approached pregnant women in Amsterdam (n = 12 373), 8266 women returned the pregnancy-questionnaire (response rate 67%) and 4389 women also participated in the biomarker study. Of the 7050 women who had given permission for follow-up, 5218 returned the baby-questionnaire (response rate 73%). An overview of the subsamples available for each of the chapters of this thesis is presented in a flowchart in Figure 1.2.
15
_________________________________________________________________ General introduction
1
Research questions
The following research questions will be addressed in the remaining chapters of this thesis:
Ethnicity and perinatal health outcomes1. (a) Do the birthweights of term-born infants differ between ethnic groups; and (b) to
what extent can the ethnic differences in birthweight be explained by constitutional vs. environmental risk factors? (Chapter 2)
2. (a) Does the prevalence of preterm birth (PTB), divided into spontaneous and iatrogenic preterm births, differ between ethnic groups; (b) to what extent can ethnic differences in PTB prevalence be explained by conventional risk factors; and (c) to what extent can a cumulation of conventional risk factors explain ethnic differences in PTB prevalence? (Chapter 3)
Maternal psychosocial and behavioral health 3. How do levels of maternal psychosocial problems during pregnancy relate to whether or not
a woman continues to smoke during pregnancy? (Chapter 4)
Psychobiological pathways into perinatal health outcomes4. (a) How does maternal depressive symptomatology during pregnancy relate to the prevalence
of four major perinatal health outcomes, i.e. PTB, small-for-gestational-age (SGA), child loss and a low Apgar score; (b) does maternal smoking during pregnancy mediate these associations; and (c) does ethnic background modify these associations? (Chapter 5)
5. (a) How does maternal cortisol concentration during pregnancy relate to fetal growth as measured by offspring birthweight and SGA risk at term; and (b) to what extent do maternal cortisol levels mediate the association between maternal psychosocial problems during pregnancy and fetal growth? (Chapter 6)
Psychobiological pathways into infant health outcomes6. (a) How do maternal vitamin B-12 and folate concentrations during pregnancy relate to
the prevalence of excessive infant crying; and (b) in what way are maternal psychological problems during pregnancy involved in the association of vitamin B-12 and folate status with excessive infant crying? (Chapter 7)
In the general discussion (Chapter 8), the answers to the abovementioned research questions are summarized and discussed. Furthermore, methodological considerations, implications for interventions and recommendations for future research are discussed.
_________________________________________________________________ General introduction
1
References
1. Barker DJ. The origins of the developmental origins theory. J Intern Med 2007;261:412-7.2. Seckl JR, Holmes MC. Mechanisms of disease: glucocorticoids, their placental metabolism and fetal ‘programming’ of adult pathophysiology. Nat Clin Pract Endocrinol Metab 2007;3:479-88.3. Gluckman PD, Hanson MA, Cooper C, Thornburg KL. Effect of in utero and early-life conditions on adult
health and disease. N Engl J Med 2008;359:61-73.4. EURO-PERISTAT project, with SCPE, EUROCAT, EURONEOSTAT. European perinatal health report.
Better statistics for better health in pregnant women and their babies. Paris: EURO-PERISTAT; 2008. Re-trieved from: http://www.europeristat.com.
5. Mohangoo AD, Buitendijk SE, Hukkelhoven CWPM, Ravelli ACJ, Rijninks-van Driel GC, Tamminga P, et al. Higher perinatal mortality in the Netherlands than in other European countries: the Peristat-II study. [In Dutch]. Ned Tijdschr Geneeskd 2008;152:2718-27.
6. De Graaf JP, Ravelli ACJ, Wildschut HIJ, Denktas S, Voorham AJJ, Bonsel GJ, et al. Perinatal outcomes in the four largest cities and in deprived neighbourhoods in the Netherlands. [In Dutch]. Ned Tijdschr Geneeskd 2008;152:2734-40.
7. Thorngren-Jerneck K, Herbst A. Low 5-minute Apgar score: a population-based register study of 1 million term births. Obstet Gynecol 2001;98:65-70.
8. Romo A, Carceller R, Tobajas J. Intrauterine growth retardation (IUGR): epidemiology and etiology. Pediatr Endocrinol Rev 2009;6(Suppl 3):332-6.9. McFarlin BL. Solving the puzzle of prematurity. Am J Nurs 2009;109:60-3.10. Gravlee CC. How race becomes biology: embodiment of social inequality. Am J Phys Anthropol
2009;139:47-57.11. Kuzawa CW. The development origins of adult health: intergenerational inertia in adaptation and disease.
In: Trevathan WR, McKenna JJ, editors. Evolutionary medicine and health: new perspectives. New York: Oxford University Press; 2008, pp. 325-49.
12. Gennaro S. Overview of current state of research on pregnancy outcomes in minority populations. Am J Obstet Gynecol 2005;192:S3-10.13. Gagnon AJ, Zimbeck M, Zeitlin J, Alexander S, Blondel B, Buitendijk S, et al. Migration to western indus-
trialised countries and perinatal health: a systematic review. Soc Sci Med 2009;69:934-46.14. Harding S, Rosato MG, Cruickshank JK. Lack of change in birthweights of infants by generational status
among Indian, Pakistani, Bangladeshi, Black Caribbean, and Black African mothers in a British cohort study. Int J Epidemiol 2004;33:1279-85.
15. Drooger JC, Troe JW, Borsboom GJ, Hofman A, Mackenbach JP, Moll HA, et al. Ethnic differences in prenatal growth and the association with maternal and fetal characteristics. Ultrasound Obstet Gynecol 2005;26:115-22.
16. Schulpen TW, van Steenbergen JE, van Driel HF. Influences of ethnicity on perinatal and child mortal-ity in the Netherlands. Arch Dis Child 2001;84:222-6.
17. Van der Wal MF, Uitenbroek DG, van Buuren S. Birthweight of children in Amsterdam by ethnic origin. [In Dutch]. Tijdschrift voor Gezondheidswetenschappen 2000;78:15-20.
18. Misra DP, Guyer B, Allston A. Integrated perinatal health framework. A multiple determinants model with a life span approach. Am J Prev Med 2003;25:65-75.
19. Green NS, Damus K, Simpson JL, Iams J, Reece EA, Hobel CJ, et al. Research agenda for preterm birth: recommendations from the March of Dimes. Am J Obstet Gynecol 2005;193:626-35.
20. Cogswell ME, Yip R. The influence of fetal and maternal factors on the distribution of birthweight. Semin Perinatol 1995;19:222-40.
21. Ergaz Z, Avgil M, Ornoy A. Intrauterine growth restriction-etiology and consequences: what do we know about the human situation and experimental animal models? Reprod Toxicol 2005;20:301-22.
22. Ramakrishnan U, Manjrekar R, Rivera J, Gonzales-Cossio T, Martorell R. Micronutrients and preg-nancy outcome: a review of the literature. Nutr Res 1999;19:103-59.
23. Iams JD. The epidemiology of preterm birth. Clin Perinatol 2003;30:651-64.24. Leung BM, Kaplan BJ. Perinatal depression: prevalence, risks, and the nutrition link - A review of the
literature. J Am Diet Assoc 2009;109:1566-75.25. Hobel CJ, Goldstein A, Barrett ES. Psychosocial stress and pregnancy outcome. Clin Obstet Gynecol
2008;51:333-48.26. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population.
Appl Psychol Meas 1977;1:385-401.27. Spielberger CD, Gorsuch RL, Lushene RE. STAI Manual for the State-Trait Anxiety Inventory. Palo Alto,
CA: Consulting Psychologists Press; 1970.28. Alder J, Fink N, Bitzer J, Hosli I, Holzgreve W. Depression and anxiety during pregnancy: a risk factor
for obstetric, fetal and neonatal outcome? A critical review of the literature. J Matern Fetal Neonatal Med 2007;20:189-209.
29. Kramer MS, Lydon J, Seguin L, Goulet L, Kahn SR, McNamara H, et al. Stress pathways to spontane-ous preterm birth: the role of stressors, psychological distress, and stress hormones. Am J Epidemiol 2009;169:1319-26.
30. Khashan AS, McNamee R, Abel KM, Mortensen PB, Kenny LC, Pedersen MG, et al. Rates of preterm birth following antenatal maternal exposure to severe life events: a population-based cohort study. Hum Reprod 2009;24:429-37.
31. Murphy CC, Schei B, Myhr TL, Du MJ. Abuse: a risk factor for low birth weight? A systematic review and meta-analysis. CMAJ 2001;164:1567-72.
32. Vrijkotte TG, van der Wal MF, van Eijsden M, Bonsel GJ. First-trimester working conditions and birth-weight: a prospective cohort study. Am J Public Health 2009;99:1409-16.
33. Khashan AS, McNamee R, Abel KM, Pedersen MG, Webb RT, Kenny LC, et al. Reduced infant birth-weight consequent upon maternal exposure to severe life events. Psychosom Med 2008;70:688-94.
34. Dole N, Savitz DA, Siega-Riz AM, Hertz-Picciotto I, McMahon MJ, Buekens P. Psychosocial factors and preterm birth among African American and White women in central North Carolina. Am J Public Health 2004;94:1358-65.
35. Dayan J, Creveuil C, Herlicoviez M, Herbel C, Baranger E, Savoye C, et al. Role of anxiety and depression in the onset of spontaneous preterm labor. Am J Epidemiol 2002;155:293-301.
36. Orr ST, James SA, Blackmore PC. Maternal prenatal depressive symptoms and spontaneous preterm births among African-American women in Baltimore, Maryland. Am J Epidemiol 2002;156:797-802.
37. Carmichael SL, Shaw GM. Maternal life event stress and congenital anomalies. Epidemiology 2000;11:30-5.38. Berle JO, Mykletun A, Daltveit AK, Rasmussen S, Holsten F, Dahl AA. Neonatal outcomes in offspring of
women with anxiety and depression during pregnancy. A linkage study from The Nord-Trondelag Health Study (HUNT) and Medical Birth Registry of Norway. Arch Womens Ment Health 2005;8:181-9.
39. Jablensky AV, Morgan V, Zubrick SR, Bower C, Yellachich L. Pregnancy, delivery, and neonatal complica-tions in a population cohort of women with schizophrenia and major affective disorders. Am J Psychiatry 2005;162:79-91.
40. Paarlberg KM, Vingerhoets AJ, Passchier J, Dekker GA, van Geijn HP. Psychosocial factors and pregnancy outcome: a review with emphasis on methodological issues. J Psychosom Res 1995;39:563-95.
41. Cliver SP, Goldenberg RL, Cutter GR, Hoffman HJ, Copper RL, Gotlieb SJ, et al. The relationships among psychosocial profile, maternal size, and smoking in predicting fetal growth retardation. Obstet Gynecol 1992;80:262-7.
42. Davis RL, Rubanowice D, McPhillips H, Raebel MA, Andrade SE, Smith D, et al. Risks of congenital mal-formations and perinatal events among infants exposed to antidepressant medications during pregnancy. Pharmacoepidemiol Drug Saf 2007;16:1086-94.
43. Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CA. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosom Med 1996;58:432-46.
44. French NP, Hagan R, Evans SF, Godfrey M, Newnham JP. Repeated antenatal corticosteroids: size at birth and subsequent development. Am J Obstet Gynecol 1999;180:114-21.
45. Rowland JM, Hendrickx AG. Corticosteroid teratogenicity. Adv Vet Sci Comp Med 1983;27:99-128.
19
_________________________________________________________________ General introduction
1
46. Seckl JR. Prenatal glucocorticoids and long-term programming. Eur J Endocrinol 2004;151(Suppl 3):U49-62.47. Diego MA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, Gonzalez-Quintero VH. Prenatal depres-
sion restricts fetal growth. Early Hum Dev 2009;85:65-70.48. Giurgescu C. Are maternal cortisol levels related to preterm birth? J Obstet Gynecol Neonatal Nurs
2009;38:377-90.49. De Wit MA, Tuinebreijer WC, Dekker J, Beekman AJ, Gorissen WH, Schrier AC, et al. Depressive and
anxiety disorders in different ethnic groups: a population based study among native Dutch, and Turkish, Moroccan and Surinamese migrants in Amsterdam. Soc Psychiatry Psychiatr Epidemiol 2008;43:905-12.
50. Hogue CJ, Bremner JD. Stress model for research into preterm delivery among black women. Am J Obstet Gynecol 2005;192:S47-55.
51. Giscombe CL, Lobel M. Explaining disproportionately high rates of adverse birth outcomes among African Americans: the impact of stress, racism, and related factors in pregnancy. Psychol Bull 2005;131:662-83.
52. Helseth S, Begnum S. A comprehensive definition of infant colic: parents’ and nurses’ perspectives. J Clin Nurs 2002;11:672-80.
53. Lucassen PL, Assendelft WJ, Gubbels JW, van Eijk JT, van Geldrop WJ, Neven AK. Effectiveness of treat-ments for infantile colic: systematic review. BMJ 1998;316:1563-9.
54. Black MM. Effects of vitamin B12 and folate deficiency on brain development in children. Food Nutr Bull 2008;29:S126-31.
55. Ikeda M, Asai M, Moriya T, Sagara M, Inoue S, Shibata S. Methylcobalamin amplifies melatonin-induced circadian phase shifts by facilitation of melatonin synthesis in the rat pineal gland. Brain Res 1998;795:98-104.
56. Dror DK, Allen LH. Effect of vitamin B12 deficiency on neurodevelopment in infants: current knowledge and possible mechanisms. Nutr Rev 2008;66:250-5.
57. Hanewald GJFP. CES-D: The Dutch version. A study on reliability and validity. [In Dutch]. Amsterdam: Universiteit van Amsterdam, Vakgroep Klinische Psychologie; 1987.
58. Van der Ploeg HM. Manual for the Self Evaluation Questionnaire (ZBV). [In Dutch]. Lisse: Swets & Zeitlin-ger; 2000.
59. Huizink AC, Mulder EJ, Robles de Medina PG, Visser GH, Buitelaar JK. Is pregnancy anxiety a distinctive syndrome? Early Hum Dev 2004;79:81-91.
60. Van den Bergh B. The influence of maternal emotions during pregnancy on fetal and neonatal behavior. Pre- Peri-Nat Psychol J 1990;5:119-30.
61. Crnic KA, Greenberg MT. Minor parenting stresses with young children. Child Dev 1990;61:1628-37.62. Groenendaal JHA, Gerrits LAW. Daily Worries List (DBL). [In Dutch]. Utrecht: Universiteit Utrecht; 1996.63. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire
(JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322-55.64. Van Veldhoven M, Meijman TF, Broersen JPJ, Fortuin RJ. Manual VBBA. [In Dutch]. Amsterdam: SKB
Vragenlijst Services; 2002.
Ethnic differences in term birthweight: the role of constitutional and environmental factors
Paediatric and Perinatal Epidemiology 2008;22(4):360-8.
It is not clear to what extent ethnic differences in the term birthweight distribution are constitutional or pathological. This study explored term birthweight heterogeneity between ethnic groups and the explanatory role of constitutional and environmental factors. As part of a prospective cohort study, the Amsterdam Born Children and their Development study, 8266 pregnant women filled out a questionnaire during early pregnancy. Ethnic groups were categorised as: native Dutch group; first and second generation Surinamese, Antillean, Turkish, Moroccan, Ghanaian and other non-Dutch groups. Only singleton live births with ≥37.0 weeks of gestation and with complete data were included for analysis (n = 7118). We performed linear regression analyses to estimate the association between ethnicity and, for gestational age, standardised birthweight at term, adjusted for constitutional (fetal gender, parity, maternal age, maternal height) and environmental (education, cohabitation status, maternal body mass index, smoking, alcohol consumption, depression, work stress) determinants respectively. Mean birthweight ranged from 3223 g (second generation Surinamese newborns) to 3548 g (Dutch newborns). Adjustment for constitutional factors substantially reduced the ethnic differences in birthweight, while adjustment for environmental factors provided little additional explanation. Surinamese [first generation: regression coefficient (b) = -98.3 g, p < 0.001; second generation: b = -159.3 g, p < 0.001], first generation Antillean (b = -102.0 g, p = 0.037), and Ghanaian newborns (b = -120.7 g, p = 0.001) remained significantly smaller than Dutch newborns after adjustment for all determinants. Term birthweight differences between Dutch newborns and Turkish, Moroccan and other non-Dutch newborns were largely explained by constitutional rather than environmental determinants, limiting the need for prevention. Surinamese, Antillean and Ghanaian (mainly black) newborns remained unexplainably smaller after adjustment, leaving the possibility of either unknown constitutional or pathological underlying mechanisms.
23
____________________________________________________ Ethnic differences in term birthweight
2
Introduction
For decades, researchers worldwide have investigated the incidence of low birthweight (LBW; <2500 g) and its association with adverse health outcomes.1 Currently, LBW is considered a poor indicator for health outcomes at the population level.2 This is because newborns defined as LBW are a mixture of preterm, growth restricted, and constitutionally small babies; all three groups show quite different infant health outcomes. Wilcox1 proposed separating preterm from term births (≥37.0 weeks’ gestation) when studying health outcomes at the population level. The proportion of preterm births should be used as an indicator of a population’s infant mortality risk and the birthweight distribution of the remaining term births as a valid measure of fetal growth. Useful as this distinction may be, term birthweight heterogeneity between groups can still be the result of either pathological growth restriction or a ‘natural’ limited growth potential for constitutional profile.
Worldwide, large ethnic differences in the term birthweight distribution are observed.1,3 Previous studies explained some differences by adjusting for major determinants, such as smoking.4 However, from a preventive point of view, it is important to explore to what extent ethnic differences in the birthweight distribution can be explained by two main determinant groups: (i) constitutional determinants (e.g. maternal height), which are difficult to modify and which physiologically influence fetal growth in a natural way, or (ii) environmental determinants (e.g. smoking), which are more or less modifiable determinants that impair fetal growth.5 A predominant role for constitutional factors would add to a natural explanation of birthweight heterogeneity, with no need for preventive actions, whereas predominance of environmental factors would suggest acquired pathology with, theoretically, a higher probability of short- and long-term adverse health outcomes.
In the present study we explored, in a large prospective multi-ethnic cohort of pregnant women, (i) term birthweight heterogeneity among ethnic groups, and (ii) the roles of constitutional and environmental factors, using a hierarchical model. Taking advantage of the presence of sufficiently sized groups of first and second generation ethnic minority groups, we additionally explored intergenerational differences in the birthweight distribution at term.
Methods
The present study is part of the Amsterdam Born Children and their Development (ABCD) study (www.abcd-study.nl). The ABCD-study is a prospective, unselected and community-based cohort study, which examines the relationship between maternal lifestyle and psychosocial conditions during pregnancy and the child’s health at birth and in later life. The primary focus is on the explanation of the existing ethnic disparities in maternal and child health. Approval of the study was obtained from the Central Committee on Research involving Human Subjects
in the Netherlands, the Medical Ethical Committees of participating hospitals, and from the Registration Committee of the Municipality of Amsterdam.
Study populationBetween January 2003 and March 2004, all pregnant women living in Amsterdam were invited to enrol in the ABCD study at their first prenatal visit (around the 12th week of gestation) to participating obstetric care providers (general practitioners, midwives and hospital gynaecologists; overall participation rate 96%). Two weeks after their first prenatal visit, a pregnancy questionnaire was mailed to all women who had been approached (n = 12 373), to be returned by prepaid mail. The questionnaire was in Dutch, and accompanied by an English, Turkish or Arabic copy depending on the woman’s country of birth. Reminders were sent 2 weeks after the initial mailing to improve response. The questionnaire was filled out by 8266 women (response rate 67%). For this study, only singleton liveborns with a pregnancy duration of 37.0 or more weeks were included (n = 7318). We did not exclude the pregnancies with fetal malformations as there was no evidence of a specific relationship between birthweight of infants with malformations and ethnicity. Overall, 7118 women had complete data on all the study variables. Mean birthweight of the excluded term births with incomplete data was not significantly different from the mean birthweight of the cases included.
Outcome variableThe primary outcome variable was birthweight (in grams) of neonates born at term (gestational duration ≥37.0 weeks). Information on the pregnancy outcomes [birthweight, gestational age (GA) and fetal gender] were obtained from the Youth Health Care registration at the Municipal Health Service in Amsterdam. Duration of pregnancy (in weeks and days) was based on ultrasound or, when unavailable (<10%), on the first day of the last menstrual period (calculation by the obstetric care provider).
DeterminantsEthnicity was based on the country of birth of the pregnant woman and her mother. Country of birth included the following categories, based on the Amsterdam main ethnic populations: The Netherlands, Surinam, The Antilles/Aruba, Turkey, Morocco, Ghana, and other non-Dutch countries. Ethnic minority groups were divided into first (born outside the Netherlands) and second (born in the Netherlands, but with a mother born in another country) generation women. Because only two women were classified as second generation Ghanaian, these women were merged with the first generation Ghanaians. The native Dutch group (pregnant woman and her mother born in the Netherlands) was used as the reference group.
Determinants of fetal growth, known to be associated with both birthweight and ethnicity, were collected in the pregnancy questionnaire, except for fetal gender and GA (see above). Birthweight was standardised for GA (linear and quadratic term) in all logistic regression analyses. We distinguished two groups of determinants: constitutional and environmental determinants. Constitutional determinants were: fetal gender, parity (0 vs. ≥1), maternal age (≤24, 25-34, ≥35 year) and maternal height (cm). Environmental determinants were: educational
25
____________________________________________________ Ethnic differences in term birthweight
2
attainment (years of education after primary school: ≤5, 6-10, ≥11 year), cohabitation status (living together with partner vs. not living together/single), maternal pre-pregnancy body mass index (BMI) (underweight: BMI <18 kg/m2, normal weight: 18-24.9, overweight: 25-29.9, obesity: ≥30), smoking (average number of cigarettes per day in early pregnancy: none, <1, 1-5, >5 cigarettes), alcohol consumption during early pregnancy (no vs. yes), depression [a score on the Center for Epidemiologic Studies Depression Scale (CES-D scale)6 of lower vs. higher than the 90th percentile], and work stress (no paid job, paid job without high work stress, paid job with high work stress). High work stress is defined as ≥32 working hours per week combined with having high job strain.7 Job strain was measured by the validated Dutch version of the Job Content Questionnaire.8 Missing values of maternal weight and height were imputed by means of a random imputation method using linear regression,9 accounting for the differences among the ethnic groups.
Statistical analysisDifferences in the distribution of constitutional and environmental characteristics between the Dutch and ethnic minority groups and between generations were tested with either the chi-square test for categorical variables or ANOVA with Bonferroni correction for continuous variables. Descriptive statistics were used to obtain the mean birthweights, birthweight percentiles and mean GAs per ethnic group.
Linear regression analyses were performed to estimate the size of differences in birthweight, standardised for GA, among ethnic groups, adjusting for above-mentioned determinants of birthweight. All covariates, except for maternal height, were treated as categorical variables. Analyses followed a predefined hierarchical format. First, univariable analyses were performed to obtain the associations of ethnicity and all determinants separately with birthweight for GA. Second, two multiple linear regression models (forced entry method) were explored. The first model examined the association of ethnicity with birthweight for GA, adjusted for the constitutional determinants. The second model additionally adjusted for the environmental determinants. Data were analysed using SPSS version 14.0. In all analyses, a p-value <0.05 was considered statistically significant.
Results
Large differences in the prevalence of constitutional and environmental determinants were observed between Dutch and all ethnic minority groups (Table 2.1). Compared with the ethnic minority women, Dutch pregnant women were older, taller, higher educated, less often depressed, had a lower BMI and were more likely to have a paid job in the first trimester of pregnancy. Dutch women were more frequently expecting their first child than the first generation ethnic minority women. Large differences were found for cohabitation status: while 50-64% of the Surinamese, Antillean and Ghanaian women lived together with a partner, about 90% of the other women did.
Smoking in the first trimester of pregnancy was most common among second generation Turkish women, while alcohol consumption was most common among Dutch women. Second generation ethnic groups were on average younger, taller, more highly educated and more frequently expecting their first child than the first generation women.
Mean birthweight ranged from 3223 g (second generation Surinamese newborns) to 3548 g (Dutch newborns) (Table 2.2). Birthweight distributions of all ethnic minority groups were shifted to the left compared with the Dutch birthweight distribution. Comparisons with the Bonferroni test showed a significantly lower mean birthweight for first and second generation Surinamese and Ghanaian newborns, and for first generation Antillean and other non-Dutch newborns. The Surinamese groups showed a significantly lower mean GA than the Dutch group. Within ethnic groups, no significant differences were found between the mean birthweights and GAs of infants born to first and second generation women.
Univariable linear regression analysis showed a significant association between all determinants and birthweight for GA (Table 2.3). All ethnic minority groups had lower mean birthweights for GA compared with the Dutch group, although not all group differences were significant. After adjustment for constitutional determinants (Model 1), ethnic differences in birthweight decreased substantially for all groups. Further adjustment for environmental determinants (Model 2) provided a little additional decrease or for some groups an increase in the ethnic differences in birthweight. The adjusted birthweights of Surinamese newborns remained lower than the Dutch birthweights, with a significant difference in weight of 98 g for the first generation and 159 g for the second generation (p < 0.001). Also Ghanaian and first generation Antillean women had significantly smaller newborns after adjustment [a difference of 121 g (p = 0.001) and 102 g (p = 0.037) respectively].
Discussion
In this study, term birthweight distribution of all ethnic minority groups was shifted to lower birthweights than that of the native Dutch group. Constitutional determinants largely explained these ethnic disparities, while environmental determinants provided only a slight additional explanation. An unexplained difference in birthweight persisted for black (i.e. Surinamese, Ghanaian, and first generation Antillean) newborns compared with Dutch newborns.
Previous studies in the Netherlands found similar patterns of crude birthweights in Dutch and ethnic minority groups, with the birthweight distributions of minority groups shifted to the left.10-13 In other countries, such as the USA and UK, large and unexplainable differences in the birthweights of newborns with a different ethnic background (white vs. black) have also been observed.14,15 Most studies, however, present birthweights over the whole GA spectrum, leaving room for confusion between fetal growth and preterm birth. The Wilcox-Russell approach in our view is a valid tool to prevent such ambiguities.
Our distinction between constitutional and environmental determinants has few precedents. This distinction, in our view, is justified given the different consequences of lower birthweight
27
____________________________________________________ Ethnic differences in term birthweight
Other 1st generation 3466 (469)* 3140 3780 39.7 (1.3)
2nd generation 3511 (474) 3195 3773 39.7 (1.3)
*Significantly different from Dutch group (p < 0.05)
depending on the underlying mechanism; if the normal birthweight distribution is shifted to lower birthweights without affecting mortality risk, preventive actions are less relevant. Furthermore, the general claim for ethnic disparities should be specified to those groups really in need for action (here the black group). The distinction between constitutional and environmental determinants can be debated. In contrast to other studies,5 we did not include pre-pregnancy maternal weight as a constitutional determinant of fetal growth, because the size of the fetus is determined more by maternal height than by maternal weight.16 Only weights at the extremes, i.e. under- and overweight for height, influence fetal growth, resulting in either pathologically restricted or enlarged newborns.17
At first sight, the birthweight distribution of all ethnic minority groups was shifted to lower birthweights compared with the Dutch group. After standardisation for GA and adjustment for constitutional determinants of fetal growth, Turkish, Moroccan and other non-Dutch newborns appeared to have a comparable size at birth to Dutch newborns. This suggests a limited constitutional growth potential rather than a pathological growth restriction for the majority of newborns in these groups. Maternal height was by far the most important determinant of birthweight at term (R2 = 0.048; followed by parity: R2 = 0.022), stressing the dominance of constitutional rather than environmental determinants in ethnic differences in birthweight. This minimal role of environmental determinants interferes with the commonly held preventive view that much can be gained here. Smoking, maternal BMI and high work-related stress are significant and modifiable environmental determinants of fetal growth at the individual level, and are appropriate for preventive actions, but their potential to reduce the ethnic disparities in birthweight is low.
Table 2.3 Results of the hierarchical linear regression analyses: differences in birthweight (g) - standardised for gestational age - between the ethnic groups, adjusted for constitutional and environmental determinants.
High -116.8 -165.5, -68.2 -82.6 -128.0, -37.1aUnivariable analysis (b: unstandardised regression coefficient; 95% confidence interval).bModel 1: multiple regression, adjusted for constitutional determinants.cModel 2: multiple regression, adjusted for constitutional and environmental determinants.
Although constitutional rather than environmental determinants were also responsible for a large shift in the birthweight distribution of the Surinamese, Ghanaian and Antillean group, these newborns remained substantially smaller than the Dutch newborns. Additional explanation by other environmental factors, such as diabetic18 and hypertensive disorders,19 or nutritional habits20,21 (e.g. folic acid, fatty acids) must be investigated. The role of nutrition in this cohort is currently being examined in detailed analysis. Another explanation could probably be found in the field of cumulative risk.22 A cumulative exposure to multiple risk factors during pregnancy may substantially increase the risk of lower birthweight at term, especially among the vulnerable ethnic minority groups. Instead of such pathological explanations of the lower birthweight among black newborns, a broader constitutional explanation (including genetics) may also be appropriate.23 Lower population birthweights at term are not necessarily related to higher perinatal morbidity and mortality,24-26 suggesting some genetic influences. Whether the majority of black newborns are pathologically growth restricted or constitutionally small is not yet clear and needs further investigation.
We expected the ethnic disparities in birthweight to decrease across generations as a result of the higher educational attainment of the second generation, however, we found the opposite for most groups. Previous studies in the UK and the USA have also reported lower or similar birthweights in second generation ethnic minority groups compared with the first generation.27-30 This lack of an intergenerational increase in birthweight at term supports the minimal role of environmental determinants. That both first and second Surinamese newborns were substantially smaller than the Dutch newborns after adjustment can either be the result of genetic influences,31,32 and/or the result of unspecified socio-cultural determinants shared by the different generations.30 Because the latter explanation requires the identification of new, strong and independent environmental risk factors,4,33 a genetic role seems more likely to us. More evidence for a genetic component in fetal growth should be explored by comparing the birthweights of third and next generation ethnic groups, or by exploring DNA markers using admixture mapping.34
Our study population consisted of a large, unselected cohort of pregnant women, which adequately represented the main ethnic groups in Amsterdam.35 Although there was selective participation, with a higher participation rate among the Western groups, selection bias was absent.36 The exclusion of women with incomplete data may also have led to a relatively healthier sample for analysis. The small number of (second generation) Antillean women limited the power for detailed analysis, although the numbers were still larger than comparable studies.11,12 We could not split the Surinamese group into a Creole and Hindustani group as detailed information was available only for a sub-sample of women. However, as we did not observe significant differences in birthweight between known Hindustani and Creole newborns (results not shown), we felt it appropriate to include both in the Surinamese group.
In conclusion, the birthweight differences between Dutch newborns and Turkish, Moroccan and other non-Dutch newborns were largely explained by constitutional determinants, limiting the need for preventive actions. The substantially lower birthweights of black newborns, however, remains to be elucidated. Future research should focus on the underlying mechanism of this disparity, by searching for unexplored environmental risk factors or for cumulative risk effects, and by investigating the weight-related population mortality rates. Within the ABCD study, future research will focus on whether constitutional and environmental risk factors explain the ethnic differences in preterm birth. Considering the small influence of the investigated modifiable environmental factors, the potential to reduce ethnic disparities in term birthweight at a population level seems limited. Nevertheless, prevention of smoking, high maternal BMI and work-related stress remain important in order to improve the individual health of newborn infants.
33
____________________________________________________ Ethnic differences in term birthweight
2
References
1. Wilcox AJ. On the importance - and the unimportance - of birthweight. Int J Epidemiol 2001;30:1233-41.2. Adams M, Andersen AM, Andersen PK, Haig D, Henriksen TB, Hertz-Picciotto I, et al. Sostrup statement
on low birthweight. Int J Epidemiol 2003;32:884-5.3. Evans S, Alberman E, Pashley J, Hampton B. International Collaborative Effort (ICE) on birthweight;
plurality; and perinatal and infant mortality. II: Comparisons between birthweight distributions of births in member countries from 1970 to 1984. Acta Obstet Gynecol Scand 1989;68:11-7.
4. Goldenberg RL, Cliver SP, Mulvihill FX, Hickey CA, Hoffman HJ, Klerman LV, et al. Medical, psycho-social, and behavioral risk factors do not explain the increased risk for low birth weight among black women. Am J Obstet Gynecol 1996;175:1317-24.
5. Mamelle N, Cochet V, Claris O. Definition of fetal growth restriction according to constitutional growth potential. Biol Neonate 2001;80:277-85.
6. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401.
7. Vrijkotte TGM, van der Wal MF, van Eijsden M, Gonsel GJ. First-trimester working conditions and birth-weight: a prospective cohort study. Am J Public Health 2009;99:1409-16.
8. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics.
university papers series on quantitative applications in the social sciences. Thousand Oaks, CA: Sage Publications; 2001, pp. 27-41.10. Doornbos JP, Nordbeck HJ, van Enk AE, Muller AS, Treffers PE. Differential birthweights and the clinical
relevance of birthweight standards in a multiethnic society. Int J Gynaecol Obstet 1991;34:319-24.11. Verkerk PH, Zaadstra BM, Reerink JD, Herngreen WP, Verloove-Vanhorick SP. Social class, ethnicity
and other risk factors for small for gestational age and preterm delivery in The Netherlands. Eur J Obstet Gynecol Reprod Biol 1994;53:129-34.
12. Drooger JC, Troe JW, Borsboom GJ, Hofman A, Mackenbach JP, Moll HA, et al. Ethnic differences in prenatal growth and the association with maternal and fetal characteristics. Ultrasound Obstet Gynecol 2005;26:115-22.
13. Van der Wal MF, Uitenbroek DG, van Buuren S. Birthweight of children in Amsterdam by ethnic origin [In Dutch]. Tijdschrift voor Gezondheidswetenschappen 2000;78:15-20.14. Wilcox M, Gardosi J, Mongelli M, Ray C, Johnson I. Birth weight from pregnancies dated by ultrasono-
graphy in a multicultural British population. BMJ 1993;307:588-91.15. Shiono PH, Rauh VA, Park M, Lederman SA, Zuskar D. Ethnic differences in birthweight: the role of lifestyle and other factors. Am J Public Health 1997;87:787-93.16. Blair EM, Liu Y, de Klerk NH, Lawrence DM. Optimal fetal growth for the Caucasian singleton and assess-
ment of appropriateness of fetal growth: an analysis of a total population perinatal database. BMC Pediatr 2005;5:13.
17. Catalano PM, Ehrenberg HM. The short- and long-term implications of maternal obesity on the mother and her offspring. BJOG 2006;113:1126-33.
18. Rosenberg TJ, Garbers S, Lipkind H, Chiasson MA. Maternal obesity and diabetes as risk factors for ad-verse pregnancy outcomes: differences among 4 racial/ethnic groups. Am J Public Health 2005;95:1545-51.
19. Fang J, Madhavan S, Alderman MH. The influence of maternal hypertension on low birth weight: dif-ferences among ethnic populations. Ethn Dis 1999;9:369-76.
20. Lumey LH, Stein AD. Offspring birth weights after maternal intrauterine undernutrition: a comparison within sibships. Am J Epidemiol 1997;146:810-9.
21. Ramakrishnan U, Manjrekar R, Rivera J, Gonzales-Cossio T, Martorell R. Micronutrients and pregnancy outcome: a review of the literature. Nutr Res 1999;19:103-59.
22. Burchinal MR, Roberts JE, Hooper S, Zeisel SA. Cumulative risk and early cognitive development: a com-parison of statistical risk models. Dev Psychol 2000;36:793-807.
23. Van den Oord EJ. Ethnic differences in birth weight: maternal effects emerge from an analysis involving mixed-race us couples. Ethn Dis 2006;16:706-11.
24. Vangen S, Stoltenberg C, Skjaerven R, Magnus P, Harris JR, Stray-Pedersen B. The heavier the better? Birthweight and perinatal mortality in different ethnic groups. Int J Epidemiol 2002;31:654-60.
25. Wilcox AJ, Russell IT. Birthweight and perinatal mortality: II. On weight-specific mortality. Int J Epide-miol 1983;12:319-25.
26. Graafmans WC, Richardus JH, Borsboom GJ, Bakketeig L, Langhoff-Roos J, Bergsjo P, et al. Birth weight and perinatal mortality: a comparison of ‘optimal’ birth weight in seven Western European countries. Epidemiology 2002;13:569-74.
27. Harding S, Rosato MG, Cruickshank JK. Lack of change in birthweights of infants by generational status among Indian, Pakistani, Bangladeshi, Black Caribbean, and Black African mothers in a British cohort study. Int J Epidemiol 2004;33:1279-85.
28. Draper ES, Abrams KR, Clarke M. Fall in birth weight of third generation Asian infants. BMJ 1995;311:876.29. Margetts BM, Mohd YS, Al Dallal Z, Jackson AA. Persistence of lower birth weight in second generation
South Asian babies born in the United Kingdom. J Epidemiol Community Health 2002;56:684-7.30. David RJ, Collins RW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks,
and U.S.-born whites. N Engl J Med 1997;337:1209-14.31. Amante A, Borgiani P, Gimelfarb A, Gloria-Bottini F. Interethnic variability in birth weight and genetic
background: a study of placental alkaline phosphatase. Am J Phys Anthropol 1996;101:449-53.32. Dunger DB, Petry CJ, Ong KK. Genetic Variations and Normal Fetal Growth. Horm Res 2006;65(Suppl
3):34-40.33. Foster HW, Wu L, Bracken MB, Semenya K, Thomas J, Thomas J. Intergenerational effects of high socio-
economic status on low birthweight and preterm birth in African Americans. J Natl Med Assoc 2000;92:213-21.
34. Frank R. What to make of it? The (Re)emergence of a biological conceptualization of race in health dispa-rities research. Soc Sci Med 2007;64:1977-83.
35. Municipality of Amsterdam. Department for Research and Statistics. Population. In: Van Zee W, Hylkema C, editors. Key figures Amsterdam 2004 [In Dutch]. Amsterdam: Stadsdrukkerij Amsterdam; 2004, pp. 1-4.
36. Tromp M, van Eijsden M, Ravelli ACJ, Bonsel GJ. Anonymous non-response analysis in the ABCD-cohort study enabled by probabilistic record linkage. Paediatr Perinat Epidemiol 2009;23:264-72.
Ethnic differences in preterm birth and its subtypes: the effect of a cumulative risk profile.
Objective To explore the effect of potentially explanatory risk factors on ethnic differences in the prevalence of preterm birth (PTB) and its subtypes.
Design Prospective population-based cohort study.
Setting and population Pregnant women from Amsterdam attending their first antenatal visit for obstetric care. A total of 8266 women participated (response rate 67%). Ethnicity was based on the country of birth of the pregnant woman’s mother: the Netherlands, Surinam, the Antilles, Turkey, Morocco, Ghana and other non-Dutch countries. Exclusion criteria were multiple births and gestational age at delivery less than 24 weeks.
Methods Risk factors were obtained using a multilingual questionnaire and from the Dutch Perinatal Registration. Risk factors were summed into a cumulative risk score. Multiple logistic regression analyses were performed.
Main outcome measures Odds ratios with 95% CIs were calculated for total, spontaneous and iatrogenic (medically indicated) preterm births for the ethnic minority groups versus the Dutch reference group.
Results After adjustment for all risk factors, the Surinamese (OR 1.6, 95% CI 1.2-2.4), Ghanaian (OR 2.0, 95% CI 1.1-3.6) and Antillean (OR 1.6, 95% CI 0.8-3.3) women had a higher risk of PTB compared with the Dutch women, in particular for iatrogenic preterm birth (OR 2.1, 95% CI 1.0-4.4; OR 3.2, 95% CI 1.0-10.4; OR 3.6, 95% CI 1.1-11.2, respectively). The ethnic minority groups had a higher cumulative risk score (ranging from 2.1 to 3.7) compared with the Dutch group (1.8). Adjustment for the cumulative risk score considerably decreased the risk of PTB among the Surinamese (OR 1.2, 95% CI 0.9-1.7), Ghanaian (OR 1.3, 95% CI 0.8-2.3) and Antillean (OR 1.2, 95% CI 0.6-2.4) women.
Conclusions A cumulation of risk factors, mainly observed among the ethnic minority groups, contributes to the explanation of ethnic differences in PTB prevalence.
37
_______________________________________________________ Ethnic differences in preterm birth
3
Introduction
Preterm birth (PTB) is one of the main causes of perinatal mortality, morbidity and consequent medical costs. Despite the improvement in medical care in the past decades, the prevalence of PTB has increased.1 This problem is the result of incomplete knowledge of the aetiology of PTB and a subsequent scarcity of specific preventive and therapeutic options. Administration of progesterone to prevent PTB has provoked interest, but given the burden of treatment and the risk of adverse effects, it should only be given to high-risk groups.2 In a review study, Savitz et al.3 argued that PTB will be better understood and subsequently can be better treated if the causal pathways are divided into a pathway leading to spontaneous preterm birth (SPB) and a pathway leading to iatrogenic (medically indicated) preterm birth (IPB). From a population’s health point of view, thorough analysis of SPB and IPB is also justified as large differences in the prevalence of PTB are observed between ethnic groups, with consequent differences in perinatal and adult health outcomes.4 Whether the ethnic differences are present in both subtypes of PTB is, however, rarely explored. While a part of the ethnic differences in PTB may be attributed to genetic traits,5 different risk profiles are also thought to be involved. However, previous research has not been able to demonstrate that the ethnic disparities in PTB result from different prevalence rates of specific risk factors. We hypothesised that, instead of using single independent risk factors, using a cumulation of risk factors might explain more of the ethnic differences in PTB prevalence.1,4,6-8
The present large multi-ethnic prospective cohort study among pregnant women in Amsterdam aimed to investigate (a) which factors (i.e. maternal characteristics, lifestyles and obstetric history) are related to SPB and IPB, (b) to what extent ethnic differences in the prevalence of PTB can be explained by these risk factors and (c) whether a cumulation of risk factors contributes to the explanation of the ethnic differences in PTB prevalence.
Methods
We used data from the Amsterdam Born Children and their Development (ABCD) study. The ABCD study is a prospective, unselected and population-based cohort study aimed at examining the relationship between maternal lifestyle and psychosocial conditions during pregnancy and the child’s health at birth and in later life. Approval of the study was obtained from the Central Committee on Research involving Human Subjects, the Medical Ethical Committees of participating hospitals and the Registration Committee of Amsterdam.
Setting and populationBetween January 2003 and March 2004, all pregnant women in Amsterdam were invited to enrol in the ABCD study at their first antenatal visit (around the 12th week of gestation) to participating obstetric care providers (GPs, midwives and hospital gynaecologists; overall
participation rate 96%). All approached women (n = 12 373) received a pregnancy questionnaire, which was filled out by 8266 women (response rate 67%). For this study, we excluded multiple births and births before 24 weeks of gestation. In total, 7604 women had complete data on all relevant variables and were included in our analysis.
Data collectionTwo weeks after their first antenatal visit, a pregnancy questionnaire was sent to the women, to be returned by prepaid mail. Reminders were sent 2 weeks after the initial mailing. Questionnaires were in Dutch and accompanied by an English, Turkish or Arabic copy depending on the woman’s country of birth. The questionnaire covered socio-demographic data, maternal health, lifestyle, psychosocial stress and obstetric history. Three to six months after delivery, the women received a baby questionnaire covering the health of the mother and her baby.
Information on pregnancy outcomes was obtained from two sources. First, gestational duration (based on ultrasound or, when unavailable, timing of last menstrual period) was obtained from the Youth Health Care Registration at the Municipal Health Service in Amsterdam. Second, the Dutch Perinatal Registration (PRN) provided comprehensive data on pregnancy, obstetric history and pregnancy outcomes for 80% of our sample.
Definition of outcome variablesThe Dutch Perinatal Registration (PRN) registered the onset of deliveries (e.g. spontaneous, induction and section) only when women delivered under the supervision of a gynaecologist. Based on these data, we divided PTB (defined as a birth before 37 completed weeks of gestation) into SPB and IPB. SPB was defined as delivery onset by spontaneous preterm labour or premature rupture of membranes. IPB was defined as delivery onset through induction or primary caesarean section. PTBs with unknown type of delivery onset were classified as SPB if (a) a woman not specifically reported in the baby questionnaire to have had an iatrogenic delivery or (b) if a woman had not been under the supervision of a gynaecologist.
Risk factorsEthnicity was based on the country of birth of the participant’s mother: the Netherlands, Surinam, The Antilles/Aruba, Turkey, Morocco, Ghana and other non-Dutch countries. Other studied risk factors for PTB were: maternal age; parity; maternal pre-pregnancy body mass index (BMI) [underweight (BMI < 18 kg/m2), normal weight (18-24.9 kg/m2), overweight (25-29.9 kg/m2), obesity (≥ 30 kg/m2)]; education (years of education after primary school); cohabitation status (living together with partner, not living together/single); smoking during pregnancy (average number of cigarettes per day); alcohol use during pregnancy; depressive symptoms, measured with the CES-D scale9 (a score of ≥16 was labelled ‘depression’); physical heavy work, measured by a seven-item subscale of the Dutch version of the Job Content Questionnaire10 [never (score = 7), sometimes (score 8-20), or often/always (score >20) performing physical heavy work, no paid job during pregnancy]; previous PTB as recorded in the PRN registration; previous miscarriages or stillbirths; previous termination
39
_______________________________________________________ Ethnic differences in preterm birth
3
of pregnancy (abortion); hypertensive disorders; vaginal douching; excessive vaginal discharge and vaginal itching. Missing values of maternal weight (n = 721; 9.5%) and height (n = 256; 3.4%) were imputed by means of a random imputation method using linear regression,11 accounting for the differences in height and weight between the ethnic groups. Chronic and pregnancy-induced hypertension were defined combining self-reported data from the questionnaires and data from the PRN registration. Chronic hypertension was the case if (a) pre-existent hypertension was recorded in the PRN, or if (b) women reported high blood pressure and/or the use of medicines against high blood pressure before pregnancy or in the first 20 weeks of pregnancy. Pregnancy-induced hypertension was the case if (a) pregnancy-related hypertension, eclampsia or preeclampsia was recorded in the PRN or if (b) women without pre-existent hypertension reported high blood pressure and/or the use of medicines against high blood pressure during pregnancy.
Statistical analysisDescriptive and chi-square statistics were used to explore the ethnic differences in the prevalence rates of PTB, SPB, IPB, and PTBs with unknown subtype, as well as to explore the ethnic differences in the prevalence rates of risk factors for PTB. Subsequently, logistic regression analyses were performed to examine whether ethnic differences in the prevalence of risk factors could explain the ethnic differences in PTB and its subtypes. First, we explored whether risk factors for SPB, started by premature rupture of membranes, or SPB, started by preterm labour, differed. Because we found almost no differences (results not shown), we chose to combine both components into one category of SPB for further analysis. Univariate logistic regression analyses were used to calculate the crude odds ratios for the association between ethnicity and PTB, SPB and IPB (with native Dutch women as the reference group), and for the association between the other risk factors and PTB, SPB and IPB. Second, multiple logistic regression analyses were performed to calculate the odds ratios for the association between ethnicity and PTB, SPB and IPB adjusted for the other risk factors. PTBs with unknown subtype were only included in the risk analyses for total PTB. Most risk factors were treated as categorical variables, except for education, smoking, depressive symptoms and previous miscarriages. Additionally, interaction effects between ethnicity and other risk factors for PTB were explored.
Risk factors with a significant odds ratio >1.0 in the univariate regression model for PTB were summed into a cumulative risk score. This risk score has an ordinal level, which means that risk factors were only counted for their presence (yes/no), without taking into account their weight. The six, seven, eight and nine risk factor groups were combined due to small group numbers. Logistic regression analyses were performed to examine whether the cumulative risk score contributes to the explanation of the ethnic differences in the prevalence of PTB. First, the odds ratios for the association between the cumulative risk score and PTB were estimated, crude as well as adjusted for ethnicity. Subsequently, the odds ratios for the association between ethnicity and PTB were estimated, crude as well as adjusted for the risk score. A risk score of 0 was used as the reference category.
To assess the predictive accuracy of the logistic regression models, we used receiver operating characteristic (ROC) curves. In an ROC curve, the sensitivity of all cutoff points
of a certain (diagnostic) test is plotted against 1 minus specificity. The greater the area under the curve (AUC), the greater the predictive accuracy of a model (AUC ≤ 0.5 means no predictive ability at all; 1.0 means perfect prediction).12 To estimate the AUC values, we used the predicted probabilities obtained from the multiple logistic regression models for PTB, SPB and IPB and from the risk score model for PTB. Data were analysed using SPSS version 15.0 (SPSS Inc., Chicago, IL, USA). In all analyses, a p-value ≤ 0.05 was considered significant.
Results
In our sample, 5.5% of the women experienced a PTB, with the lowest PTB rate among the Moroccan women (4.1%) and the highest rate among the Ghanaian women (11%) (Table 3.1). Two-third (67.9%) of the PTBs were SPBs, 19.9% were IPBs, and for 12.2% the type of delivery onset was unknown. These proportions varied among the ethnic groups (p < 0.001), with a relatively high proportion of IPB among the Antillean and Ghanaian group. The prevalence rates of the risk factors for PTB were in general highly divergent between the Dutch and the ethnic minority groups (Table 3.2). The Dutch women were older, less parous, higher educated, more often had a paid job, less obese, had fewer symptoms of depression, more often consumed alcohol during pregnancy and reported vaginal douching and vaginal itching less often compared with the minority women. Significant risk factors in the multiple regression analysis for total PTB were: nulliparity (OR 2.1, 95% CI 1.6-2.6), smoking (OR 1.06, 95% CI 1.02-1.09), chronic hypertension (OR 2.4, 95% CI 1.6-3.6), pregnancy-induced hypertension (OR 1.6, 95% CI 1.2-2.3), previous PTB (OR 6.1, 95% CI 3.4-11.0) and previous miscarriage/stillbirth (OR 1.21, 95% CI 1.07-1.37) (Table 3.3). Vaginal itching was a protective factor for PTB (OR 0.6, 95% CI 0.4-0.9). SPB and IPB had some risk factors in common, but others were specifically related to either SPB or IPB. A previous PTB was more strongly related to SPB, while hypertension, maternal age ≥30 year and obesity were specifically related to IPB. The Surinamese (OR 1.6, 95% CI 1.2-2.4) and Ghanaian (OR 2.0, 95% CI 1.1-3.6) groups showed a higher risk for PTB, and specifically for IPB, compared with the Dutch group. The Antillean group showed only a significantly higher risk for IPB (OR 3.6, 95% CI 1.1-11.2). Moroccan women had a significantly lower risk for SPB (OR 0.4, 95% CI 0.2-0.8). (For other ethnic details, see Table 3.3). Interaction effects between ethnicity and other risk factors for PTB were not significant (results not shown). To examine whether a cumulation of risk factors contributed to the explanation of ethnic differences in the prevalence of PTB, we tested the data in three steps. First, the cumulative risk score was tested for its association with PTB: the risk for PTB increased with each additional risk factor a woman experienced during pregnancy, with adjusted odds ratios ranging from 2.1 (95% CI 1.2-3.7) for having one risk factor to 9.7 (95% CI 5.0-18.6) for being exposed to six or more risk factors (prevalence: 0,4%) (see lower half of Table 3.4). Second, the cumulative risk score was tested for its association with ethnicity: the ethnic minority groups had on average a higher cumulation of
41
_______________________________________________________ Ethnic differences in preterm birth
3
risk factors during pregnancy (risk score ranging from 2.1 to 3.7) than the Dutch group (risk score of 1.8) (see upper half of Table 3.4). Third, after adjustment for the risk score, the risk of PTB (and of SPB and IPB; results not shown) decreased substantially for all the ethnic minority groups (see upper half of Table 3.4). Only the PTB risk for the Moroccan group was significantly different from that of the Dutch women. The predictive accuracies of the multiple regression models for PTB, SPB and IPB and the risk score model for PTB were all significantly higher than an AUC-value of 0.5 (Table 3.5). The regression model for IPB had the largest area under the ROC curve of 0.81 (95% CI 0.75-0.86). The predictive ability of the risk score model was lower than that of the multiple regression model for PTB.
Table 3.1 Prevalence rates of total PTB, SPB and IPB per ethnic group.Ethnicity Number
Unknown subtype, subtype of PTB unknown.ap < 0.001 (chi-square test) for ethnic differences in the proportion of SPB, IPB and unknown subtype among the PTBs.
Table 3.2 Prevalence rates of risk factors for PTB, per ethnic group.
Risk factors Dutch Surinamese Antillean Turkish Moroccan Ghanaian Other
Yes 0.7 (0.5-1.0) 0.6 (0.4-0.9) 0.6 (0.4-1.0) 0.8 (0.4-1.6)aAdjusted for all the risk factors in the table.
Discussion
In this large multi-ethnic prospective cohort study, we found that SPB and IPB had both shared and specific risk factors. Nulliparity, smoking and a previous miscarriage were risk factors that SPB and IPB had in common. A more specific risk factor for SPB was a history of PTB, while hypertension, advanced maternal age and obesity were more specifically related to IPB. Even after adjustment for risk factors, Surinamese and Ghanaian women were at a higher risk of PTB, in particular IPB, compared with Dutch women. While Antillean women also seemed to have a higher risk of PTB, their risk was only statistically significantly higher for IPB. This was probably due to the small number of Antillean women in our sample. The large effect of a cumulation of risk factors – mainly observed among the ethnic minority groups – on PTB contributed to the explanation of the ethnic differences in the prevalence of PTB. The predictive accuracy of the cumulative risk score was, however, too low to allow for individual risk selection.
Table 3.3 Continued
45
_______________________________________________________ Ethnic differences in preterm birth
3
Table 3.4 The association between ethnicity and PTB, adjusted for a cumulation of risk factors (risk score).Crude OR (95% CI)
Adjusteda OR (95% CI)
Ethnicity (mean risk score) Dutch (1.8) 1.0 1.0
Surinamese (3.3) 1.9 (1.4-2.6) 1.2 (0.9-1.7)
Antillean (3.1) 1.8 (0.9-3.5) 1.2 (0.6-2.4)
Turkish (2.8) 1.0 (0.6-1.6) 0.8 (0.5-1.2)
Moroccan (2.5) 0.8 (0.5-1.2) 0.6 (0.4-1.0)
Ghanaian (3.7) 2.3 (1.4-3.9) 1.3 (0.8-2.3)
Other non-Dutch (2.1) 1.0 (0.8-1.3) 0.9 (0.7-1.2)
Risk scoreb (n PTB /total n)
0 (15/743) 1.0 1.0
1 (85/2093) 2.1 (1.2-3.6) 2.1 (1.2-3.7)
2 (110/2099) 2.7 (1.6-4.6) 2.8 (1.6-4.8)
3 (84/1411) 3.1 (1.8-5.4) 3.2 (1.8-5.6)
4 (56/720) 4.1 (2.3-7.3) 4.1 (2.3-7.3)
5 (34/351) 5.2 (2.8-9.7) 5.0 (2.7-9.5)
≥6 (33/187) 10.4 (5.5-19.6) 9.7 (5.0-18.6)
aAdjusted for the other factors in the table.bRisk factors are: nulliparity, obesity, education ≤5 years, not living together with a partner, smoking, depression, often/always doing physical heavy work, pregnancy-induced hypertension or chronic hypertension, (almost) daily vaginal douching, excessive vaginal discharge, previous preterm birth, previous miscarriage or stillbirth (>1) and previous abortion.
Table 3.5 ROC curve analyses of the multiple regression models for PTB, SPB and IPB and of the risk score model for total PTB.
Total PTBAUC (95% CI)
SPBAUC (95% CI)
IPBAUC (95% CI)
Multivariate modela 0.67 (0.64-0.69)* 0.67 (0.64-0.70)* 0.81 (0.75-0.86)*
Risk score modelb 0.61 (0.59-0.64)* - -aAnalyses based on the predicted probabilities from the multiple logistic regression models for PTB, SPB and IPB (Table 3.3).bAnalysis based on the predicted probabilities from the risk score model for PTB (Table 3.4).*p < 0.001 (this means significantly different from AUC = 0.5).
In line with previous studies, SPB was more common in women who smoked during pregnancy,13-15 nulliparous women13,14,16 and parous women with a history of PTB.13,17-20 A history of miscarriages and stillbirths as a risk factor for SPB has been reported before,15,21 but one study found a significant association only with IPB.14 The protective effect of vaginal itching during pregnancy on SPB was unexpected and remains unexplained. In contrast to previous studies, we did not find a statistically significant association of low maternal BMI (<18 kg/m2)13,15,19 and teenage pregnancies13,14 with SPB, probably due to the small size of these specific groups.
Frequently reported risk factors for IPB, in line with our results, are chronic and pregnancy-related hypertension,13,22 nulliparity14,16 and smoking.16,20 Similar trends in the effect of advanced maternal age on IPB have been reported before,13,14,17 although one other study has reported an inverse trend.16 Previous studies have also reported a positive, although not statistically significant, association between obesity and IPB.17,20 When a history of PTB is divided into a previous SPB or IPB, a previous IPB is found to be related to recurrence of an IPB.18,23 We were, however, not able to examine this in our study, because we could not divide the variable previous PTB into previous SPB or IPB. The inconsistent patterns of reported risk factors for IPB across PTB studies may be the consequence of the different practices of obstetric care providers throughout the world and over the years24 or variations in patient’s preferences25 regarding medical intervention of pregnancy. These diverse obstetric practices hamper the search for universal risk factors for IPB.20
One might argue that dividing PTB into SPB and IPB is statistically inefficient.20 However, we prefer the disentanglement as it reflects shared versus specific pathways to SPB and IPB. One of the shared pathways rests on the mediating role of intrauterine growth restriction (IUGR), for example caused by smoking during pregnancy. When IUGR is undetected by antenatal screening, several physiological mechanisms can trigger an SPB, while detection of IUGR will often be a stimulus for preterm intervention in the pregnancy (IPB).26 This mechanism makes prevention of IUGR one of the main potential ways to reduce PTB, although as yet a technique for doing so remains elusive.27 A specific pathway to IPB may, for example, lie in the mechanism of ischaemic placental disease (e.g. pre-eclampsia). However, ischaemic placental diseases can also be a primary cause for SPB.28 Ethnic differences in the prevalence of PTB were only observed for the Surinamese, Ghanaian and Antillean women compared with the Dutch women. Turkish and Moroccan women rather had a decreased than increased risk for both SPB and IPB compared with the Dutch women, although they did not have a more favourable risk profile. It is not clear whether environmental and/or genetic factors are responsible for the low risk of PTB among the Turkish and the Moroccan group. Surinamese, Ghanaian and Antillean women had an excess risk of IPB rather than SPB compared with Dutch women. Several pathways may explain this higher risk. First of all, the cumulative risk score analysis we presented provides an uncommon yet potentially fruitful approach to explain the ethnic disparities in PTB. The clustering of both major (e.g. chronic hypertension) and minor (e.g. depression) risk factors among the ethnic minority groups substantially increased their risk of PTB. Apparently, cumulation of risk factors during pregnancy initiated multiplicative effects on the risk of PTB. A related hypothesis postulates that ethnic minority groups are more susceptible to a cumulation of risk factors (e.g. racism, poor socio-economic position), not only during pregnancy but also before pregnancy.29 This pre-pregnancy cumulative risk is hypothesised to trigger early parturition through neuroendocrine and inflammatory pathways.30 Second, other risk factors may be involved. Various micronutrients (e.g. fatty acids)31 may play a role in the association between ethnicity and PTB,32 as well as low maternal gestational
47
_______________________________________________________ Ethnic differences in preterm birth
3
weight gain.33 We are currently examining the association between these variables and pregnancy outcomes in more detail. A third explanation for the ethnic disparities in the prevalence of PTB relates to obstetric care practices. On average, the ethnic minority groups in our cohort had a delayed start to their antenatal care.34 Subsequent inadequate antenatal care may put these groups at a higher risk for PTB.35,36 Further ethnic differences in healthcare use are seen around the time of delivery. Most minority women in the Netherlands prefer to deliver in hospital, while Dutch women are used to home births and only attend the hospital in case of higher risk of pregnancy complications. In our study cohort, about 70% of the minority women and 40% of the Dutch women self-reported a preference to deliver in hospital. Even among low-risk pregnancies, opting for hospital delivery is related to more medical interventions, including preterm caesarean sections, than opting for home delivery.37 Besides the influence of the pregnant women themselves, the delivery practices of the obstetric care providers may also depend on the ethnic background of a pregnant woman.38
A final explanation concerns the influence of genetic traits on ethnic disparities in PTB prevalence as the highest risk groups (Surinamese, Antillean and Ghanaian women) share a black ancestry, while the lowest risk groups (Dutch, Turkish and Moroccan women) share a Caucasoid ancestry.39 Previous studies have found significantly shorter gestations among black compared with white women, possibly related to earlier fetal maturation and a subsequent survival advantage.40,41 Identifying the level of fetal maturation would consequently be a more appropriate measure for prematurity than a fixed cutoff point of 37 completed weeks, especially when dealing with different ethnic groups. Another indication for the influence of genetic traits on racial disparities is the contribution of paternal race to the risk of PTB, probably by fetal inheritance of paternal genes. Biracial couples appear to have a different PTB risk than monoracial couples.42,43 In our study, we were not able to identify the biracial couples; these couples may have biased the association between ethnicity and the risk of PTB. Although the evidence for a genetic contribution to ethnic disparities in PTB prevalence has accumulated, direct evidence of racial differences in specific genetic factors, which have been linked to PTB, is still limited.29 This search is complicated by emerging evidence for gene-environment interactions that may be relevant to racial disparity in the prevalence of PTB. For example, the higher rates of chronic hypertension among black women combined with racial differences in genes affecting the infection/inflammation and vascular processes could theoretically contribute to an excess risk of PTB among black compared with white women.29
The primary strengths of this cohort study are the prospective, unselected and population-based nature of the study, which represented in reasonable numbers the main ethnic groups in Amsterdam.44 Although there was selective participation, with a higher participation rate among the Western groups, selection bias did not occur.45 The exclusion of women with incomplete data might also have led to a relative healthier sample for analysis. We could not split the Surinamese group into a Creole and Hindustani group as detailed information was available only for a subsample of women. However, as we did not observe significant differences in the prevalence of PTB between known Hindustani and Creole newborns (results not shown), we felt it appropriate to include both in the Surinamese group.
We collected a comprehensive set of risk factors at the time of the first antenatal visit. However, the study lacked specific clinical tests to predict PTB. Although so far no study including fluid samples has been able to either adequately identify women at high risk for PTB or subsequently to offer an effective intervention,15,46,47 such data are necessary, for example to diagnose vaginal infections (we did not find a significant association between PTB and self-reported vaginal symptoms). Our results may have been influenced by inaccuracies in the classification of PTB and its subtypes. The overall classification of PTB is, however, accurate because most gestational age data were based on ultrasound measurements. Furthermore, to determine whether time of birth was term or preterm, we used two sources that were highly consistent. For some PTBs, mainly among the ethnic minority groups, information on the type of delivery onset was unavailable, which probably resulted in a decreased risk for SPB and/or IPB among the minority groups. In conclusion, risk factors for PTB often cluster in the ethnic minority groups, putting them at an additional higher risk of PTB, and in particular IPB, compared with the Dutch group. Consequently, the ethnic minority groups should be a main target group for PTB prevention programmes, covering a wide range of risk factors. Although our cumulative risk results suggest that we should focus our preventive efforts on pregnant women with the highest cumulative risk profile, preventive efforts would have a larger impact when targeting a larger population with a moderate cumulative risk profile (e.g. cumulative risk score of 3).48 Individual risk factors most amenable for preventive efforts are smoking, obesity and advanced maternal age (by encouraging pregnancy at an earlier age). The remaining ethnicity-related risks appear restricted to blacks in whom gestation might be slightly shorter, either by gene-environment interactions or, less problematic, by earlier fetal maturation.
49
_______________________________________________________ Ethnic differences in preterm birth
3
References
1. Green NS, Damus K, Simpson JL, Iams J, Reece EA, Hobel CJ, et al. Research agenda for preterm birth: recommendations from the March of Dimes. Am J Obstet Gynecol 2005;193:626-35.
2. Ness A, Dias T, Damus K, Burd I, Berghella V. Impact of the recent randomized trials on the use of progesterone to prevent preterm birth: a 2005 follow-up survey. Am J Obstet Gynecol 2006;195:1174-9.
3. Savitz DA, Blackmore CA, Thorp JM. Epidemiologic characteristics of preterm delivery: etiologic heterogeneity. Am J Obstet Gynecol 1991;164:467-71.
4. Gennaro S. Overview of current state of research on pregnancy outcomes in minority populations. Am J Obstet Gynecol 2005;192(Suppl 5):S3-10.
5. Menon R, Velez DR, Thorsen P, Vogel I, Jacobsson B, Williams SM, et al. Ethnic differences in key can-didate genes for spontaneous preterm birth: TNF-alpha and its receptors. Hum Hered 2006;62:107-18.
6. Appleyard K, Egeland B, van Dulmen MH, Sroufe LA. When more is not better: the role of cumula-tive risk in child behavior outcomes. J Child Psychol Psychiatry 2005;46:235-45.
7. Dew PC, Guillory VJ, Okah FA, Cai J, Hoff GL. The effect of health compromising behaviors on pre-term births. Matern Child Health J 2007;11:227-33.
8. Callahan MA, Sexton K. If cumulative risk assessment is the answer, what is the question? Environ Health Perspect 2007;115:799-806.
9. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401.
10. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally comparative assessments of psychosocial job characteris-tics. J Occup Health Psychol 1998;3:322-55.
11. Allison PD. Multiple Imputation: Basics. In: Laughton CD, Carr E, Santoyo D, editors. Missing data. Sage university papers series on quantitative applications in the social sciences. Thousand Oaks, CA: Sage Publications; 2001, pp. 27-41.
12. Goodwin LK, Iannacchione MA, Hammond WE, Crockett P, Maher S, Schlitz K. Data mining me-thods find demographic predictors of preterm birth. Nurs Res 2001;50:340-5.
13. Berkowitz GS, Blackmore-Prince C, Lapinski RH, Savitz DA. Risk factors for preterm birth subtypes. Epidemiology 1998;9:279-85.
14. Meis PJ, Michielutte R, Peters TJ, Wells HB, Sands RE, Coles EC, et al. Factors associated with pre-term birth in Cardiff, Wales. II. Indicated and spontaneous preterm birth. Am J Obstet Gynecol 1995;173:597-602.
15. Smith GC, Shah I, White IR, Pell JP, Crossley JA, Dobbie R. Maternal and biochemical predictors of spontaneous preterm birth among nulliparous women: a systematic analysis in relation to the degree of prematurity. Int J Epidemiol 2006;35:1169-77.
16. Harlow BL, Frigoletto FD, Cramer DW, Evans JK, LeFevre ML, Bain RP, et al. Determinants of pre-term delivery in low-risk pregnancies. The RADIUS Study Group. J Clin Epidemiol 1996;49:441-8.
17. Adams MM, Sarno AP, Harlass FE, Rawlings JS, Read JA. Risk factors for preterm delivery in a healthy cohort. Epidemiology 1995;6:525-32.
18. Ananth CV, Getahun D, Peltier MR, Salihu HM, Vintzileos AM. Recurrence of spontaneous versus medically indicated preterm birth. Am J Obstet Gynecol 2006;195:643-50.
19. Mercer BM, Goldenberg RL, Das A, Moawad AH, Iams JD, Meis PJ, et al. The preterm prediction study: a clinical risk assessment system. Am J Obstet Gynecol 1996;174:1885-93.
20. Savitz DA, Dole N, Herring AH, Kaczor D, Murphy J, Siega-Riz AM, et al. Should spontaneous and medically indicated preterm births be separated for studying aetiology? Paediatr Perinat Epidemiol 2005;19:97-105.
21. Buchmayer SM, Sparen P, Cnattingius S. Previous pregnancy loss: risks related to severity of preterm delivery. Am J Obstet Gynecol 2004;191:1225-31.
22. Ananth CV, Vintzileos AM. Maternal-fetal conditions necessitating a medical intervention resulting in preterm birth. Am J Obstet Gynecol 2006;195:1557-63.
23. Meis PJ, Goldenberg RL, Mercer BM, Iams JD, Moawad AH, Miodovnik M, et al. The preterm predic-tion study: risk factors for indicated preterm births. Maternal-Fetal Medicine Units Network of the National Institute of Child Health and Human Development. Am J Obstet Gynecol 1998;178:562-7.
24. Savitz DA, Hertz-Picciotto I, Poole C, Olshan AF. Epidemiologic measures of the course and outcome of pregnancy. Epidemiol Rev 2002;24:91-101.
25. Gareen IF, Morgenstern H, Greenland S, Gifford DS. Explaining the association of maternal age with cesarean delivery for nulliparous and parous women. J Clin Epidemiol 2003;56:1100-10.
26. Morken NH, Kallen K, Jacobsson B. Fetal growth and onset of delivery: a nationwide population-based study of preterm infants. Am J Obstet Gynecol 2006;195:154-61.
27. Papiernik E. Fetal growth retardation: a limit for the further reduction of preterm births. Matern Child Health J 1999;3:63-9.
28. Ananth CV, Vintzileos AM. Epidemiology of preterm birth and its clinical subtypes. J Matern Fetal Neonatal Med 2006;19:773-82.
29. Fiscella K. Race, genes and preterm delivery. J Natl Med Assoc 2005;97:1516-26.30. Hogue CJ, Bremner JD. Stress model for research into preterm delivery among black women. Am J
Obstet Gynecol 2005;192(Suppl 5):S47-55.31. McGregor JA, Allen KG, Harris MA, Reece M, Wheeler M, French JI, et al. The omega-3 story: nu-
tritional prevention of preterm birth and other adverse pregnancy outcomes. Obstet Gynecol Surv 2001;56(Suppl 1):S1-13.
32. Ramakrishnan U, Manjrekar R, Rivera J, Gonzales-Cossio T, Martorell R. Micronutrients and preg-nancy outcome: a review of the literature. Nutr Res 1999;19:103-59.
33. Stotland NE, Caughey AB, Lahiff M, Abrams B. Weight gain and spontaneous preterm birth: the role of race or ethnicity and previous preterm birth. Obstet Gynecol 2006;108:1448-55.
34. Alderliesten ME, Vrijkotte TG, van der Wal MF, Bonsel GJ. Late start of antenatal care among ethnic minorities in a large cohort of pregnant women. BJOG 2007;114:1232-9.
35. Vintzileos AM, Ananth CV, Smulian JC, Scorza WE, Knuppel RA. The impact of prenatal care in the United States on preterm births in the presence and absence of antenatal high-risk conditions. Am J Obstet Gynecol 2002;187:1254-7.
36. Barros H, Tavares M, Rodrigues T. Role of prenatal care in preterm birth and low birthweight in Portugal. J Public Health Med 1996;18:321-8.
37. Van der Hulst LA, van Teijlingen ER, Bonsel GJ, Eskes M, Bleker OP. Does a pregnant woman’s in-tended place of birth influence her attitudes toward and occurrence of obstetric interventions? Birth 2004;31:28-33.
38. Braveman P, Egerter S, Edmonston F, Verdon M. Racial/ethnic differences in the likelihood of cesar-ean delivery, California. Am J Public Health 1995;85:625-30.
39. Risch N, Burchard E, Ziv E, Tang H. Categorization of humans in biomedical research: genes, race and disease. Genome Biol 2002;3(7):comment 2007:1-12.
40. Kistka ZA, Palomar L, Lee KA, Boslaugh SE, Wangler MF, Cole FS, et al. Racial disparity in the fre-quency of recurrence of preterm birth. Am J Obstet Gynecol 2007;196:131-6.
41. Patel RR, Steer P, Doyle P, Little MP, Elliott P. Does gestation vary by ethnic group? A London-based study of over 122 000 pregnancies with spontaneous onset of labour. Int J Epidemiol 2003;33:107-13.
42. Getahun D, Ananth CV, Selvam N, Demissie K. Adverse perinatal outcomes among interracial cou-ples in the United States. Obstet Gynecol 2005;106:81-8.
51
_______________________________________________________ Ethnic differences in preterm birth
3
43. Palomar L, DeFranco EA, Lee KA, Allsworth JE, Muglia LJ. Paternal race is a risk factor for preterm birth. Am J Obstet Gynecol 2007;197:152-7.
44. Municipality of Amsterdam. Department for Research and Statistics. Population. In: van Zee W, Hyl-kema C, editors. Key figures Amsterdam 2004 [In Dutch]. Amsterdam, the Netherlands: Stadsdruk-kerij Amsterdam; 2004, pp. 1-4.
45. Tromp M, van Eijsden M, Ravelli ACJ, Bonsel GJ. Anonymous non-response analysis in the ABCD-cohort study enabled by probabilistic record linkage. Paediatr Perinat Epidemiol 2009;23:264-72.
46. Goldenberg RL, Iams JD, Mercer BM, Meis PJ, Moawad A, Das A, et al. The Preterm Prediction Study: toward a multiple-marker test for spontaneous preterm birth. Am J Obstet Gynecol 2001;185:643-51.
47. Iams JD. Prediction and early detection of preterm labor. Obstet Gynecol 2003;101:402-12.48. Davis CH, MacKinnon DP, Schultz A, Sandler I. Cumulative risk and population attributable fraction
in prevention. J Clin Child Adolesc Psychol 2003;32:228-35.
Geertje GoedhartMarcel F. van der Wal
Pim CuijpersGouke J. Bonsel
Chapter 4Psychosocial problems and continued smoking during pregnancy.
The present study examined the association of several psychosocial problems with continued smoking during pregnancy. Based on a population-based cohort study among pregnant women in Amsterdam (n = 8266), women who smoked before pregnancy were included in this study (n = 1947). Women completed a questionnaire around the 12th week of gestation. Based on whether they smoked in the past week, participants were categorized as quitters or non-quitters. Depressive symptoms (CES-D), anxiety (STAI), pregnancy-related anxiety, job strain, parenting stress and physical/sexual violence were measured. Multiple logistic regression analyses were performed. After adjustment for sociodemographic and smoking-related covariates, low and high levels of pregnancy-related anxiety, exposure to physical/sexual violence, and high job strain were significantly associated with continued smoking during pregnancy. Intensive and comprehensive smoking cessation programs are required for pregnant women, which includes the management of psychosocial problems.
55
______________________________________________________ Psychosocial problems and smoking
4
Introduction
Maternal prenatal smoking has found to be associated with adverse infant health outcomes, such as intrauterine growth restriction,1 preterm delivery,2 perinatal mortality,3 and impaired brain development.4 Today, most pregnant women seem to be aware of these health risks,5 however, awareness alone is not sufficient to prompt women to quit smoking. In 2002, an USA and a Dutch survey showed respectively a smoking prevalence of 22% and 17% among pregnant women.1,6
Apart from physical and psychological addiction to cigarettes, psychosocial stress is reported to be one of the main motives for continued smoking during pregnancy.7 Evidence on the association of psychosocial problems with continued smoking during pregnancy is however limited by methodological limitations of previous studies, such as a retrospective design,8 small sample size,9 heterogeneity of the reference group,10 or a clinical sample.11 Furthermore, the main subject of previous research has been the effect of depression, ignoring the effect of other psychosocial problems. As part of a large population-based study among pregnant women in Amsterdam, the association between a wide range of psychosocial problems and continued smoking during pregnancy was examined, eventually to provide indications for the improvement of smoking cessation programs for pregnant women.
Methods
Study populationThe present study is part of the Amsterdam Born Children and their Development (ABCD) study. Details on the study design are published elsewhere.2 A group of 12 373 urban pregnant women received a pregnancy-questionnaire after their first prenatal visit (around the 12th week of gestation); 8266 women completed the questionnaire. Compared to the initially approached group, the response group was older, more often nulliparae, higher educated, and more often from Dutch origin. For this study, only women who smoked before pregnancy (24%) and who had complete data on all relevant variables were included (n = 1947).
MeasurementsThe outcome measure - smoking status during early pregnancy - was defined as follows: women who reported in the questionnaire to have changed their smoking habits since they knew they were pregnant and who did not smoke in the past week were categorized as quitters; all other women were categorized as non-quitters.
Depressive symptoms were measured by the Center for Epidemiologic Studies Depression Scale (CES-D) (α = 0.90).12 Anxiety was measured by the state-scale of the State-Trait Anxiety Inventory (STAI) (α = 0.94).13 Pregnancy-related anxiety was measured by the revised version of the Pregnancy Related Anxiety Questionnaire (PRAQ-R) (α = 0.81).14 Parenting stress was measured by the Frequency scale of the Parenting Daily Hassles (PDH) (α = 0.85).15 For descriptive analysis, the sum scores of
these four scales were trichotomised around the 50th and 90th percentiles into low, moderate and high. For parenting stress an additional category was made that included women with no children at home to take care of. Job strain was measured by the Job Content Questionnaire.16 Job strain was defined as a combination of workload (25 items; α = 0.85) and job control (11 items; α = 0.92) and categorized into low, moderate and high job strain. High job strain refers to a combination of high workload and low job control.17 An additional category was made for unemployed women. The occurrence of physical and/or sexual violence during pregnancy was retrospectively asked with 2 items in a second questionnaire 3 months after delivery. Because only 5132 of the 8266 women completed this questionnaire, an additional category was made for system-missings.
Sociodemographic covariates included maternal age, parity, ethnicity (based on the country of birth of the pregnant woman’s mother), educational attainment, and cohabitation status. Smoking-related covariates included the amount of smoking before pregnancy, and whether or not people were smoking in the direct environment of the woman. The desirability of the pregnancy was measured by a 4-item scale constructed for this study (α = 0.75). Women were asked if they were happy with the pregnancy and if they wanted the pregnancy. The sum score of the items was trichotomised around the 50th and 90th percentiles into high, moderate and low desirability.
Statistical analysisChi-square tests were used to examine the associations of covariates and psychosocial problems with continued smoking during pregnancy. Logistic regression analyses were performed to estimate the independent associations between each of the psychosocial problems and continued smoking during pregnancy (non-quitters versus quitters), following a predefined hierarchical format of adjustment in three steps: respectively the sociodemographic covariates, the other covariates and the other psychosocial problems were added to the model. Because of high collinearity between the variables depressive symptoms and anxiety (r = 0.89), these variables were not entered into the same model. Depressive symptoms and anxiety (both linear terms), and pregnancy-related anxiety (linear and quadratic term) were entered as continuous variables into the regression analyses, whereas the other psychosocial problems were treated as categorical variables.
Results
Thirty-eight percent of the women continued to smoke during early pregnancy (Table 4.1). These non-quitters were more often primi-/multiparae, low educated, not-cohabiting, from Turks or Moroccan origin, and/or heavy smokers before pregnancy compared to the quitters. The non-quitters also had a higher level of psychosocial problems.
Univariate logistic regression analyses showed a significant association of depressive symptoms, anxiety, pregnancy-related anxiety, physical/sexual violence, high job strain, and unemployment during pregnancy with continued smoking during early pregnancy (Table 4.2). After all steps of adjustment, in which the odds ratios changed most after adjustment for sociodemographic covariates,
57
______________________________________________________ Psychosocial problems and smoking
4
women with high and low levels of pregnancy-related anxiety (linear term: OR 0.75, p < 0.001; quadratic term: OR 1.01, p < 0.001), with exposure to physical/sexual violence (OR 3.13, p = 0.021), and with high job strain (OR 1.66, p = 0.047) remained at a significantly higher risk of continued smoking during pregnancy.
Discussion
In this large population-based cohort of pregnant women, significant independent associations between several psychosocial problems and continued smoking during early pregnancy were observed: women with high levels of pregnancy-related anxiety, high job strain, and exposure to physical/sexual violence more often continued to smoke during pregnancy. This study is the first to report an association between pregnancy-related anxiety and continued smoking. Remarkably, we also observed a higher frequency of continued smoking among women with very low levels of pregnancy-related anxiety; this can possibly be explained by indifference about their own and the child’s health. Only one earlier study examined the association between physical/sexual violence and continued smoking during pregnancy, but they did not found a significant effect.8 The association between high job strain and continued smoking replicated previous findings.18,19 Unlike previous studies,9,10,20 a significant association of depressive and anxiety symptoms with smoking during pregnancy was not observed. Differences in level of adjustment or in the composition of the reference group could partly account for these inconsistent findings.21 Although the CES-D and STAI are valid and widely used scales for the measurement of respectively depressive and anxiety symptoms, DSM-IV diagnosed depressive and anxiety disorders may be better predictors of smoking during pregnancy.11
Several underlying mechanisms can explain the observed association between psychosocial problems and smoking.9,22 First, smoking enhances the sense of well-being and is used as a tool to cope with negative mood or stress experiences. Second, women with psychosocial problems are less confident about their chances of successful quitting. Third, smoking provides quick and direct reinforcement to depressed women who have a reduced capacity to initiate other reinforcing activities.
A limitation of this study is the cross-sectional design, which keeps from answering the question of causality. The prospective study of Blalock et al.11 provided first evidence on the assumed causal direction that psychosocial problems predict continued smoking during pregnancy. Another limitation concerns the validity of self-reported smoking status, which is often unreliable and which in this study is limited to early pregnancy. However, the actual associations are therefore probably even higher, because women with psychosocial problems usually are overrepresented among the group of false quitters.11
In order for smoking cessation programs among pregnant women to be more effective, an intensive and comprehensive approach is required,22 which includes the management of psychosocial problems like pregnancy-related anxiety or job strain. Such programs may provide a huge improvement in perinatal health outcomes, as the offspring of stressed pregnant women who smoke are at an increased risk by the direct and indirect effects of both smoking and stress.23
High 1.30 (0.79-2.15) 1.17 (0.68-2.01) 1.22 (0.60-2.09) 0.93 (0.50-1.76)aOR1 = univariate logistic regression, crude odds ratio.bOR2 = odds ratio, adjusted for age, parity, ethnicity, education and cohabitation status.cOR3 = OR2, additionally adjusted for amount of smoking before pregnancy, smokers in environment and desirability of the pregnancy. dOR4 = OR3, additionally adjusted for anxiety, pregnancy-related anxiety, job strain, violence and parenting stress.eDepressive symptoms and anxiety not adjusted for each other, because of high collinearity.
61
______________________________________________________ Psychosocial problems and smoking
4
References
1. Jaddoe VWV, Verburg BO, de Ridder MAJ, Hofman A, Mackenbach JP, Moll HA, et al. Maternal smoking and fetal growth characteristics in different periods of pregnancy: the Generation R Study. Am J Epidemiol 2007;165:1207-15.
2. Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in preterm birth and its sub-types: the effect of a cumulative risk profile. BJOG 2008;115:710-9.
3. Salihu HM, Aliyu MH, Pierre-Louis BJ, Alexander GR. Levels of excess infant deaths attributable to ma-ternal smoking during pregnancy in the United States. Matern Child Health J 2003;7:219-27.
4. Roza SJ, Verburg BO, Jaddoe VWV, Hofman A, Mackenbach JP, Steegers EAP, et al. Effects of mater-nal smoking in pregnancy on prenatal brain development. The Generation R Study. Eur J Neurosci 2007;25:611-7.
5. Hymowitz N, Schwab M, McNerney C, Schwab J, Eckholdt H, Haddock K. Postpartum relapse to cigarette smoking in inner city women. J Natl Med Assoc 2003;95:461-74.
6. Goodwin RD, Keyes K, Simuro N. Mental disorders and nicotine dependence among pregnant women in the United States. Obstet Gynecol 2007;109:875-83.
7. Haslam C, Draper ES. A qualitative study of smoking during pregnancy. Psychol Health Med 2001;6:95-9.8. Bullock LF, Mears JLC, Woodcock C, Record R. Retrospective study of the association of stress and smo-
king during pregnancy in rural women. Addict Behav 2001;26:405-13.9. Zhu SH, Valbo A. Depression and smoking during pregnancy. Addict Behav 2002;27:649-58.10. Orr ST, Newton E, Tarwater PM, Weismiller D. Factors associated with prenatal smoking among black
women in eastern North Carolina. Matern Child Health J 2005;9:245-52.11. Blalock JA, Robinson JD, Wetter DW, Cinciripini PM. Relationship of DSM-IV-based depressive disorders
to smoking cessation and smoking reduction in pregnant smokers. Am J Addict 2006;15:268-77.12. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl
Psychol Meas 1977;1:385-401.13. Spielberger CD, Gorsuch RL, Lushene RE. STAI Manual for the Stait-Trait Anxiety Inventory. Palo Alto,
CA: Consulting Psychologists Press; 1970.14. Huizink AC, Mulder EJH, Robles de Medina PG, Visser GHA, Buitelaar JK. Is pregnancy anxiety a distinc-
tive syndrome? Early Hum Dev 2004;79:81-91.15. Crnic KA, Greenberg MT. Minor parenting stresses with young children. Child Dev 1990;61:1628-37.16. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire
(JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322-55.
17. Vrijkotte TGM, van der Wal MF, van Eijsden M, Bonsel GJ. First-trimester working conditions and birth-weight: a prospective cohort study. Am J Public Health 2009;99:1409-16.
18. Dejin-Karlsson E, Hanson BS, Östergren PO, Ranstam J, Isacsson SO, Sjöberg NO. Psychosocial resources and persistent smoking in early pregnancy - a population study of women in their first pregnancy in Swe-den. J Epidemiol Community Health 1996;50:33-9.
19. Wergeland E, Strand K, Bjerkedal T. Smoking in pregnancy: a way to cope with excessive workload? Scand J Prim Health Care 1996;14:21-8.
20. Hanna EZ, Faden VB, Dufour MC. The motivational correlates of drinking, smoking, and illicit drug use during pregnancy. J Subst Abuse 1994;6:155-67.
21. Ludman EJ, McBride CM, Nelson JC, Curry SJ, Grothaus LC, Lando HA, et al. Stress, depressive symp-toms, and smoking cessation among pregnant women. Health Psychol 2000;19:21-7.
22. Borrelli B, Bock B, King T, Pinto B, Marcus BH. The impact of depression on smoking cessation in women. Am J Prev Med 1996;12:378-87.
23. Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in term birthweight: the role of constitutional and environmental factors. Paediatr Perinat Epidemiol 2008;22:360-8.
Geertje GoedhartAnne C. Snijders
Arlette E. HesselinkMireille N. van Poppel
Gouke J. BonselTanja G.M. Vrijkotte
Chapter 5Maternal depressive symptoms in relation to perinatal mortality and morbidity: results from a large multi-ethnic cohort study.
Objective This study explored (i) whether maternal depressive symptoms during pregnancy are associated with preterm birth (PTB), small-for-gestational-age (SGA), a low Apgar score and child loss, (ii) whether maternal smoking mediates the associations and (iii) whether the associations differ by ethnic background.
Methods Pregnant women in Amsterdam were approached during their first prenatal visit to par-ticipate in the Amsterdam Born Children and their Development (ABCD) study. They filled out a pregnancy questionnaire covering sociodemographic data, lifestyle and (psychosocial) health. Depressive symptoms were assessed with the CES-D scale. The baseline sample consisted of 8052 women; main ethnic groups were: Dutch, Creole, Turkish and Moroccan. Multiple logistic regres-sion analysis was performed.
Results The prevalence of perinatal outcomes was: 5.4% (PTB), 12.3% (SGA), 1.5% (low Apgar score) and 1.4% (child loss). The prevalence of high depressive symptomatology was 30.6%. After adjustment for maternal age, parity, education, ethnicity, pre-pregnancy BMI, hypertension, alcohol and drug use and a small mediation effect of maternal smoking, high vs. low levels of depressive symptoms were associated with SGA (OR 1.19, p = .02) and a low Apgar score (OR 1.74, p = .01), but not with PTB (OR 1.16, p = .18) and child loss (OR 1.28, p = .24). Stratified analyses by ethnic background showed a tendency towards higher risks, though insignificant, among Creole women.
Conclusions Several pathways may explain the detrimental effects of maternal depressive symp-tomatology on perinatal health outcomes, including a psychoendocrinological pathway involving the hormone cortisol or mediation effects by maternal risk behaviors. Further research should explore the underlying pathways, in particular among ethnic subgroups.
65
_______________________________________________ Maternal depression and perinatal outcomes
5
Introduction
While perinatal mortality rates decreased over the last decades due to improved perinatal health care, perinatal morbidity rates increased.1-4 This is not without long-term clinical, social and economical consequences; major indicators of perinatal morbidity, including small-for-gestational-age (SGA), preterm birth (PTB) and a low Apgar score, have been shown to be related to adverse health outcomes in adult life. PTB, for example, may result in cognitive, behavioral or emotional problems.4, 5 Babies born SGA are at an increased risk for cardiovascular disease, type 2 diabetes and psychomotor and intellectual impairment.6, 7 A low Apgar score is related to an increased risk for cerebral palsy, epilepsy, neurologic disability and low cognitive function.2, 8 This emphasizes the urgent need for prevention of not only perinatal mortality but also perinatal morbidity, by exploring and acting on their causal factors.
One of the causal factors for perinatal mortality and morbidity may be maternal depression during pregnancy. It is hypothesized that depression results in neuroendocrine alterations, including a hyper- or hypoactivity of the hypothalamic-pituitary-adrenal (HPA) axis.9-11 The hormonal end products of the HPA-axis, such as cortisol and norepinephrine, may in turn affect uterine artery blood flow, fetal development and growth, and parturition. Maternal depression may also indirectly affect perinatal health outcomes through mediation by maternal risk behaviors. Depressed women, for example, more often continue to smoke during pregnancy12 and in turn, smoking is an important risk factor for adverse perinatal health outcomes;13-16 hence, maternal smoking is a potential mediator.
It has been observed that fetuses of depressed women are more likely to have a reduced growth17-21 – resulting in SGA babies – and to be born preterm.17-20, 22-25 Other studies, however, did not observe an association between maternal depression and SGA or PTB.17, 21, 22, 26-28 One of the explanations for the inconsistent results is the limited comparability of the studies; they differ with respect to several methodological issues like the measurement of maternal depression (mental disorder21, 27 vs. depressive symptomatology20, 24, 28), the time window during pregnancy when maternal depression is measured (first24, second27 or third26 trimester), the study population (teenagers20, 21 vs. adults20, white vs. black25), and the study design (retrospective21 vs. prospective24). It is also hypothesized that offspring of depressed women are more likely to have an impaired general condition at birth, as measured by the Apgar score,29 or to die during the perinatal period, but this is – so far – hardly confirmed.21, 26, 28, 30, 31 As the global prevalence of mental illness, including prenatal depression, is rising,32 it becomes more and more important to study the adverse effects of maternal depression on perinatal mortality and morbidity in large prospective birth cohort studies.
Ethnic minority groups often have a high prevalence of both prenatal depression and adverse perinatal health outcomes. Although previous studies showed that ethnic disparities in perinatal health outcomes could only to a very small part be explained by levels of depressive symptoms,33-35 there might be interactive effects of depression with ethnic background. It has for example been suggested that, compared to white women, black women are more susceptible for stress-induced neuroendocrine and inflammatory pathways into adverse perinatal outcomes, possibly as a result of higher vascular (re)activity.35-38
As part of a prospective multi-ethnic cohort study among pregnant women in Amsterdam, we explored (i) whether maternal depressive symptoms during pregnancy are associated with four major perinatal health outcomes, i.e. PTB, SGA, a low Apgar score and child loss, (ii) whether maternal smoking during pregnancy mediates these associations and (iii) whether the associations are modified by ethnic background. As far as we know, this is the largest study in this specific research field.
Methods
The present study is part of the Amsterdam Born Children and their Development (ABCD) study. The ABCD study is a prospective community-based cohort study aimed at examining the relationship between maternal lifestyle and psychosocial conditions during pregnancy and the child’s health at birth as well as in later life. Approval of the study was obtained from the Central Committee on Research involving Human Subjects, the Medical Ethical Committees of participating hospitals and the Registration Committee of Amsterdam.
Setting and populationBetween January 2003 and March 2004, all pregnant women in Amsterdam were invited to enroll in the ABCD study at their first prenatal visit to participating obstetric care providers (around the 12th week of gestation). All approached women (n = 12 373) received a pregnancy questionnaire; 8266 women returned the questionnaire (response rate 67%). For the current study, women with multiple births were excluded. The baseline sample consisted of 8050 women who had complete data on the independent variable (CES-D score) and relevant covariates.
Data collectionTwo weeks after their first prenatal visit, a pregnancy questionnaire was sent to the women’s home address, to be returned by prepaid mail. The questionnaire covered sociodemographic data, lifestyle and (psychosocial) health. Reminders were sent 2 weeks after the initial mailing. Questionnaires were in Dutch and, depending on the woman’s country of birth, accompanied by an English, Turkish or Arabic copy. Information on pregnancy outcomes was obtained from two sources: (1) the Youth Health Care Registration at the Public Health Service in Amsterdam and (2) the Dutch Perinatal Registration (PRN) (individual linkage possible for 80% of our sample), which include the national obstetric databases for midwives and gynecologists, and the national neonatal database for pediatricians and neonatologists. Additional information on child loss was gathered from the mother and health care providers.
Maternal depressive symptomsThe level of maternal depressive symptoms during pregnancy was measured by self-report in the pregnancy questionnaire using the Center for Epidemiologic Studies Depression scale (CES-D). The CES-D is designed to determine depressive symptomatology in the week previous to the
67
_______________________________________________ Maternal depression and perinatal outcomes
5
acquisition of the questionnaire and does not aim to identify chronic or clinical depression.39 Nevertheless, scores on the CES-D scale correlate quite well with clinical assessments and a score of 16 or higher is commonly used to define high-risk groups or possible cases of clinical depression.24,
39, 40 In this study, we also applied the cut-off point of 16 to dichotomize the group into low and high depressive symptomatology. The CES-D is a 20-item scale which was found to have good validity, reliability and factor structure.39, 41 The scale has been translated into different languages; in this study, the validated Dutch version was used.42 Although the CES-D was initially designed for use in the general population, it has been demonstrated to be a valid and reliable measurement tool in different subgroups, including pregnant24, 43 and black39, 44, 45 women. Reliability analysis in our sample of pregnant women resulted in a Cronbach’s Alpha of 0.90 for the total sample of pregnant women; reliability was comparable among ethnic subgroups (Cronbach’s Alpha ranging from 0.89 to 0.91).
Perinatal health outcomesFour major perinatal health outcomes were explored, i.e. preterm birth (PTB), small-for-gestational-age (SGA), low Apgar score, and child loss. PTB was defined as a delivery between 24.0 and 36.6 weeks of completed gestation. Data on gestational duration were based on ultrasound or, when unavailable (<10%), timing of last menstrual period. Complete data for PTB analysis were available for 7550 women. Newborns were categorized as SGA when they had a birthweight below the 10th percentile for gestational age on the basis of gender- and parity-specific standards from the PRN. For SGA analysis, we excluded births below 24.0 weeks of gestation; finally, 7530 women had complete data for this analysis. The Apgar score is commonly used to measure the general condition of a neonate immediately after birth. The obstetric care provider measures the newborn’s condition at 1, 5 and 10 minutes after birth by paying attention to the appearance, pulse, grimace, activity and respiration of the baby.29 At every item, a score of 0, 1 or 2 points can be obtained, with a maximum total score of 10. A low Apgar score was defined as below 7. Only Apgar scores measured at 5 minutes after birth were analyzed, since this score is regarded as the best predictor of survival in infancy.46, 47 For this specific analysis, fetal deaths were excluded; 6390 women had complete data for the analysis on Apgar scores. Missing data for Apgar scores were mainly due to incomplete linkage with PRN data. Child loss included miscarriages (<22 weeks of gestation), fetal deaths (22 weeks of gestation – delivery) and early neonatal deaths (0 – 7 days after delivery). Non-spontaneous abortions (n = 14) and unknown child losses (n = 158) were excluded, leaving 7901 women with complete data for the child loss analysis.
CovariatesThe following covariates were measured in the pregnancy questionnaire and included in the analysis: maternal age, parity (0, ≥1), education (years of education after primary school), ethnicity [based on the country of birth of the participant’s mother: Dutch, Creole (Surinamese-Creole, Antillean/Aruban and Ghanaian), Turkish, Moroccan and other non-Dutch countries], maternal pre-pregnancy body mass index (BMI), hypertensive disorder (no/unknown, chronic hypertension
and pregnancy-induced hypertension), alcohol consumption (no/yes) and drug use (no/yes) during pregnancy. Maternal smoking during pregnancy (no/yes) was included as a potential mediator. Chronic and pregnancy-induced hypertension were defined combining self-reported data from the questionnaires and data from the PRN registration. Chronic hypertension was the case if (a) pre-existent hypertension was recorded in the PRN, or if (b) women reported high blood pressure and/or the use of medicines against high blood pressure before pregnancy or in the first 20 weeks of pregnancy. Pregnancy-induced hypertension was the case if (a) pregnancy-related hypertension, eclampsia or preeclampsia was recorded in the PRN, or if (b) women without pre-existent hypertension reported high blood pressure and/or the use of medicines against high blood pressure during pregnancy.
Statistical analysisDescriptive statistics were used to profile the sample characteristics according to the level of maternal depressive symptoms; differences were tested with ANOVA for continuous variables and chi-square tests for categorical variables. To examine the association between maternal depressive symptoms (high vs. low) and perinatal health outcomes, multiple logistic regression analyses were performed, adjusting for maternal age, parity, education, ethnicity, pre-pregnancy BMI, hypertension, alcohol and drug use (Model 1); maternal age, education and pre-pregnancy BMI were included as continuous variables. For SGA analysis, the quadratic terms for maternal age and BMI were also included, while parity was excluded (the definition of SGA already account for the parity effect).
To examine whether maternal smoking during pregnancy is a mediator in the associations between maternal depressive symptoms and perinatal health outcomes, the following conditions must hold: (1) the independent variable must affect the mediator, (2) the independent variable must affect the outcome variable, (3) the mediator must affect the outcome variable, (4) the effect of the independent variable on the outcome variable must be less when controlling for the effect of the mediator.48 Condition 1 and 3 were tested with descriptive statistics. Condition 2 and 4 were tested with logistic regression analysis by additionally entering maternal smoking to the adjusted model (Model 2).
To examine a possible interaction effect of maternal depressive symptoms with ethnic background, interaction terms were tested by logistic regression analysis.48 In addition, stratified analyses by the main ethnic groups (Dutch, Creole, Turkish and Moroccan) were performed for all four perinatal health outcomes. Data were analyzed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). In all analyses, a p-value <0.05 was considered significant.
Results
Of the 8050 women in the baseline sample, 2465 (30.6%) reported high levels of depressive symptoms during pregnancy (Table 5.1). These women were significantly (p < .001) younger,
69
_______________________________________________ Maternal depression and perinatal outcomes
5
less often nulliparous and lower educated compared to the women who reported low levels of depressive symptoms. The average pre-pregnancy BMI (23.7 vs. 22.9 kg/m2) and the use of cigarettes (14.4% vs. 7.4%) and drugs (3.9% vs. 1.8%) during pregnancy was furthermore higher among these women. The use of alcohol during pregnancy was, however, higher among the women with low depressive symptomatology (23.0% vs. 16.4%). In our sample, 4270 women (53%) were of native Dutch origin, 640 women (8%) had a Creole background, 402 (5%) a Turkish background, 690 (8.5%) a Moroccan background and the final 2048 women (25.5%) had an origin in another non-Dutch country. While already 22% of the Dutch women reported high levels of depressive symptoms, this percentage was even higher among the Creole (42.8%), Turkish (55.7%), Moroccan (42.0%) and other non-Dutch group (35.6%).
In our sample, we observed a prevalence of 5.4% for PTB, 12.3% for SGA, 1.5% for a low Apgar score and 1.4% for child loss. Univariate logistic regression analysis showed that the prevalence of all four outcomes was significantly higher among the group of women with high levels of depressive symptoms (Table 5.2). After adjustment for relevant covariates (maternal age, parity, education, ethnicity, pre-pregnancy BMI, hypertension, alcohol and drug use; Model 1), the significant association of high vs. low depressive symptoms with PTB (OR 1.21, p = .10) and child loss (OR 1.32, p = .18) disappeared, while the associations with SGA (OR 1.25, p = .004) and a low Apgar score (OR 1.77, p = .01) remained statistically significant.
To examine whether maternal smoking during pregnancy mediates the association between maternal depressive symptoms and perinatal outcomes, we tested the conditions for mediation.48 Maternal depressive symptomatology was associated with maternal smoking prevalence during pregnancy (p < .001, Table 1). Maternal smoking was associated with PTB (p < .001) and SGA (p < .001), but not with a low Apgar score (p = .11) and child loss (p = .06) (results not shown). By adding maternal smoking to the adjusted logistic regression models, the effect of maternal depressive symptoms on the perinatal outcomes slightly decreased (Model 2, Table 5.2). All together, maternal smoking can be considered to be a partial mediator in the association between maternal depressive symptoms and both PTB and SGA. After adjustment for maternal smoking, the association of high vs. low depressive symptoms with SGA (OR 1.19, p = .02) and a low Apgar score (OR 1.74, p = .01) was statistically significant, while the association with PTB (OR 1.61, p = .18) and child loss (OR 1.28, p = .24) was insignificant.
The inclusion of an interaction term between depressive symptoms and ethnicity to the full logistic regression models resulted in insignificant interaction effects (p-values ranging from .367 to .903; results not shown). Stratified analyses by ethnic group, however, showed a tendency towards a stronger association of high vs. low levels of maternal depressive symptoms with all four perinatal health outcomes among the Creole group compared to the other ethnic groups (Table 5.3a-d).
a Adjusted for maternal age, parity, education, ethnicity, pre-pregnancy BMI, hypertension, alcohol and drug use; SGA analysis: additional inclusion of quadratic term for maternal age and pre-pregnancy BMI, exclusion of parity because SGA is by definition corrected for parity.
bAdditionally adjusted for maternal smoking during pregnancy (potential mediator).
71
_______________________________________________ Maternal depression and perinatal outcomes
5
Table 5.3a Association between maternal depressive symptoms and SGA, separated by ethnicity.Ethnicity Depressive
aAdjusted for maternal age (linear & quadratic term), education, pre-pregnancy BMI (linear & quadratic term), hypertension, alcohol and drug use.bAdditionally adjusted for maternal smoking during pregnancy.
Table 5.3b Association between maternal depressive symptoms and PTB, separated by ethnicity.Ethnicity Depressive
aAdjusted for maternal age, parity, education, pre-pregnancy BMI, hypertension, alcohol and drug use.bAdditionally adjusted for maternal smoking during pregnancy.
Table 5.3c Association between maternal depressive symptoms and a low Apgar score, separated by ethnicity.Ethnicity Depressive
aAdjusted for maternal age, parity, education, pre-pregnancy BMI, hypertension, alcohol and drug use.bAdditionally adjusted for maternal smoking during pregnancy.
aAdjusted for maternal age, parity, education, pre-pregnancy BMI, hypertension, alcohol and drug use.bAdditionally adjusted for maternal smoking during pregnancy.
Discussion
The purpose of the present study was to explore whether maternal depressive symptoms during pregnancy are associated with major perinatal health outcomes. In general, higher rates of PTB, SGA, a low Apgar score and child loss were observed among women with high levels of depressive symptoms during pregnancy. After adjustment for covariates and a small mediation effect of maternal smoking, offspring of mothers with high depressive symptoms had a statistically significant higher risk for SGA and a low Apgar score. Although we did not find statistically significant interaction effects of depressive symptoms with ethnic background, a tendency for a higher risk was observed among women with a Creole background.
In contrast with previous smaller studies,21, 26-28 we observed a significant association between maternal depressive symptoms and a low Apgar score at 5 minutes after birth. The significant association we found with SGA was consistent with some previous studies,19, 20 but inconsistent with others.21, 27, 28 While several studies observed a higher PTB risk among depressed women,18-24 we could not prove this in our sample. This might be explained by the various definitions used for PTB. Although all studies defined PTB as a birth before 37.0 weeks of gestation, only one study23 restricted the PTB group by defining a minimum weeks of gestation. We excluded births before 24.0 weeks of gestation because they are mostly not viable. Including these births would have resulted in increased and significant odds ratios for the association between maternal depressive symptoms and PTB (OR 1.25, p = .03 after full adjustment), supporting previous studies. We, however, feel that including all births before 37.0 weeks of gestation results in overestimation of the PTB risk, partly reflecting a potential association between maternal depression and miscarriages/fetal death. The few studies that explored the association of maternal depression with child loss are hard to compare, because of different definitions of child loss. We did not observe a significant association with child loss, including both miscarriages and perinatal deaths, which was consistent with a previous study exploring stillbirths over the whole range of pregnancy.21 Several small studies exploring the
73
_______________________________________________ Maternal depression and perinatal outcomes
5
association with miscarriages, however, presented significant30, 49 as well as insignificant31 results. The number of child losses in our sample were too small to perform separate analyses for subtypes of child loss; moreover, the data on miscarriages were most likely incomplete.
Stratified analysis by ethnic background suggested that the association of high vs. low levels of maternal depressive symptoms with perinatal health outcomes was stronger among the Creole women compared to the other ethnic groups. A similar difference between African-American and white women was reported before by Orr and Miller.25 These differences are not likely to be explained by higher risk behaviors such as maternal smoking or a lack of prenatal health care visits, because these risk behaviors are also seen among the Turkish and Moroccan groups.33, 50, 51 A more likely explanation is the higher vascular activity and reactivity to stress among blacks.36, 37 Previous research exploring ethnic differences in vascular characteristics of pregnant women observed greater resistance artery contractility, higher vascular creatine kinase activity and lower nitric oxide bioavailability among blacks compared to whites, which may contribute to higher blood pressure levels and hypertensive disorders.52 In case of depression or stress, this may lead to exaggerated cardiovascular reactivity among blacks, implying a higher risk for adverse pregnancy outcomes.35
The comparison of studies on the detrimental effects of maternal depression during pregnancy is hampered by the use of various kinds of measurement tools to assess either maternal depressive disorders or depressive symptomatology. While most studies used a self-report questionnaire to assess the level of depressive symptoms,20, 22, 24, 28 some used a diagnostic interview to assess mental disorders based on DSM-IV criteria.21, 27 Observed associations seemed, however, independent of the measurement tool. The CES-D scale, which was used in this study to measure depressive symptomatology, was found to have good validity, internal consistency and factor structure in the general population39, 41 as well as in various subpopulations, including pregnant women24, 43 and different ethnic groups.24, 39, 44, 45 Nonetheless, it is questionable whether the commonly used cut-off point of a CES-D score of ≥16 to define high-risk groups can similarly be applied among pregnant women and certain ethnic groups, which tend to score higher on the somatic component of the CES-D scale.24, 41, 44, 45, 53, 54 To address this problem, several approaches can be followed, including the omission of the somatic subscale of the CES-D54 or by raising the cut-off point to the upper 10 percent.24 We explored both approaches by performing additional analyses; while we hypothesized that the association of maternal depressive symptoms with perinatal health outcomes would be stronger, the opposite appeared to be true (results not shown). This was also observed in a previous study by Kabir et al.54 which compared the psychometric properties and predictive capacity for perinatal outcomes of the full CES-D scale with the CES-D scale without the somatic component. Apparently, the full CES-D scale and customary cut-off point of 16 can appropriately be applied among pregnant women; whether this is also true for specific ethnic subgroups remains, however, unclear and needs to be further explored. In our sample, 30.6% of the women had a CES-D score of 16 or higher, meaning that they had elevated levels of depressive symptoms and that they were at high risk for clinical depression.24 This prevalence was even higher among the ethnic minority groups (42-55%). In the 2008 general health survey among the Amsterdam population, 46.6% of the women reported mild depressive symptoms and 8% severe depressive symptoms [measured
with the Kessler Psychological Distress scale (K10)].55 In the 2004 general health survey in Amsterdam, structured interviews were performed using the Composite International Diagnostic Interview (CIDI); the Turkish (14.9%) and Moroccan (6.6%) group were found to have a higher 1-month prevalence of depressive disorders compared to the Dutch (4.4%) and Surinamese (1.1%) group.56 These depression prevalences are, however, hard to compare as they are assessed using different psychometric measurement tools.
Several pathways may explain the association between maternal depressive symptomatology during pregnancy and offspring risk for SGA and a low Apgar score. First, maternal depression may have an indirect effect on the fetus through its promoting influence on other risk factors for adverse perinatal health outcomes. Examples of risk factors associated with both maternal depression and offspring health are poor nutrition,32, 57 hypertension,38, 58 use of cigarettes, alcohol and drugs,59, 60 and lack of prenatal health care visits.50 Most studies in this research field included smoking as a covariate in the statistical models though suggested that it would be more likely for smoking to be on the causal pathway, i.e. being a mediator.21, 23, 24, 28 Many people, including pregnant women, smoke to relieve depression, unhappiness or stress,12, 61 and maternal smoking, in turn, is an important risk factor for fetal growth and development.13, 16 We have shown through mediation analysis that smoking was a partial mediator in the association between maternal depressive symptoms and, in particular, PTB and SGA. As a result, reducing the prevalence of depression during pregnancy will not only directly improve perinatal health outcomes, but also indirectly through a reduction in the smoking prevalence. The use of antidepressants might also mediate an association between maternal depression and offspring outcome.19, 62, 63 In our sample, only 0,8% of the women reported the use of antidepressants during pregnancy. Antidepressant use was more often, but not exclusively, reported by women with high depressive symptomatology and by Dutch women. We did, however, not observe significant associations between antidepressant use and perinatal health outcomes (results not shown). Second, a direct pathway through which maternal depressive symptoms affect fetal development, growth and parturition involves the activity of the HPA-axis. Depression is found to be related to increased HPA activity which results, among others, in enhanced levels of the hormone cortisol.17 Maternal cortisol has the ability to cross the placenta. Normally, the placental barrier [including 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2)] limits fetal exposure to maternal cortisol.64 However, in the case of elevated maternal cortisol levels and/or placental 11β-HSD2 deficiency, increasing levels of maternal cortisol cross the placenta resulting in elevated fetal cortisol levels.9 This mechanism is supported by the observation that maternal cortisol levels are highly correlated to fetal cortisol levels.65 It is hypothesized that fetal exposure to excess cortisol levels may impair fetal development and growth and may trigger early parturition through several mechanisms, including dysregulation of the fetal autonomic nervous system activity,66 increased uterine artery vasoconstriction resulting in reduced uterine blood flow67 or increased vulnerability to infectious agents.68 Major strengths of the present study include the large sample size, the community-based sample and the prospective study design. Selective response was present in the ABCD study, with
75
_______________________________________________ Maternal depression and perinatal outcomes
5
a higher participation rate among Western women, but selection bias could not be proven.69 Even though the ethnic groups had reasonable sizes, ethnic minority groups were too small to prove a potential interaction effect with maternal depression on perinatal outcomes. Maternal depressive symptoms were measured during the first trimester of pregnancy, which is suggested to be the most critical window to affect fetal development.70 Nonetheless, a different level of depressive symptoms at the end of pregnancy might still increase or decrease the odds for adverse perinatal outcomes.17, 70, 71 We therefore recommend for future research to measure depressive symptoms at several time points during pregnancy.
In conclusion, in – to our knowledge – the largest study exploring the detrimental effects of maternal depressive symptomatology during pregnancy on perinatal mortality and morbidity, we observed a higher risk to be born SGA or with a low Apgar score for offspring of mothers who reported high levels of depressive symptoms during pregnancy. Although the effect sizes are moderate, the extremely high prevalence of high depressive symptoms during pregnancy in our sample makes it a major public health problem. Women should be screened for depressive symptomatology during early pregnancy or preferably, during the preconceptional period. They should subsequently be urged to follow a comprehensive intervention program which addresses both psychosocial well-being and risk behaviors like smoking. Further research should explore whether certain groups (e.g. black women) are more sensitive to the detrimental effects of high maternal depressive symptoms during pregnancy. Moreover, future research should explore the underlying mechanisms to explain the observed associations.
1. EURO-PERISTAT project with SCPE, EUROCAT, EURONEOSTAT. European perinatal health report. Better statistics for better health in pregnant women and their babies. Paris: EURO-PERISTAT; 2008. Retrieved from http://www.europeristat.com.
2. Thorngren-Jerneck K, Herbst A. Low 5-minute Apgar score: a population-based register study of 1 million term births. Obstet Gynecol 2001;98:65-70.
3. Romo A, Carceller R, Tobajas J. Intrauterine growth retardation (IUGR): epidemiology and etiology. Pediatr Endocrinol Rev 2009;6(Suppl 3):332-6.
4. Ramachandrappa A, Jain L. Health issues of the late preterm infant. Pediatr Clin North Am 2009;56:565-77.
5. McFarlin BL. Solving the puzzle of prematurity. Am J Nurs 2009;109:60-3.6. Barker DJ. Adult consequences of fetal growth restriction. Clin Obstet Gynecol 2006;49:270-83.7. Puga B, Puga PG, de Arriba A, Armendariz Y, Labarta JI, Longas AF. Psychomotor and intellectual
development (Neurocognitive Function) of children born small for gestational age (SGA). Transversal and longitudinal study. Pediatr Endocrinol Rev 2009;6(Suppl 3):358-70.
8. Ehrenstein V, Pedersen L, Grijota M, Nielsen GL, Rothman KJ, Sorensen HT. Association of Apgar score at five minutes with long-term neurologic disability and cognitive function in a prevalence study of Danish conscripts. BMC Pregnancy Childbirth 2009;9:14.
9. Seckl JR, Holmes MC. Mechanisms of disease: glucocorticoids, their placental metabolism and fetal ‘programming’ of adult pathophysiology. Nat Clin Pract Endocrinol Metab 2007;3:479-88.
10. Chida Y, Steptoe A. Cortisol awakening response and psychosocial factors: a systematic review and meta-analysis. Biol Psychol 2009;80:265-78.
11. Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CA. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosom Med 1996;58:432-46.
12. Borrelli B, Bock B, King T, Pinto B, Marcus BH. The impact of depression on smoking cessation in women. Am J Prev Med 1996;12:378-87.
13. Jaddoe VW, Troe EJ, Hofman A, Mackenbach JP, Moll HA, Steegers EA, et al. Active and passive maternal smoking during pregnancy and the risks of low birthweight and preterm birth: the Generation R Study. Paediatr Perinat Epidemiol 2008;22:162-71.
14. Salihu HM, Aliyu MH, Pierre-Louis BJ, Alexander GR. Levels of excess infant deaths attributable to maternal smoking during pregnancy in the United States. Matern Child Health J 2003;7:219-27.
15. Shah NR, Bracken MB. A systematic review and meta-analysis of prospective studies on the association between maternal cigarette smoking and preterm delivery. Am J Obstet Gynecol 2000;182:465-72.
16. Roza SJ, Verburg BO, Jaddoe VW, Hofman A, Mackenbach JP, Steegers EA, et al. Effects of maternal smoking in pregnancy on prenatal brain development. The Generation R Study. Eur J Neurosci 2007;25:611-7.
17. Alder J, Fink N, Bitzer J, Hosli I, Holzgreve W. Depression and anxiety during pregnancy: a risk factor for obstetric, fetal and neonatal outcome? A critical review of the literature. J Matern Fetal Neonatal Med 2007;20:189-209.
18. Field T, Diego M, Dieter J, Hernandez-Reif M, Schanberg S, Kuhn C, et al. Prenatal depression effects on the fetus and the newborn. Infant Behav Dev 2004;27:216-29.
19. Oberlander TF, Warburton W, Misri S, Aghajanian J, Hertzman C. Neonatal outcomes after prenatal exposure to selective serotonin reuptake inhibitor antidepressants and maternal depression using population-based linked health data. Arch Gen Psychiatry 2006;63:898-906.
20. Steer RA, Scholl TO, Hediger ML, Fischer RL. Self-reported depression and negative pregnancy outcomes. J Clin Epidemiol 1992;45:1093-9.
21. Ferri CP, Mitsuhiro SS, Barros MC, Chalem E, Guinsburg R, Patel V, et al. The impact of maternal experience of violence and common mental disorders on neonatal outcomes: a survey of adolescent mothers in Sao Paulo, Brazil. BMC Public Health 2007;7:209.
77
_______________________________________________ Maternal depression and perinatal outcomes
5
22. Kramer MS, Lydon J, Seguin L, Goulet L, Kahn SR, McNamara H, et al. Stress pathways to spontaneous preterm birth: the role of stressors, psychological distress, and stress hormones. Am J Epidemiol 2009;169:1319-26.
23. Jesse DE, Seaver W, Wallace DC. Maternal psychosocial risks predict preterm birth in a group of women from Appalachia. Midwifery 2003;19:191-202.
24. Orr ST, James SA, Blackmore PC. Maternal prenatal depressive symptoms and spontaneous preterm births among African-American women in Baltimore, Maryland. Am J Epidemiol 2002;156:797-802.
25. Orr ST, Miller CA. Maternal depressive symptoms and the risk of poor pregnancy outcome. Review of the literature and preliminary findings. Epidemiol Rev 1995;17:165-71.
26. Lundy BL, Jones NA, Field T, Nearing G, Davalos M, Pietro PA, et al. Prenatal depression effects on neonates. Infant Behav Dev 1999;22:119-29.
27. Andersson L, Sundstrom-Poromaa I, Wulff M, Astrom M, Bixo M. Neonatal outcome following maternal antenatal depression and anxiety: a population-based study. Am J Epidemiol 2004;159:872-81.
28. Berle JO, Mykletun A, Daltveit AK, Rasmussen S, Holsten F, Dahl AA. Neonatal outcomes in offspring of women with anxiety and depression during pregnancy. A linkage study from The Nord-Trondelag Health Study (HUNT) and Medical Birth Registry of Norway. Arch Womens Ment Health 2005;8:181-9.
29. Sykes GS, Molloy PM, Johnson P, Gu W, Ashworth F, Stirrat GM, et al. Do Apgar scores indicate asphyxia? Lancet 1982;1:494-6.
30. Nakano Y, Oshima M, Sugiura-Ogasawara M, Aoki K, Kitamura T, Furukawa TA. Psychosocial predictors of successful delivery after unexplained recurrent spontaneous abortions: a cohort study. Acta Psychiatr Scand 2004;109:440-6.
31. Nelson DB, McMahon K, Joffe M, Brensinger C. The effect of depressive symptoms and optimism on the risk of spontaneous abortion among innercity women. J Womens Health (Larchmt) 2003;12:569-76.
32. Leung BM, Kaplan BJ. Perinatal depression: prevalence, risks, and the nutrition link - A review of the literature. J Am Diet Assoc 2009;109:1566-75.
33. Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in term birthweight: the role of constitutional and environmental factors. Paediatr Perinat Epidemiol 2008;22:360-8.
34. Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in preterm birth and its subtypes: the effect of a cumulative risk profile. BJOG 2008;115:710-9.
35. Giscombe CL, Lobel M. Explaining disproportionately high rates of adverse birth outcomes among African Americans: the impact of stress, racism, and related factors in pregnancy. Psychol Bull 2005;131:662-83.
36. Barnes VA, Treiber FA, Musante L, Turner JR, Davis H, Strong WB. Ethnicity and socioeconomic status: impact on cardiovascular activity at rest and during stress in youth with a family history of hypertension. Ethn Dis 2000;10:4-16.
37. Dysart JM, Treiber FA, Pflieger K, Davis H, Strong WB. Ethnic differences in the myocardial and vascular reactivity to stress in normotensive girls. Am J Hypertens 1994;7:15-22.
38. Hogue CJ, Bremner JD. Stress model for research into preterm delivery among black women. Am J Obstet Gynecol 2005;192(5 Suppl):S47-55.
39. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401.
40. Weissman MM, Sholomskas D, Pottenger M, Prusoff BA, Locke BZ. Assessing depressive symptoms in five psychiatric populations: a validation study. Am J Epidemiol 1977;106:203-14.
41. Shafer AB. Meta-analysis of the factor structures of four depression questionnaires: Beck, CES-D, Hamilton, and Zung. J Clin Psychol 2006;62:123-46.
42. Bouma J, Ranchor AV, Sanderman R, van Sonderen E. The measurement of depressive symptoms with the CES-D: a manual. [In Dutch]. Groningen: Noordelijk Centrum voor Gezondheidsvraagstukken, Rijksuniversiteit Groningen; 1995.
43. Field T, Hernandez-Reif M, Diego M, Schanberg S, Kuhn C. Stability of mood states and biochemistry across pregnancy. Infant Behav Dev 2006;29:262-7.
44. Roberts RE. Reliability of the CES-D Scale in different ethnic contexts. Psychiatry Res 1980;2:125-34.45. Spijker J, van der Wurff FB, Poort EC, Smits CH, Verhoeff AP, Beekman AT. Depression in first generation
labour migrants in Western Europe: the utility of the Center for Epidemiologic Studies Depression Scale (CES-D). Int J Geriatr Psychiatry 2004;19:538-44.
46. Casey BM, McIntire DD, Leveno KJ. The continuing value of the Apgar score for the assessment of newborn infants. N Engl J Med 2001;344:467-71.
47. Drage JS, Kennedy C, Schwarz BK. The Apgar score as an index of neonatal mortality. A report from the collaborative study of cerebral palsy. Obstet Gynecol 1964;24:222-30.
48. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173-82.
49. Sugiura-Ogasawara M, Furukawa TA, Nakano Y, Hori S, Aoki K, Kitamura T. Depression as a potential causal factor in subsequent miscarriage in recurrent spontaneous aborters. Hum Reprod 2002;17:2580-4.
50. Alderliesten ME, Vrijkotte TG, van der Wal MF, Bonsel GJ. Late start of antenatal care among ethnic minorities in a large cohort of pregnant women. BJOG 2007;114:1232-9.
51. Troe EJ, Raat H, Jaddoe VW, Hofman A, Steegers EA, Verhulst FC, et al. Smoking during pregnancy in ethnic populations: the Generation R study. Nicotine Tob Res 2008;10:1373-84.
52. Brewster LM, Taherzadeh Z, Volger S, Clark JF, Rolf T, Wolf H, et al. Ethnic differences in resistance artery contractility of normotensive pregnant women. Am J Physiol Heart Circ Physiol 2010. doi:10.1152/ajpheart.00919.2009
53. Simon GE, Goldberg DP, Von Korff M, Ustün TB. Understanding cross-national differences in depression prevalence. Psychol Med 2002;32:585-94.
54. Kabir K, Sheeder J, Stevens-Simon C. Depression, weight gain, and low birth weight adolescent delivery: do somatic symptoms strengthen or weaken the relationship? J Pediatr Adolesc Gynecol 2008;21:335-42.
55. GGD Amsterdam. This healthy is Amsterdam! Final report Amsterdam Health Monitor 2008. [In Dutch]. Amsterdam: GGD Amsterdam; 2009.
56. De Wit MA, Tuinebreijer WC, Dekker J, Beekman AJ, Gorissen WH, Schrier AC, et al. Depressive and anxiety disorders in different ethnic groups: a population based study among native Dutch, and Turkish, Moroccan and Surinamese migrants in Amsterdam. Soc Psychiatry Psychiatr Epidemiol 2008;43:905-12.
57. Ramakrishnan U, Manjrekar R, Rivera J, Gonzales-Cossio T, Martorell R. Micronutrients and pregnancy outcome: a review of the literature. Nutr Res 1999;19:103-59.
58. Lau TK, Pang MW, Sahota DS, Leung TN. Impact of hypertensive disorders of pregnancy at term on infant birth weight. Acta Obstet Gynecol Scand 2005;84:875-7.
59. Dew PC, Guillory VJ, Okah FA, Cai J, Hoff GL. The effect of health compromising behaviors on preterm births. Matern Child Health J 2007;11:227-33.
60. Hanna EZ, Faden VB, Dufour MC. The motivational correlates of drinking, smoking, and illicit drug use during pregnancy. J Subst Abuse 1994;6:155-67.
61. Goedhart G, van der Wal MF, Cuijpers P, Bonsel GJ. Psychosocial problems and continued smoking during pregnancy. Addict Behav 2009;34:403-6.
62. Yonkers KA, Wisner KL, Stewart DE, Oberlander TF, Dell DL, Stotland N, et al. The management of depression during pregnancy: a report from the American Psychiatric Association and the American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114:703-13.
63. Suri R, Altshuler L, Hellemann G, Burt VK, Aquino A, Mintz J. Effects of antenatal depression and antidepressant treatment on gestational age at birth and risk of preterm birth. Am J Psychiatry 2007;164:1206-13.
64. Mark PJ, Waddell BJ. P-glycoprotein restricts access of cortisol and dexamethasone to the glucocorticoid receptor in placental BeWo cells. Endocrinology 2006;147:5147-52.
79
_______________________________________________ Maternal depression and perinatal outcomes
5
65. Glover V, Teixeira J, Gitau R, Fisk NM. Mechanisms by which maternal mood in pregnancy may affect the fetus. Contemp Rev Obstet Gynecol 1999;155-60.
66. Diego MA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, Gonzalez-Quintero VH. Prenatal depression restricts fetal growth. Early Hum Dev 2009;85(1):65-70.
67. Field T, Diego M. Cortisol: the culprit prenatal stress variable. Int J Neurosci 2008;118:1181-205.68. Wadhwa PD, Culhane JF, Rauh V, Barve SS, Hogan V, Sandman CA, et al. Stress, infection and preterm
birth: a biobehavioural perspective. Paediatr Perinat Epidemiol 2001;15(Suppl 2):17-29.69. Tromp M, van Eijsden M, Ravelli ACJ, Bonsel GJ. Anonymous non-response analysis in the ABCD
cohort study enabled by probabilistic record linkage. Paediatr Perinat Epidemiol 2009;23:264-72.70. Sandman CA, Glynn L, Schetter CD, Wadhwa P, Garite T, Chicz-DeMet A, et al. Elevated maternal
cortisol early in pregnancy predicts third trimester levels of placental corticotropin releasing hormone (CRH): priming the placental clock. Peptides 2006;27:1457-63.
71. Roesch SC, Dunkel-Schetter C, Woo G, Hobel CJ. Modeling the types and timing of stress in pregnancy. Anxiety, Stress, and Coping 2004;17:87-102.
Geertje GoedhartTanja G.M. Vrijkotte
Tessa J. RoseboomMarcel F. van der Wal
Pim CuijpersGouke J. Bonsel
Chapter 6Maternal cortisol and offspring birthweight: results from a large prospective cohort study.
Maternal psychosocial problems may affect fetal growth through maternal cortisol. This large prospective cohort study examined among 2810 women (1) the association of maternal cortisol levels with offspring birthweight and small-for-gestational-age (SGA) risk and (2) the mediating role of maternal cortisol on the relation between maternal psychosocial problems and fetal growth. Pregnant women in Amsterdam were approached during their first prepartum visit (±13 weeks gestation). Total maternal cortisol level was determined in serum and maternal psychosocial indicators were collected through a questionnaire. Maternal cortisol levels were negatively related to offspring birthweight (b = -0.35; p < 0.001) and positively to SGA (OR = 1.00; p = 0.027); after adjustment (for gestational age at birth, infant gender, ethnicity, maternal age, parity, BMI, and smoking), these effects were statistically insignificant. Post hoc analysis revealed a moderation effect by time of day: only in those women who provided a blood sample ≤09:00h (n = 94), higher maternal cortisol levels were independently related to lower birthweights (b = -0.94; p = 0.025) and a higher SGA risk (OR = 1.01; p = 0.032). Maternal psychosocial problems were not associated with cortisol levels. In conclusion, although an independent association between maternal cortisol levels in early pregnancy and offspring birthweight and SGA risk was not observed, exploratory post hoc analysis suggested that the association was moderated by time of day, such that the association was only present in the early morning. The hypothesis that maternal psychosocial problems affect fetal growth through elevated maternal cortisol levels could not be supported.
83
________________________________________________ Maternal cortisol and offspring birthweight
6
Introduction
An adverse intrauterine environment is an important determinant of postnatal pathology (e.g. cardiovascular, metabolic or psychiatric disease), through fetal programming.1 It is hypothesized that, besides environmental factors like malnutrition and smoking, maternal psychosocial problems are a major factor underlying fetal programming. A common perinatal marker of an adverse intrauterine environment - predisposing newborns to negative health outcomes in later life - is birthweight.1,2 Numerous studies examined the association between maternal psychosocial problems and low birthweight or small-for-gestational-age (SGA), however, evidence is inconclusive.3-7
The psychophysiology underlying the influence of maternal psychosocial problems is based upon the hypothesis that psychosocial stressors dysregulate the hypothalamic-pituitary-adrenal (HPA) axis resulting in hypersecretion of the hormone cortisol. Elevated levels of cortisol, in turn, may reduce fetal growth.2,5,8 Mainly among animals, but also among humans, prepartum treatment with glucocorticoids has been found to reduce offspring birthweight.9-12 Yet only a few small-scale observational studies have examined the association between maternal cortisol levels and fetal growth, and in addition their association with maternal psychosocial problems; their results were inconsistent.13-15 More evidence is needed to support the hypothesis on the psychoendocrinological pathway of maternal psychosocial problems, in order to support the screening and treatment of highly (physiologically) distressed pregnant women. In a large prospective cohort study among pregnant women in Amsterdam, we examined the association of maternal cortisol with offspring birthweight and SGA, thereby applying multivariate analyses aimed at an adequate correction for covariates such as maternal age and parity. Additionally, we tested the mediation effect of maternal cortisol on the relation between maternal psychosocial problems and offspring birthweight.
Methods
The present study is part of the Amsterdam Born Children and their Development (ABCD) study. The ABCD study is a prospective community-based cohort study that examines the relationship between maternal lifestyle and psychosocial conditions during pregnancy and the child’s health at birth as well as in later life. Approval of the study was obtained from the Central Committee on Research involving Human Subjects in the Netherlands, the Medical Ethical Committees of participating hospitals, and from the Registration Committee of Amsterdam.
Study populationBetween January 2003 and March 2004, all pregnant women living in Amsterdam were invited to enroll in the ABCD study at their first visit to participating obstetric care providers (around the 13th week of gestation). Two weeks after this visit a pregnancy-questionnaire, covering sociodemographic data, obstetric history, lifestyle and psychosocial health, was mailed to all approached women
(n = 12 373), to be returned by prepaid mail. The questionnaire was in Dutch, and accompanied by an English, Turkish or Arabic copy depending on the woman’s country of birth. In addition, women were invited to participate in the biomarker study, for which an additional blood sample was taken during routine blood collection at the first visit. Women who gave permission for follow-up received a baby-questionnaire 3 months after delivery, covering lifestyle, emotional problems and the babies’ health.
The pregnancy-questionnaire was filled out by 8266 women (response rate 67%). Of these respondents, 5131 (62%) also completed the baby-questionnaire and 4389 (53%) participated in the biomarker study. After biochemical analyses, 4252 women had a useable cortisol measurement; for 3211 of these women matching data were available on the time of day and gestational age at blood sampling. From this group, 3092 women gave birth to a liveborn singleton for whom data on birthweight and gestational age were available. Exclusion criteria were: preterm birth (gestational age <37.0 weeks) (n = 159), pre-gestational diabetes mellitus (n = 19), steroid medication (n = 11), and gestational age at blood sampling >20.0 weeks (n = 105). Finally, 2810 women were available for analysis.
Outcome variablesPrimary outcome variables were birthweight (in g) and SGA (yes, no) at term. SGA was defined as a birthweight below the 10th percentile for gestational age on the basis of gender- and parity-specific standards from the Dutch Perinatal Registry. Information on infant gender, birthweight, and gestational age [based on ultrasound or, when unavailable (< 10%), on the first day of the last menstrual period] were obtained from the Youth Health Care registration at the Public Health Service in Amsterdam.
Maternal cortisolFor each participant of the biomarker study, one blood sample was taken in a 9-mL Vacuette (Greiner BV, Alphen aan de Rijn, the Netherlands) for the preparation of serum. All samples were sent to the Regional Laboratory of Amsterdam for processing. At the laboratory, 1-mL plasma and serum aliquots were prepared by centrifugation (1600 x g for 10 min at room temperature) and stored at -80°C until analysis [mean: 22 months (range: 13-29)]. Cortisol analysis was performed at the National Institute for Public Health and the Environment, Bilthoven, the Netherlands. Total cortisol in serum was determined by radio-immunoassay and defined in µg/L; this was converted into nmol/L by multiplying the values by 2.759. The interassay coefficient of variation (CV) was 10.2% for low values and 4.9% for high values.
Cortisol secretion is characterized by a diurnal rhythm with peak levels shortly after awakening and a more or less declining pattern thereafter.16 In the present study, time of day at blood sampling was not fixed, but ranged from 08:00 to 19:25h. To adjust for the diurnal variation in cortisol secretion, cortisol concentrations were standardized for the time of day at blood sampling, and additionally, for gestational age at blood sampling17 (calculated with the available information on date of blood sampling, date of birth, and gestational age at birth; ranging from 40 to 140 days). The standardization of cortisol was performed by regressing time of day (categorical scale) and gestational age (categorical scale) at blood sampling to cortisol values (continuous scale). For each participant, we then calculated the estimated cortisol value at 08:00-09:00h and at a gestational age of 40-83 days.
85
________________________________________________ Maternal cortisol and offspring birthweight
6
Psychosocial indicatorsDepressive symptoms were measured by the Center for Epidemiologic Studies Depression Scale (CES-D).18 The 20-item CES-D scale assesses the self-reported frequency of depressive symptoms experienced over the past week. In this study, the validated Dutch version of the CES-D was used (α = 0.90).19
Anxiety was measured by the state-scale of the State-Trait Anxiety Inventory (STAI).20 The validated Dutch version of the STAI by Van der Ploeg21 was used (α = 0.94). State anxiety is conceptualized as a transient emotional condition, characterized by subjectively experienced tension, together with an enhanced activity of the autonomous nervous system. The 20-item state anxiety scale measures the anxiety feelings in the past week.
Pregnancy-related anxiety was measured by three scales (α = 0.81),22 based on the Pregnancy Related Anxiety Questionnaire (PRAQ-R). The three scales were ‘fear of giving birth’ (3 items), ‘fear of bearing a handicapped child’ (4 items), and ‘concern about one’s appearance’ (3 items).
Parenting stress was measured with the Frequency scale of the Parenting Daily Hassles (PDH),23 which was translated in Dutch by Groenendaal and Gerrits24 (α = 0.85). The PDH scale lists 20 typical everyday events in parenting young children, for example ‘Being busy cleaning things up all day’. Women who were responsible for the care of children at home had to rate the frequency of occurrence of each item in the past month.
For the descriptive analyses, the sum scores of the scales measuring depressive symptoms, anxiety, pregnancy-related anxiety and parenting stress (only for primi-/multiparous women) were trichotomised into low (<50th percentile), moderate (50th - 90th percentile) and high (>90th percentile) levels, based on the distribution of the study sample.
Job strain was measured by four scales of the Work Experience and Appreciation Questionnaire (VBBA),25 a validated Dutch questionnaire based on the Job Content Questionnaire of Karasek et al.26 The scales measured workload (25 items; α = 0.85) and job control (11 items; α = 0.92). Job strain was defined as a combination of workload and job control (a detailed description on the categorization of job strain can be found in Vrijkotte et al.7). An additional category was made for unemployed women.
Physical and/or sexual violence was measured retrospectively in the baby-questionnaire with the following questions: ‘Were you beaten, kicked or punched, or otherwise physically abused by anyone during your pregnancy?’, and ‘Did anyone force you to engage in sexual acts during the pregnancy?’. Women who answered yes on at least one of these questions were categorized as having experienced physical and/or sexual violence.
CovariatesInformation on sociodemographic, physiological, psychological and lifestyle characteristics of the women was obtained by the pregnancy-questionnaire. Ethnicity was based on the country of birth of the pregnant woman’s mother. Country of birth included the following categories, based on the Amsterdam main ethnic populations: The Netherlands, Surinam, The Antilles/Aruba, Turkey, Morocco, Ghana, and other non-Dutch countries. The native Dutch group was used as the reference group. Other covariates included maternal age, parity, smoking during pregnancy
(average number of cigarettes per day), and pre-pregnancy BMI. Missing values of maternal weight and height were imputed by means of a random imputation method using linear regression analysis,27 accounting for the differences among the ethnic groups.
Statistical analysisDescriptive statistics were used to plot standardized maternal cortisol levels against offspring birthweight and to calculate the median cortisol levels and interquartile ranges. The Kruskal-Wallis test was performed to test the associations of the covariates and offspring outcomes with maternal cortisol levels. To examine the independent association between maternal cortisol levels and birthweight, linear regression analyses were performed, adjusting for gestational age (linear and quadratic term), infant gender, ethnicity, maternal age (linear and quadratic term), parity (0, ≥1), BMI (linear and quadratic term), and smoking (linear term). The independent association between maternal cortisol levels and SGA was estimated by logistic regression analyses, adjusting for ethnicity, maternal age, BMI, and smoking (the term SGA already takes differences in gestational age, infant gender and parity into account). Based on the descriptive statistics, standardized maternal cortisol levels were included in the analyses in two different ways: (1) continuous and (2) categorical, in which cortisol level was trichotomised around the 50th and 90th percentile into low (reference group), moderate and high cortisol levels. Analyses were repeated stratified for different time-of-day groups (post hoc analysis). To test whether maternal cortisol is a mediator in the association between maternal psychosocial problems and birthweight, according to Baron and Kenny28 the following regression equations should be estimated: (1) the association between the independent variable and the mediator, (2) the association between the independent variable and the outcome variable, (3) the association between the mediator and the outcome variable, controlling for the effect of the independent variable. Data were analyzed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). In all analyses a p-value < 0.05 was considered significant.
Results
The women in the present study had a mean age of 31.4 years; 57.5% was nulliparae and 65.6% was of Dutch origin. Mean birthweight in this sample was 3526.1 g, mean gestational age 39.7 weeks, and 10.9% of the newborns were categorized as SGA. Standardized maternal cortisol levels ranged from 184.88 to 977.27 nmol/L. On average, women with higher age, parity and BMI had significantly lower maternal cortisol levels (Table 6.1). Furthermore, a higher maternal cortisol level was observed among mothers whose infant had a lower birthweight, was SGA or was a girl.
87
________________________________________________ Maternal cortisol and offspring birthweight
6
Figure 6.1 Observed values in the study sample (n = 2810): mean birthweight (g) and 95% confidence interval as a function of standardized maternal cortisol levels (categorized into deciles for presentation purposes).
Table 6.1 Sample characteristics and their association with maternal cortisol level (n = 2810).
a Median [interquartile range (IQR)] cortisol level in nmol/L, standardized for time of day and gestational age at blood sampling.
b Kruskal-Wallis test.c Data on psychosocial problems were not available for all 2810 women included in the regression analyses.d Only primi-/multiparous women.
Table 6.1 Continued
89
________________________________________________ Maternal cortisol and offspring birthweight
6
Maternal cortisol and offspring birthweight Figure 6.1 shows the raw association between standardized maternal cortisol levels (categorized into deciles) and mean birthweight. Overall, birthweight decreased with increasing cortisol levels, especially among the highest cortisol levels. Univariate linear regression results for the whole sample (n = 2810) showed that standardized maternal cortisol level was negatively associated with birthweight (b = -0.35; p < 0.001) (Table 6.2). When dividing cortisol level into three categories, both moderate (b = -44.48; p = 0.022) and high (b = -94.50; p = 0.003) cortisol levels were significantly associated with lower birthweights compared with the category of low cortisol level. After adjustment for the covariates (gestational age, infant gender, ethnicity, maternal age, parity, BMI, and smoking), cortisol level (continuous as well as categorical) was not significantly associated with birthweight. The logistic regression results for the association between cortisol level and SGA were quite similar to the birthweight results. Only univariately, an increasing cortisol level was associated with an increasing risk for SGA (OR = 1.00; p = 0.027); regression results for moderate (OR = 1.16; p = 0.248) and high (OR = 1.37; p = 0.111) vs. low cortisol level showed increasing but insignificant odds ratios.
Post hoc analysis: moderation effect by time of dayBased on the dynamic pattern of the cortisol diurnal rhythm,15,29 post hoc analysis was performed to explore whether time of day was a moderator in the association between maternal cortisol levels and fetal growth. Regression analysis was stratified by the following time groups: 08:00-09:00h (T1, n = 94), 09:01-12:00h (T2, n = 1564), 12:01-16:00h (T3, n = 970), and 16:01-19:25h (T4, n = 182). Multiple linear regression results for the - for covariates adjusted - association between maternal cortisol levels (continuous scale) and offspring birthweight were as follows: T1 [b (95% CI)] = -0.94 (-1.75, -0.12), T2 = -0.01 (-0.21, 0.19), T3 = -0.07 (-0.37, 0.24), T4 = -0.20 (-0.92, 0.52). These results revealed only a significant association among the group who provided a blood sample ≤09:00h. For this subsample, further analyses are presented in Table 6.2. Compared with the whole sample (n = 2810), maternal cortisol levels in the subsample were more strongly related to birthweight, univariately (b = -1.05; p = 0.012) as well as after adjustment for the covariates (b = -0.94; p = 0.025). In the categorical model, only the offspring of women with the highest cortisol levels had significantly lower birthweights compared with the offspring of women with the lowest cortisol levels (adjusted model: b = -373.64; p = 0.031). In the regression model for SGA, cortisol levels were associated with SGA, univariately (OR = 1.01; p = 0.031) as well as multivariately (OR = 1.01; p = 0.032). Again, only the category with the highest cortisol levels had a significantly higher number of SGA babies compared with the category with the lowest cortisol levels (adjusted model: OR = 13.48; p = 0.016). Additional Chi-square tests showed that the sample characteristics of the whole sample (for the purpose of analysis: minus the subsample) did not differ from the characteristics of the subsample (results not shown). It should be mentioned that stratification could impact the interpretation of p-values due to the issue of multiple testing. A possible correction method is the conservative Bonferroni correction: dividing the p-value by the number of tests performed. Applying this method to the stratified analysis leads to a corrected
p-value of 0.05/4=0.0125. Subsample results are then insignificant. In fact, the subgroup analysis was hypothesis- rather than data-driven and one could argue that no correction for multiple testing is required.30 Nevertheless, the stratified analysis results should be interpreted with caution.
Mediation analysisWomen with high levels of depressive symptoms and pregnancy-related anxiety or with exposure to physical/sexual violence had slightly higher median cortisol levels compared with women with low levels of these psychosocial problems; the differences were however not statistically significant (Table 6.1). Across the various levels of the anxiety, parenting stress and job strain scales, the median cortisol level neither showed a consistent increasing nor decreasing trend. Because the first condition for establishing a mediation effect (association between independent variable and mediator) did not hold, there was no need to estimate the other regression equations necessary for testing the mediation effect. Among the subsample of women who provided a blood sample ≤09:00h, mediation analysis also stagnated after testing the first condition (results not shown).
Table 6.2 Regression results on the association between maternal cortisol level and birthweight and small-for-gestational-
a Regression coefficient (b), adjusted for gestational age, infant gender, ethnicity, maternal age, parity, BMI, smoking.b Odds ratio (OR), adjusted for ethnicity, maternal age, BMI, smoking.*p < 0.05.
91
________________________________________________ Maternal cortisol and offspring birthweight
6
Discussion
The present study showed that maternal cortisol levels in early pregnancy are negatively associated with offspring birthweight and positively with the risk for SGA, however, the associations were statistically insignificant after controlling for relevant covariates. Additionally, we did not observe a mediation effect of maternal cortisol on the association between maternal psychosocial problems and offspring birthweight. Our results are not in line with the previous small-scale studies which observed an association between maternal cortisol levels and fetal growth,13 estimated fetal weight14,31,32 or birthweight.14,15 The inconsistent findings can, at first, be ascribed to the variety in the time of day maternal cortisol was collected.17 Exploratory post hoc analysis suggested a moderation effect by time of day: only among a subsample of women who provided a blood sample ≤09:00h, a significant independent effect of maternal cortisol levels on offspring birthweight and SGA risk was found. This time-of-day effect is in line with the only previous study that collected multiple cortisol measurements throughout the day; they observed an association of both morning cortisol levels at 08:00h and the morning cortisol decline with offspring birthweight, in contrast to the afternoon cortisol decline.15 It is hypothesized that the circadian cortisol rhythm reflects various HPA-axis functioning with especially the cortisol awakening response (CAR; the period of cortisol secretion in the first hour immediately after awakening) being a good predictor for maternal and fetal pathophysiology.16,33 So far, evidence for this hypothesis is only provided by animal studies which observed that the greatest maternal-fetal glucocorticoid transfer occurs at the peak of the circadian rhythm due to saturation of the 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2) enzyme.34 A second explanation for the inconsistent observations between our study and previous studies can be ascribed to the influence of maternal and environmental covariates. Previous studies did not or incompletely adjust for relevant covariates. The confounding effect of the covariates may also be moderated by time of day, in that the association between maternal cortisol levels and offspring birthweight was less disturbed by the covariates among the early morning subsample. Remarkably, this again resembles characteristics of the CAR, which appears to be influenced by genetic factors rather than factors like age, BMI and exercise.16 It can be speculated that maternal and environmental factors affect fetal growth through rather than outside the effect of maternal cortisol. In that case, the association between maternal cortisol levels and fetal growth would have been overadjusted. Third, maternal cortisol was collected at various time points during gestation. While this study assessed maternal cortisol in the late first or early second trimester of pregnancy, previous studies assessed cortisol in the second13,14 or third trimester.15 The time window of the influence of maternal cortisol on fetal growth is, however, unknown. While it is suggested that early gestation may be a particular sensitive period for maternal-placental-fetal pathophysiology,35 late pregnancy is a critical period as it is known for the accelerated fetal somatic growth.15
A final issue concerns the assessment of total vs. free cortisol levels. While it is hypothesized that free or unbound cortisol is the only biologically active part, it has not been ruled out that the bound part also still has physiological effects.36 Furthermore, the use of free rather than total cortisol levels in examining its physiological effects would only be preferable among certain pathological cases with a different ratio of bound vs. unbound cortisol levels. It might be possible that psychosocial problems are stronger associated with the unbound cortisol fraction, because previous studies suggested that stress decreases the corticotrophin-binding globulin (CBG) levels, resulting in increased levels of unbound cortisol.37 While it is strongly suggested that maternal cortisol mediates the association between maternal psychosocial problems and offspring birthweight, evidence is very limited. Diego et al.13 presented a significant mediation model, while Kivlighan et al.15 did not. Besides the lack of an association between maternal cortisol and offspring birthweight, another assumption of the mediation pathway, i.e. the association between maternal psychosocial problems and maternal cortisol level, was also not met in the present study. The link between psychosocial factors and HPA-axis activation is a complex interplay of psychobiological events, especially during pregnancy.8,38 Therefore it is not surprising that evidence is inconsistent. Previous research suggests that the strength and direction of the association may be modulated by (1) the nature of the stressor: chronic vs. episodic, mild vs. severe, positive vs. negative affect,8,33,39-42 (2) the level of exhaustion of the HPA-axis activity,33,39 (3) the subjective perception of the stressor and coping abilities,43,44 (4) gestational age: early vs. late pregnancy,8,45 and (5) the time of day: morning vs. evening cortisol levels.15,42,45,46 The mechanisms by which maternal cortisol levels may influence fetal growth have yet to be determined. The main potential pathway is based on the ability of maternal cortisol to directly cross the placenta. Normally, the placental barrier (including 11β-HSD2 and P-glycoprotein) limits fetal exposure to maternal cortisol.47 However, in the case of elevated maternal cortisol levels and/or placental 11β-HSD2 deficiency, increasing levels of biologically active maternal cortisol cross the placenta resulting in elevated fetal cortisol levels.2 This mechanism is supported by the observation that maternal cortisol levels are highly correlated to fetal cortisol levels.48 It is hypothesized that fetal exposure to excess cortisol levels may restrict fetal growth by dysregulating fetal autonomic nervous system activity and by mobilizing fetal energy stores via glycogenolysis resulting in a high degree of calorie expenditure.13 Another potential pathway suggests that cortisol, in cooperation with norepinephrine, induces uterine artery vasoconstriction resulting in reduced uterine blood flow. The restriction of oxygen and nutrient delivery to the fetus subsequently affects fetal growth and development.35
Strengths of the study include a large community-based sample within a prospective study design. Women who participated in the biomarker study were, however, older, taller, higher educated, more often nulliparous, more often of Dutch origin, and experienced less psychosocial problems compared with the women who did not provide a blood sample (n = 3564; data not shown). Their infants had also higher birthweights and were less often born SGA. To our knowledge, the present study is the largest study measuring maternal cortisol,
93
________________________________________________ Maternal cortisol and offspring birthweight
6
maternal characteristics (including psychosocial problems), as well as pregnancy outcomes. Large studies, however, often go with limited possibilities for large-scale expensive and time-consuming measurements. For this reason, our measurement of cortisol leaves much to be desired. Considering the diurnal rhythm of cortisol secretion, multiple measurements of cortisol during the day, possibly over several days, would be preferable over a single cortisol measurement.49 Ideally, cortisol values should be controlled for the effect of behavioral factors such as smoking, nutrition or activity measures.50,51 In conclusion, the results of this large-scale study could not support the hypothesis that maternal psychosocial problems affect fetal growth through maternal cortisol levels, by lack of a distinct association between cortisol and birthweight and between cortisol and psychosocial problems. As exploratory post hoc analysis suggested a moderation effect by time of day, further research on the adverse effects of disturbed maternal cortisol levels on offspring health should include multiple cortisol measurements at different time frames of the circadian cortisol rhythm.
1. Barker DJP, Eriksson JG, Forsen T, Osmond C. Fetal origins of adult disease: strength of effects and bio-logical basis. Int J Epidemiol 2002;31:1235-9.
2. Seckl JR, Holmes MC. Mechanisms of disease: glucocorticoids, their placental metabolism and fetal ‘pro-gramming’ of adult pathophysiology. Nat Clin Pract Endocrinol Metab 2007;3:479-88.
3. Alder J, Fink N, Bitzer J, Hosli I, Holzgreve W. Depression and anxiety during pregnancy: a risk factor for obstetric, fetal and neonatal outcome? A critical review of the literature. J Matern Fetal Neonatal Med 2007;20:189-209.
4. Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in term birthweight: the role of constitutional and environmental factors. Paediatr Perinat Epidemiol 2008;22:360-8.
5. Hobel CJ, Goldstein A, Barrett ES. Psychosocial stress and pregnancy outcome. Clin Obstet Gynecol 2008;51:333-48.
6. Murphy CC, Schei B, Myhr TL, Du Mont J. Abuse: a risk factor for low birth weight? A systematic review and meta-analysis. CMAJ 2001;164:1567-72.
7. Vrijkotte TGM, van der Wal MF, van Eijsden M, Bonsel GJ. First-trimester working conditions and birth-weight: a prospective cohort study. Am J Public Health 2009;99:1409-16.
8. Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CA. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosom Med 1996;58:432-46.
10. French NP, Hagan R, Evans SF, Godfrey M, Newnham JP. Repeated antenatal corticosteroids: size at birth and subsequent development. Am J Obstet Gynecol 1999;180:114-21.
11. Newnham JP, Moss TJ. Antenatal glucocorticoids and growth: single versus multiple doses in animal and human studies. Semin Neonatol 2001;6:285-92.
12. Seckl JR. Prenatal glucocorticoids and long-term programming. Eur J Endocrinol 2004;151(Suppl 3):U49-62.
13. Diego MA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, Gonzalez-Quintero VH. Prenatal depres-sion restricts fetal growth. Early Hum Dev 2009;85:65-70.
14. Field T, Hernandez-Reif M, Diego M, Figueiredo B, Schanberg S, Kuhn C. Prenatal cortisol, prematurity and low birthweight. Infant Behav Dev 2006;29:268-75.
15. Kivlighan KT, DiPietro JA, Costigan KA, Laudenslager ML. Diurnal rhythm of cortisol during late preg-nancy: associations with maternal psychological well-being and fetal growth. Psychoneuroendocrinology 2008;33:1225-35.
16. Clow A, Thorn L, Evans P, Hucklebridge F. The awakening cortisol response: methodological issues and significance. Stress 2004;7:29-37.
17. Allolio B, Hoffmann J, Linton EA, Winkelmann W, Kusche M, Schulte HM. Diurnal salivary cortisol pat-terns during pregnancy and after delivery: relationship to plasma corticotrophin-releasing-hormone. Clin Endocrinol (Oxf.) 1990;33:279-89.
18. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401.
19. Bouma J, Ranchor AV, Sanderman R, van Sonderen E. The measurement of depressive symptoms with the CES-D: a manual. [In Dutch]. Groningen: Noordelijk Centrum voor Gezondheidsvraagstukken, Rijks-universiteit Groningen; 1995.
21. Van der Ploeg HM. Manual for the Self Evaluation Questionnaire. [In Dutch]. Lisse: Swets & Zeitlinger; 2000.
22. Huizink AC, Mulder EJH, Robles de Medina PG, Visser GHA, Buitelaar JK. Is pregnancy anxiety a dis-tinctive syndrome? Early Hum Dev 2004;79:81-91.
95
________________________________________________ Maternal cortisol and offspring birthweight
6
23. Crnic KA, Greenberg MT. Minor parenting stresses with young children. Child Dev 1990;61:1628-37.24. Groenendaal JHA, Gerrits LAW. Daily Worries List. [In Dutch]. Utrecht: Universiteit Utrecht; 1996.25. Van Velhoven M, Meijman TF, Broersen JPJ, Fortuin RJ. Manual VBBA. [In Dutch]. Amsterdam: SKB
Vragenlijst Services; 2002.26. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire
(JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322-55.
27. Allison PD. Multiple imputation: basics. In: Laughton CD, Carr E, Santoyo D, editors. Missing data. Sage university papers series on quantitative applications in the social sciences. Thousand Oaks, CA: Sage Pub-lications; 2001, pp. 27-41.
28. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: con-ceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173-82.
29. Weitzman ED, Fukushima D, Nogeire C, Roffwarg H, Gallagher TF, Hellman L. Twenty-four hour pattern of the episodic secretion of cortisol in normal subjects. J Clin Endocr 1971;33:14-22.
30. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ 1998;316:1236-8.31. Diego MA, Jones NA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, et al. Maternal psychological
distress, prenatal cortisol, and fetal weight. Psychosom Med 2006;68:747-53.32. Field T, Diego M, Hernandez-Reif M, Gil K, Vera Y. Prenatal maternal cortisol, fetal activity and growth.
Int J Neurosci 2005;115:423-9.33. Chida Y, Steptoe A. Cortisol awakening response and psychosocial factors: a systematic review and meta-
analysis. Biol Psychol 2009;80:265-78.34. Venihaki M, Carrigan A, Dikkes P, Majzoub JA. Circadian rise in maternal glucocorticoid prevents pul-
monary dysplasia in fetal mice with adrenal insufficiency. Proc Natl Acad Sci USA 2000;97:7336-41.35. Field T, Diego M. Cortisol: the culprit prenatal stress variable. Int J Neurosci 2008;118:1181-1205.36. Levine A, Zagoory-Sharon O, Feldman R, Lewis JG, Weller A. Measuring cortisol in human psychobiolo-
gical studies. Physiol Behav 2007;90:43-53.37. Fleshner M, Deak T, Spencer RL, Laudenslager ML, Watkins LR, Maier SF. A long-term increase in basal
levels of corticosterone and a decrease in corticosteroid-binding globulin after acute stressor exposure. Endocrinology 1995;136:5336-42.
38. Hellhammer DH, Wust S, Kudielka BM. Salivary cortisol as a biomarker in stress research. Psychoneuro-endocrinology 2009;34:163-71.
39. Pruessner JC, Hellhammer DH, Kirschbaum C. Burnout, perceived stress, and cortisol responses to awak-ening. Psychosom Med 1999;61:197-204.
40. Pruessner M, Hellhammer DH, Pruessner JC, Lupien SJ. Self-reported depressive symptoms and stress levels in healthy young men: associations with the cortisol response to awakening. Psychosom Med 2003;65:92-9.
41. Vreeburg SA, Hoogendijk WJG, van Pelt J, DeRijk RH, Verhagen JCM, van Dyck R, et al. Major depressive disorder and hypothalamic-pituitary-adrenal axis activity. Arch Gen Psychiatry 2009;66:617-26.
42. Wüst S, Federenko I, Hellhammer DH, Kirschbaum C. Genetic factors, perceived chronic stress, and the free cortisol response to awakening. Psychoneuroendocrinology 2000;25:707-20.
43. Biondi M, Picardi A. Psychological stress and neuroendocrine function in humans: the last two decades of research. Psychother Psychosom 1999;68:114-50.
44. Dickerson SS, Kemeny ME. Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychol Bull 2004;130:355-91.
45. Obel C, Hedegaard M, Henriksen TB, Secher NJ, Olsen J, Levine S. Stress and salivary cortisol during pregnancy. Psychoneuroendocrinology 2005;30:647-56.
46. Vedhara K, Miles J, Bennett P, Plummer S, Tallon D, Brooks E, et al. An investigation into the relationship between salivary cortisol, stress, anxiety and depression. Biol Psychol 2003;62:89-96.
47. Mark PJ, Waddell BJ. P-glycoprotein restricts access of cortisol and dexamethasone to the glucocorticoid receptor in placental BeWo cells. Endocrinology 2006;147:5147-52.
48. Glover V, Teixeira J, Gitau R, Fisk NM. Mechanisms by which maternal mood in pregnancy may affect the fetus. Contemp Rev Obstet Gynecol 1999;155-60.
49. Harville EW, Savitz DA, Dole N, Herring AH, Thorp JM, Light KC. Patterns of salivary cortisol secretion in pregnancy and implications for assessment protocols. Biol Psychol 2007;74:85-91.
50. Steptoe A, Ussher M. Smoking, cortisol and nicotine. Int J Psychophysiol 2006;59:228-35.51. Ukkola O, Gagnon J, Rankinen T, Thompson PA, Hong Y, Leon AS, et al. Age, body mass index, race
and other determinants of steroid hormone variability: the HERITAGE Family Study. Eur J Endocrinol 2001;145:1-9.
Geertje GoedhartMarcel F. van der Wal
Manon van EijsdenGouke J. Bonsel
Chapter 7Maternal vitamin B-12 and folate status during pregnancy and excessive infant crying.
Background The etiology of excessive infant crying is largely unknown. We hypothesize that ex-cessive infant crying may have an early nutritional origin during fetal development.
Aims This study is the first to explore whether (1) maternal vitamin B-12 and folate status during pregnancy are associated with excessive infant crying, and (2) whether and how maternal psycho-logical well-being during pregnancy affects these associations.
Study design Women were approached around the 12th pregnancy week to complete a question-naire (n = 8266) and to donate a blood sample (n = 4389); vitamin B-12 and folate concentrations were determined in serum. Infant crying behavior was measured through a postpartum question-naire (±3 months; n = 5218).
Subjects Pregnant women living in Amsterdam and their newborn child.
Outcome measures Excessive infant crying, defined as crying ≥3 hours/day on average in the past week.
Results Multiple logistic regression analysis was performed for 2921 (vitamin B-12) and 2622 (folate) women.Vitamin B-12 concentration (categorized into quintiles) was associated with exces-sive infant crying after adjustment for maternal age, parity, ethnicity, education, maternal smoking and psychological problems [OR (95%CI): Q1 = 3.31 (1.48-7.41); Q2 = 2.50 (1.08-5.77); Q3 = 2.59 (1.12-6.00); Q4 = 2.77 (1.20-6.40); Q5 = reference]. Stratified analysis suggested a stronger association among women with high levels of psychological problems during pregnancy. Folate concentration was not associated with excessive infant crying.
Conclusions First evidence is provided for an early nutritional origin in excessive infant crying. A low maternal vitamin B-12 status during pregnancy could, in theory, affect infant crying behavior through two potential mechanisms: the methionine-homocysteine metabolism and/or the matu-ration of the sleep-wake rhythm.
99
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
Introduction
Excessive infant crying during the first months of life is a common and serious problem. Parents often experience a lot of stress and concern, which may result in poor parent-child interactions or even child abuse.1-3 These children may in the long term experience behavioral, psychological and cognitive dysfunctioning, like hyperactivity and discipline problems.4-6 Many theories exist on the origins of excessive infant crying, yet there is still no general agreement. It is suggested to be a multi-causal condition that may reflect either gastrointestinal disorders, dysregulation of the central nervous system, endocrine dysregulation through the hypothalamic-pituitary-adrenal (HPA) axis, poor parent-child interactions, infant’s personality characteristics, or just an extreme clinical manifestation of normal emotional development.3,7,8 As the proposed explana-tions vary widely in their amenability for prevention and treatment,9 it is urgent to expand the knowledge on the etiology of excessive infant crying in order to improve infant health.
Potential prenatal risk factors for excessive infant crying are maternal psychosocial stress and heavy smoking behavior during pregnancy.10-15 So far, another prenatal behavioral risk fac-tor – maternal nutritional status – has not been explored in relation to infant crying behavior. Recently, the effects of maternal vitamin B-12 (cobalamin) and folate deficiency on fetal devel-opment have received attention.16 We can think of two potential mechanisms through which maternal vitamin B-12 and/or folate deficiency – other than through mediation by maternal psychological problems17 – may affect infant crying behavior. First, vitamin B-12 and folate are involved in the methionine-homocysteine metabolism, which is essential in fetal neurodevelop-ment.18,19 Second, vitamin B-12 is involved in melatonin synthesis and the sleep-wake rhythm;20,21 previous studies suggested that excessive infant crying could be regarded as a symptom of an immature sleep-wake rhythm.22,23 In addition, we hypothesize that this could be amplified by an immature cortisol rhythm,8 which in theory may originate from maternal psychopathology during pregnancy.24 As part of a large prospective cohort study among pregnant women in Am-sterdam, we were the first to explore whether maternal vitamin B-12 and folate status during pregnancy are associated with excessive infant crying, and in addition, whether maternal psy-chological problems during pregnancy mediate or modify these associations.
Methods
The present study is part of the Amsterdam Born Children and their Development (ABCD) study. The ABCD study is a prospective community-based cohort study that examines the relationship between maternal lifestyle and psychosocial conditions during pregnancy and the child’s health at birth as well as in later life. Approval of the study was obtained from the Central Committee on Research involving Human Subjects in the Netherlands, the Medical Ethical Committees of participating hospitals, and from the Registration Committee of Amsterdam.
Study population and designBetween January 2003 and March 2004, all pregnant women in Amsterdam were invited to enroll in the ABCD study at their first prenatal visit to participating obstetric care providers (median 13 pregnancy weeks, interquartile range 3 weeks). Two weeks after this visit a pregnancy-questionnaire, covering sociodemographic data, obstetric history, lifestyle and (psychosocial) health, was mailed to all approached women (n = 12 373), to be returned by prepaid mail. The questionnaire was in Dutch, and accompanied by an English, Turkish or Arabic copy depending on the woman’s country of birth. In addition, women were invited to participate in the biomarker study, for which an additional blood sample was taken during routine blood collection at the first prenatal visit. Women who gave permission for follow-up received a baby-questionnaire after delivery (median 13 weeks postpartum, interquartile range 1 week), covering lifestyle, maternal (psychosocial) health and infant’s health. The pregnancy-questionnaire was filled out by 8266 women (response rate 67%) and 4389 women also participated in the biomarker study. A total of 5218 women completed the baby-questionnaire. For this study, multiple births (n = 87) were excluded.
Excessive infant cryingThe amount of infant crying was measured in the baby-questionnaire with the question: ‘How many hours per day (24 hours) on average did your baby cry in the past week?’ Excessive crying was defined as crying on average ≥3 hours per day in the past week.
Maternal vitamin B-12 and folate concentrationsMaternal concentrations of vitamin B-12 and folate were determined in blood samples collected at the first prenatal visit. All samples were sent to the Regional Laboratory of Amsterdam for processing, either by courier or by overnight mail in special envelopes. This delay was shown not to compromise the validity of specific biomarkers.25 At the laboratory, plasma and serum were prepared by centrifugation and stored as 1-mL aliquots at -80°C until analysis. Vitamin B-12 analysis was performed at the National Institute for Public Health and the Environment, Bilthoven, the Netherlands. Vitamin B-12 concentration was determined by immunoassay with chemiluminescence detection on the Access Immunoassay System (Beckman Coulter), with an inter-assay coefficient of variation (CV) ≤7.8% and an upper detection limit of 1500 pg/mL. Two measurement were above this limit and therefore excluded, leaving 4255 usable vitamin B-12 measurements. Folate analysis was performed at the Medical Laboratory Dr. Stein & Collegae, Maastricht, the Netherlands, by immunoassay with chemiluminescence detection on the ADVIA Centaur System (Bayer Group). Inter-assay CV was ≤6.1%. One measurement was below lower detection limit (0.8 nmol/L) and 356 measurements (8%) were above upper detection limit (54.4 nmol/L); these measurements were excluded, leaving 3938 usable folate measurements. Because the concentrations of vitamin B-12 and folate decrease during pregnancy,26 we standardized the vitamin B-12 and folate concentrations for gestational age at blood sampling by regressing gestational age (categorical scale) both to vitamin B-12 and folate concentrations (continuous scales). For each participant, we then calculated the estimated vitamin
101
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
B-12 and folate concentrations at a gestational age of 30-73 days. The standardized vitamin B-12 and folate concentrations were categorized into quintiles (Q1-Q5).
Maternal psychological problemsMaternal psychological problems were measured in the pregnancy-questionnaire and included depressive symptoms, anxiety and pregnancy-related anxiety. Depressive symptoms were measured by the validated Dutch version of the Center for Epidemiologic Studies Depression Scale (CES-D).27 The 20-item CES-D scale (α = 0.90) assesses the self-reported frequency of depressive symptoms experienced over the past week. Anxiety was measured by the validated Dutch version of the state-scale of the State-Trait Anxiety Inventory (STAI).28 The 20-item state anxiety scale (α = 0.94) measures the anxiety feelings in the past week. Pregnancy-related anxiety was measured by three scales (α = 0.81), based on the Pregnancy Related Anxiety Questionnaire – Revised version (PRAQ-R).29 The three scales were ‘fear of giving birth’ (3 items), ‘fear of bearing a handicapped child’ (4 items), and ‘concern about one’s appearance’ (3 items). For the descriptive analyses, the sum scores of the three scales were trichotomised into low (<50th percentile), moderate (50th - 90th percentile) and high (>90th percentile) levels, based on the distribution of the study sample.
CovariatesPotential covariates were included in the analysis. Ethnicity was based on the country of birth of the pregnant woman’s mother. Country of birth included the following categories, based on the Amsterdam main ethnic populations: The Netherlands (reference group), Surinam, Turkey, Morocco, and other non-Dutch countries. Other covariates included maternal age, parity (0, ≥1), education (years of education after primary school), and maternal smoking during pregnancy (average number of cigarettes per day).
Statistical analysisMatching data on maternal vitamin B-12 and folate concentrations and infant crying behavior was available for respectively 3007 and 2702 women. Twenty two women had missing data on gestational age at blood sampling and 64 women had missing data on one of the covariates. The final sample for analysis consisted of 2921 (vitamin B-12) and 2622 (folate) women.
Descriptive statistics were performed to calculate and test the prevalence of excessive infant crying (tested with Chi-square analysis) and the median and interquartile ranges of vitamin B-12 and folate concentrations (tested with the Kruskal-Wallis test) according to the levels of the covariates and maternal psychological problems. To examine the independent associations of vitamin B-12 and folate concentration with excessive infant crying, univariate and multiple logistic regression analyses were performed, adjusting for the covariates maternal age, parity, ethnicity, education and smoking (Model 1). The variables vitamin B-12, folate, parity and ethnicity were entered as categorical variables in the analysis, whereas the other variables were included on a continuous scale.
To examine whether the presence of maternal psychological problems is a mediator in the associations, the following conditions must hold: (1) the independent variable must affect the mediator, (2) the independent variable must affect the outcome variable, (3) the mediator must affect the outcome
variable, (4) the effect of the independent variable on the outcome variable must be less when controlling for the effect of the mediator.30 Condition 1 and 3 were tested with descriptive statistics. Condition 2 and 4 were tested with logistic regression analysis; the scales depressive symptoms and pregnancy-related anxiety were entered to Model 2. Because of high collinearity between the scales depressive symptoms and anxiety (r = 0.89), anxiety was not included in the model. To examine whether the presence of maternal psychological problems is a moderator, first, an interaction between the independent variable and the moderator should be tested.30 Second, the effects of the independent variable across levels of the moderator could be measured. Therefore, we stratified the multiple logistic regression analysis by two groups: women who reported high levels (>90th percentile) of depressive symptoms, anxiety and/or pregnancy-related anxiety during pregnancy versus women who did not. For this stratified analysis, data were available for 2911 women. Data were analyzed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). In all analyses a p-value < 0.05 was considered significant.
Results
The prevalence of excessive infant crying in the study sample was 3.4%; the prevalence was significantly higher among young women (<25 year), women with a non-Dutch background, low-educated women (<6 years of education after primary school), and among women with high levels (>90th percentile) of psychological problems (Table 7.1). The median vitamin B-12 and folate concentrations significantly differed across the categories of the covariates. The prevalence of excessive infant crying increased with decreasing vitamin B-12 and folate concentrations (Table 7.2). Univariate logistic regression analysis showed that, compared to infants born to women in the highest vitamin B-12 category (Q5), infants born to women in the lower quintiles (Q1-Q4) had a significantly higher risk for excessive crying behavior, with OR’s ranging from 2.56 to 3.91. This effect remained after adjustment for the covariates maternal age, parity, ethnicity, education and maternal smoking (Model 1). The univariate regression results for folate showed that only infants born to mothers in the lowest quintile had a significantly higher risk for excessive crying behavior compared with infants born to mothers in the highest quintile (OR = 2.66). This effect disappeared after adjustment. The first three conditions for a possible mediation effect of maternal psychological problems were hold (Table 7.1 and 7.2), however the 4th condition was violated: additional inclusion of the scales depressive symptoms and pregnancy-related anxiety to the multiple regression model hardly changed the association between vitamin B-12 status and excessive infant crying (Model 2, Table 7.2). The interaction between vitamin B-12 status and the presence of maternal psychological problems during pregnancy was insignificant (p = 0.265). Stratified multiple regression analysis revealed that the association between vitamin B-12 and excessive infant crying was much stronger among the group of women who reported high levels of psychological problems during pregnancy (OR’s ranging from 4.27 to 12.21) compared with the group of women who did not (OR’s ranging from 1.53 to 2.47), however, confidence intervals were large (Table 7.3).
103
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
Table 7.1 Sample characteristics according to the prevalence of excessive infant crying and the vitamin B-12 (n = 2921)
and folate (n = 2622) concentrations.
nExcessive crying (%)
Vitamin B-12 Folate
Median (IQR)a n Median (IQR)b
Maternal age (years)
<25 235 6.8* 291.0 (130)† 230 20.9 (12)†
25-34 1981 3.5 320.7 (135) 1774 28.7 (17)
>34 705 2.0 344.4 (143) 618 30.0 (17)
Parity
0 1721 2.8 319.0 (135)† 1541 29.1 (17)†
≥1 1200 4.2 332.2 (145) 1081 26.6 (16)
Ethnicity
Dutch 1973 2.1* 325.5 (135)† 1759 30.0 (16)†
Turkish 83 7.2 245.5 (112) 81 22.5 (12)
Moroccan 105 10.5 321.6 (138) 102 19.6 (13)
Surinamese 162 7.4 308.0 (138) 156 22.1 (12)
Other 598 4.8 328.1 (152) 524 27.9 (17)
Education (years)
<6 371 8.9* 300.1 (164)† 345 21.8 (15)†
6-10 1100 3.6 322.3 (141) 988 27.0 (16)
>10 1450 1.8 330.7 (131) 1289 30.7 (16)
Smoking (cigarettes/day)
<1 2739 3.2 325.3 (137)† 2451 28.4 (17)†
≥1 182 5.5 293.2 (142) 171 22.7 (17)
Depressive symptoms
Low 1455 2.3* 325.6 (139)† 1297 29.3 (16)†
Moderate 1166 3.7 323.4 (138) 1048 27.4 (18)
High 300 7.7 314.6 (130) 276 25.6 (15)
Anxiety
Low 1379 1.9* 330.8 (138)† 1228 29.3 (16)†
Moderate 1222 3.9 318.7 (140) 1099 27.5 (17)
High 310 7.4 315.3 (146) 284 25.8 (16)
Pregnancy-related anxiety
Low 1417 2.7 330.1 (140)† 1279 28.2 (17)†
Moderate 1150 3.7 320.8 (139) 1026 28.6 (17)
High 354 5.1 308.7 (131) 316 26.4 (17)a Median and interquartile range (IQR) of vitamin B-12 concentration (pg/mL), standardized for gestational age at blood sampling.
b Median and interquartile range (IQR) of folate concentration (nmol/L), standardized for gestational age at blood sampling.
Q=Quintile of total group.a Vitamin B-12 concentration (pg/mL) and folate concentration (nmol/L), standardized for gestational age at blood sampling.b Odds ratio (OR) and 95% confidence interval (95% CI), adjusted for maternal age, parity, ethnicity, education, smoking.c Additionally adjusted for depressive symptoms and pregnancy-related anxiety (anxiety was not included, because of high collinearity with depressive symptoms).
Table 7.3 Association between maternal vitamin B-12 concentration during pregnancy and excessive infant crying, stratified by the presence of maternal psychological problems during pregnancy (n = 2911).
Excessive infant crying
Q n % Univariate modelOR (95% CI)
Model 1a
OR (95% CI)
No psychological problemsb Q1 427 3.5 2.45 (0.99-6.05) 2.12 (0.84-5.37)
Q=Quintile of total group.a Odds ratio (OR) and 95% confidence interval (95% CI), adjusted for maternal age, parity, ethnicity, education, smoking.b Presence of psychological problems is defined as the presence of high depressive symptoms, high anxiety and/or high pregnancy-related anxiety during pregnancy.
105
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
Discussion
This prospective cohort study is the first to explore the potential association between maternal nutritional status during pregnancy, in particular vitamin B-12 and folate status, and infant crying behavior. The results showed that maternal vitamin B-12 status during pregnancy, but not maternal folate status, is related to excessive infant crying in the first months after birth. Additional analyses suggested that this relation was not mediated, but moderated by the presence of maternal psychological problems during pregnancy, i.e. infants born to mothers with both a low vitamin B-12 concentration and psychological problems during pregnancy appeared to have a very high risk for excessive crying behavior.
Two potential mechanisms may explain the association between maternal vitamin B-12 status and excessive infant crying. First, maternal vitamin B-12 and folate deficiencies affect, among others, fetal neurodevelopment. Although the biological mechanism remains in question, a commonly used theory refers to the involvement of vitamin B-12 and folate in the methionine-homocysteine metabolism.18,19 This metabolism is essential for the myelination-process of the brain: myelin is a supportive tissue that surrounds and protects the nerve cells and facilitates communication between the cells. Myelination of the brain is most active in the first 6 months of life. Disruptions in myelination can have significant effects on the central nervous system functioning. Infants born to vitamin B-12 and folate deficient mothers often display developmental delay and neurological and physical symptoms like irritability and failure-to-thrive.16,18,19,31 It is not unthinkable that the prevalence of excessive infant crying is also higher among those infants. Nevertheless, the observation that maternal folate status was not associated with excessive infant crying puts some doubt on this explanatory mechanism. Moreover, the possible moderation effect by maternal psychological problems increases the likelihood for the second mechanism, i.e. the maturation of circadian rhythms. Perhaps the high prevalence of excessive crying among infants born to women with both high levels of psychological problems and a low vitamin B-12 concentration reflects a stage in early infancy of incoordination among multiple immature circadian systems.8 Besides the hypothetical influence of maternal psychological problems on the maturation of the cortisol circadian rhythm, vitamin B-12 potentially affects the circadian rhythm of melatonin. The hormone melatonin is secreted by the pineal gland in response to darkness, and controls, among others, the sleep-wake cycle.23 The development of the sleep-wake cycle starts prenatally: the maternal circadian rhythm of melatonin secretion entrains the fetal circadian system. After birth, infant’s retina-mediated melatonin secretion begins to develop in response to light and darkness, which requires about 3 months to develop. During these months, disagreement between prenatal entrainment and postnatal signals causes uncomfortable overlapping states of wakefulness and sleep. Additionally, the developing melatonin circadian rhythm is responsible for a temporary imbalanced serotonin-melatonin system, which is involved in gastrointestinal muscle contractions. It is suggested that during the first 3 postnatal months, infants may suffer more from intestinal cramps by high serotonin and low melatonin concentrations.32 Because it is also during these months that excessive infant
crying mostly occurs, it is hypothesized that this crying behavior is a symptom of an immature melatonin circadian rhythm.8,23,32 Vitamin B-12, in turn, is involved in the melatonin secretion system. First evidence, provided by a few case studies on the vitamin B-12 treatment effect on sleep disorders, suggests that vitamin B-12 promotes melatonin synthesis21,33 and increases the sensitivity of melatonin to bright light.20,34 Whether and how the maturation of the melatonin circadian rhythm in newborns infants is affected by vitamin B-12 is however unexplored and needs further research.
Vitamin B-12 deficiency is usually caused by pernicious anemia (i.e. impaired absorption of vitamin B-12) or by a low intake of animal-source foods, mainly seen among vegetarians or people with a low socioeconomic status.31,35 Infants born to deficient mothers are also at a higher risk of vitamin B-12 deficiency, because the fetus is totally dependent on the maternal-fetal transfer of vitamin B-12.26 In the present study, young and low educated women had a lower vitamin B-12 concentration. Remarkably, Turkish women had an extremely low vitamin B-12 concentration compared with the concentrations among the other ethnic groups. This was also observed among pregnant women in Turkey36 and among Turkish depressive patients in the Netherlands.37 A possible explanation could be that Turkish groups have a high prevalence of Helicobacter pylori infection (75-82%),38,39 which is a risk factor for vitamin B-12 deficiency.40,41 This may be amplified by the lower meat and fish consumption observed among Turkish women compared with the indigenous women in the Netherlands.42 Besides the high-risk Turkish women, all pregnant women consuming a lower amount of animal-source foods should be aware of the potential health risks for mother and child, as our results suggest that even a marginally lower maternal vitamin B-12 concentration puts infants at a higher risk for excessive crying behavior.
The strengths of the present study include the large number of women with available blood and questionnaire data, the prospective nature of the study, and the prespecified multivariate analysis design. Compared with the women who did not participate in the ABCD biomarker study, the studied sample was a relatively healthy sample; they were older, higher educated, more often nulliparous, more often of Dutch origin, and they experienced less psychological problems. The prevalence of excessive crying was also lower among their infants (data not shown). Worldwide, no consensus has been reached for the definition of excessive infant crying, also called infant colic. Criteria for excessive crying range from crying >3 hours/day on >3 days/week for >3 weeks43 till an unquantified ‘cries a lot’.3 Depending on the criteria used different groups of infants are included, resulting in a wide prevalence range from 2.5% to 40%.44,45 This is a problem for the comparison of research as well as for clinical practice. The amount of infant crying should preferably be collected prospectively with diaries or audiotape recordings rather than retrospectively with a questionnaire, however, such methods are difficult to achieve in large-scale studies. We were able to collect maternal serum vitamin B-12 and folate concentrations on a large scale, but only during early pregnancy. Our results are, however, not likely to be influenced by changes in late gestation; longitudinal studies did show a decrease of maternal serum vitamin B-12 and folate concentrations during pregnancy – for which we corrected – but this reflects
107
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
a physiological rather than a dietary change.26,46 Moreover, the theoretical background of our study design was based on the hypothesis that the maternal nutritional status affects the fetus as early as the periconceptional period.18,26
In conclusion, this study provides first evidence for an early nutritional origin in infant crying behavior, as the results suggest that infants born to women with a low vitamin B-12 status during pregnancy are at a higher risk for excessive crying behavior in their first months of life. Maternal psychological problems may possibly modify this association. Further research is necessary to support this preliminary evidence and to elucidate the underlying mechanisms. If confirmed, prenatal checks of pregnant women should be extended with the diagnosis and treatment of women with a low vitamin B-12 concentration.
1. Reijneveld SA, van der Wal MF, Brugman E, Sing RA, Verloove-Vanhorick SP. Infant crying and abuse. Lancet 2004;364:1340-2.
2. Raiha H, Lehtonen L, Huhtala V, Saleva K, Korvenranta H. Excessively crying infant in the family: mother-infant, father-infant and mother-father interaction. Child Care Health Dev 2002;28:419-29.
3. Helseth S, Begnum S. A comprehensive definition of infant colic: parents’ and nurses’ perspectives. J Clin Nurs 2002;11:672-80.
4. Burgess KB, Marshall PJ, Rubin KH, Fox NA. Infant attachment and temperament as predictors of subsequent externalizing problems and cardiac physiology. J Child Psychol Psychiatry 2003;44:819-31.
5. Rao MR, Brenner RA, Schisterman EF, Vik T, Mills JL. Long term cognitive development in children with prolonged crying. Arch Dis Child 2004;89:989-92.
6. Wolke D, Rizzo P, Woods S. Persistent infant crying and hyperactivity problems in middle childhood. Pediatrics 2002;109:1054-60.
7. Barr RG. Colic and crying syndromes in infants. Pediatrics 1998;102(5 Suppl E):1282-6.8. White BP, Gunnar MR, Larson MC, Donzella B, Barr RG. Behavioral and physiological responsivity, sleep, and
patterns of daily cortisol production in infants with and without colic. Child Dev 2000;71:862-77.9. Lucassen PL, Assendelft WJ, Gubbels JW, van Eijk JT, van Geldrop WJ, Neven AK. Effectiveness of treatments
for infantile colic: systematic review. BMJ 1998;316:1563-9.10. Van der Wal MF, van Eijsden M, Bonsel GJ. Stress and emotional problems during pregnancy and excessive
infant crying. J Dev Behav Pediatr 2007;28:431-7.11. Canivet CA, Ostergren PO, Jakobsson IL, Dejin-Karlsson E, Hagander BM. Infantile colic, maternal smoking
and infant feeding at 5 weeks of age. Scand J Public Health 2008;36:284-91.12. Sondergaard C, Henriksen TB, Obel C, Wisborg K. Smoking during pregnancy and infantile colic. Pediatrics
2001;108:342-6.13. Wurmser H, Rieger M, Domogalla C, Kahnt A, Buchwald J, Kowatsch M, et al. Association between life stress
during pregnancy and infant crying in the first six months postpartum: a prospective longitudinal study. Early Hum Dev 2006;82:341-9.
14. Canivet CA, Ostergren PO, Rosen AS, Jakobsson IL, Hagander BM. Infantile colic and the role of trait anxiety during pregnancy in relation to psychosocial and socioeconomic factors. Scand J Public Health 2005;33:26-34.
15. De Weerth C, van Hees Y, Buitelaar JK. Prenatal maternal cortisol levels and infant behavior during the first 5 months. Early Hum Dev 2003;74:139-51.
16. Molloy AM, Kirke PN, Brody LC, Scott JM, Mills JL. Effects of folate and vitamin B12 deficiencies during preg-nancy on fetal, infant, and child development. Food Nutr Bull 2008;29(2 Suppl):S101-11.
17. Bottiglieri T. Folate, vitamin B12, and neuropsychiatric disorders. Nutr Rev 1996;54:382-90.18. Black MM. Effects of vitamin B12 and folate deficiency on brain development in children. Food Nutr Bull
2008;29(2 Suppl):S126-31.19. Dror DK, Allen LH. Effect of vitamin B12 deficiency on neurodevelopment in infants: current knowledge and
possible mechanisms. Nutr Rev 2008;66:250-5.20. Honma K, Kohsaka M, Fukuda N, Morita N, Honma S. Effects of vitamin B12 on plasma melatonin rhythm in
humans: increased light sensitivity phase-advances the circadian clock? Experientia 1992;48:716-20.21. Ikeda M, Asai M, Moriya T, Sagara M, Inoue S, Shibata S. Methylcobalamin amplifies melatonin-induced cir-
cadian phase shifts by facilitation of melatonin synthesis in the rat pineal gland. Brain Res 1998;795:98-104.22. Kirjavainen J, Lehtonen L, Kirjavainen T, Kero P. Sleep of excessively crying infants: a 24-Hour Ambulatory
Sleep Polygraphy study. Pediatrics 2004;114:592-600.23. Weissbluth M, Weissbluth L. Colic, sleep inertia, melatonin and circannual rhythms. Med Hypotheses
1992;38:224-8.24. Buitelaar JK, Huizink AC, Mulder EJ, de Medina PG, Visser GH. Prenatal stress and cognitive development and
temperament in infants. Neurobiol Aging 2003;24(Suppl 1):S53-60.
109
_________________________________________ Maternal vitamin B-12 and folate and infant crying
7
25. Van Eijsden M, van der Wal MF, Hornstra G, Bonsel GJ. Can whole-blood samples be stored over 24 hours without compromising stability of C-reactive protein, retinol, ferritin, folic acid, and fatty acids in epidemio-logic research? Clinical Chemistry 2005;51:230-2.
26. Cikot RJ, Steegers-Theunissen RP, Thomas CM, de Boo TM, Merkus HM, Steegers EA. Longitudinal vitamin and homocysteine levels in normal pregnancy. Br J Nutr 2001;85:49-58.
27. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psy-chol Meas 1977;1:385-401.
29. Huizink AC, Mulder EJ, Robles de Medina PG, Visser GH, Buitelaar JK. Is pregnancy anxiety a distinctive syndrome? Early Hum Dev 2004;79:81-91.
30. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: concep-tual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173-82.
31. Chalouhi C, Faesch S, Anthoine-Milhomme MC, Fulla Y, Dulac O, Cheron G. Neurological consequences of vitamin B12 deficiency and its treatment. Pediatr Emerg Care 2008;24:538-41.
32. Weissbluth L, Weissbluth M. Infant colic: the effect of serotonin and melatonin circadian rhythms on the intes-tinal smooth muscle. Med Hypotheses 1992;39:164-7.
33. Tomoda A, Miike T, Matsukura M. Circadian rhythm abnormalities in adrenoleukodystrophy and methyl B12 treatment. Brain Dev 1995;17:428-31.
34. Ikeda M, Honda K, Inoue S. Vitamin B12 amplifies circadian phase shifts induced by a light pulse in rats. Expe-rientia 1996;52:691-4.
35. Allen LH. How common is vitamin B-12 deficiency? Am J Clin Nutr 2009;89:693S-6.36. Ackurt F, Wetherilt H, Loker M, Hacibekiroglu M. Biochemical assessment of nutritional status in pre- and
post-natal Turkish women and outcome of pregnancy. Eur J Clin Nutr 1995;49:613-22.37. Guzelcan Y, van Loon P. Vitamin B12 status in patients of Turkish and Dutch descent with depression: a com-
parative cross-sectional study. Ann Gen Psychiatry 2009;8:18.38. De Vries AC, van Driel HF, Richardus JH, Ouwendijk M, van Vuuren AJ, de Man RA, et al. Migrant communi-
ties constitute a possible target population for primary prevention of Helicobacter pylori-related complications in low incidence countries. Scand J Gastroenterol 2008;43:403-9.
39. Loffeld RJ, van der Putten AB. Changes in prevalence of Helicobacter pylori infection in two groups of patients undergoing endoscopy and living in the same region in the Netherlands. Scand J Gastroenterol 2003;38:938-41.
40. Serin E, Gumurdulu Y, Ozer B, Kayaselcuk F, Yilmaz U, Kocak R. Impact of Helicobacter pylori on the develop-ment of vitamin B12 deficiency in the absence of gastric atrophy. Helicobacter 2002;7:337-41.
41. Kountouras J, Gavalas E, Boziki M, Zavos C. Helicobacter pylori may be involved in cognitive impairment and dementia development through induction of atrophic gastritis, vitamin B-12 folate deficiency, and hyperhomo-cysteinemia sequence. Am J Clin Nutr 2007;86:805-6.
42. Palsma AH, Nicolaou M, van Dam RM, Stronks K. Nutritional habits of Dutch residents of Turkish and Mo-roccan origin aged 18-30 years. Priorities for nutritional interventions. [In Dutch]. Tijdschrift voor Gezond-heidswetenschappen 2006;7:415-22.
43. Wessel MA, Cobb JC, Jackson EB, Harris GS, Jr., Detwiler AC. Paroxysmal fussing in infancy, sometimes called colic. Pediatrics 1954;14:421-35.
44. Reijneveld SA, Brugman E, Hirasing RA. Excessive infant crying: the impact of varying definitions. Pediatrics 2001;108:893-7.
45. Lucassen PL, Assendelft WJ, van Eijk JT, Gubbels JW, Douwes AC, van Geldrop WJ. Systematic review of the occurrence of infantile colic in the community. Arch Dis Child 2001;84:398-403.
46. Koebnick C, Heins UA, Dagnelie PC, Wickramasinghe SN, Ratnayaka ID, Hothorn T, et al. Longitudinal con-centrations of vitamin B(12) and vitamin B(12)-binding proteins during uncomplicated pregnancy. Clin Chem 2002;48(6 Pt 1):928-33.
Main aims of this thesis were (I) to explore and explain disparities in perinatal health outcomes among ethnic groups in Amsterdam, the Netherlands, and (II) to elucidate psychobiological pathways into (ethnicity-related) perinatal and infant health outcomes. In this chapter, we will reflect on our main results and discuss them in the light of methodological issues and implications for clinical practice and further research.
Ethnicity and perinatal health outcomesMain indicators of perinatal health outcomes are perinatal mortality and perinatal morbidity, including congenital disorders, a low birthweight as indicated by small-for-gestational-age (SGA), preterm birth (PTB) and a low Apgar score. We observed different prevalences of these perinatal health outcomes among the main ethnic groups in Amsterdam (i.e. the Dutch, Surinamese, Antillean, Ghanaian, Turkish and Moroccan group) (Figures 8.1-8.3). Compared to the Dutch newborns, newborns of all ethnic minority groups had on average a lower birthweight and a higher SGA risk. The prevalence of PTB was higher among the Surinamese, Antillean and Ghanaian group, but not among the Turkish and Moroccan group. When comparing the outcomes of first and second generation groups, different patterns were observed; while for the PTB prevalence a certain amount of convergence was noted between the second generation groups and the Dutch group, this was largely absent for the birthweights. The prevalence of a low Apgar score (score <7) and child loss (miscarriages and perinatal deaths included; abortions excluded) was highest among the Surinamese, Antillean and Ghanaian group. In contrast, no significant ethnic differences were found in the prevalence of congenital disorders; this prevalence may, however, be an underestimation in that child losses – which most likely include the most serious congenital disorders – were excluded from prevalence estimation. A similar phenomenon was probably responsible for the low prevalence of congenital disorders observed in the four largest cities in the Netherlands.1 While ethnic disparities in the prevalence of child loss are – without doubt – substantial, the consequences of ethnic disparities in perinatal morbidity rates are less clear. In general, SGA, a low Apgar score and – in particular – PTB have important consequences for a newborn’s future health and survival.2-7 However, when comparing populations, differences in mean birthweight and gestational age are not necessarily related to higher population mortality and morbidity rates; the entire population distribution of birthweight and gestational age may be ‘naturally’ shifted as a result of smaller stature (e.g. Surinam-Hindustani) or earlier fetal maturation (e.g. blacks).8-12 Nevertheless, we observed the highest rates of child loss and low Apgar scores among those ethnic groups (i.e. the Surinamese, Antillean and Ghanaian group) with the lowest birthweights and highest PTB prevalence. As the perinatal morbidity rates increased in the last decades, it becomes more and more important to explore its etiology; understanding the ethnic disparities in perinatal morbidity rates will be a huge step forward. This will – most likely – not only improve worldwide health but will also reduce societal and economical costs following adverse perinatal health outcomes.4,7,13-16
113
___________________________________________________________________ General discussion
8
To reduce the ethnic disparities in perinatal health outcomes we have to understand how the broad concept of ethnicity is related to perinatal outcomes. Ethnicity in itself is a ‘distal’ factor that can only affect health through mediation of ‘proximal’ risk factors;17 proximal risk factors (e.g. smoking, stress or nutrition) can directly affect health through pathophysiological mechanisms. We explored whether a set of conventional, mainly proximal risk factors mediated the observed association of ethnicity with birthweight and PTB risk. Lower birthweights among ethnic minority groups can be the result of either pathological growth restriction or a more ‘natural’ limited growth potential due to constitutional factors like maternal size, parity and maternal age.18,19 We concluded that, in our sample, an important part of the newborns of ethnic minority groups was constitutionally smaller due to ethnic differences in maternal height. In general, smaller mothers get smaller babies; this is, in fact, of evolutionary importance to secure safe vaginal deliveries. Paternal/fetal genome has therefore to be overruled by maternal constraint. Whether this has any negative consequences for future offspring health is, however, unknown.18 Another part of the newborns of ethnic minority groups was probably pathologically smaller, although higher prevalences of (working) stress and slightly higher prevalences of smoking and underweight among ethnic minority groups only appeared to be to a small extent responsible for the lower birthweights. With respect to ethnic disparities in PTB risk, we observed that ethnic differences in the prevalence of chronic hypertension, smoking, obesity and previous induced abortions partly explained the higher PTB risk among certain ethnic minority groups. At the end, compared to the Dutch group the Turkish and Moroccan group did not seem to have unfavorable perinatal health outcomes as their newborns were not at a higher PTB risk and were mainly constitutionally smaller due to smaller mothers (Figures 8.1 & 8.2). Similar findings have been reported in studies from Rotterdam (the Netherlands), Belgium, France and the USA.20-23 In contrast, the Surinamese, Antillean and Ghanaian group did seem to have unfavorable perinatal health outcomes, which could partly, but not fully, be explained by mediation of a set of conventional risk factors. Reducing this set of conventional risk factors will hence reduce, but not dissolve the higher perinatal morbidity rates among these groups. Studies in Rotterdam, France, the USA and the UK also reported the worst perinatal health outcomes among Afro-Caribbean and African women.20,21,24-29 While the Surinamese-Hindustani newborns possibly were constitutionally smaller, the newborns with an African ancestry probably had an excess risk explained by unexplored risk factors like maternal nutritional status30,31 and infections32 or, which we believe is more likely, by a cumulation of risk factors and/or (epi)genetic influences.33-36 Currently, research on (epi)genetic effects is rapidly growing, whereas research on cumulative risk profiles deserves more attention.
Cumulation of risksIn contrast to, for instance, the period of the Dutch famine,37 pregnant women are nowadays exposed to numerous ‘small’ risk factors. After years of research, it should be admitted that ethnic disparities in perinatal health outcomes are not likely to be explained by single effects of such small risk factors. Researchers therefore more and more emphasize to search for cumulative effects of multiple risk factors.15,34,38,39 Multiple pathophysiological processes probably cause much more
damage through interaction effects than a single harmful agent. We provided preliminary evidence for this cumulative risk hypothesis by presenting a cumulative effect of multiple risk factors (both proximal and distal risk factors) on PTB risk; the highest PTB risk was observed among the group of women with the highest number of risk factors. In addition, we observed a higher cumulation of risk factors among ethnic minority groups compared to the Dutch group, which appeared to be largely responsible for the ethnic disparities in PTB risk. These results seem promising, however, further research is necessary to support the preliminary findings and to develop an improved cumulative risk score with a higher predictive value. Only then it could be an effective tool for clinical practice for the purpose of screening pregnant women for their cumulative risk profile and addressing antenatal intervention programs to those women with the highest risk profile. Recently, a new framework has been proposed that expands the cumulative risk hypothesis by stating that ethnic disparities in perinatal health outcomes not only result from differential cumulative risk profiles during pregnancy, but moreover from differential life course trajectories.17,35 This framework hypothesizes that a woman’s reproductive potential is largely set by fetal programming during her own fetal development.13,35 Subsequently, her reproductive potential is influenced by exposure to risk factors (e.g. infections, poor diets, chronic disease, stress) and protective factors (e.g. social support, coping) across the life course.17,35 Translating this life course approach to ethnic disparities in perinatal health outcomes, it is suggested that ethnic minority groups have a disadvantaged reproductive potential as a result of chronic exposure to poor socioeconomic conditions, infections, psychosocial stress and racism passed on across generations through fetal programming.34,35,40,41 Nonetheless, this life course framework is still in its early gestation and biological evidence from human studies is lacking.
Psychobiological pathways into ethnicity-related perinatal health outcomesThe potentially negative effects of maternal psychosocial stress on perinatal health outcomes increasingly get attention, especially in explaining ethnic disparities. In 2005, the March of Dimes (USA), for instance, recommended six research priorities for investigating the etiology of PTB; one of the priorities is to delineate the role of stress and accompanying psychobiological pathways in the risk for PTB and in ethnic disparities in PTB risk.15 Numerous research studies on the negative effects of maternal psychosocial stress on perinatal health outcomes have already been performed, however, results are very inconsistent.42-44 In contrast, research on psychobiological pathways into perinatal health outcomes is scarce.32,45 With the advantage of a large sample of pregnant women, we explored how maternal psychosocial well-being during pregnancy, as indicated by the amount of depressive symptoms, was related to major perinatal health outcomes and furthermore, how maternal cortisol concentration during pregnancy, as a biomarker of psychosocial stress, was related to fetal growth. We observed the highest prevalence of PTB, SGA, a low Apgar score and child loss among the group of women who reported high levels of depressive symptomatology during pregnancy. Only the prevalences of SGA and a low Apgar score remained, however, statistically significantly higher after correction for the effects of relevant covariates. Various indirect and/or direct pathways could be responsible for the higher risk of perinatal health problems among
115
___________________________________________________________________ General discussion
8
women with high levels of depressive symptomatology during pregnancy. As we observed a higher smoking prevalence during pregnancy among women with psychosocial problems, these women could be indirectly at a higher risk for perinatal health problems through the toxic effects of cigarette smoking.46 Alternatively, these women could be directly at a higher risk through a psychobiological pathway involving altered concentrations of the hormone cortisol. Our data did, however, not seem to support such a pathway as we did not observe a statistically significant association of maternal cortisol concentration with both maternal psychosocial problems and offspring birthweight. The analysis was, however, hindered by methodological difficulties like the complex nature of the circadian cortisol rhythm.
Overall, we think the detrimental effects of maternal psychosocial stress during pregnancy on perinatal health outcomes are hard to elucidate because they depend on several intra- and interindividual factors. Normally, the human body responds to stress with the fight-or-flight response, i.e. complex feedback mechanisms of the cardiovascular, metabolic, immune and neuroendocrine system.47 Hereafter, the human body has to regain physiological stability or homeostasis through a mechanism called allostasis. This involves the regulation of biomarkers of stress, such as the corticotrophin-releasing hormone (CRH), cortisol, heart rate activity and blood pressure. In case of ‘allostatic load’, for example by chronic exposure to stress, the ‘set points’ of these biomarkers change in order to maintain physiological balance over time. This may result in pathophysiological effects like an increased blood pressure or heart rate or an abnormal neuroendocrinological response to stress.47-49 We hypothesize that pregnant women with allostatic load (also called ‘stress age’ or ‘weathering’40,44,48) as a results of chronic stress exposure before and/or during pregnancy are at a higher risk for adverse perinatal health outcomes. In addition, the susceptibility of pregnant women to detrimental effects of psychosocial stress will also depend on their coping strategies, for instance the access to and utilization of social support.47,50 Future research is challenged by the search for those women who are most susceptible to detrimental effects of psychosocial stress during pregnancy on perinatal health outcomes; screening women for allostatic load by monitoring blood pressure, heart rate activity and the concentration of stress hormones (e.g. cortisol, CRH) seems a new and promising way to do so.
The theory of allostatic load is suspected to be highly relevant for women from ethnic minority groups.40,48 They are often exposed to chronic stress across the life course due to racism, low socioeconomic status and language and cultural differences. Within the ABCD cohort, ethnic minority groups reported higher levels of maternal psychosocial problems during pregnancy compared to the Dutch group. The prevalence of ‘depression’, for example, as operationalized by a score of 16 or higher on the Center for Epidemiologic Studies Depression Scale (CES-D), was respectively 44%, 41%, 55%, 42%, and 34% for the Surinamese, Antillean, Turkish, Moroccan and Ghanaian group compared to 22% for the Dutch group. And while 0.5% of the Dutch group reported exposure to physical and/or sexual violence during pregnancy, this ranged from 0.6% to 3.1% among the ethnic minority groups. In additional analyses, we explored whether the higher prevalence of psychosocial stress during pregnancy among ethnic minority groups actually was responsible for their higher prevalence of adverse perinatal outcomes; we therefore expanded
the risk models for low birthweight and PTB with maternal psychosocial problems (depressive symptoms, pregnancy-related anxiety, parenting stress, work stress and physical/sexual violence). The results are shown in Figures 8.1 and 8.2, represented in the third bar for each ethnic group. In contrast to what expected, ethnic disparities in birthweight and PTB risk could not be explained by ethnic differences in maternal psychosocial problems. In theory, the explored associations might be more complex in that psychosocial stress may interact with ethnic background in affecting perinatal health outcomes. Such an interaction effect could potentially be attributed to higher allostatic load among ethnic minority groups and/or higher vascular activity and reactivity to stress among blacks.40,51 This needs, however, to be elucidated in future research.
Psychobiological pathways into infant health outcomesBeyond the importance to elucidate psychobiological pathways that connect maternal psychosocial well-being during pregnancy with perinatal health outcomes, it is of growing importance to elucidate the psychobiological pathways that connect maternal health during pregnancy with infant psychosocial well-being. In fact, some risk factors during pregnancy may not affect fetal growth or the time of parturition, meanwhile damaging certain developing physiological systems. A poorly understood indicator of infant psychosocial stress is excessive infant crying or infant colic. While excessive infant crying can be partly attributed to postnatal factors like parental caring or milk allergy,52 further etiology may originate in early fetal development. Several maternal risk factors during pregnancy, including maternal psychosocial stress and smoking, have been linked to excessive infant cring.53,54 A potential link with the nutritional status of the mother during pregnancy is yet missing, while it is known that nutrients are essential for fetal neurodevelopment.55,56 We hypothesized that maternal folate and vitamin B-12 status during pregnancy could be related to infant crying behavior. Our data confirmed this hypothesis for vitamin B-12 but not for folate. The psychobiological pathway explaining the observed association most likely involves the fetal development of the circadian sleep-wake rhythm. The development of this rhythm depends on the hormone melatonin; melatonin synthesis, in turn, depends – among others – on vitamin B-12.57 Excessive infant crying in the first months of life could be a symptom of an immature sleep-wake rhythm.58,59 These data provide first evidence for an early nutritional origin in excessive infant crying. The results, however, require confirmation. Further research should explore the hypothesized psychobiological pathway, the threshold level at which maternal vitamin B-12 status becomes alarmingly low, and the possible moderating role of maternal psychosocial problems. If confirmed, this could have important implications for (i) clinical practice, in that screening and treating pregnant women for vitamin B-12 deficiency will decrease the number of excessively crying infants, and (ii) for public health, in that the psychosocial well-being of infants and their parents will improve.
117
___________________________________________________________________ General discussion
The research questions of this thesis were examined within the context of the Amsterdam Born Children and their Development (ABCD) study. The ABCD study is a large prospective birth cohort study aimed at exploring the association between (ethnic differences in) maternal risk factors during pregnancy and (ethnic disparities in) perinatal and infant health outcomes. In this paragraph, we will discuss some methodological issues regarding the ABCD study design and sample and the measurement of major variables mentioned in this thesis.
ABCD study: design and sample One of the main strengths of the ABCD study is the large unselected community-based sample. During the inclusion period, all pregnant women in Amsterdam were approached by participating obstetric caregivers at their first regular prenatal visit. Because women were approached by a reliable source, the response rate was high (67%) and the sample diverse with respect to socioeconomic and ethnic background and health status. The actual participation was still somewhat selective, with a higher participation rate among Western groups, especially for the ABCD biomarker study (collection of an extra blood sample for research purposes). Despite the selective participation, selection bias appeared to be acceptably low as the association between a number of conventional risk factors and perinatal health outcomes was similar for respondents and non-respondents.60 Hence we assume that the observed associations reported in this thesis will be hardly influenced by selective participation.
Another strength of the ABCD study is the collection of a comprehensive set of maternal risk factors through maternal blood samples and questionnaire data. The maternal risk factors were measured around the end of the first trimester and beginning of the second trimester of pregnancy. This period of early gestation is suggested to be a particular sensitive period for maternal-placental-fetal pathophysiology and fetal programming because of the rapid organ development.61 Some risk factors, for instance cigarette smoking, infections or acute severe stress, however, might have non-programming effects on fetal growth and parturition in late pregnancy; unmeasured levels of such risk factors in late pregnancy could have influenced our risk models for adverse perinatal health outcomes.
EthnicityEthnicity was based on the self-reported country of birth of the pregnant woman and her mother – to distinguish between first and second generation – and categorized into the main ethnic groups in Amsterdam: Dutch, Surinamese, Antillean/Aruban, Ghanaian, Turkish, Moroccan and other non-Dutch (Western and non-Western) group. The ABCD study sample was representative for the ethnic distribution in Amsterdam.62 The use of the variable ethnicity in epidemiological research is highly debated, but nonetheless very useful when searching for explanations of health disparities.63-65 Classification by country of birth is widely used to define ethnic groups for it is an objective and stable measure. In addition, it often reflects both racial ancestry and cultural habits.
121
___________________________________________________________________ General discussion
8
It could, however, not differentiate between ethnic subgroups, for example Surinamese-Creole and Surinamese-Hindustani groups.63,66 For this purpose, self-identification of ethnicity would be more useful. Within the ABCD study, an additional question on self-identification provided data on ethnic subgroups. Nevertheless, we preferred to base ethnicity mainly on country of birth of the pregnant woman and her mother rather than on self-identification, because the latest measure is not stable over time and does not reflect racial ancestry,63,66 which we think may be important in explaining ethnic disparities in perinatal health outcomes. The operationalisation of ethnicity in future research among next generations will yet be more complicated as country of birth will then not be a reflection anymore of racial ancestry and cultural background.
Measurement of maternal psychosocial healthMaternal psychosocial health during pregnancy was measured in the pregnancy-questionnaire through several well-validated psychosocial scales: the Center for Epidemiologic Studies Depression scale (CES-D)67 measured depressive symptomatology, the state-scale of the State-Trait Anxiety Inventory (STAI)68 measured anxiety, the Pregnancy Related Anxiety Questionnaire - Revised version (PRAQ-R)69,70 measured pregnancy-related anxiety, the Work Experience and Appreciation Questionnaire (VBBA)71,72 measured job strain and finally, the frequency scale of the Parenting Daily Hassles (PDH)73 measured parenting stress. Additionally, women were retrospectively asked in the baby-questionnaire whether they were exposed to physical and/or sexual violence during pregnancy. Cronbach’s alphas for the psychosocial scales ranged from 0.81 to 0.94. Though the scales have good internal consistency, are well validated and are more or less used in previous research, they have several methodological limitations. First, the CES-D and STAI can not be used to diagnose respectively a depressive or anxiety disorder according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Though it might be possible that there is a dose-response relationship with mental disorders having a more negative impact on perinatal health outcomes than subclinical levels of depression or anxiety, this is not yet proved.42,74 Second, it is questionable whether certain psychosocial scales are as suitable in a pregnant population as in the general population, because of the risk for somatic confounding. Certain symptoms of pregnancy are, for example, similar to depressive symptoms (e.g. tiredness, loss of appetite).75 Restricting our analysis to non-somatic symptoms did, however, not improve the predictive capacity of the CES-D scale for perinatal health outcomes, which supported previous research.76 Third, it is not clear whether psychosocial scales have the same validity across ethnic groups, because some ethnic groups tend to somatize psychosocial stress much more. The few studies which explored this issue observed good validity and internal consistency of psychometric scales across ethnic groups, but suggested that cut-off points for detecting psychiatric disorders could be different.77-80 Further research on the validity and diagnostic thresholds of psychosocial scales across various subpopulations is necessary for the purpose of epidemiological research as well as for screening purposes in clinical practice. One of the best ways of assessing the pathophysiological effects of maternal psychosocial stress during pregnancy on fetal health is by assessing biomarkers of stress. A well-accepted
biomarker of stress is the hormone cortisol, which is released by the hypothalamic-pituitary-adrenal (HPA) axis in response to a stressor.81 As part of the ABCD biomarker study, we measured total cortisol concentration in serum in a large sample of more than 4000 pregnant women. Nonetheless, the use of cortisol in epidemiological research involves some methodological difficulties. First, general cortisol secretion is characterized by a circadian rhythm with peak levels shortly after awakening and a more or less declining pattern thereafter. It is hypothesized that the cortisol awakening response (CAR) reflects a different HPA-axis functioning than the cortisol secretion throughout the rest of the day. The CAR is suspected to be a good marker of psychosocial stress.82,83 Whether an altered CAR of the mother also has the most detrimental effects on fetal health is, however, unclear; so far, preliminary evidence is only provided by animal research.84 With only one cortisol measurement for each pregnant woman, we were not able to make a statement on the specific relationship of various parts of the diurnal cortisol rhythm with fetal health. We did, however, observe explicitly lower offspring birthweights among mothers with high versus low cortisol concentrations in the morning compared to other time points across the day. Second, it is hypothesized that free cortisol concentration is a better marker of potential pathophysiological effects than total cortisol concentration.85 Preliminary evidence suggests that exposure to stress decreases corticotrophin-binding globulin (CBG) concentration, resulting in a different ratio of total vs. free cortisol.86 If confirmed, free cortisol concentration should preferable be used in exploring the pathophysiological effects of stress on fetal health. In conclusion, our cortisol measurement suffered from serious limitations leaving the lack of an observed association between maternal cortisol concentration and both maternal psychosocial stress and offspring birthweight open to doubt. Much need to be done to reveal the best way to operationalize maternal cortisol concentration in exploring its pathophysiological effects on perinatal health outcomes. Until then, screening pregnant women for altered cortisol concentrations will not be effective.
Measurement of perinatal and infant health outcomesData on gestational age and offspring birthweight were highly available from the Youth Health Care registration at the Public Health Service in Amsterdam. An 80% linkage (higher for Western groups) with the Dutch Perinatal Registration (PRN) provided a second source of information on perinatal health outcomes. To define whether neonates were SGA, reference data from the PRN, matched on the period of deliveries within the ABCD cohort, were used. Although the SGA categorization accounted for the non-pathological effects of fetal gender and parity on birthweight, it did not account for the effects of maternal height. As a result, newborns of very small mothers (especially seen among some ethnic minority groups) could have been wrongly categorized as SGA. Gestational age was mostly based on ultrasound measurements; when unavailable (<10%), gestational age was based on the first day of the last menstrual period (calculation by obstetric care provider). Perinatal outcome data were incomplete (particularly for non-Western groups) for the type of preterm delivery (spontaneous vs. iatrogenic) and Apgar scores, because these data were only available from the PRN. Data on fetal/infant mortality was most likely incomplete for miscarriages.
123
___________________________________________________________________ General discussion
8
The amount of infant crying was measured in the baby-questionnaire (±3 months after birth) with the question: ‘How many hours per day (24 hours) on average did your baby cry in the past week?’ We defined excessive infant crying as crying on average ≥3 hours per day in the past week. Worldwide, no consensus has been reached for the definition of excessive infant crying, also called infant colic.52 Used criteria for excessive crying range from crying >3 hours/day on >3 days/week for >3 weeks to an unquantified ‘cries a lot’. Depending on the criteria used, different groups of infants are included.87 This is a problem for the comparison of research as well as for clinical practice. The reason for infants to cry excessively has different etiologies that are hard to distinguish.52 The association we observed between low maternal vitamin B-12 status during pregnancy and excessive infant crying may therefore be different depending on the homogeneity of cases. In future research, prospective measurements with diaries and audiotape recordings rather than retrospective questionnaires should preferably be used to diagnose various types of excessive infant crying, however, such measurements are difficult to achieve in large-scale epidemiological studies.
Implications for interventions
The period of pregnancy is one of the best opportunities for health interventions. Pregnant women are usually more willing to make an effort for healthy behaviors in order to restrict the possible health risks for their unborn child. In high income countries, antenatal educational programs are largely attended by pregnant women. There is, however, a huge gap in our knowledge about the efficacy of these programs.88 Though pregnant women evaluate antenatal programs as highly valuable and as an important source of information on risk behaviors, it is unclear whether there is an actual benefit for the health of the mother and her child. The efficacy of antenatal programs should hence be better evaluated in order to improve maternal and child health. Important risk factors for perinatal and infant health outcomes that are amenable for intervention are maternal smoking, obesity, nutritional status and psychosocial stress. First, the toxic effects of maternal smoking during pregnancy are believed to be one of the most harmful effects on fetal development. Although most women seem to be aware of the fetal health risks from smoking,89 38% of the pre-pregnancy smokers in our sample continued to smoke during pregnancy, resulting in a smoking prevalence of 9% during pregnancy. As the detrimental effects of smoking can immediately be stopped by quitting smoking, effective smoking cessation programs for pregnant women are crucial. Smoking cessation programs like cognitive-behavioral and nicotine replacement therapy have been shown to reduce smoking and to improve perinatal health outcomes.90-92 In contrast to smoking, immediate intervention effects are more difficult to achieve for obese pregnant women. As the risk for obesity often accumulates over the life course as a result of multiple determinants (e.g. rapid weight gain during infancy, poor nutritional habits, sedentary lifestyle),93,94 comprehensive interventions across the life course are necessary to ensure a preconceptional healthy body weight. Nevertheless, it will still be useful to give advice on healthy nutrition and physical activity during
antenatal educational programs. In fact, nutritional advice on, for example, folic acid and vitamin B-12 intake can have immediate positive effects on fetal development. The rising prevalence of psychosocial problems is becoming a public health problem. Obstetric caregivers should be aware of the impact of this problem on maternal and fetal health. The negative consequences of maternal psychosocial stress during pregnancy on fetal development more and more emerge, stressing the need to develop and implement intervention programs for pregnant women with psychosocial stress. Screening women for psychosocial stress should be part of routine antenatal care, accompanied by well-planned treatment referral preferably at the same site as the obstetrical provider.95 Promising treatments for psychosocial stress include cognitive-behavioral therapy, relaxation techniques and social support.44,96 For mental disorders like depression, pharmacologic treatments are proved to be efficacious, however, many pregnant women are reluctant to use medication because of concerns about the impact on the fetus. There is yet a striking need for more research on efficacious and safe (non)pharmacologic interventions for pregnant women with psychosocial problems. An important shortcoming of many antenatal health promotion programs is the focus on single rather than multiple risk behaviors, while the majority of pregnant women face more than one risk factor simultaneously.97-99 Moreover, risk behaviors are often highly connected; stressed women, for example, tend to smoke more often and also show poorer nutritional habits. Proximal risk behaviors like stress, smoking and nutritional habits often share the same distal risk factors like financial problems or lack of social support.17 Comprehensive antenatal intervention programs that cover not only important proximal risk behaviors but also distal factors like economic conditions or social environment will therefore be most effective to reduce adverse perinatal health outcomes. The Dutch ‘VoorZorg’ program,100 in which low-educated teenage pregnant women receive intensive support during and after pregnancy, is a good example of a perinatal education program that covers both proximal and distal risk factors.
Another shortcoming of antenatal programs is the limited reach of ethnic minority groups, while those groups are most in need for intervention programs because of their high-risk profile. They often have a high cumulation of both distal and proximal risk factors, and furthermore, they tend to start late with antenatal care visits to obstetric care providers.101 To reduce ethnic disparities in perinatal health outcomes, easy accessible antenatal programs are needed that are adjusted to specific ethnic groups to account for cultural and language differences. It is, for example, observed that folic acid use before and during pregnancy is much lower among non-Dutch-speaking women.102 Furthermore, cultural-specific beliefs and norms about, for instance, smoking, nutrition, physical activity and emotional problems are important determinants of risk behaviors in certain ethnic groups.98,103,104 An example of an ethnic-specific antenatal intervention program is the Dutch program ‘Blije Moeders, Blije Babies’ (BMBB) which focused on Turkish pregnant women. BMBB covered multiple risk factors including maternal smoking behavior and depressive feelings. Although the BMBB program did not seem to have an effect on smoking prevalence while only a small effect on the level of depressive feelings, the program was very successful in reaching an underserved minority group through Turkish community health workers.105 While in the next
125
___________________________________________________________________ General discussion
8
generations language problems most likely will disappear, it will depend on the future level of acculturation whether cultural-specific beliefs and norms should be accounted for in antenatal health promotion programs.
Ideally, prevention of (ethnic disparities in) adverse perinatal health outcomes starts preconceptionally rather than at the first prenatal visit when it is largely too late. In the Netherlands, the Erasmus Medical Center in Rotterdam was one of the firsts to start with a pilot study on preconceptional counseling.106 Couples planning to be pregnant were screened and educated on modifiable risk factors. After three months, significant improvements were observed on the prevalence of obesity, folic acid use, alcohol use and physical activity, but not on the use of cigarettes and drugs. To reduce addictive behaviors, a more intensive approach will probably be needed. Nonetheless, the pilot on preconceptional counseling seems promising and should be further developed.
Final conclusion
The main aims of this thesis were to explore and explain the ethnic disparities in perinatal health outcomes in Amsterdam and furthermore, to explore the psychobiological pathways into perinatal and infant health outcomes. Based on observations within the ABCD cohort and a reflection on these observations, the following concluding points could be drawn:
(1) Ethnic disparities in perinatal health outcomes were observed among the main ethnic groups in Amsterdam, i.e. the Dutch, Surinamese, Antillean, Ghanaian, Turkish and Moroccan group: mean offspring birthweight was lower among all ethnic minority groups compared to the Dutch group while the risk for PTB, a low Apgar score and child loss was only higher for the Surinamese, Antillean and Ghanaian group. Although the ethnic disparities could, to some extent, be explained by single conventional maternal risk factors during pregnancy, a cumulation effect of multiple risk factors – especially seen among the ethnic minority groups – seemed to provide a better explanation. It seems furthermore reasonable that (epi)genetic influences are involved as offspring from mainly African descent had unexplainable lower birthweights and a higher PTB risk. To reduce ethnic disparities in perinatal health outcomes, comprehensive antenatal health promotion programs should be developed and implemented that cover not only important proximal risk behaviors like smoking, poor nutrition and work stress, but also distal factors like lack of social support or financial problems. As long as there is no convergence in risk behaviors and their determinants between ethnic minority groups and the native Dutch group, we recommend the implementation of antenatal programs that account for cultural-specific beliefs and norms.
(2) One of the important risk factors for adverse perinatal health outcomes appeared to be maternal psychosocial stress. The prevalence of perinatal mortality and morbidity was higher among women who reported high levels of depressive symptoms during pregnancy, though only the prevalence
of SGA and a low Apgar score appeared to be statistically significantly higher when adjusting for relevant covariates. The relationship between maternal psychosocial stress during pregnancy and offspring outcome could at first be ascribed to a small mediation effect of maternal smoking behavior. Alternatively, elevated levels of the stress hormone cortisol potentially influence fetal development, however, only limited evidence for this psychobiological pathway could be provided because of methodological difficulties related to the measurement of cortisol. While ethnic minority groups in our sample had a high prevalence of both psychosocial problems and adverse perinatal health outcomes, differences in psychosocial well-being did not seem to be responsible for ethnic disparities in perinatal health outcomes. We hypothesized that pregnant women from African descent might be more susceptible for detrimental effects of psychosocial stress on offspring outcomes through ‘allostatic load’ and/or higher vascular (re)activity. As psychosocial stress is becoming a huge public health problem, obstetric caregivers should be aware of the potentially negative effects on fetal health. We recommend to screen pregnant women for psychosocial stress as part of routine antenatal care or preferably, as part of preconceptional care, provided that efficacious treatment programs are available.
(3) Psychosocial stress in infancy as expressed by excessive crying behavior may have its origin during fetal development. Maternal nutritional status during pregnancy appeared to be of importance as an association was observed between a low vs. high maternal vitamin B-12 status during pregnancy and excessive crying behavior of the infant in the first months of life; this might, in theory, be ascribed to fetal (dys)maturation of the sleep-wake / melatonin rhythm. Although these observations offer opportunities for prevention and treatment of excessively crying infants, the preliminary results first need to be replicated, preferably with prospective measurements of infant crying behavior through diaries or audiotape recordings.
127
___________________________________________________________________ General discussion
8
References
1. De Graaf JP, Ravelli ACJ, Wildschut HIJ, Denktas S, Voorham AJJ, Bonsel GJ, et al. Perinatal outcomes in the four largest cities and in deprived neighbourhoods in the Netherlands. [In Dutch]. Ned Tijdschr Geneeskd 2008;152:2734-40.
2. Callaghan WM, MacDorman MF, Rasmussen SA, Qin C, Lackritz EM. The contribution of preterm birth to infant mortality rates in the United States. Pediatrics 2006;118:1566-73.
3. Casey BM, McIntire DD, Leveno KJ. The continuing value of the Apgar score for the assessment of new-born infants. N Engl J Med 2001;344:467-71.
4. Thorngren-Jerneck K, Herbst A. Low 5-minute Apgar score: a population-based register study of 1 million term births. Obstet Gynecol 2001;98:65-70.
5. Figueras F, Figueras J, Meler E, Eixarch E, Coll O, Gratacos E, et al. Customised birthweight standards accurately predict perinatal morbidity. Arch Dis Child Fetal Neonatal Ed 2007;92:F277-80.
6. Barker DJ. Adult consequences of fetal growth restriction. Clin Obstet Gynecol 2006;49:270-83.7. Ramachandrappa A, Jain L. Health issues of the late preterm infant. Pediatr Clin North Am 2009;56:565-
77.8. Kistka ZA, Palomar L, Lee KA, Boslaugh SE, Wangler MF, Cole FS, et al. Racial disparity in the frequency
of recurrence of preterm birth. Am J Obstet Gynecol 2007;196:131.e1-6.9. Graafmans WC, Richardus JH, Borsboom GJ, Bakketeig L, Langhoff-Roos J, Bergsjo P, et al. Birth weight
and perinatal mortality: a comparison of “optimal” birth weight in seven Western European countries. Epidemiology 2002;13:569-74.
10. Vangen S, Stoltenberg C, Skjaerven R, Magnus P, Harris JR, Stray-Pedersen B. The heavier the better? Birthweight and perinatal mortality in different ethnic groups. Int J Epidemiol 2002;31:654-60.
11. Patel RR, Steer P, Doyle P, Little MP, Elliott P. Does gestation vary by ethnic group? A London-based study of over 122,000 pregnancies with spontaneous onset of labour. Int J Epidemiol 2003;33:107-13.
12. Wilcox AJ, Russell IT. Birthweight and perinatal mortality: II. On weight-specific mortality. Int J Epide-miol 1983;12:319-25.
13. Barker DJ. The origins of the developmental origins theory. J Intern Med 2007;261:412-7.14. Xu X, Grigorescu V, Siefert KA, Lori JR, Ransom SB. Cost of racial disparity in preterm birth: evidence
from Michigan. J Health Care Poor Underserved 2009;20:729-47.15. Green NS, Damus K, Simpson JL, Iams J, Reece EA, Hobel CJ, et al. Research agenda for preterm birth:
recommendations from the March of Dimes. Am J Obstet Gynecol 2005;193:626-35.16. Romo A, Carceller R, Tobajas J. Intrauterine growth retardation (IUGR): epidemiology and etiology. Pe-
diatr Endocrinol Rev 2009;6(Suppl 3):332-6.17. Misra DP, Guyer B, Allston A. Integrated perinatal health framework. A multiple determinants model
with a life span approach. Am J Prev Med 2003;25:65-75.18. Gluckman PD, Hanson MA. Maternal constraint of fetal growth and its consequences. Semin Fetal Neona-
tal Med 2004;9:419-25.19. Mamelle N, Cochet V, Claris O. Definition of fetal growth restriction according to constitutional growth
potential. Biol Neonate 2001;80:277-85.20. Troe EJ, Raat H, Jaddoe VW, Hofman A, Looman CW, Moll HA, et al. Explaining differences in birth-
weight between ethnic populations. The Generation R Study. BJOG 2007;114:1557-65.21. Zeitlin J, Bucourt M, Rivera L, Topuz B, Papiernik E. Preterm birth and maternal country of birth in a
French district with a multiethnic population. BJOG 2004;111:849-55.22. El-Sayed AM, Galea S. Explaining the low risk of preterm birth among arab americans in the United
States: an analysis of 617451 births. Pediatrics 2009;123:e438-45.23. Buekens P, Masuy-Stroobant G, Delvaux T. High birthweights among infants of north African immigrants
in Belgium. Am J Public Health 1998;88:808-11.24. Kelly Y, Panico L, Bartley M, Marmot M, Nazroo J, Sacker A. Why does birthweight vary among ethnic
groups in the UK? Findings from the Millennium Cohort Study. J Public Health 2009;31:131-7.
25. Howard DL, Marshall SS, Kaufman JS, Savitz DA. Variations in low birth weight and preterm delivery among blacks in relation to ancestry and nativity: New York City, 1998-2002. Pediatrics 2006;118:e1399-405.
26. David RJ, Collins RW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks, and U.S.-born whites. N Engl J Med 1997;337:1209-14.
27. Shiono PH, Rauh VA, Park M, Lederman SA, Zuskar D. Ethnic differences in birthweight: the role of lifestyle and other factors. Am J Public Health 1997;87:787-93.
28. Aveyard P, Cheng KK, Manaseki S, Gardosi J. The risk of preterm delivery in women from different ethnic groups. BJOG 2002;109:894-9.
29. Wilcox M, Gardosi J, Mongelli M, Ray C, Johnson I. Birth weight from pregnancies dated by ultrasono-graphy in a multicultural British population. BMJ 1993;307:588-91.
30. Van Eijsden, M. Ethnicity, nutrition, and pregnancy: food for thought. [Dissertation]. Amsterdam: Uni-versity of Amsterdam; 2008.
31. Ramakrishnan U, Manjrekar R, Rivera J, Gonzales-Cossio T, Martorell R. Micronutrients and pregnancy outcome: a review of the literature. Nutr Res 1999;19:103-59.
32. Wadhwa PD, Culhane JF, Rauh V, Barve SS, Hogan V, Sandman CA, et al. Stress, infection and preterm birth: a biobehavioural perspective. Paediatr Perinat Epidemiol 2001;15(Suppl 2):17-29.
33. Amante A, Borgiani P, Gimelfarb A, Gloria-Bottini F. Interethnic variability in birth weight and genetic background: a study of placental alkaline phosphatase. Am J Phys Anthropol 1996;101:449-53.
34. Gennaro S. Overview of current state of research on pregnancy outcomes in minority populations. Am J Obstet Gynecol 2005;192:S3-10.
35. Lu MC, Halfon N. Racial and ethnic disparities in birth outcomes: a life-course perspective. Matern Child Health J 2003;7:13-30.
36. Menon R. Spontaneous preterm birth, a clinical dilemma: etiologic, pathophysiologic and genetic hetero-geneities and racial disparity. Acta Obstet Gynecol Scand 2008;87:590-600.
37. Roseboom T, de Rooij S, Painter R. The Dutch famine and its long-term consequences for adult health. Early Hum Dev 2006;82:485-91.
38. Callahan MA, Sexton K. If cumulative risk assessment is the answer, what is the question? Environ Health Perspect 2007;115:799-806.
39. Dew PC, Guillory VJ, Okah FA, Cai J, Hoff GL. The effect of health compromising behaviors on preterm births. Matern Child Health J 2007;11:227-33.
40. Dominguez TP. Race, racism, and racial disparities in adverse birth outcomes. Clin Obstet Gynecol 2008;51:360-70.
41. Giscombe CL, Lobel M. Explaining disproportionately high rates of adverse birth outcomes among Afri-can Americans: the impact of stress, racism, and related factors in pregnancy. Psychol Bull 2005;131:662-83.
42. Alder J, Fink N, Bitzer J, Hosli I, Holzgreve W. Depression and anxiety during pregnancy: a risk factor for obstetric, fetal and neonatal outcome? A critical review of the literature. J Matern Fetal Neonatal Med 2007;20:189-209.
43. Vrijkotte TG, van der Wal MF, van Eijsden M, Bonsel GJ. First-trimester working conditions and birth-weight: a prospective cohort study. Am J Public Health 2009;99:1409-16.
44. Hobel CJ, Goldstein A, Barrett ES. Psychosocial stress and pregnancy outcome. Clin Obstet Gynecol 2008;51:333-48.
45. Diego MA, Field T, Hernandez-Reif M, Schanberg S, Kuhn C, Gonzalez-Quintero VH. Prenatal depres-sion restricts fetal growth. Early Hum Dev 2009;85:65-70.
46. Jaddoe VW, Troe EJ, Hofman A, Mackenbach JP, Moll HA, Steegers EA, et al. Active and passive maternal smoking during pregnancy and the risks of low birthweight and preterm birth: the Generation R Study. Paediatr Perinat Epidemiol 2008;22:162-71.
129
___________________________________________________________________ General discussion
8
47. Shannon M, King TL, Kennedy HP. Allostasis: a theoretical framework for understanding and evaluating perinatal health outcomes. J Obstet Gynecol Neonatal Nurs 2007;36:125-34.
48. Hogue CJ, Bremner JD. Stress model for research into preterm delivery among black women. Am J Obstet Gynecol 2005;192:S47-55.
49. McEwen BS. Protective and damaging effects of stress mediators. N Engl J Med 1998;338:171-9.50. Biondi M, Picardi A. Psychological stress and neuroendocrine function in humans: the last two decades
of research. Psychother Psychosom 1999;68:114-50.51. Dysart JM, Treiber FA, Pflieger K, Davis H, Strong WB. Ethnic differences in the myocardial and vascular
reactivity to stress in normotensive girls. Am J Hypertens 1994;7:15-22.52. Helseth S, Begnum S. A comprehensive definition of infant colic: parents’ and nurses’ perspectives. J Clin
Nurs 2002;11:672-80.53. Canivet CA, Ostergren PO, Jakobsson IL, Dejin-Karlsson E, Hagander BM. Infantile colic, maternal
smoking and infant feeding at 5 weeks of age. Scand J Public Health 2008;36:284-91.54. Van der Wal MF, van Eijsden M, Bonsel GJ. Stress and emotional problems during pregnancy and exces-
sive infant crying. J Dev Behav Pediatr 2007;28:431-7.55. Black MM. Effects of vitamin B12 and folate deficiency on brain development in children. Food Nutr Bull
2008;29:S126-31.56. Dror DK, Allen LH. Effect of vitamin B12 deficiency on neurodevelopment in infants: current knowledge
and possible mechanisms. Nutr Rev 2008;66:250-5.57. Ikeda M, Asai M, Moriya T, Sagara M, Inoue S, Shibata S. Methylcobalamin amplifies melatonin-induced
circadian phase shifts by facilitation of melatonin synthesis in the rat pineal gland. Brain Res 1998;795:98-104.
58. Weissbluth M, Weissbluth L. Colic, sleep inertia, melatonin and circannual rhythms. Med Hypotheses 1992;38:224-8.
59. White BP, Gunnar MR, Larson MC, Donzella B, Barr RG. Behavioral and physiological responsivity, sleep, and patterns of daily cortisol production in infants with and without colic. Child Dev 2000;71:862-77.
60. Tromp M, van Eijsden M, Ravelli ACJ, Bonsel GJ. Anonymous non-response analysis in the ABCD cohort study enabled by probabilistic record linkage. Paediatr Perinat Epidemiol 2009;23:264-72.
61. Godfrey KM, Barker DJ. Fetal programming and adult health. Public Health Nutr 2001;4:611-24.62. Municipality of Amsterdam. Department for Research and Statistics. Population. In: van Zee W, Hylkema
63. Bhopal R. Glossary of terms relating to ethnicity and race: for reflection and debate. J Epidemiol Commu-nity Health 2004;58:441-5.
64. Risch N, Burchard E, Ziv E, Tang H. Categorization of humans in biomedical research: genes, race and disease. Genome Biol 2002;3:comment2007.1-12.
65. Fassaert T, de Wit M, Goedhart G, Verhoeff A. Ethnicity as a variable in health research. Considerations from an epidemiological perspective. [In Dutch]. CMG 2009;6:22-9.
66. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of ‘country of birth’ for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14:1-14.
67. Radloff LS. The CES-D scale: a self-reported depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401.
71. Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3:322-55.
72. Van Veldhoven M, Meijman TF, Broersen JPJ, Fortuin RJ. Manual VBBA. [In Dutch]. Amsterdam: SKB Vragenlijst Services; 2002.
73. Crnic KA, Greenberg MT. Minor parenting stresses with young children. Child Dev 1990;61:1628-37.74. Andersson L, Sundstrom-Poromaa I, Wulff M, Astrom M, Bixo M. Neonatal outcome following maternal
antenatal depression and anxiety: a population-based study. Am J Epidemiol 2004;159:872-81.75. Orr ST, James SA, Blackmore PC. Maternal prenatal depressive symptoms and spontaneous preterm
births among African-American women in Baltimore, Maryland. Am J Epidemiol 2002;156:797-802.76. Kabir K, Sheeder J, Stevens-Simon C. Depression, weight gain, and low birth weight adolescent delivery:
do somatic symptoms strengthen or weaken the relationship? J Pediatr Adolesc Gynecol 2008;21:335-42.77. Fassaert T, de Wit MA, Tuinebreijer WC, Wouters H, Verhoeff AP, Beekman AT, et al. Psychometric
properties of an interviewer-administered version of the Kessler Psychological Distress scale (K10) among Dutch, Moroccan and Turkish respondents. Int J Methods Psychiatr Res 2009;18:159-68.
78. Simon GE, Goldberg DP, Von KM, Ustun TB. Understanding cross-national differences in depression prevalence. Psychol Med 2002;32:585-94.
79. Roberts RE. Reliability of the CES-D Scale in different ethnic contexts. Psychiatry Res 1980;2:125-34.80. Spijker J, van der Wurff FB, Poort EC, Smits CH, Verhoeff AP, Beekman AT. Depression in first generation
labour migrants in Western Europe: the utility of the Center for Epidemiologic Studies Depression Scale (CES-D). Int J Geriatr Psychiatry 2004;19:538-44.
81. Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CA. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosom Med 1996;58:432-46.
82. Chida Y, Steptoe A. Cortisol awakening response and psychosocial factors: a systematic review and meta-analysis. Biol Psychol 2009;80:265-78.
83. Clow A, Thorn L, Evans P, Hucklebridge F. The awakening cortisol response: methodological issues and significance. Stress 2004;7:29-37.
84. Venihaki M, Carrigan A, Dikkes P, Majzoub JA. Circadian rise in maternal glucocorticoid prevents pul-monary dysplasia in fetal mice with adrenal insufficiency. Proc Natl Acad Sci U S A 2000;97:7336-41.
85. Levine A, Zagoory-Sharon O, Feldman R, Lewis JG, Weller A. Measuring cortisol in human psychobiolo-gical studies. Physiol Behav 2007;90:43-53.
86. Fleshner M, Deak T, Spencer RL, Laudenslager ML, Watkins LR, Maier SF. A long-term increase in basal levels of corticosterone and a decrease in corticosteroid-binding globulin after acute stressor exposure. Endocrinology 1995;136:5336-42.
87. Reijneveld SA, Brugman E, Hirasing RA. Excessive infant crying: the impact of varying definitions. Pedi-atrics 2001;108:893-7.
88. Gagnon AJ, Sandall J. Individual or group antenatal education for childbirth or parenthood, or both. Co-chrane Database Syst Rev 2007:CD002869.
89. Hymowitz N, Schwab M, McNerney C, Schwab J, Eckholdt H, Haddock K. Postpartum relapse to cigarette smoking in inner city women. J Natl Med Assoc 2003;95:461-74.
90. Lumley J, Oliver SS, Chamberlain C, Oakley L. Interventions for promoting smoking cessation during pregnancy. Cochrane Database Syst Rev 2004:CD001055.
91. Kelley K, Bond R, Abraham C. Effective approaches to persuading pregnant women to quit smoking: a meta-analysis of intervention evaluation studies. Br J Health Psychol 2001;6:207-28.
92. Crawford JT, Tolosa JE, Goldenberg RL. Smoking cessation in pregnancy: why, how, and what next. Clin Obstet Gynecol 2008;51:419-35.
93. Derbyshire E. Low maternal weight: effects on maternal and infant health during pregnancy. Nurs Stand 2007;22:43-6.
131
___________________________________________________________________ General discussion
8
94. Johnson DB, Gerstein DE, Evans AE, Woodward-Lopez G. Preventing obesity: a life cycle perspective. J Am Diet Assoc 2006;106:97-102.
95. Yonkers KA, Smith MV, Lin H, Howell HB, Shao L, Rosenheck RA. Depression screening of perinatal women: an evaluation of the healthy start depression initiative. Psychiatr Serv 2009;60:322-8.
96. Dimidjian S, Goodman S. Nonpharmacologic intervention and prevention strategies for depression dur-ing pregnancy and the postpartum. Clin Obstet Gynecol 2009;52:498-515.
97. Joseph JG, El-Mohandes AA, Kiely M, El-Khorazaty MN, Gantz MG, Johnson AA, et al. Reducing psy-chosocial and behavioral pregnancy risk factors: results of a randomized clinical trial among high-risk pregnant african american women. Am J Public Health 2009;99:1053-61.
98. Katz KS, Blake SM, Milligan RA, Sharps PW, White DB, Rodan MF, et al. The design, implementation and acceptability of an integrated intervention to address multiple behavioral and psychosocial risk factors among pregnant African American women. BMC Pregnancy Childbirth 2008;8:22.
99. Hyman DJ, Pavlik VN, Taylor WC, Goodrick GK, Moye L. Simultaneous vs sequential counseling for multiple behavior change. Arch Intern Med 2007;167:1152-8.
100. Kooijman K, Struijf E, van Coeverden S, Crijnen A, van Leerdam F. VoorZorg: educational assistance from pregnancy to infancy. [In Dutch]. Tijdschrift voor Jeugdgezondheidszorg 2008;40:60-2.
101. Alderliesten ME, Vrijkotte TG, van der Wal MF, Bonsel GJ. Late start of antenatal care among ethnic minorities in a large cohort of pregnant women. BJOG 2007;114:1232-9.
102. Van Eijsden M, van der Wal MF, Bonsel GJ. Folic acid knowledge and use in a multi-ethnic pregnancy cohort: the role of language proficiency. BJOG 2006;113:1446-51.
103. Nierkens V, Stronks K, van Oel CJ, de Vries H. Beliefs of Turkish and Moroccan immigrants in The Neth-erlands about smoking cessation: implications for prevention. Health Educ Res 2005;20:622-34.
104. Palsma AH, Nicolaou M, van Dam RM, Stronks K. Nutritional habits of Dutch residents of Turkish and Moroccan origin aged 18-30 years. Priorities for nutritional interventions. [In Dutch]. Tijdschrift voor Gezondheidswetenschappen 2006;7:415-22.
105. Hesselink AE, van Poppel MN, van Eijsden M, Twisk JWR, van der Wal MF. The effectiveness of a peri-natal education program on smoking, child care and psychosocial health for Turkish women. A quasi randomized controlled trial. [Unpublished work]. 2009.
106. Van Mil NH, Hammiche F, Laven JSE, Wildhagen M, Bonsel GJ, Steegers EAP, et al. Preconception coun-seling on modifiable lifestyle risk factors. The Rotterdam Reproduction Risk score (R3 score). J Dev Orig Health Dis 2009;1:S65-6.
One of the main aims of this thesis was to explore perinatal health epidemiology in the multi-ethnic city of Amsterdam, the Netherlands. As described in Chapter 1, perinatal health epidemiology studies the health of mother and baby during pregnancy, birth and the early postpartum period. Important markers of an unhealthy intrauterine environment are perinatal mortality and morbidity; perinatal morbidity (e.g. preterm birth, low birthweight) could have important consequences for long-term health outcomes such as cardiovascular disease. Worldwide, perinatal mortality and morbidity rates differ between ethnic groups; this can, in theory, be explained by ethnic differences in a broad range of maternal risk factors during pregnancy. The first objectives of this thesis were to examine (a) the perinatal health outcomes among the main ethnic groups in Amsterdam, and (b) to what extent ethnic disparities in perinatal health outcomes could be explained by ethnic differences in maternal risk factors during pregnancy.
This thesis further focused on psychobiological processes involved in (ethnic disparities in) perinatal health outcomes. Maternal psychosocial problems during pregnancy may be an important risk factor for adverse perinatal health outcomes; a potential psychobiological pathway involves the hypothalamic-pituitary-adrenal (HPA) axis and the hormone cortisol. Evidence for the potentially adverse effects of maternal psychosocial problems is, however, inconsistent and research on the underlying psychoendocrinological pathway is scarce. Psychobiological processes involved in fetal and infant health outcomes also include fetal origins of infant psychosocial health. Excessive infant crying can be considered a reasonable indicator of infant psychosocial stress; fetal origins of excessive infant crying are largely unknown and more research is required. The third and fourth objectives of this thesis were to examine (c) the association between maternal psychosocial well-being during pregnancy (as indicated by self-report scales and the biomarker cortisol) and perinatal health outcomes, and (d) the association between maternal risk factors during pregnancy, in particular maternal nutritional status, and infant psychosocial well-being (as indicated by infant crying behavior).
To study the objectives of this thesis, data from the Amsterdam Born Children and their Development (ABCD) study were used. The ABCD study is a prospective cohort study that aims to explore the association between (ethnic differences in) maternal risk factors during pregnancy and (ethnic disparities in) perinatal and infant health outcomes. From January 2003 till March 2004, all pregnant women in Amsterdam (n = 12 373) were approached to enroll in the ABCD study at their first prenatal visit to an obstetric caregiver (± 12th pregnancy week). Women were asked to fill out a pregnancy-questionnaire covering sociodemographic information, lifestyle and (psychosocial) health (n = 8266; response rate 67%), and to donate a blood sample for the analysis of nutrients and hormones (n = 4389). Three months after delivery, women received a baby-questionnaire covering the (psychosocial) health of mother and baby (n = 5218). To enhance the inclusion of pregnant women from the main ethnic groups in Amsterdam, i.e. the Dutch, Surinamese, Antillean/Aruban, Turkish, Moroccan and Ghanaian group, questionnaires
were in Dutch, English, Turkish or Arabic and Turkish- or Arabic-speaking trained female interviewers were available for oral administration of the questionnaire. Perinatal health outcomes were available from the Amsterdam Youth Health Care Registration and the Dutch Perinatal Registration.
We started by studying perinatal health outcomes and potential risk factors among the main ethnic groups in Amsterdam. In Chapter 2, ethnic disparities in the birthweight distribution of term-born infants were explored and to what extent the disparities could be explained by ethnic differences in maternal risk factors during pregnancy. Because birthweight differences can be either constitutional (‘natural’ limited growth potential) or pathological (growth restricted), respectively the explanatory role of constitutional and environmental factors was examined. Ethnic groups were categorised into a Dutch group and first and second generation Surinamese, Antillean, Turkish, Moroccan, Ghanaian and other non-Dutch groups. Only singleton live births with ≥37.0 weeks gestation were included for analysis (n = 7118). Mean birthweight ranged from 3223 g (second generation Surinamese newborns) to 3548 g (Dutch newborns). Linear regression analysis showed that the ethnic minority groups had significantly lower mean birthweights compared to the Dutch group. Adjustment for constitutional factors (fetal gender, maternal age, parity, maternal height) substantially reduced the ethnic disparities in birthweight (corrected for gestational age), while adjustment for environmental factors (education, cohabitation status, maternal BMI, smoking, alcohol consumption, depressive symptoms, work stress) provided little additional explanation. The birthweight disparities between Dutch newborns and Turkish, Moroccan and other non-Dutch newborns could largely be explained by, in particular, the constitutional determinants, limiting the opportunities for prevention. On the contrary, birthweight disparities between the Dutch group and the first and second generation Surinamese and Ghanaian and first generation Antillean group could not be fully explained by the explored maternal risk factors; mean birthweights remained respectively 98, 159, 121 and 102 g lower. Future research should explore whether the unexplained birthweight disparities reflect pathological growth restriction or a constitutional limited growth potential, by searching for unexplored environmental factors, cumulative risk effects and genetic components, and by exploring weight-specific mortality and morbidity rates per ethnic group.
In Chapter 3, ethnic disparities in the prevalence of preterm birth (PTB) were explored and to what extent the disparities could be explained by ethnic differences in maternal risk factors during pregnancy. In addition, the effect of a cumulation of risk factors on ethnic disparities in PTB prevalence was explored. PTB (<37.0 weeks gestation) was divided into spontaneous preterm births (SPB) and iatrogenic (medically indicated) preterm births (IPB). Only singleton live births with ≥24.0 weeks gestation were included for analysis (n = 7604). Ethnicity was based on the country of birth of the pregnant woman’s mother: Dutch, Surinamese, Antillean, Turkish, Moroccan, Ghanaian, and other non-Dutch. In our sample, a mean PTB prevalence of 5.5% was observed, ranging from 4.1% (Moroccan group) to 11.0% (Ghanaian group); 68% was SPB, 20% IPB and 12% had an unknown subtype. Compared to the Dutch group, significant
ethnic disparities in PTB prevalence were not observed for the Turkish, Moroccan and other non-Dutch group; these groups even had a lower PTB risk. On the contrary, Surinamese, Ghanaian and Antillean women had a higher PTB risk, in particular for IPB with odds ratios of 2.1, 3.2 and 3.6 respectively after full adjustment [risk factors: maternal age, parity, maternal BMI, education, cohabitation status, maternal smoking, alcohol consumption, depressive symptoms, physical heavy work, previous obstetric complications (PTB, abortion, miscarriage/stillbirth), hypertensive disorder, and indicators of vaginal infectious disease]. Ethnic minority groups had a higher cumulation of risk factors (‘cumulative risk score’ ranging from 2.1 to 3.7) compared to the Dutch group (score of 1.8). Adjustment for the cumulative risk score considerably decreased the PTB risk among Surinamese, Ghanaian and Antillean women. The predictive accuracy of the cumulative risk score was, however, too low to be used for individual risk selection. In conclusion, the PTB risk of Dutch, Turkish, Moroccan and other non-Dutch women seemed favourable compared to the PTB risk of Surinamese, Antillean and Ghanaian women; this could potentially be explained by cumulative risk effects, unexplored risk factors, obstetric care practices or (epi)genetic influences.
Next, we studied the potentially negative effects of maternal psychosocial problems during pregnancy on perinatal health outcomes. In Chapter 4, we explored how maternal psychosocial problems relate to whether or not a woman continues to smoke during pregnancy. Only women who smoked before pregnancy were included in this study (n = 1947). Women were categorized as quitters (62%) or non-quitters (38%) based on their self-report of smoking behavior during early pregnancy. Non-quitters reported significantly higher levels of depressive symptoms, anxiety, pregnancy-related anxiety, job strain and physical/sexual violence, but not of parenting stress. After adjustment for maternal age, parity, ethnicity, education, cohabitation status, the amount of smoking before pregnancy, smokers in the environment and desirability of pregnancy, women with very low or high levels of pregnancy-related anxiety, exposure to physical/sexual violence or high job strain had a significantly higher risk for continued smoking during pregnancy. Although a causal relationship could not be verified because of the cross-sectional design of data collection, if confirmed, smoking cessation programs for pregnant women should include the management of psychosocial problems.
In Chapter 5, the association between maternal depressive symptoms during pregnancy and major perinatal health outcomes, i.e. preterm birth (PTB), small-for-gestational-age (SGA), Apgar score and child loss, was examined. In addition, it was examined whether the associations were mediated by maternal smoking during pregnancy and whether the associations differed between ethnic groups. Ethnic groups were categorized into a Dutch, Creole, Turkish and Moroccan group. Maternal depressive symptoms were measured by the validated Center for Epidemiologic Studies Depression scale (CES-D). Multiple births were excluded leaving a baseline sample of 8050 women for analysis. Thirty percent (n = 2465) of the women reported high levels of depressive symptoms. The prevalence of perinatal outcomes in our sample was: 5.4% (PTB), 12.3% (SGA), 1.5% (low Apgar score) and 1.4% (child loss), with a higher prevalence among women with high levels of depressive symptoms. After adjustment for maternal age, parity, ethnicity, education, pre-pregnancy BMI,
hypertension, alcohol and drug use and maternal smoking, high levels of maternal depressive symptoms were significantly associated with SGA and a low Apgar score, but not with PTB and child loss. The observed associations could possibly be explained by a psychoendocrinological pathway or through mediation by maternal risk behaviours such as smoking, poor nutrition, a lack of prenatal health care visits or the use of antidepressants. Interaction effects between maternal depressive symptoms and ethnic background on perinatal health outcomes were insignificant, however, stratified analysis by ethnic background suggested an excess risk for Creole women, possibly explained by higher vascular activity and reactivity to stress among blacks.
In Chapter 6, a potential psychobiological pathway into adverse perinatal health outcomes involving the hormone cortisol was explored. We examined the association of maternal cortisol concentration during pregnancy with maternal psychosocial problems and with fetal growth, as indicated by offspring birthweight and SGA risk. Total maternal cortisol concentration was determined in serum and standardized for time of day and gestational age at blood sampling. After excluding multiple births, preterm births, pre-gestational diabetes mellitus, steroid medication and gestational age at blood sampling >20.0 weeks, 2810 women were included in the analysis. We observed that a higher maternal cortisol concentration was associated with lower offspring birthweight and a higher SGA risk; the associations were, however, insignificant after adjustment for gestational age at birth, infant gender, ethnicity, maternal age, parity, maternal BMI and smoking. Stratified analysis for time of day at blood sampling suggested that the association was only present among cortisol concentrations in blood samples collected ≤09:00h (n = 94). Maternal cortisol concentration was unrelated to the level of maternal depressive symptoms, anxiety, pregnancy-related anxiety, parenting stress and job strain and exposure to physical/sexual violence. In conclusion, our data could not support the hypothesis that maternal psychosocial problems affect fetal growth through maternal cortisol levels. Our measurement of cortisol, however, left much to be desired. Future research should include multiple cortisol measurements at different time frames of the diurnal cortisol rhythm.
Lastly, we explored fetal origins of infant psychosocial well-being, as indicated by infant crying behavior. The origins of excessive infant crying are largely unknown; an early nutritional origin has not yet been explored. In Chapter 7, we examined the association of maternal vitamin B-12 and folate status during pregnancy with excessive infant crying. The amount of infant crying was reported by the mother a few months after birth and excessive infant crying was defined as crying ≥3 hours/day on average in the past week. Vitamin B-12 and folate concentrations were determined in serum and standardized for gestational age at blood sampling. Multiple births were excluded leaving 2921 (vitamin B-12) and 2622 (folate) women for analysis. In our sample, the prevalence of excessive infant crying was 3.4%. We observed a significant association between a low maternal vitamin B-12 concentration and excessive infant crying, also after adjustment for maternal age, parity, ethnicity, education, smoking and maternal psychological problems. Stratified analysis suggested a stronger association among women with high levels of depressive symptoms, anxiety and/or pregnancy-related anxiety during pregnancy. Maternal folate concentration was
not associated with excessive infant crying. We discussed two potential mechanisms through which vitamin B-12 may affect infant crying behavior: (1) vitamin B-12 and folate are involved in the methionine-homocysteine metabolism, which is essential in fetal neurodevelopment; (2) vitamin B-12 is involved in melatonin synthesis; excessive infant crying could be a symptom of an immature melatonin/sleep-wake rhythm; this could hypothetically be amplified by an immature cortisol rhythm caused by maternal psychological problems. In conclusion, this study provided first evidence for an early nutritional origin in excessive infant crying; further research is, however, necessary to support this preliminary evidence. If confirmed, prenatal checks of pregnant women should be extended with diagnosing and treating for vitamin B-12 deficiency.
In Chapter 8, we discussed the main findings presented in this thesis, methodological issues such as the operationalisation of ethnicity and maternal and infant psychosocial well-being, and implications for clinical practice such as the need for interventions covering psychosocial problems, multiple risk factors and cultural and language differences. Three main concluding points were made:
First, ethnic disparities in perinatal health outcomes were observed among the main ethnic groups in Amsterdam, i.e. the Dutch, Surinamese, Antillean, Ghanaian, Turkish and Moroccan group: mean offspring birthweight was lower among all ethnic minority groups compared to the Dutch group while the risk for PTB, a low Apgar score and child loss was only higher for the Surinamese, Antillean and Ghanaian group. Although the ethnic disparities could, to some extent, be explained by single conventional maternal risk factors during pregnancy, a cumulation effect of multiple risk factors – especially seen among the ethnic minority groups – seemed to provide a better explanation. It seems furthermore reasonable that (epi)genetic influences are involved as offspring from mainly African descent had unexplainable lower birthweights and a higher PTB risk. To reduce ethnic disparities in perinatal health outcomes, comprehensive antenatal health promotion programs should be developed and implemented that cover not only important proximal risk behaviors like smoking, poor nutrition and work stress, but also distal factors like lack of social support or financial problems. As long as there is no convergence in risk behaviors and their determinants between ethnic minority groups and the native Dutch group, we recommend the implementation of antenatal programs that account for cultural-specific beliefs and norms.
Second, one of the important risk factors for adverse perinatal health outcomes appeared to be maternal psychosocial stress. The prevalence of perinatal mortality and morbidity was higher among women who reported high levels of depressive symptoms during pregnancy, though only the prevalence of SGA and a low Apgar score appeared to be statistically significantly higher when adjusting for relevant covariates. The relationship between maternal psychosocial stress during pregnancy and offspring outcome could at first be ascribed to a small mediation effect of maternal smoking behavior. Alternatively, elevated levels of the stress hormone cortisol potentially influence fetal development, however, only limited evidence for this psychobiological pathway could be provided because of methodological difficulties related to the measurement of cortisol. While ethnic minority groups in our sample had a high prevalence of both psychosocial problems
and adverse perinatal health outcomes, differences in psychosocial well-being did not seem to be responsible for ethnic disparities in perinatal health outcomes. We hypothesized that pregnant women from African descent might be more susceptible for detrimental effects of psychosocial stress on offspring outcomes through ‘allostatic load’ and/or higher vascular (re)activity. As psychosocial stress is becoming a huge public health problem, obstetric caregivers should be aware of the potentially negative effects on fetal health. We recommend to screen pregnant women for psychosocial stress as part of routine antenatal care or preferably, as part of preconceptional care, provided that efficacious treatment programs are available.
Third, psychosocial stress in infancy as expressed by excessive crying behavior may have its origin during fetal development. Maternal nutritional status during pregnancy appeared to be of importance as an association was observed between a low vs. high maternal vitamin B-12 status during pregnancy and excessive crying behavior of the infant in the first months of life; this might, in theory, be ascribed to fetal (dys)maturation of the sleep-wake / melatonin rhythm. Although these observations offer opportunities for prevention and treatment of excessively crying infants, the preliminary results first need to be replicated, preferably with prospective measurements of infant crying behavior through diaries or audiotape recordings.
Een van de hoofddoelen van dit proefschrift was het bestuderen van de perinatale gezondheid in de multi-etnische stad Amsterdam in Nederland. Zoals beschreven staat in hoofdstuk 1 bestudeert de perinatale gezondheidsepidemiologie de gezondheid van moeder en kind tijdens de zwangerschap, de geboorte en de periode vlak na de geboorte. Belangrijke indicatoren van een ongezonde intra-uteriene (d.w.z. in de baarmoeder) omgeving zijn perinatale sterfte en ziekte van het kind; perinatale ziekte (bijv. vroeggeboorte, laag geboortegewicht) kan ingrijpende gevolgen hebben voor de gezondheid op lange termijn, zoals een verhoogd risico op hart- en vaatziekten. Wereldwijd verschilt de prevalentie van perinatale sterfte en ziekte tussen etnische groepen. Dit kan – in theorie – worden verklaard door etnische verschillen in risicofactoren bij de moeder tijdens de zwangerschap. De eerste doelstellingen van dit proefschrift waren (a) het onderzoeken van perinatale gezondheidsuitkomsten bij de belangrijkste etnische groepen in Amsterdam, en (b) in welke mate etnische verschillen in perinatale gezondheidsuitkomsten verklaard zouden kunnen worden door etnische verschillen in maternale risicofactoren tijdens de zwangerschap. Dit proefschrift richtte zich verder op het onderzoeken van psychobiologische processen die van invloed kunnen zijn op (etnische verschillen in) perinatale gezondheidsuitkomsten. Psychosociale problemen bij de moeder tijdens de zwangerschap kunnen een belangrijke risicofactor zijn voor slechtere perinatale gezondheidsuitkomsten; bij een mogelijk psychobiologisch proces dat hieraan ten grondslag kan liggen zijn de hypothalamic-pituitary-adrenal (HPA) as en het hormoon cortisol betrokken. Bewijs voor eventuele schadelijke effecten van maternale psychosociale problemen is echter inconsistent en onderzoek naar het onderliggende psycho-endocrinologische proces is schaars. Wanneer we kijken naar psychobiologische processen die een rol spelen bij de gezondheid van de foetus en het kind valt de foetale oorsprong van de psychosociale gezondheid van het kind hier ook onder. Overmatig huilgedrag van een kind kan wel gezien worden als een aannemelijke graadmeter van psychosociale stress bij het kind; risicofactoren tijdens de foetale ontwikkeling voor overmatig huilgedrag van een kind zijn grotendeels onbekend en meer onderzoek is nodig. De derde en vierde doelstelling van dit proefschrift waren (c) het onderzoeken van het verband tussen de psychosociale gezondheid van de moeder tijdens de zwangerschap (gemeten door middel van zelfrapportage en de biomarker cortisol) en perinatale gezondheidsuitkomsten, en (d) het verband tussen maternale risicofactoren tijdens de zwangerschap, met name de voedingsstatus van de moeder, en de psychosociale gezondheid van het kind (met als indicator het huilgedrag van het kind). Om de doelstellingen van dit proefschrift te onderzoeken zijn data van de Amsterdam Born Children and their Development (ABCD) studie gebruikt. De ABCD-studie is een grootschalig, prospectief cohortonderzoek dat als doel heeft om het verband tussen (etnische verschillen in) maternale risicofactoren tijdens de zwangerschap en (etnische verschillen in) de gezondheid van het kind bij geboorte en op latere leeftijd te onderzoeken. Van januari 2003 tot maart 2004 werden alle zwangere vrouwen in Amsterdam (n = 12 373) tijdens hun eerste bezoek aan de verloskundige, gynaecoloog of huisarts (± 12e zwangerschapweek) benaderd om mee te doen
aan de ABCD-studie. De vrouwen werden gevraagd om een uitgebreide vragenlijst in te vullen over sociaaldemografische factoren, leefstijl en (psychosociale) gezondheid, en een bloedmonster af te staan voor de analyse van voedingsstoffen en hormonen; 8266 vrouwen (67%) vulden de vragenlijst in en van deze vrouwen stonden 4389 (53%) een bloedmonster af. Drie maanden na de bevalling ontvingen de vrouwen een vragenlijst over de (psychosociale) gezondheid van de moeder en de baby; deze werd ingevuld door 5218 vrouwen. Om verzekerd te zijn van de participatie van zwangere vrouwen vanuit de belangrijkste etnische groepen in Amsterdam, namelijk de Nederlandse, Surinaamse, Antilliaanse/Arubaanse, Turkse, Marokkaanse en Ghanese groep, waren de vragenlijsten in het Nederlands, Engels, Turks of Arabisch en waren er Turks- of Arabischsprekende vrouwelijke interviewers beschikbaar voor een mondelinge afname van de vragenlijst. De perinatale gezondheidsuitkomsten werden beschikbaar gesteld door de Jeugdgezondheidszorg in Amsterdam en de Perinatale Registratie Nederland.
Allereerst hebben we de perinatale gezondheidsuitkomsten en potentiële risicofactoren bij de belangrijkste etnische groepen in Amsterdam onderzocht. In hoofdstuk 2 werden de etnische verschillen in het geboortegewicht van à term geboren kinderen (geboren vanaf de 37e zwangerschapsweek) onderzocht en in welke mate de verschillen konden worden verklaard door etnische verschillen in maternale risicofactoren tijdens de zwangerschap. Omdat verschillen in het geboortegewicht zowel constitutioneel (‘natuurlijk’ verminderd groeipotentieel) als pathologisch (beperkte groei door externe factoren) kunnen zijn, werd de verklarende rol van achtereenvolgens constitutionele en omgevingsfactoren onderzocht. De etnische groepen werden gecategoriseerd in een Nederlandse groep en 1e en 2e generatie Surinaamse, Antilliaanse, Turkse, Marokkaanse, Ghanese en overige niet-Nederlandse groepen. Alleen levend geboren eenlingen, geboren vanaf de 37e zwangerschapsweek, werden meegenomen in de analyse (n = 7118). Het gemiddeld geboortegewicht varieerde van 3223 gr. (2e generatie Surinaamse pasgeborenen) tot 3548 gr. (Nederlandse pasgeborenen). Lineaire regressie analyse toonde aan dat de etnische minderheidsgroepen een – rekening houdend met de verschillen in zwangerschapsduur – statistisch significant lager geboortegewicht hadden dan de Nederlandse groep. Na correctie voor constitutionele factoren [geslacht van het kind, leeftijd van de moeder, pariteit (d.w.z. aantal eerdere zwangerschappen) en de lengte van de moeder] werden de etnische verschillen in geboortegewicht substantieel kleiner, terwijl correctie voor omgevingsfactoren [opleidingsniveau, thuissituatie (alleenwonend of samenwonend), body mass index (BMI) van de moeder, rookgedrag, alcoholgebruik, depressieve symptomen, en werkstress] weinig aanvullende verklaring bood. De verschillen in het geboortegewicht tussen de Nederlandse baby’s en Turkse, Marokkaanse en overige niet-Nederlandse baby’s konden grotendeels worden verklaard door in het bijzonder de constitutionele factoren; hierdoor zijn de mogelijkheden voor preventie van een laag geboortegewicht onder deze groepen beperkt. Daarentegen konden de verschillen in het geboortegewicht tussen de Nederlandse groep en de 1e en 2e generatie Surinaamse en Ghanese en 1e generatie Antilliaanse groep niet volledig worden verklaard door de onderzochte maternale risicofactoren; de geboortegewichten bleven respectievelijk 98, 159, 121 en 102 gr. lager.
Toekomstig onderzoek zal moeten uitwijzen of de onverklaarde verschillen in geboortegewicht een pathologische groeibeperking of een natuurlijk verminderd groeipotentieel weergeven, door te kijken naar de mogelijke effecten van nog niet eerder onderzochte omgevingsfactoren, naar de cumulatieve effecten van meerdere risicofactoren, naar genetische componenten, en naar geboortegewichtspecifieke sterfte- en ziektecijfers per etnische groep. In hoofdstuk 3 werden de etnische verschillen in de prevalentie van vroeggeboorte onderzocht en in welke mate de verschillen verklaard konden worden door etnische verschillen in maternale risicofactoren tijdens de zwangerschap. Verder werd ook het effect van een cumulatie van meerdere risicofactoren op de etnische verschillen in vroeggeboorte onderzocht. De vroeggeboortes (d.w.z. geboren voor de 37e zwangerschapsweek) werden ingedeeld in twee categorieën: spontane vroeggeboortes en medisch ingezette vroeggeboortes. Alleen levend geboren eenlingen, geboren na een zwangerschapsduur van minstens 24 weken, werden meegenomen in de analyse (n = 7604). Etniciteit werd gebaseerd op het geboorteland van de moeder van de zwangere vrouw: Nederlands, Surinaams, Antilliaans, Turks, Marokkaans, Ghanees en overig niet-Nederlands. In onze studiegroep eindigde 5,5% van de zwangerschappen in een vroeggeboorte, variërend van 4,1% bij de Marokkaanse groep tot 11% bij de Ghanese groep; 68% van de vroeggeboortes was een spontane vroeggeboorte, 20% een medisch ingezette vroeggeboorte en van 12% was het type vroeggeboorte onbekend. Vergeleken met de Nederlandse groep werden er geen significante etnische verschillen in de prevalentie van vroeggeboortes waargenomen voor de Turkse, Marokkaanse en overig niet-Nederlandse groep; deze groepen hadden zelfs een lager risico op een vroeggeboorte. Daarentegen hadden de Surinaamse, Ghanese en Antilliaanse vrouwen een hoger risico op een vroeggeboorte, vooral op medisch ingezette vroeggeboortes met odds ratios (een maat voor het risico) van respectievelijk 2.1, 3.2 en 3.6 na correctie voor alle onderzochte risicofactoren [maternale leeftijd, pariteit, BMI, opleidingsniveau, thuissituatie, rookgedrag, alcoholgebruik, depressieve symptomen, lichamelijk zwaar werk, eerdere zwangerschapscomplicaties (vroeggeboorte, abortus, miskraam, doodgeboren kind), hoge bloeddruk/pre-eclampsie en aanwijzingen voor een vaginale infectieziekte]. De etnische minderheidsgroepen hadden een hogere cumulatie van de risicofactoren (met een ‘cumulatieve risicoscore’ van 2.1 tot 3.7) vergeleken met de Nederlandse groep (score van 1.8). Na correctie voor de cumulatieve risicoscore nam het risico op een vroeggeboorte aanzienlijk af bij de Surinaamse, Ghanese en Antilliaanse vrouwen. De voorspellende waarde van de cumulatieve risicoscore was echter te laag om te kunnen gebruiken voor individuele risico selectie. Concluderend kunnen we zeggen dat het risico op een vroeggeboorte bij Nederlandse, Turkse, Marokkaanse en overig niet-Nederlandse vrouwen gunstig is vergeleken met het risico bij Surinaamse, Antilliaanse en Ghanese vrouwen; dit kan mogelijk verklaard worden door cumulatieve effecten van meerdere risicofactoren, door niet eerder onderzochte risicofactoren, door de praktijk van de obstetrische gezondheidszorg, of door (epi)genetische invloeden.
Vervolgens hebben we de mogelijk negatieve effecten van psychosociale problemen bij de moeder tijdens de zwangerschap op perinatale gezondheidsuitkomsten bestudeerd. In hoofdstuk 4 onderzochten we of maternale psychosociale problemen in verband stonden met of een vrouw
doorgaat met roken tijdens de zwangerschap of niet. Alleen vrouwen die voor hun zwangerschap rookten werden meegenomen in dit onderzoek (n = 1947). De vrouwen werden ingedeeld als stoppers (62%) of doorrokers (38%), gebaseerd op hun rookgedrag in het begin van de zwangerschap dat ze rapporteerden in de vragenlijst. De doorrokers rapporteerden in de vragenlijst significant hogere niveaus van depressieve symptomen, angst, zwangerschapsgerelateerde angst, werkstress en lichamelijk/seksueel geweld, maar niet van opvoedingsstress. Na correctie voor maternale leeftijd, pariteit, etniciteit, opleidingsniveau, thuissituatie, de hoeveelheid sigaretten voor de zwangerschap, rokers in de omgeving, en de wenselijkheid van de zwangerschap, bleek dat vrouwen met een hele lage of hoge score voor zwangerschapsgerelateerde angst, met blootstelling aan lichamelijk/seksueel geweld of met een hoge werkstress een significant hoger risico hadden op doorroken tijdens de zwangerschap. Een causaal verband kan echter niet hard worden gemaakt vanwege het cross-sectionele design van de dataverzameling. Wanneer het onderzochte verband bevestigd wordt is het belangrijk voor de praktijk dat stoppen-met-roken cursussen voor zwangere vrouwen aandacht schenken aan het omgaan met psychosociale problemen. In hoofdstuk 5 werd het verband tussen depressieve symptomen bij de moeder tijdens de zwangerschap en belangrijke perinatale gezondheidsuitkomsten, namelijk vroeggeboorte, small-for-gestational-age (SGA, geboortes waarbij het geboortegewicht valt binnen de laagste 10 procent van de geboortegewichtverdeling, rekening houdend met het geslacht van het kind, de zwangerschapsduur en het aantal eerdere zwangerschappen van de moeder), lage Apgar-score (maat voor de algemene toestand van een pasgeborene) en sterfte van het kind, onderzocht. Daarnaast werd ook onderzocht of de onderzochte verbanden via het rookgedrag van de moeder tijdens de zwangerschap liepen en of de verbanden verschilden tussen de etnische groepen. De etnische groepen werden verdeeld in een Nederlandse, Creoolse, Turkse en Marokkaanse groep. Maternale depressieve symptomen werden gemeten in de vragenlijst aan de hand van de Center for Epidemiologic Studies Depression scale (CES-D). Na het excluderen van meerlingen uit het onderzoek bleven er 8050 vrouwen over voor de analyse. Dertig procent van de vrouwen (n = 2465) rapporteerde een hoog niveau van depressieve symptomen. De prevalentie van perinatale uitkomsten in onze studiegroep was: 5,4% (vroeggeboorte), 12,3% (SGA), 1,5% (lage Apgar-score) en 1,4% (sterfte van het kind), met een hogere prevalentie bij de vrouwen met een hoog niveau van depressieve symptomen. Na correctie voor de leeftijd van de moeder, pariteit, etniciteit, opleidingsniveau, BMI, hoge bloeddruk / pre-eclampsie, alcohol en drugsgebruik, en het rookgedrag van de moeder, bleek dat vrouwen met een hoog niveau van depressieve symptomen een significant hoger risico hadden op een SGA-baby of een baby met een lage Apgar-score, maar niet op een vroeggeboorte of sterfte van het kind. De waargenomen verbanden kunnen mogelijk worden verklaard door een psycho-endocrinologisch proces of via risicogedrag van de moeder zoals roken, ongezonde voedingsgewoontes, een gebrek aan prenatale gezondheidscontroles of het gebruik van antidepressiva. Interactie-effecten (d.w.z. dat het verband tussen twee variabelen wordt bepaald door een derde variabele) van maternale depressieve symptomen en etniciteit op perinatale gezondheidsuitkomsten waren niet significant, maar gestratificeerde analyses (d.w.z. aparte analyses voor verschillende categorieën van een bepaalde variabele) per etnische groep suggereerden een
extra hoog risico voor Creoolse vrouwen op slechte perinatale gezondheidsuitkomsten bij een hoog niveau van depressieve symptomen; dit kan mogelijk verklaard worden door een hoge vasculaire activiteit en reactiviteit volgend op stress bij blacks (vrouwen van Afrikaanse oorsprong met een donkere huidskleur). In hoofdstuk 6 werd onderzocht of een psychobiologisch proces waarbij het hormoon cortisol (ook wel het stresshormoon genoemd) betrokken is mogelijk kan leiden tot slechtere perinatale gezondheidsuitkomsten. We onderzochten het verband tussen de cortisolconcentratie van de moeder tijdens de zwangerschap en zowel maternale psychosociale problemen als foetale groei (met als indicatoren het geboortegewicht van het kind en het risico op SGA). De totale cortisolconcentratie van de moeder werd gemeten in serum en gestandaardiseerd voor het tijdstip en de zwangerschapsduur op het moment van bloedafname. Na het excluderen van meerlingen, vroeggeboortes, vrouwen met diabetes mellitus voor de zwangerschap, gebruik van steroïden en vrouwen met een zwangerschapsduur van meer dan 20 weken op moment van bloedafname, werden 2810 vrouwen meegenomen in de analyse. Uit de analyse bleek dat een hogere cortisolconcentratie van de moeder gerelateerd was aan een lager geboortegewicht en een hoger risico op een SGA-kind; deze verbanden waren echter niet statistisch significant na correctie voor zwangerschapsduur bij de geboorte, geslacht van het kind, etniciteit en leeftijd, pariteit, BMI en rookgedrag van de moeder. Gestratificeerde analyses voor het tijdstip waarop de bloedafname plaatsvond suggereerden dat het verband alleen aanwezig was wanneer gekeken werd naar cortisolconcentraties in bloedmonsters die voor of om 09:00uur afgenomen waren (n = 94). De maternale cortisolconcentratie was niet gerelateerd aan het niveau van maternale depressieve symptomen, angst, zwangerschapsgerelateerde angst, opvoedingsstress, werkstress of blootstelling aan lichamelijk/seksueel geweld. Concluderend konden onze data niet de hypothese ondersteunen dat maternale psychosociale problemen de foetale groei beïnvloeden door middel van het maternale cortisolniveau. Onze meting van de cortisolconcentratie liet echter nog veel te wensen over. Toekomstig onderzoek zou meerdere cortisolmetingen op verschillende momenten binnen het dagritme van cortisol afgifte moeten meenemen.
Tenslotte onderzochten we de foetale oorsprong van de psychosociale gezondheid van het kind, met als indicator het huilgedrag van het kind. De oorsprong van overmatig huilgedrag is grotendeels onbekend; de mogelijkheid dat voeding in een zeer vroeg stadium al van invloed kan zijn is nog niet onderzocht. In hoofdstuk 7 onderzochten we het verband tussen de vitamine B12 en foliumzuurstatus van de moeder tijdens de zwangerschap en overmatig huilgedrag van het kind. De moeder rapporteerde in de vragenlijst een paar maanden na de geboorte hoeveel haar kind huilde. Overmatig huilgedrag werd gedefinieerd als gemiddeld ≥ 3 uur per dag huilen in de afgelopen week. Vitamine B12 en foliumzuurconcentraties werden gemeten in serum en gestandaardiseerd voor de zwangerschapsduur op het moment van bloedafname. Na het excluderen van meerlingen bleven er 2921 (vitamine B12) en 2622 (foliumzuur) vrouwen over voor de analyse. In onze studiegroep was de prevalentie van overmatig huilgedrag 3,4%. De analyses lieten een significant verband zien tussen een laag vitamine B12 gehalte van de moeder en overmatig huilgedrag van het kind,
ook na correctie voor leeftijd, pariteit, etniciteit, opleidingsniveau, rookgedrag en psychologische problemen van de moeder. Gestratificeerde analyses suggereerden een sterker verband bij vrouwen met een hoog niveau van depressieve symptomen, angst en/of zwangerschapsgerelateerde angst. Er werd geen verband gevonden tussen maternale foliumzuurconcentratie en overmatig huilgedrag van het kind. We bediscussieerden twee potentiële verklaringen voor hoe vitamine B12 invloed kan hebben op het huilgedrag van het kind: (1) vitamine B12 en foliumzuur zijn betrokken bij de methionine-homocysteïne stofwisseling, welke essentieel is voor de foetale neuro-ontwikkeling, en (2) vitamine B12 is betrokken bij de melatonine synthese; overmatig huilgedrag van een kind kan een symptoom zijn van een onontwikkeld melatonine/slaap-waak ritme; dit kan hypothetisch gezien weer versterkt worden door een onontwikkeld cortisol ritme veroorzaakt door maternale psychosociale problemen. Samengevat gaf deze studie de eerste aanwijzingen dat voeding in een zeer vroeg stadium al van invloed kan zijn op overmatig huilgedrag van een kind; verder onderzoek is echter wel nodig om deze aanwijzingen te bevestigen. Wanneer dit het geval zal blijken te zijn zou dit voor de praktijk betekenen dat prenatale controles bij zwangere vrouwen uitgebreid zouden moeten worden met de diagnose en behandeling van vrouwen met een vitamine B12 tekort.
In hoofdstuk 8 bediscussieerden we de belangrijkste bevindingen die in dit proefschrift gepresenteerd zijn, methodologische kwesties zoals de operationalisatie van etniciteit en de psychosociale gezondheid van moeder en kind, en toepassingen voor de klinische praktijk zoals de behoefte aan interventies op het gebied van psychosociale problemen en vrouwen met meerdere risicofactoren tegelijkertijd, waarbij er aandacht is voor verschillen in cultuur en taal. Onze resultaten leidden tot de volgende drie hoofdconclusies: (1) De perinatale gezondheidsuitkomsten verschilden tussen de belangrijkste etnische groepen in Amsterdam, namelijk de Nederlandse, Surinaamse, Antilliaanse, Ghanese, Turkse en Marokkaanse groep: het gemiddelde geboortegewicht van de kinderen was lager bij alle etnische minderheidsgroepen vergeleken met de Nederlandse groep, terwijl het risico op een vroeggeboorte, een lage Apgar-score en sterfte van het kind alleen hoger was voor de Surinaamse, Antilliaanse en Ghanese groep. Hoewel de etnische verschillen in zekere mate verklaard konden worden door individuele risicofactoren bij de moeder tijdens de zwangerschap, een cumulatie van meerdere risicofactoren – wat vooral werd gezien bij de etnische minderheidsgroepen – leek een betere verklaring te bieden. Verder lijkt het aannemelijk dat er ook (epi)genetische invloeden bij betrokken zijn, aangezien vooral de kinderen van Afrikaanse afkomst onverklaarbare lage geboortegewichten en een hoger risico op vroeggeboorte hadden. Om etnische verschillen in perinatale gezondheidsuitkomsten te verminderen zouden er brede prenatale gezondheidsprogramma’s ontwikkeld en geïmplementeerd moeten worden, die niet alleen ingaan op belangrijke proximale (‘dichtbij’) risicofactoren zoals roken, slechte voedingsgewoontes en werkstress, maar ook op distale (‘veraf ’) factoren zoals een gebrek aan sociale ondersteuning of financiële problemen. Zolang het risicogedrag en de determinanten van het gedrag tussen de etnische minderheidsgroepen en de Nederlandse groep niet gelijk is raden
we aan om prenatale programma’s te implementeren die rekening houden met cultuurspecifieke normen en attitudes. (2) Een van de belangrijke risicofactoren voor slechtere perinatale gezondheidsuitkomsten blijkt psychosociale stress bij de moeder te zijn. De prevalentie van perinatale sterfte en ziekte was hoger bij vrouwen die een hoog niveau van depressieve symptomen tijdens de zwangerschap rapporteerden, echter na correctie voor relevante factoren waren alleen de prevalentie van een SGA-geboorte en een lage Apgar-score statistisch significant hoger. Het verband tussen psychosociale stress bij de moeder tijdens de zwangerschap en de gezondheid van het kind kan allereerst worden toegeschreven aan een klein mediërend effect van het rookgedrag van de moeder tijdens de zwangerschap. Een andere optie is dat een hoog niveau van het stresshormoon cortisol de foetale ontwikkeling mogelijk kan beïnvloeden, maar bewijs voor dit psychobiologisch proces was beperkt vanwege de methodologische moeilijkheden bij het meten van cortisol. Hoewel etnische minderheidsgroepen in ons onderzoek een hoge prevalentie hadden van zowel psychosociale problemen als slechte perinatale gezondheidsuitkomsten, leken de verschillen in de psychosociale gezondheid niet verantwoordelijk voor de etnische verschillen in perinatale gezondheidsuitkomsten. We stelden de hypothese dat zwangere vrouwen van Afrikaanse afkomst misschien meer gevoelig zijn voor de schadelijke effecten van psychosociale stress op de gezondheid van hun kind middels een proces genaamd ‘allostatic load’ (d.w.z. negatieve lichamelijke reacties als gevolg van chronische blootstelling aan stress) en/of een hogere vasculaire (re)activiteit. Omdat psychosociale stress steeds meer een groot algemeen gezondheidsprobleem aan het worden is zouden verloskundigen alert moeten zijn op de eventuele negatieve effecten ervan op de gezondheid van het kind. We adviseren om zwangere vrouwen te screenen op psychosociale stress als onderdeel van de standaard prenatale zorg of nog liever, als onderdeel van de preconceptionele zorg, op voorwaarde dat er effectieve behandelprogramma’s beschikbaar zijn. (3) Psychosociale stress bij het kind zoals geuit door middel van overmatig huilgedrag kan zijn oorsprong al hebben tijdens de foetale ontwikkeling. De maternale voedingsstatus tijdens de zwangerschap blijkt belangrijk te zijn aangezien er een verband werd gevonden tussen een lage vs. hoge maternale vitamine B12 status tijdens de zwangerschap en overmatig huilgedrag van het kind tijdens de eerste maanden van zijn leven; dit kan, in theorie, worden toegeschreven aan de foetale (onder)ontwikkeling van het slaap-waak/melatonine ritme. Hoewel deze observaties mogelijkheden bieden voor het voorkomen en behandelen van overmatig huilgedrag bij kinderen moeten deze resultaten eerst worden herhaald, het liefst met behulp van prospectieve metingen van het huilgedrag van het kind middels dagboeken of geluidsopnames.
11ß-HSD2 11β-hydroxysteroid dehydrogenase type 2ABCD Amsterdam Born Children and their DevelopmentANOVA Analysis of varianceb Beta (unstandardized regression coefficient)BMI Body Mass IndexCAR Cortisol awakening responseCBG Corticotrophin-binding globulinCES-D Center for Epidemiologic Studies Depression scaleCI Confidence intervalCV Coefficient of variationDNA Deoxyribonucleic acidDSM-IV Diagnostic and Statistical Manual of Mental Disorders – version IVGA Gestational ageGP General practitionerHPA-axis Hypothalamic-pituitary-adrenal axisIPB Iatrogenic preterm birthIQR Interquartile rangeIUGR Intrauterine growth restrictionLBW Low birthweightOR Odds ratioPDH Parenting Daily HasslesPRAQ-R Pregnancy Related Anxiety Questionnaire – Revised versionPRN Dutch Perinatal Registration (Perinatale Registratie Nederland)PTB Preterm birthQ QuintileSD Standard deviationSGA Small-for-gestational-ageSPB Spontaneous preterm birthSPSS Statistical Package for the Social SciencesSTAI State-Trait Anxiety InventoryVBBA Work Experience and Appreciation Questionnaire (Vragenlijst
Affiliations of the authors who contributed to chapters of this thesis:
Prof. dr. G.J. Bonsel Erasmus Medical Center, Department of Obstetrics and Prenatal Care, Rotterdam, The Netherlands; Academic Medical Center, University of Amsterdam, Department of Social Medicine, Amsterdam, The Netherlands
Prof.dr. P. Cuijpers VU University Amsterdam, Department of Clinical Psychology, Amsterdam, The Netherlands
Dr.ir. M. van Eijsden Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, Amsterdam, The Netherlands
Dr. A.E. Hesselink Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, Amsterdam, The Netherlands
Dr. M.N.M. van Poppel VU University Medical Center, EMGO Institute, Department of Public and Occupational Health, Amsterdam, The Netherlands
Dr. T.J. Roseboom Academic Medical Center, University of Amsterdam, Department of Clinical Epidemiology and Biostatistics, Amsterdam, The Netherlands
Drs. A.C. Snijders Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, Amsterdam, The Netherlands
Dr. T.G.M. Vrijkotte Academic Medical Center, University of Amsterdam, Department of Social Medicine, Amsterdam, The Netherlands
Dr. M.F. van der Wal Public Health Service of Amsterdam, Department of Epidemiology, Documentation and Health Promotion, Amsterdam, The Netherlands
‘Natus est!’ ‘Het is geboren!’ Eindelijk heeft u dan mijn proefschrift in handen. In zekere zin kan het promotietraject wel vergeleken worden met een periode van zwangerschap en bevalling (in ieder geval met de mijne). Na wat opstartproblemen in het eerste trimester zijn het tweede en derde trimester een verademing; hoewel het belangrijk is om de voortgang enigszins te bevorderen en te bewaken en om bij tijd en wijlen een bezoek te brengen aan de promotor (‘iemand die voor iets ijvert’; dit gaat ook op voor de verloskundige), is het vooral ook zaak om voldoende rust te nemen en te genieten. Deze rust wordt opeens bruut verstoord door het opkomen van de weeën, oftewel het afronden van het manuscript. Dan dringt opeens het besef door dat je in de periode ervoor wel degelijk naar iets hebt toegewerkt. Zonder planning ga je de bevalling in, want het moment van de geboorte staat nog niet vast. De weeën worden sterker en de pijn wordt botgevierd op de promotor. Er ontstaat drang om te persen: tijd om af te ronden. En dan is daar de geboorte! Zielsgelukkig wordt het proefschrift vastgehouden en bewonderd; de pijn is vergeten. Nou ja, voor eventjes dan, want er volgen nog wat naweeën: de promotieplechtigheid. Daar zal de navelstreng worden doorgeknipt en wordt de kraamvrouw doctor.
Dit proefschrift was niet tot stand gekomen zonder de hulp van vele personen. Allereerst wil ik alle zwangere vrouwen, verloskundigen, gynaecologen, huisartsen en laboranten bedanken die onbaatzuchtig hebben meegewerkt aan de ABCD-studie. Mooi hoe we samen kunnen werken aan een betere gezondheid van het (on)geboren kind!
Dan wil ik mijn (co)promotoren bedanken. Gouke, ik heb enorm veel van je mogen leren in de afgelopen jaren. Dankzij jouw onderwijs ben ik gegroeid van een pas afgestudeerd groentje naar een wetenschapper. Jij hebt me geleerd hoe een juiste onderzoeksbril eruit ziet. Ondanks dat het je regelmatig ontbrak aan tijd en structuur, waren de momenten dat we elkaar spraken nuttig en plezierig. Ik zal je wijze raad (en niet te vergeten je stroopwafels) missen. Marcel, bedankt dat ik in de afgelopen jaren altijd bij je terecht kon voor vragen of advies. Je ontspannen manier van werken waren voor mij fijn en vertrouwd. Met een hart voor preventie heb jij me geleerd om m’n onderzoeksbril ook scherp te stellen op het belang voor de dagelijkse praktijk. Pim, ondanks dat we denk ik meer uit onze samenwerking hadden kunnen halen, vond ik de contacten die we hebben gehad fijn; bedankt voor je vertrouwen in mij.
Naast mijn promotoren heb ik enorm veel geleerd van mijn ABCD-collega’s Tanja en Manon. Tanja, ik heb het enorm gewaardeerd dat je altijd tijd vond om met me mee te denken, of het nu in Amsterdam was of op een vrije dag om de hoek in Leidsche Rijn. Ik heb genoten van onze gesprekken in de trein, op de fiets en natuurlijk in Chili! Manon, het was erg fijn om jou als collega-ABCD-promovenda en co-auteur te hebben; dankzij je heldere inzicht en schrijfwijze wist je mijn artikelen altijd inzichtelijker te maken. Dank ook voor al je hoe-maak-ik-een-proefschrift-adviezen.
Reinoud, Aimée, Marieke H., Marieke B. en Eva, het was leuk om met jullie op te trekken binnen het ABCD-team en in Chili; dames, ik kijk uit naar jullie proefschriften! Anne, Arlette, Tessa en Mireille, bedankt voor het meedenken en meeschrijven! Verder wil ik iedereen bedanken die op wat voor manier heeft meegewerkt aan de ABCD-studie en dus indirect aan dit proefschrift: Swaas voor de secretariële klussen en meer, Hans als datamanager, Miranda en Anita voor de koppeling van de ABCD-data met de data van de Perinatale Registratie Nederland, promovenda en stagiaires voor het uitpluizen, kopiëren en invoeren van data uit medische dossiers en groeiboekjes, en vele anderen.
Ik wil de leescommissie bedanken voor het lezen en kritisch beoordelen van mijn proefschrift: prof. dr. Bea van den Bergh, prof. dr. Joris van der Post, dr. Tessa Roseboom, prof. dr. Régine Steegers-Theunissen, prof. dr. Karien Stronks, prof. dr. Arnoud Verhoeff en dr. Carolina de Weerth.
Ik wil alle EDG collega’s en in het bijzonder al mijn gezellige kamergenootjes (Annemarie, Thijs, Joanne, Floor, Fatima, Adèle, Henriëtte) bedanken voor de geweldige tijd die ik bij de GGD Amsterdam heb gehad; zonder al die gezellige koffie-met-taart-kletspraatjes, lunches, pepernoten en bemoedigende woorden weet ik niet of ik het had volgehouden. Speciaal wil ik mijn (ex-)collega’s, vriendinnen en paranimfen Lotte en Femke bedanken; wat heerlijk dat ik altijd bij jullie terecht kon om sam en te lachen, te huilen, te balen of te stralen! Daarom heb ik jullie graag op 30 september aan mijn zijde.
Dear Aida, thank you for your creativity and sensitivity that you used in designing the cover and invitation belonging to this thesis. I am really pleased with the result! Lieve Hanna, wat fijn dat ik gebruik mocht maken van je vertaal-skills! Misschien is dit een mooi opstapje naar een baan als ZZP-vertaler.
Beste IRAS-collega’s, bedankt dat ik bij jullie een nieuw plekje heb mogen vinden en dat ik de mogelijkheid kreeg om rustig mijn promotie af te maken. Lieve vrienden, vriendinnen, familieleden, RijnWaardianen en een ieder die ik niet bij name heb genoemd maar waar ik wel mee ben opgetrokken in de afgelopen jaren, bedankt voor jullie vriendschap, gezelligheid, en meeleven! Tjebbe & Tas, ik heb genoten op Mallorca en in Toscane; thanks voor jullie vriendschap, bemoedigingen en de zorg voor mijn lieve dochter!
Mijn laatste dankwoorden wil ik graag richten aan hen die mij het meest dierbaar zijn. Michiel, Mirjam & Neriah, Hanna & Matthieu, opa & oma Roebers, pa & ma Goedhart, Meninka & Paulo, veel dank voor jullie liefde, enthousiasme en betrokkenheid! Pap en mam, ik ben jullie geweldig dankbaar voor alle liefde en steun, niet alleen in de afgelopen jaren, maar ook daarvoor. Het is altijd weer heerlijk om ‘thuis’ te komen!
Mijn allerliefste Anna, jij bent het zonnetje in mijn leven! Jouw schaterlach en mooie koppie doen alle zorgen als sneeuw voor de zon verdwijnen. Jij zet me altijd weer even met beide benen op de grond en laat me zien wat het belangrijkste is in het leven: samen genieten van de simpelste dingen. Dit heb je trouwens niet van een vreemde, want je papa is er ook zo een; jullie tweetjes zijn echt het beste wat me is overkomen! Nu dit proefschrift klaar is heb ik weer meer rust om gezellig tijd met z’n drieën door te brengen (America, here we come!). Lieve Frank, heel veel dank voor je liefde, steun en vertrouwen in alle goede en kwade dagen! Niet alleen hief jij met mij het glas tijdens de vreugdevolle momenten, ook voorzag jij mij van een glaasje lekkers tijdens de momenten van stress of vermoeidheid. Dank ook voor alle nachtelijke uurtjes dat we nog samen hebben zitten printen of dat je mij ervan weerhield om mijn laptop uit het raam te gooien (toch handig zo’n ICT-mannetje). Je vrolijke, optimistische en ontspannen karakter maken dat het elke dag weer genieten is om samen met jou te zijn. En wanneer ik wat teveel ontspande gaf je me gewoon een liefdevol schopje onder m’n kont; zo is dit proefschrift uiteindelijk toch nog klaar gekomen.
Geertje de Wolf zag op 28 juni 1982 het levenslicht op 19 hoog te Zeist. Na de basisschool in Utrecht met goed gevolg te hebben doorlopen vertrok ze naar het oosten van het land om daar in 2000 aan het Greijdanus College te Zwolle haar VWO diploma te behalen. Vervolgens vertrok ze naar de zuidelijkste stad van Nederland Trajectum ad Mosam, Maastricht, om daar Gezondheidswetenschappen te gaan studeren aan de Universiteit Maastricht. Haar stageperiode bracht ze door aan de University of KwaZulu-Natal in Durban, Zuid-Afrika, om daar onderzoek te doen naar de relatie tussen de kwaliteit van de zorg voor tuberculose patiënten in de ziekenhuizen in KwaZulu-Natal en de behandelingsuitkomsten bij deze patiënten. Op 28 juni 2005 ontving ze haar bul voor de masters Gezondheidsvoorlichting en Epidemiologie. Tijdens haar studieperiode was ze actief lid van diverse commissies binnen de studentenverenigingen Trajectum ad Mosam en Ichthus en binnen de kerkelijke gemeente. Verder werkte ze als student-assistent bij de vakgroep Gezondheidsvoorlichting aan de Universiteit Maastricht. Vanaf april 2005 ging ze aan de slag als promovenda bij de GGD Amsterdam (cluster Epidemiologie, Documentatie en Gezondheidsbevordering) en het Academisch Medisch Centrum (afdeling Sociale Geneeskunde) te Amsterdam, waar ze betrokken was bij de Amsterdam Born Children and their Development (ABCD) studie. Binnen deze geboortecohortstudie deed ze voornamelijk onderzoek naar etnische verschillen in geboorteuitkomsten en de effecten van maternale stress tijdens de zwangerschap op geboorteuitkomsten. Deze onderzoeken zijn gebundeld in dit proefschrift getiteld ‘Perinatal health epidemiology in multi-ethnic Amsterdam: psychobiological processes.’ Tijdens het promotietraject heeft Geertje haar onderzoeken op verschillende internationale en nationale congressen gepresenteerd. Vanaf januari 2010 is ze werkzaam als postdoc bij het Institute for Risk Assessment Sciences (IRAS), Universiteit Utrecht, waar ze betrokken is bij een groot multinationaal onderzoek naar de mogelijke relatie tussen blootstelling aan draadloze communicatietechnologieën en het risico op hersentumoren bij jonge mensen. Op 29 april 2005 is ze getrouwd met Frank Goedhart. Samen hebben ze een dochter genaamd Anna (2008).
157
________________________________________________________ Curriculum Vitae - Publications
Publications
Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in preterm birth and its subtypes: the effect of a cumulative risk profile. BJOG 2008;115(6):710-19.
Goedhart G, van Eijsden M, van der Wal MF, Bonsel GJ. Ethnic differences in term birthweight: the role of constitutional and environmental factors. Paediatric and Perinatal Epidemiology 2008;22(4):360-8.
Goedhart G, van Eijsden M, van der Wal MF, Vrijkotte TGM, Bonsel GJ. Prematuriteit en laag geboortegewicht: wat zijn de risicofactoren? (Prematurity and low birthweight: what are the risk factors?). Vroeg 2007;24(1):6-9.
Goedhart G, van der Wal MF, Cuijpers P, Bonsel GJ. Psychosocial problems and continued smoking during pregnancy. Addictive Behaviors 2009;34(4):403-6.
Goedhart G, Vrijkotte TGM, Roseboom TJ, van der Wal MF, Cuijpers P, Bonsel GJ. Maternal cortisol and offspring birthweight: results from a large prospective cohort study. Psychoneuroendocrinology 2010;35(5):644-52.
Fassaert T, de Wit MAS, Goedhart G, Verhoeff AP. Etniciteit als variabele in onderzoek: beschouwing vanuit epidemiologisch perspectief. (Ethnicity as a variable in health research: considerations from an epidemiological perspective). Cultuur Migratie Gezondheid 2009;6(1):22-9.
Goedhart G, Snijders AC, Hesselink AE, van Poppel MN, Bonsel GJ, Vrijkotte TGM. Maternal depressive symptoms in relation to perinatal mortality and morbidity: results from a large multi-ethnic cohort study. Psychosomatic Medicine; in press.
Goedhart G, van der Wal MF, van Eijsden M, Bonsel GJ. Maternal vitamin B-12 and folate status during pregnancy and excessive infant crying. Submitted.
