Randomized Trial of a Parenting Intervention for Very Preterm Infants: Outcome at 2 Years Samantha Johnson, PhD, CPsychol, Andrew Whitelaw, MD, FRCPCH, Cris Glazebrook, PhD, CPsychol, Chrissie Israel, RN, RSCN, Rebecca Turner, PhD, Ian R. White, MSc, Tim Croudace, PhD, Franca Davenport, MSc, and Neil Marlow, DM, FMedSci Objectives To determine the efficacy of a neonatal parenting intervention for improving development in very preterm infants. Study design A cluster-randomized, controlled trial with a cross-over design and washout period was conducted in 6 neonatal centers. Two hundred thirty-three babies <32 weeks’ gestation were recruited (intervention = 112; control = 121). Intervention families received weekly Parent Baby Interaction Programme (PBIP) sessions during neonatal intensive care unit admission and up to 6 weeks after discharge. Control families received standard care. All 195 infants remaining in the study at 24 months’ corrected age were assessed by psychologists blinded to group allocation. Results There was no significant difference in Mental Development Index (0.9 points; 95% CI, 5.0, 3.2) or Psy- chomotor Development Index (2.5; 3.3, 8.4) scores between the intervention and control groups and no significant effect of intervention on Mental Development Index or Psychomotor Development Index scores for subgroups dichotomized by gestational age (<28 weeks/$28 weeks), parity (1st/other child) or mother’s cohabiting status (supported/unsupported). Conclusions There was no effect of PBIP on infant development at 2 years’ corrected age. Parenting interven- tions may be better delivered after discharge or targeted for preterm infants with high biological and social risk. (J Pediatr 2009;155:488-94). P reterm birth places a child at high risk for neuropsychological impairments, learning difficulties, and behavior problems later in life. 1,2 Although preterm interruption to the developing brain and associated perinatal insults can account for much of the impairment observed, 3,4 the developmental vulnerability conferred by preterm birth may be mediated by environmental experience. 5,6 Hence, developmental care practices are often implemented by clinical staff to minimize the ad- verse impact of the neonatal intensive care unit (NICU) environment. 7 Early maternal factors also may moderate outcomes for these children. The psychological stress associated with preterm birth 8-10 adversely affects attachment, maternal sensitivity to baby’s cues, and mother-infant interaction, 8,9 factors that are predictive of infant development and long term outcome. 11,13-16 Therefore, interventions designed to enhance the parent-infant relationship may have positive effects on infant cognitive outcomes. Results regarding the efficacy of early parenting interventions for preterm infants are conflicting and inconclusive, however. 10 Although some trials report modest beneficial effects of intervention on cognitive development, 11,12 others have failed to detect an effect 13 or have shown benefits only after adjustment for confounders. 21,22 A recent meta-analysis found a significant effect of early parenting programs on cognitive outcomes in infancy and early childhood, with some evidence of a greater effect for interventions that were focused on facilitation of the parent-infant relationship. 10 We conducted a randomized, controlled trial of a neonatal parenting intervention for very preterm infants and have previously reported that this did not improve maternal outcomes at three months corrected age. 14 We have subsequently fol- lowed up this cohort to determine whether there are longer-term effects of intervention on infant outcomes. We hypothesized that the Parent Baby Interaction Programme (PBIP) would improve cognitive development at 2 years’ corrected age in very preterm infants. From the University of Nottingham (S.J.), Nottingham, United Kingdom; Department of Clinical Science (A.W., C.I., F.D.), University of Bristol, Bristol, United Kingdom; School of Community Health Sciences (C.G.), University of Nottingham, Nottingham, United Kingdom; MRC Biostatistics Unit (R.T., I.R.W.), Cambridge, United Kingdom; Department of Psychiatry (T.C.), University of Cambridge, Cambridge, United Kingdom; and Institute for Women’s Health (S.J., N.M.), University College London, London, United Kingdom. Supported by the Health Foundation, London, United Kingdom The authors have no conflicts of interest to declare. 0022-3476/$ - see front matter. Copyright Ó 2009 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2009.04.013 BSID-II Bayley Scales of Infant Development, 2nd Edition IMD Index of Multiple Deprivation MDI Mental Development Index NICU Neonatal intensive care unit PDI Psychomotor Development Index PBIP Parent Baby Interaction Programme PSI Parenting Stress Index RN Research nurse 488
Transcript
Randomized Trial of a Parenting Intervention for Very Preterm Infants:Outcome at 2 Years
RSCN, Rebecca Turner, PhD, Ian R. White, MSc, Tim Croudace, PhD, Franca Davenport, MSc, and Neil Marlow, DM, FMedSci
Objectives To determine the efficacy of a neonatal parenting intervention for improving development in verypreterm infants.Study design A cluster-randomized, controlled trial with a cross-over design and washout period was conductedin 6 neonatal centers. Two hundred thirty-three babies <32 weeks’ gestation were recruited (intervention = 112;control = 121). Intervention families received weekly Parent Baby Interaction Programme (PBIP) sessions duringneonatal intensive care unit admission and up to 6 weeks after discharge. Control families received standardcare. All 195 infants remaining in the study at 24 months’ corrected age were assessed by psychologists blindedto group allocation.Results There was no significant difference in Mental Development Index (�0.9 points; 95% CI,�5.0, 3.2) or Psy-chomotor Development Index (2.5;�3.3, 8.4) scores between the intervention and control groups and no significanteffect of intervention on Mental Development Index or Psychomotor Development Index scores for subgroupsdichotomized by gestational age (<28 weeks/$28 weeks), parity (1st/other child) or mother’s cohabiting status(supported/unsupported).Conclusions There was no effect of PBIP on infant development at 2 years’ corrected age. Parenting interven-tions may be better delivered after discharge or targeted for preterm infants with high biological and social risk.(J Pediatr 2009;155:488-94).
Preterm birth places a child at high risk for neuropsychological impairments, learning difficulties, and behavior problemslater in life.1,2 Although preterm interruption to the developing brain and associated perinatal insults can account formuch of the impairment observed,3,4 the developmental vulnerability conferred by preterm birth may be mediated by
environmental experience.5,6 Hence, developmental care practices are often implemented by clinical staff to minimize the ad-verse impact of the neonatal intensive care unit (NICU) environment.7 Early maternal factors also may moderate outcomes forthese children. The psychological stress associated with preterm birth8-10 adversely affects attachment, maternal sensitivity tobaby’s cues, and mother-infant interaction,8,9 factors that are predictive of infant development and long term outcome.11,13-16
Therefore, interventions designed to enhance the parent-infant relationship may have positive effects on infant cognitiveoutcomes.
Results regarding the efficacy of early parenting interventions for preterm infants are conflicting and inconclusive,however.10 Although some trials report modest beneficial effects of intervention on cognitive development,11,12 others havefailed to detect an effect13 or have shown benefits only after adjustment for confounders.21,22 A recent meta-analysis found asignificant effect of early parenting programs on cognitive outcomes in infancy and early childhood, with some evidence ofa greater effect for interventions that were focused on facilitation of the parent-infant relationship.10
We conducted a randomized, controlled trial of a neonatal parenting intervention for very preterm infants and havepreviously reported that this did not improve maternal outcomes at three months corrected age.14 We have subsequently fol-lowed up this cohort to determine whether there are longer-term effects of intervention on infant outcomes. We hypothesizedthat the Parent Baby Interaction Programme (PBIP) would improve cognitive development at 2 years’ corrected age in verypreterm infants.
From the University of Nottingham (S.J.), Nottingham,United Kingdom; Department of Clinical Science (A.W.,C.I., F.D.), University of Bristol, Bristol, United Kingdom;School of Community Health Sciences (C.G.), Universityof Nottingham, Nottingham, United Kingdom; MRCBiostatistics Unit (R.T., I.R.W.), Cambridge, UnitedKingdom; Department of Psychiatry (T.C.), University ofCambridge, Cambridge, United Kingdom; and Institutefor Women’s Health (S.J., N.M.), University CollegeLondon, London, United Kingdom. Supported by theHealth Foundation, London, United Kingdom
The authors have no conflicts of interest to declare.
0022-3476/$ - see front matter. Copyright � 2009 Mosby Inc.
All rights reserved. 10.1016/j.jpeds.2009.04.013
BSID-II Bayley Scales of Infant Development, 2nd Edition
IMD Index of Multiple Deprivation
MDI Mental Development Index
NICU Neonatal intensive care unit
PDI Psychomotor Development Index
PBIP Parent Baby Interaction Programme
PSI Parenting Stress Index
RN Research nurse
488
Vol. 155, No. 4 � October 2009
Methods
The study design (ISRCTN56341521) has been described indetail previously.14 Briefly, a cluster-randomized, controlledtrial with a cross-over design was conducted in 6 neonatalcenters (incorporating 7 NICUs) in 2 regions of the UnitedKingdom (3 each in the South West and Trent regions).Approval was obtained from the South West multi-centerethics committee and the local research ethics committeesat each center. Within regions, 2 centers were paired on thebasis of deprivation indices, and the third center from eachregion formed the final pair. As a measure of deprivation,an Index of Multiple Deprivation (IMD)15 score was assignedfor each family that is derived from data on deprivation at thesmall area level in 7 domains including income, employment,and crime. Scores for England range from 0.59 (leastdeprived) to 86.36 (most deprived), with a median of17.02. Recruitment was conducted in 2 phases. For the firstphase, 1 center from each pair was randomized by the tossof a coin to the intervention with the other center random-ized to control. During this phase participants were recruitedfor 6 months in the Trent region and 7 months in the SouthWest to achieve target recruitment. After a 3-month washoutperiod implemented to eliminate contamination of the con-trol group, each center crossed over to the opposite treatmentcondition. During this second phase, participants wererecruited for 7 months in Trent and 6 months in the SouthWest.
Babies born at <32 weeks’ gestational age in participatingcenters were recruited. Babies with illness incompatible withlife and families resident outside the study catchment areawere excluded. Recruitment was conducted by 7 researchnurses (RNs) who obtained written informed consent fromparents as soon as possible after birth. Given the study design,neither RNs nor parents were blinded to group allocationbefore recruitment. Babies were followed up at dischargeand at 3-months’14 and 24 months’ corrected age. At 24months’ corrected age, 1 of 2 psychologists who were blindedto treatment group allocation contacted parents to schedulea home visit during which a standardised developmental testwas administered. One week after the home visit, parentswere sent a letter detailing the child’s test results.
InterventionThe PBIP16 provides structured parental support during theneonatal period to facilitate attachment, enhance parent-infant interaction, sensitize parents to their baby’s cues, facil-itate parents’ confidence in identifying and meeting theirbaby’s needs, and to educate parents in developmental careprinciples. It is delivered through a framework of activitiesin 4 areas comprising discursive (eg, infant development),tactile (eg, handling), verbal (eg, talking to the baby), andobservational (eg, identifying behavioral states and baby’scues) sessions.
For this study, the mother was the primary recipient of theintervention. The intervention was delivered by RNs trained
in PBIP before study commencement. Regular meetings wereheld with the clinical trial manager (C.I.) to ensure skillsmaintenance throughout the intervention period. PBIPactivities were designed to be delivered in weekly 1-hoursessions beginning from the first weeks after birth up toa maximum of 6 postdischarge sessions. The interventionwas directed only to parents. Regular NICU nursing staffwere neither recipients of the intervention nor trained inPBIP principles and continued to deliver standard carethroughout the study periods at their center. This wasrequired to enable the cross-over design to be implementedin which standard care alone was provided by NICU staffduring the control period. Families in the control groupreceived standard care.
MeasuresDevelopmental outcome at 24 months’ corrected age wasassessed using the Bayley Scales of Infant Development 2ndEdition (BSID-II).17 This is a norm-referenced test that yieldsstandardised scores (mean, 100; SD, 15; range, 50 to 150) forcognitive (Mental Development Index; MDI) and motordevelopment (Psychomotor Development Index; PDI).MDI scores were the primary outcome measure for thisstudy, and PDI scores were a secondary outcome measure.
Children who could not be assessed because of severedisability and those whose index scores fell below test limitswere assigned a nominal index score of 49 (1 point belowthe basal test score) for quantifying severely delayed out-come.18 Psychologists were formally trained in test adminis-tration before study commencement and achieved excellentinter-rater reliability (MDI, 97% item-by-item agreement;PDI, 94% agreement) in 11 randomly selected assessmentsscored simultaneously by both examiners throughout theperiod of data collection. Index scores were used to classifydevelopmental delay using conventional SD-banded cut-offs (mild, �1 to �2 SD; moderate, �2 to �3 SD; severe,<� 3 SD). RNs collected clinical and demographic informa-tion from mothers’ and babies’ medical notes and throughparental interviews during NICU admission.
Statistical AnalysesThe predefined trial primary outcomes were the ParentingStress Index (PSI)19 at 3 months’ corrected age and theBSID-II MDI at 24 months’ corrected age. Results for the PSIhave been reported previously.14 In power calculations, a sam-ple size of 172 would have 90% chance of detecting a 0.5 SD dif-ference in PSI scores at P < .05. The target sample size wasincreased to 250 to allow for possible clustering.14
Data were double-entered, verified, and analyzed usingStata and SPlus. A prespecified 2-stage analysis method wasdeveloped for the study and is described in detail elsewhere.20
A brief outline of these analyses follows. In the first stage,differences (d) in mean outcome between the 2 periods ofrecruitment were calculated for each cluster. In the secondstage, t tests were used to compare the mean of d betweenclusters that received the experimental intervention in Phase1 versus Phase 2. To allow for differences in sample sizes,
489
THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 155, No. 4
t tests were weighted by n1n2/(n1 + n2), where n1 and n2 arethe number of infants in Phases 1 and 2. Subgroup analyseswere prespecified for the effect of intervention on MDI scoresby gestational age (<28 weeks vs $28 weeks), parity (firstbaby vs second or other child), and mother’s cohabiting sta-tus (supported vs unsupported mothers). Subgroup analyseswere performed using linear mixed models with fixed effectsfor intervention subgroup, cluster subgroup, and phase sub-group interactions, and random effects for period withincluster.20 In sensitivity analyses, an additional random effectwas added to the mixed models to allow for similaritiesbetween twins; this did not substantially change any results.Data were analyzed on an intention-to-treat basis. Furtheranalysis adjusted for maternal age, parity, IMD scores, occu-pational status, smoking during pregnancy and gestationalage at birth as linear effects, and cohabiting status, sex, andmultiple birth as binary variables.
Results
The derivation of this cohort has been described previously.14
Of 496 babies born at <32 weeks at participating centers, 156were excluded and 33 died before obtaining consent. Of theremaining 307 babies, 233 (76%) were recruited to the study:112 (81% of those eligible) to the intervention group and 121(72% of those eligible) to the control. There were no signifi-cant differences in sex, birth weight, gestational age andmultiple births between babies who were and were notrecruited to the study. The intervention and control groupswere well matched for baseline characteristics (Table I).
Detailed information regarding intervention delivery hasbeen published previously.14 Briefly, 108 of 112 babies inthe intervention group (96%) received at least 1 PBIP session,with median 8 sessions received (IQR, 5 to 11). More sessionswere delivered in the NICU (median, 5; IQR, 2 to 7) than athome (median, 2; IQR, 1 to 4). Gestational age was correlatednegatively with number of sessions received in the NICU (rs =�0.35, P < .001) and positively with sessions received afterdischarge (rs = 0.31, P < .001). Conversely, length of admis-sion was positively correlated with the number of predis-charge sessions (rs = 0.51, P < .001) and negatively withpostdischarge sessions (rs = �0.23, P < .014).
At 24-months’ corrected age, 92 (82%) infants remained inthe intervention group and 103 (85%) in the control group.BSID-II assessments were administered for all these infants(age at assessment: Intervention mean, 24.47 months; SD,1.22 months; range, 24 to 32 months; Control mean, 24.06months; SD, 0.52 months; range, 20 to 25 months) (Figure 1).
Primary Outcome: Cognitive Development (MDIScores)A nominal MDI score were assigned for 3 and 4 infants in theintervention and control groups, respectively, for childrenwith severe disability or those with scores below test limits.An MDI score was excluded for 1 child for whom a validassessment could not be obtained because of difficulties in
490
interpreting the child’s native language during testing.Unadjusted MDI scores were approximately 0.5 SD belowthe population mean (100), and 8% to 9% had scores inthe moderate to severely delayed range (Table II). Therewas no significant difference in MDI scores between the in-tervention and control groups: mean MDI was 0.9 pointslower in the intervention group (�0.9 points; 95% CI,�5.0 to 3.2). These results were little changed on adjustmentfor baseline covariates (�0.4 points; 95% CI, �5.8 to 5.0).
Intervention effects on MDI scores did not differ signifi-cantly by gestational age (P = .12), parity (P = .52), or cohab-iting status (P = .19). The effect of intervention was notsignificant in any of the 6 subgroups, but nonsignificantimprovements in MDI scores were seen for babies born at<28 weeks (7.3 points; 95% CI, �4.3 to 18.9) and for unsup-ported mothers (9.5 points; 95% CI, -6.8 to 25.8) (Figure 2).
Secondary Outcome: Motor Development (PDIScores)A nominal PDI score were assigned for 3 and 6 infants in theintervention and control groups, respectively, and PDI scoreswere excluded for 2 children in whom the assessment couldnot be completed due to language or behavioral difficulties(both in the intervention group). Unadjusted mean PDIscores were approximately 0.5 SD below the populationmean with 11% of the control group and 5% of the
Table I. Baseline characteristics of participants recruitedto intervention and control groups
Maternal characteristics
Intervention (n = 99) Control (n = 111)
Age, median (IQR) 30 (24-34) 29 (25-35)White/European n (%) 80 (81) 101 (91)Single/divorced, n (%) 16 (16) 20 (18)First baby, n (%) 53 (54) 64 (58)Highest educational level,* n (%)
University degree or above 18 (19) 18 (17)No further education 51 (53) 49 (47)
Smoked while pregnant,z n (%) 13 (16) 13 (15)IMD score, x median (IQR) 14 (9-23) 14 (9-22)
Infant characteristics
Intervention (n = 112) Control (n = 121)
Male, n (%) 50 (45) 61 (50)Singleton birth, n (%) 81 (72) 96 (79)Gestational age, median (range) 28.5 (23-31) 29.0 (23-31)Gestational age, <28 wk, n (%) 35 (31) 29 (24)Birth weight, median (IQR) 1121 (900-1408) 1220 (935-1499)Birth weight <1000 g, n (%) 38 (34) 38 (31)Days in hospital, median (IQR) 56 (39-86) 53 (39-78)
Occupational status was classified using the National Statistics Socio-Economic Classification,Class I/II Managerial/Professional Occupations, Class III/IV Intermediate occupations/smallemployers, Class V-VIII Lower supervisory, and routine occupations/unemployed.*Intervention, n = 96; control, n = 104.†Intervention, n = 94; control, n = 103.zIntervention, n = 81; control, n = 83.xIndex of Multiple Deprivation score.24
Johnson et al
October 2009 ORIGINAL ARTICLES
Figure 1. Participant flow to 24 months’ corrected age.
intervention group being classified with moderate to severelydelayed psychomotor development. There was no significantdifference in PDI scores between the intervention and controlgroup: Mean PDI score was 2.5 points higher in the interven-tion group than in the control group (95% CI, �3.3 to 8.4).These results were little changed on adjustment for baselinecovariates (3.1 points; 95% CI, �3.3 to 9.6). Similarly, inter-vention effects did not differ significantly between subgroupsof infants classified by gestational age (P = .14), parity (P =.63), and mother’s cohabiting status (P = .66). Although
Randomized Trial of a Parenting Intervention for Very Preterm I
the intervention effect on PDI scores was not significant forany of the subgroups, PDI scores were higher for babiesborn at <28 weeks (9.1 points; 95% CI, �2.4 to 18.6) andfor babies of single mothers (5.7 points; 95% CI, �9.1 to20.5) (Figure 2).
Discussion
In this rigorous randomized, controlled trial, there was nosignificant effect of the PBIP16 on cognitive development at
Table II. Developmental outcomes at 24 months’ corrected age
Control (n = 103) Intervention (n = 91) Mean difference (95% CI)† P value
BSID-II MDI Mean (SD)* 92.9 (18.2) 91.3 (18.6) �0.9 (�5.0 to 3.2) 0.57Normal development n (%) 69 (67%) 58 (64%)Mildly delayed n (%) 26 (25%) 24 (26%)Moderately delayed n (%) 3 (3%) 5 (5%)Severely delayed n (%) 5 (5%) 4 (4%)BSID-II PDIz Mean (SD)* 92.0 (16.8) 94.8 (15.9) 2.5 (�3.3 to 8.4) 0.30Normal development n (%) 77 (75%) 64 (71%)Mildly delayed n (%) 15 (15%) 21 (23%)Moderately delayed n (%) 5 (5%) 1 (1%)Severely delayed n (%) 6 (6%) 4 (4%)
BSID-II, Bayley Scales of Infant Development, 2nd Edition; MDI, Mental Development Index; PDI, Psychomotor Development Index. Outcomes for planned analyses are shown in bold. MDI, primaryoutcome; PDI, secondary outcome. Normal development, scores $1 SD; mildly delayed, scores �1 SD to�2 SD; moderately delayed, scores �2 SD to �3 SD; severely delayed, score < � 3 SD.*Raw mean (SD) unadjusted for cluster cross-over design.†Mean difference adjusted for cluster cross-over design.zIntervention, n = 90.
nfants: Outcome at 2 Years 491
THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 155, No. 4
2 years’ corrected age in very preterm infants. This result is per-haps not entirely unexpected because trials investigating the ef-ficacy of parenting interventions for very preterm infants havegenerally produced variable and conflicting results. Althoughsome studies have reported beneficial effects of interventionon cognitive outcomes,11,12,21 others have failed to detect an ef-fect13,22 or, at best, have found significant differences only afteradjustment for confounding factors.23,24 We also failed to de-tect an effect of PBIP on motor development at 2 years’ cor-rected age, which is commensurate with other studies.10
We previously assessed parenting stress, maternal respon-sivity to baby’s cues, and mother-infant interaction and foundno significant effect of PBIP on these outcomes at 3 months’corrected age.14 Because PBIP did not improve these variablesthat have been shown to mediate outcomes for preterminfants,11,13 it is also unsurprising that we did not detect a ben-eficial effect of intervention on long-term infant development.Because other studies have reported positive effects, it may beconcluded that this particular parenting program was ineffec-tive in reducing maternal stress, facilitating mother-infantinteraction and thus in improving infant outcomes.
However, it may be argued that PBIP is a valuable programbut that we were unable to detect an effect in this study.Although PBIP was designed to be delivered in hourly ses-sions from birth up to 6 weeks after discharge, the actualnumber of sessions received was determined by the mother’savailability on the NICU and her willingness to participate inthe intervention at home. The mean number of sessions was
Figure 2. Intervention effect on BSID-II MDI (primary out-come) and PDI scores (secondary outcome) for subgroups ofvery preterm infants at 24 months’ corrected age. Subgroupswere dichotomized on the basis of a priori criteria (gestationalage, <28 weeks vs $28 weeks; parity, first child vs other child;cohabiting status, supported vs unsupported mothers).
492
8.04 (SD, 4.34), which translated to an average of approxi-mately 6 hours in total.14 This is considerably less than the10 sessions that might be expected for a baby born at 31 com-pleted weeks plus 6 days of gestation (discharged at 36weeks), the most mature in our study. Studies that havereported beneficial results have shown a dose-dependencyeffect of intervention25,26 and have typically implementedmore extended intervention periods.26-28 The failure ofPBIP to improve outcomes might therefore be attributed toa limited intervention ‘‘dose’’ as a result of the lower than ex-pected number of sessions received. It is possible that the lowparticipation in intervention was a result of the PBIP beingdelivered by RNs who were not observed by parents to be reg-ular NICU staff. Parents receiving the intervention may havefelt ‘‘singled out’’ in comparison to other mothers on theunit, and this may have had negative effects on participationin the intervention. It is possible that the uptake and there-fore dose of such an intervention would be increased by in-tegrating it as a part of routine care and thus beingdelivered by routine NICU staff. It was not possible for theintervention to be delivered in this manner in the presentstudy given the cross-over design.
The low number of sessions received at home may beparticularly important because contemporary studies thathave reported positive effects on maternal29 and infantoutcomes have used interventions that commenced just be-fore discharge and were delivered predominantly when fam-ilies had returned home.23,27 Interventions to facilitateparent-infant interaction may therefore be best commencedat the time of discharge when babies are medically stable,have improved neurobehavioural function, and parentshave greater opportunity to interact with their infants.
Interventions centered around the provision of infant-focused developmental care during NICU admission havealso shown significant effects.30 Aspects of developmentalcare, including the use of nests, incubator covers, and posi-tioning, were also incorporated in PBIP. Despite the provi-sion of both developmental care techniques and parent-focused activities in this intervention, we did not finda beneficial effect of PBIP on either maternal14 or infant out-comes. However, many NICUs currently incorporate aspectsof developmental care as part of routine clinical practice.Because we did not implement any changes to existing clini-cal practice in participating units, it is likely that develop-mental care practices may have been incorporated as partof standard care during the study period. This may havediluted any effect of intervention in this study. Althoughwe did not document existing infant-focussed developmentalcare practices in participating NICUs, we are aware that noneof these implemented a structured program of parent-focused support.
Planned subgroup analyses for infants categorized by ges-tational age, parity, and maternal cohabiting status similarlyfailed to show significant effects of intervention, but therewere nonsignificant trends for a positive effect of PBIP oncognitive outcomes for babies born at <28 weeks’ gestationalage and those with unsupported mothers. These findings
Johnson et al
October 2009 ORIGINAL ARTICLES
should be interpreted with caution because confidence inter-vals were very wide. Previous studies have reported variableresults regarding the efficacy of parenting interventions forsubgroups of preterm infants. Although some authors havereported greater effects for smaller babies,11,24 others haveshown larger effect sizes for heavier infants.26 Althoughgreater effect sizes for more immature infants may be relatedto intervention dose, it is unlikely to be the case in this studybecause there was no significant association between thenumber of PBIP sessions and birth weight or gestationalage. Other meta-analyses have shown that parenting pro-grams are effective for families with low socioeconomic sta-tus.31 Early intervention might thus be best targeted tosubgroups of preterm infants with higher medical or socialrisk, as advocated by other authors, but this requires furthersystematic investigation.10 Follow-up for this study cohortmay be advantageous because beneficial effects may beevidenced on longer-term outcomes in these subgroups.
The strength of this study lies in the methodological rigorof the design.32 The study was adequately powered, and targetrecruitment was achieved with groups that were well matchedfor baseline maternal and infant characteristics. Recruitmentand follow-up rates were similar for the 2 groups despite par-ticipants’ and researchers’ awareness of group allocation be-fore consent. Study attrition was minimal, and 100% ofbabies remaining in the study were assessed at 2 years’ cor-rected age. A gold standard developmental assessment wasused, and study psychologists were blinded to treatmentgroup allocation. Contamination of the control group waseliminated by the implementation of a washout period.
In conclusion, the PBIP was not effective in improvingcognitive or motor development in very preterm infants at2 years’ corrected age. Planned subgroup analyses identifiednonsignificant trends for a beneficial effect of interventionon cognitive development in infants born at <28 weeks’ ges-tational age and those with unsupported mothers. Althoughthis suggests that intervention may be better targeted for pre-term infants with greatest biological and social risk, furthersystematic research is required to address this issue. Becauseresults regarding the efficacy of parenting interventions inimproving infant cognitive outcomes are variable and oftenconflicting, further research is needed to explore the mosteffective dose and timing for intervention and to identifythe components that have greatest impact on cognitiveoutcomes to refine such programs for very preterm infants. n
Acknowledgments available at www.jpeds.com.
Submitted for publication Jan 9, 2009; last revision received March 12, 2009;
accepted April 8, 2009.
Reprint requests: Dr Samantha Johnson, Institute for Women’s Health, 86-96
Babies. Palo Alto, CA: Stanford University Press; 1997.
27. Koldewijn K, Wolf M, van Wassenaer A, Beelen A, de Groot IJM,
Hedlund R. The Infant Behavioral Assessment and Intervention Pro-
gram to support preterm infants after hospital discharge: a pilot study.
Dev Med Child Neurol 2005;47:105-12.
28. Barrera ME, Rosenbaum PL, Cunningham CE. Early home intervention
with low-birth-weight infants and their parents. Child Dev 1986;57:20-33.
494
29. Kaaresen PI, Ronning JA, Ulvund SE, Dahl LB. A randomized, controlled
trial of the effectiveness of an early-intervention program in reducing
parenting stress after preterm birth. Pediatrics 2006;118:e9-19.
30. Symington A, Pinelli K. Developmental care for promoting development
and preventing morbidity in preterm infants (Cochrane Review). The
Cochrane Library, 2003(2).
31. Bakermans-Kranenburg MJ, van Ljzendoorn MH, Juffer F. Less is more:
meta-analyses of sensitivity and attachment interventions in early child-
hood. Psychol Bull 2003;129:195-215.
32. Platt MW. A good idea that doesn’t work: the Parent Baby Interaction
Programme. Arch Dis Childhood Fetal Neonatal Ed 2007;92:F427-8.
50 Years Ago in THE JOURNAL OF PEDIATRICS
Congenital Neurocutaneous Syndromes in Childhood: II. Tuberous SclerosisChao DH. J Pediatr 1959;55:447-459
‘‘But it’s no use going back to yesterday, because I was a different person then,’’ declared the heroine in LewisCarroll’s Alice’s Adventures in Wonderland. Shortly after that 1865 classic, the French neurologist Bourneville
described tuberous sclerosis. Fifty years ago in The Journal, Chao elaborated how tuberous sclerosis manifestationsdiffer longitudinally in children. Nevertheless, Chao goes down the rabbit hole when acknowledging Vogt’s triad—epilepsy, mental deficiency, and adenoma sebaceum—as requisite for diagnosis in the ‘‘defective’’ child who ‘‘rarelylives more than 25 years.’’ Pediatrics was a different field then.
Today we recognize tuberous sclerosis as an autosomal-dominant, progressive disorder that arises de novo in twothirds of cases. Defects in the TSC1 and TSC2 genes, which produce hamartin and tuberin, respectively, and inhibit thedownstream mTOR-signaling complex, can be identified in 70% of individuals. Disease penetrance is highly variable,and genetic mosaicism plays a role. Seizures are the most common herald, because of tubers in the cerebral cortex.Patients can also harbor subependymal nodules or rarely subependymal giant cell astrocytomas. As the child ages,skin manifestations can include hypomelanotic macules (ash leaf spots), cafe-au-lait spots, facial angiofibroma (ad-enoma sebaceum), shagreen patches, and periungal fibromas. Some toddlers have cardiac rhabdomyomas, and otherscan have renal cysts or angiomyolipoma.
Diagnosis and management differ vastly from 50 years ago. Ash leaf spots, seen best by today’s looking glass, theWood’s lamp, can establish the diagnosis in the toddler with seizures. More commonly, tuberous sclerosis is diag-nosed after brain magnetic resonance imaging after seizures, especially infantile spasms. Vigabtrin can abate thesespasms instead of adrenocorticotropic hormone. Epilepsy surgery can be an option for patients with refractory sei-zures localized to a single tuber. The mTOR inhibitor rapamycin can shrink subependymal giant cell astrocytomaswithout surgery. Contrary to Vogt’s triad, cognitive outcomes are known to be highly variable, and some childrennormal and others autistic.
Can we prevent a child from becoming a different person than he was then? Preliminary reports indicate that ra-pamycin might reverse intellectual disability or prevent seizures in tuberous sclerosis.1 As Alice mused, ‘‘It would be sonice if something made sense for a change.’’
Paul Graham Fisher, MDDepartments of Human Biology
Pediatrics, Neurology and NeurosurgeryStanford University
Palo Alto, California10.1016/j.jpeds.2009.05.002
Reference
1. Crino PB. Do we have a cure for tuberous sclerosis complex? Epilepsy Curr 2008;8:159-62.
Johnson et al
October 2009 ORIGINAL ARTICLES
Acknowledgments
We would like to thank the families who participated in the study fortheir cooperation and long-term support. Our thanks are also extendedto the NICU staff in participating centers (Cheltenham General Hospi-tal; City Hospital, Nottingham; Leicester Royal Infirmary; Queen’s
Randomized Trial of a Parenting Intervention for Very Preterm In
Medical Centre, Nottingham; Gloucestershire, Royal Hospital; South-mead Hospital, Bristol; St. Michael’s Hospital, Bristol). The interventionwas delivered by research nurses Julie Berry, Mandy Bond, Jane Hay-hurst, Helen Lang, Patty Mistry, Lisa Ramsey, and Helen States underthe supervision of Chrissie Israel. Outcome data were collected by Char-lotte Beer (Nottingham) and Sam Johnson (Nottingham). James Kirkbr-ide (Cambridge) coded index of multiple deprivation scores.
