Perceived Parental Burden and ServiceUse for Child and AdolescentPsychiatric Disorders

Adrian Angold, MRCPsych, Stephen C. Messer, PhD, Dalene Stangl, PhD,Elizabeth M. Z. Farmer, PhD, Elizabeth J. Costello, PhD, and Barbara J. Burns, PhD

Introduction

Platt defined burden as "the presence ofproblems, difficulties, or adverse eventswhich affect the life (lives) of the psychiatricpatient's significant other(s)." -P 383 Whileother definitions of burden have been pro-posed,' 2 all emphasize the effect the patienthas upon the family, or the impact that livingwith the patient has on the family's dailyroutines and, possibly, health. Most studieshave focused on caregivers for adults withsevere and persistent mental illness or forthe elderly.3-8 These studies have found thatcaregivers experience relatively high levelsof both objective burden (e.g., providingtransportation, assisting the patient withdaily tasks) and subjective burden (e.g.,reduced caregiver well-being, worry). Fur-thermore, caregiving frequently adds a set ofburdens to an already high level of hardshipwithin the family. Caregiver burden has alsosometimes been associated with frequencyof patient hopitalizations.9-12

Providing any care for an adult relativeis an unexpected event, but parents areexpected to be responsible for the care oftheir minor children. Emotional and finan-cial hassles are part of the normal parentingprocess. When the additional burdens asso-ciated with a child's psychiatric problemsare added to these normal hassles, mightthey be more difficult to bear? Contrari-wise, since such hassles are normal, parentsof psychologically disturbed children maybe primed to take additional difficulties intheir stride, with little sense of burden. Par-ents of children with chronic illness reportsubstantial levels of social, financial, andemotional burden.'3 9 However, similarstudies of children with psychiatric disor-ders do not seem to have been done,although many of the factors associatedwith increased burden for families of adultpatients (such as the patient's living with

family, poverty, single-parenthood andunemployment) are common in the familiesof children with psychiatric problems.3(35

There is a big difference in seeking helpfor psychiatric disorders for children and foradults in that children rarely refer them-selves for treatment. We know from severalcommunity studies3439 that only a small pro-portion of children with psychiatric disor-ders are receiving treatment at any point, sothe presence of a disorder is not sufficient toexplain treatment seeking. This paper exam-ines the level of burden experienced by par-ents on account of their children's problemsas a factor in propelling parents to seek helpfor their children's disorders.

Methods

Subjects and Proceduires

The Great Smoky Mountains Study isa longitudinal study of the development ofpsychiatric disorder and need for mentalhealth services in rural and urban youth.The details of the study design and instru-ments used can be found in Costello et al.40

Adrian Angold, Elizabeth M. Z. Farmer, and Eliza-beth J. Costello are with the DevelopmentalEpidemiology Program and Barbara J. Bums is withthe Division of Social and Community Psychiatry,Department of Psychiatry and Behavioral Sciences,Duke University Medical Center. Durham, NC.Dalene Stangl is with the lnstitute of Statistics andDecision Sciences, Department of Statistics andDecision Sciences, Duke University Medical Cen-ter. Stephen C. Messer is with the Center for FamilyResearch. Department of Psychiatry and BehavioralSciences, George Wbashington University MedicalCenter, W ashington, DC.

Requests for reprints should be sent to

Adrian Angold, MRCPsych, Box 3454, Duke Uni-versity Medical Center, Durham, NC 27702.

This paper was accepted April 16, 1997.

American Journal of Public Health 75January 1998, Vol. 88, No. 1

Angold et al.

Briefly, a representative sample of 4500 9-,11-, and 13-year-olds, identified through theStudent Information Management Systemof the public school systems of 11 countiesin western North Carolina, was selectedthrough a household equal-probabilitydesign. A screening questionnaire wasadministered to a parent (usually themother), by telephone or in person. Thisconsisted of 55 questions from the ChildBehavior Checklist about the child's behav-ioral problems, together with some basicdemographic and service-use questions. Allchildren scoring above a predetermined cut-point score of 20 (designed to include about25% of the population) on the behavioralquestions, plus a 1-in-10 random sample ofthose scoring below the cutpoint, wererecruited for the longitudinal study. Theresponse rate was 80%, resulting in aninterviewed sample of 1015. Designweights were used for the computation ofpopulation prevalence estimates.

Measures

Child and primary caretaker wereinterviewed separately about the child'spsychiatric status and service use on thebasis of the third edition of the Child andAdolescent Psychiatric Assessment,4'which generated DSM-III-R (The Diagnos-tic and Statistical Manual of Mental Disor-ders, 3rd ed, rev) diagnoses, and the Childand Adolescent Services Assessment.42 Ofprimary caretakers interviewed, 84% werebiological mothers, 6.7% were biologicalfathers; 8.6% were other females, and 0.6%were other males (all will be referred to inthis paper as "parents"). The Child andAdolescent Burden Assessment43 wasadministered to the parent only. The refer-ence period for each of these instrumentswas the 3 months prior to the interview.

The Child and Adolescent BurdenAssessment. The Child and Adolescent Bur-den Assessment (CABA) was completedafter the diagnostic interview. Parents wereasked about 20 potential perceived bur-dens-that is, problems or difficulties intheir own lives that they perceived as beingcaused or exacerbated by their child's psy-chiatric symptoms. The following areaswere covered: expenses and financial diffi-culties, problems in relationships with fam-ily or social network members, restrictionson activities, and decreased feelings ofwell-being and competence. Items werescored as 0, 1, 2, or 3 depending on thedegree of burden (possible scores rangedfrom 0 through 59). Rules for assigningthese scores are contained in the CABAschedule.

The first step in the administration ofthe CABA involved an interviewer decisionabout whether there was any potential forparental burden, since we found that askingquestions about psychopathology-relatedburdens was inappropriate when little or nochild symptomatology had been reported orwhen the parent had already indicated inresponse to detailed questioning aboutsymptoms that he or she had no significantproblems with the child. On the basis of allthe information collected in the interview,the interviewer decided whether there wasany potential for burden, and if there was,the CABA was administered. All together,349 CABAs were completed. Of these, 193indicated the presence of at least one per-ceived burden, while 156 indicated that per-ceived burden was absent. In all othercases, perceived burden was regarded asbeing absent.

Factor analyses have indicated that onemajor factor predominates in the CABA inboth general population and severely dis-turbed clinical samples, accounting forapproximately 30% of the items' variance.44Coefficient alpha for the scale in this samplewas .88. Two-week stability of the CABA ina small (n = 19) clinical sample was ade-quate (intraclass correlation coefficient =.60). Construct validity of the CABA is indi-cated by significant mean differences inCABA scores between groups with varyingrates and severity of psychopathology: 0.9(SD = 3.3) in the Great Smoky MountainStudy general population sample;7.4 (SD=7.1) in a child guidance clinicsample; and 15.7 (SD= 10.1) in a group ofseriously emotionally disturbed youth atimminent risk for out-of-home placement.44

The Child and Adolescent PsychiatricAssessment. The Child and Adolescent Psy-chiatric Assessment is an interviewer-basedinterview, that provides a structured ques-tioning scheme enabling interviewers todetermine whether symptoms, as defined inan extensive glossary, are present or absentand to code their frequency, duration, andonset.4' Diagnostic 1-week test-retest relia-bilities for child self-reports range from0.55 for conduct disorder to 1.0 for sub-stance abuse/dependence.45 Diagnoses andsymptom scores are generated by comput-erized algorithms. A symptom was countedas being present if it was reported by eitherthe parent or child. Psychosocial impair-ment secondary to psychiatric symptoma-tology in 17 areas of functioning was alsorated according to a series of definitions andrules specified in the psychiatric assess-ments glossary and the interview schedule.The intraclass correlation coefficient forlevel of psychosocial impairment by child

self-report was .77.45 In addition, theassessment contained a number of ques-tions about demographic status, past historyof parental psychiatric problems, recent lifeevents, and family relationship problems.

The Child and Adolescent ServicesAssessment. The Child and Adolescent Ser-vices Assessment collects parent and childreports on the use of mental health servicesprovided by the specialty mental healthsector, schools, child welfare, primaryhealth care, juvenile justice, and informalcommunity sources. Use of a service wascoded positively if either the parent or thechild reported it. Psychometric analysesshowed that test-retest consistency of chil-dren's responses to the services assessmentwas very good for the most intensive ser-vices (K =.82-.92 for inpatient, out-of-home, and juvenile justice services); ade-quate for moderate level services(K =.52-.58 for outpatient, crisis, and otherprofessional services); and not very goodfor the least intensive services (K =.39-.43for school and informal services). Parentsshowed a similar pattern, except thatschool, informal, and outpatient serviceswere somewhat more reliably reported.Combined parent and child reports fromthe services assessment correctly identified90% of children who were receiving ser-vices according to a community mentalhealth center's records.

Results

Details of the rate of diagnosis, impair-ment, and service use have been publishedelsewhere. The 3-month prevalence of anyDSM-III-R Axis I disorder was 20.3%(SE = 1.7). Four percent of children hadreceived specialist mental health servicesduring the preceding 3 months, and 12.4%had received services for mental health prob-lems in other than mental health settings.

Rates ofPerceived Burden

A total of 10.7% of all parents of 9-,11-, and 13-year-olds in the populationreported at least one perceived burden result-ing from child psychiatric symptomatology.The most common individual burdensinvolved effects on personal well-being,stigma, and restrictions on personal activities(see Messer et al. for details).4' Some burdenwas perceived by 4.5% of parents of chil-dren with no diagnosis or psychosocialimpairment, compared with 16.8% of thosewhose children had only a diagnosis, 17.8%of those whose children had only impair-ment, and 38.8% of those whose children

January 1998, Vol. 88, No. 176 American Journal of Public Health

Burden and Child Psychiatric Use

had both a diagnosis and impairment(weighted prevalence estimates).

Degree ofBurden

A simple measure of degree of burdenis the sum of all 20 self-reported burden itemscores in the CABA. For the whole popula-tion, the weighted mean total burden score

was 0.9 (SD = 3.3; range = 0 through 28).Those whose children had no diagnosis

or impainnent had a mean CABA total score

of 0.3 (SD = 2.1), compared with 1.4 (SD =3.3) for those with only a diagnosis, 1.2 (SD= 3.2) for those with impairment only, and4.4 (SD = 5.2) for those whose children hadboth a diagnosis and impairment.

Predictors ofBurden

We conceptualized four groups of pos-

sible causes of burden in addition to theexistence of diagnosis or impairment. Thefirst group of causes were demographic fac-tors: rurality of residence; poverty (familyincome less than $10 000 per annum); theage of the child; and the child's gender. Thesecond group consisted of stress-and-strainmeasures generated by the Child and Ado-lescent Psychiatric Assessment: total nega-

tive life events in the preceding 3 months; a

parental history of psychopathology; par-ent-child relationship problems; and a

group of "family structure" and communityproblems, comprising parental unemploy-ment, residence in substandard conditionsor a poor neighborhood, large sibship (morethan 4), and single-parenting. The thirdgroup contained two variables definingseverity of the child's psychopathology: thenumber of symptoms that would counttowards a DSM-III-R diagnosis and thetotal impairment score from the Child andAdolescent Psychiatric Assessment. Thefinal group concemed whether the child hadan anxiety or depressive disorder or had a

disruptive behavior disorder, and containedall 28 specific diagnostic categories. Thequestion addressed by this final group was

whether any specific diagnoses are associ-ated with a particularly elevated burdenwhen overall level of symptomatology andimpairment and overall type of diagnosisare already controlled.

We conducted this analysis in twostages because the error scores in initial trialregressions were not normally distributed.The first stage consisted of best-subsetlogistic regressions of the presence of any

burden, and may be thought of as addressingthe question, "What gets parents over thethreshold for experiencing burden?" In thesecond stage, best-subset, ordinary leastsquares regressions of the predictors on thelog of the total burden scores in those whohad parental perceived burden (n = 193)were conducted. In this group, the residualsfrom the regressions adequately approxi-mated a normal distribution. This stageaddressed the question, "Once the parentis over the threshold for experiencing bur-den, what determines the level of burdenexperienced?"

The final best-fitting models fromthese two sets of regressions are presentedin Tables 1 and 2. The odds ratios, parame-

ter estimates and test statistics refer tosimultaneous models in which each effect isestimated with all the other terms in themodel controlled for.

It could be argued that the correlationsobserved here do not represent burdenresulting from symptoms (although that iswhat parents were instructed to report on),but rather that the perception of burden

makes parents more likely to report symp-

toms or functional impainnent in their chil-dren. As a check on this possibility, we

regressed child-only reports of total symp-

toms and total impairment level on theprobability of the presence of any burden.Both were significantly associated withparental burden (symptoms odds ratio[OR]=1.05; P = .02; impairment OR =

1.23; P = .00007). Then, we regressed thesesame child-only reports of psychopathologyon the level of parental perceived burdenfor parents with reported burden. This time,only the level of impairment was a significantpredictor (impairment F(1,192)= 10.49;P=.001). Given the well-known low corre-

lations between parent and child reports ofsymptoms and impairments, we concludedfrom these results that parents really were

reporting on burdens attributable to theirchildren's symptomatology.

Burden as a Predictor ofService Use

We looked first at the rates of specialtymental health service use by children of par-

ents with and without parental burden. Fig-ure 1 presents rates of service use for chil-dren with and without diagnoses andimpairment.

We then explored the impact of burdenon the probability of service use, in com-

parison with the other potential predictorvariables described above, using best-subsetlogistic regression.

American Journal of Public Health 77January 1998, Vol. 88, No. 1

TABLE 1 -Predictors of the Presence of Any Burden Remaining Significant in a Logistic Model in the Whole Population:Great Smoky Mountains Study

Odds Ratio Standardized(95% Confidence Interval) Parameter Estimate SE p

Total DSM-111-R symptom score 1.15 (1.11, 1.18) .51 .016 3 x 10-17Total functional impairment score 1.23 (1.15, 1.31) .33 .033 4 x 10-10Anxiety or depression diagnosis .49 (.26, .81) -.12 .312 .02Parental history of mental health problems 1.74 (1.18, 2.56) .14 .198 .005

Note: DSM-111-R = Diagnostic and Statistical Manual of Mental Disorders, 3rd ed, rev.

TABLE 2-Predictors of the Level of Perceived Parental Burden RemainingSignificant in a Regression Model in Those with Perceived ParentalBurden in the Great Smoky Mountains Study (n = 193)

ParameterEstimate SE F (3,189) P

Total DSM-111-R symptom score .01 .008 8.6 .004Total functional impairment score .02 .016 9.2 .002Tics .17 .171 4.0 .046

Angold et al.

Table 3 shows the odds ratios forsignificant predictors in the final model.Only two significant predictors emerged, byfar the strongest of which was parental bur-den. We also observed that the effect of bur-den on the probability of service use was a

decelerating quadratic function, with little orno additional increase in the probability ofservice use above a total burden score of 8.

Since children often do not need theirparents' help to make contact with school-based services (counselors, psychologists,or special classes), we compared the effectsof parental burden and the other potentialpredictors on the use of school services

(Table 3). This time, a rather different pic-ture emerged, with the child's total symp-

tom score having the largest effect andparental burden having a smaller effect. Thepresence of depression or anxiety was asso-

ciated with less school service use than thepresence of other disorders.

Discussion

The strongest predictor of the presence

of any perceived burden was the child'stotal symptom score, but the child's level ofimpairment also made a substantial contri-

bution. Children with depressive or anxietydisorders proved less burdensome than chil-dren with other disorders, while parentswith preexisting mental health problemsperceived more burden than those without.The final model, for level of perceived bur-den (Table 2), showed that again levels ofthe child's symptomatology and functionalimpairment had the most effect on parentalperceived burden scores. In addition, thepresence of tics was associated withincreased perceived burden.

Perhaps the most important result isthe very low rate of specialty mental healthservice use when no perceived burden was

January 1998, Vol. 88, No. 1

78 American Journal of Public Health

vA

; 30-

c

ID

E 20

.C

._

100

S1

0

Diagnosis and impairmentImpairment onlyNo diagnosis or impairment

FIGURE 1-Weighted percentage of children receiving professional mental health services by diagnosis, impairment, andparental burden group in the Great Smoky Mountains Study (n = 1015).

TABLE 3-Predictors Remaining Significant in Logistic Models of Specialist Mental Health Service and School Use in theGreat Smoky Mountains Study (n = 1015).

Odds Ratio Standardized(95% Confidence Parameter

Interval) Estimate SE P

Specialist mental health service useTotal DSM-111-R symptom score 1.05 (1.01,1.08) .16 .016 .007Parental burden 1.39 (1.25, 1.55) .83 .055 2 x 104Parental burden squared .99 (.98, .99) .60 .002 2 x 10

School service useTotal symptom score 1.08 (1.05, 1.11) .28 .015 8 x 1 07Depression or anxiety .46 (.24, .89) -.13 .331 .02Parental burden 1.17 .40 .046 5 x 104Parental burden squared .99 (.99, .99) -.25 .002 .02

Diagnosis only

Burden and Child Psychiatric Use

reported. Fewer than 2% of children with-out a diagnosis, impairment, or burden sawa mental health professional. Similarly,fewer than 2% of children with only a diag-nosis or only impairment received specialtymental health services. In each of these sit-uations, the presence of perceived burdenwas associated with at least a fivefoldincrease in the rate of service use. When achild had both a diagnosis and impairment,the presence of perceived burden was asso-ciated with a threefold increase in the use ofspecialist mental health service. Thus, atevery level of psychopathology, the pres-ence of parental perceived burden was apowerful predictor of the use of specialtymental health service. When parental per-ceived burden was taken into account, theamount or type of symptomatology shownby a child had relatively little effect on theprobability of receiving services, and thelevel of impairment had no effect. In an ear-lier paper,49 we reported substantial effectsof diagnosis and impairment on specialtymental health service use in analyses whereburden was not considered. It appears thatmost of the effects of symptomatology andimpairment on service use were mediatedby perceived parental burden. We alsoobserved a dose-response relationshipbetween level of parental perceived burdenand probability of service use. School ser-vices, on the other hand, were more directlysensitive to the number of the child's symp-toms than to parental burden although it ispossible that had we measured teacher bur-den, it might have proved to be more highlyrelated to school service use.

However, this is only a single studyconducted in a largely rural area, so it isimportant to know whether similar findingsoccur in other regions (e.g., the inner city)and other service systems. In addition, weinvestigated only limited dimensions of ser-vice use. It is important to know whetherburden influences inception into services,maintenance and volume of service use,level of care received, and the costs of ser-vices. We also measured only perceivedburden in one parent, and it seems likelythat the burdens felt by other caregivers arealso important.

There is also a question as to whetherthe CABA is really measuring only the bur-den of psychiatric disorder in children. Weasked parents to report only burdens thatthey attributed to their children's psy-chopathology, but it is possible that theymisattributed some of the normal hassles ofparenting to their children's psychopathol-ogy. Our finding that symptomatology andimpainnent that were reported by the childalso predicted parental burden suggests that

such misattribution is not the whole story.But even if it were, the fact remains thatparents who attributed burden to their chil-dren's symptomatology were much morelikely to seek specialty mental health ser-vices for their children and their childrenwere also more likely to receive school ser-vices. Thus, even if the CABA measuresmisattributions in whole or in part, thosemisattributions are important for under-standing mental health service use.

Although a great deal of work hasbeen done on the effects of parental psychi-atric disorders on children and on the linksbetween parent-child relationships and psy-chopathology,5053 little attention has beenpaid to the impact of children's problems onparental mental disorder. The psychologicalburdens described by parents as resultingfrom their children's problems suggest thatthis issue is worthy of more attention. Inaddition, some parents reported that theirchildren's problems had substantial nega-tive effects on their family and social rela-tionships; this situation might be expectedto increase the risk of future psychologicalproblems.

It would be interesting to know howthe perception of burden emerges in reac-tion to the development of psychiatric dis-order, and to what degree and why it fluctu-ates over time. For instance, do parentsadjust to their children's problems overtime and experience less burden, or doescontinuing symptomatology lead to escalat-ing burden? It would also be interesting toknow what criteria parents use in decidingto attribute a difficulty of their own to achild's behavior, and why some parentsreport no burden despite having a severelydisturbed child. One also wonders whatother ways parents deal with their sense ofbeing burdened. For instance, do they seekhelp from their social networks? By whatmechanism is the perception of burdenrelated to service use, and why do someparents who report high levels of burdenstill not get help for their children? D

AcknowledgmentsThis project was supported by grant MH48085 fromthe National Institute of Mental Health. Additionalsupport was provided through faculty scholar awardsfrom the William T. Grant Foundation to Dr Angoldand Dr Costello and a center grant from the LeonLowenstein Foundation to Dr Angold.

References1. Platt S. Measuring the burden of psychiatric

illness on the family: an evaluation of somerating scales. Psychol Med. 1985;15:383-393.

2. Biegel D, Sales E, Schulz R. Family Caregiv-ing in Chronic Illness: Alzheimer's Disease,Cancer, Heart Disease, Mental Illness andStroke. Newbury Park, Calif: Sage Publica-tions; 1991.

3. Francell CG, Conn VS, Gray DP. Families'perceptions of burden of care for chronicallymentally ill relatives. Hosp Community Psy-chiatry. 1988;39:1296-1300.

4. Frank RG, German PS, Burns BJ, Johnson W,Miller N. Use of services by cognitivelyimpaired elderly persons residing in the com-munity. Hosp Community Psychiatry. 1988;39:555-557.

5. George LK, Gwyther LP. Caregiver well-being: a multidimensional examination offamily caregivers of demented adults. Geron-tologist. 1986;26:253-259.

6. Schene AH, Tessler RC, Gamache GM.Instruments measuring family or caregiverburden in severe mental illness. Soc Psychia-try Psychiatr Epidemiol. 1994;29:228-240.

7. Tessler RC, Gamache GM. Continuity ofcare, residence, and family burden in Ohio.Milbank Q. 1994;72:149-169.

8. Carpentier N, Lesage A, Goulet J, Lalonde P,Renaud M. Burden of care of families not liv-ing with young schizophrenic relatives. HospCommunity Psychiatry. 1992;43:38-43.

9. Brown G, Bone M, Dalison B, Wing J. Schiz-ophrenia and Social Care. New York NY:Oxford University Press; 1966.

10. Thompson E H, Doll W. The burden of fami-lies coping with the mentally ill: an invisiblecrisis. Fam Relations. 1982;31:379-388.

11. Crotty P, Kulys R. Are schizophrenics a bur-den to their families? Significant others'views. Health Soc Work. 1986;1 1:173-188.

12. Hoenig J, Hamilton M. The schizophrenicpatient in the community and his effect on thehousehold. Int J Soc Psychiatry. 1966; 12:165-176.

13. Benson BA, Gross AM, Messer SC, KellumG, Passmore LA. Social support networksamong families of children with craniofacialanomalies. Health Psychol. 1991;10:252-258.

14. Frank RG, Gertler PJ. An assessment ofmeasurement bias in estimating the impact ofmental distress on earnings. Hum Resources.1991;26:154-164.

15. Savoca E. Measurement error in self-evalua-tions of mental health: implications for labormarket analysis. In: Frank RG, Manning WG,eds. Economics and Mental Health. Balti-more, Md: The Johns Hopkins UniversityPress; 1992: 130-151.

16. Drotar D, Bush M. Mental health issues andservices. In: Hobbs N, Perrin JM, eds. Issuesin the Care of Children with Chronic Illness.San Francisco, Calif: Jossey-Bass Publishers;1985: 514-550.

17. Hobbs N, Perrin JM. Issues in the Care ofChildren with Chronic Illness. San FranciscoCalif: Jossey-Bass Publishers; 1985.

18. Magrab PR. Psychological Management ofPediatric Problems: Early Life Conditionsand Chronic Diseases. Baltimore, Md: Uni-versity Park Press; 1978.

19. Pless IB, Douglas IWB. Chronic illness inchildhood: I. epidemiological and clinicalcharacteristics. Pediatrics. 1972;47:405-414.

January 1998, Vol. 88, No. 1 American Journal of Public Health 79

Angold et al.

20. Holroyd J, Guthrie D. Stress in families withneuromuscular disease. J Clin Psychol.1979;35:735-739.

21. Vance JC. Effects of nephrotic syndrome onthe family: a controlled study. Pediatrics.1980;65:948-955.

22. Meyerowitz JH, Kaplan HB. Familialresponses to stress: the case of cystic fibrosis.Soc Sci Med. 1967; 1:249-266.

23. Kazak AE, Christakis DA. Caregiving issuesin families of children with chronic medicalconditions. In: Kahana E, Biegel DE, WykleML, eds. Family Caregiving across the Life-span. Thousand Oaks, Calif: Sage Publica-tions; 1994: 331-355.

24. Patterson JM. A family stress model: the familyadjustment and adaptation response. In: Ram-sey CN, ed. Family Systems in Medicine. NewYork, NY: Guilford Press; 1989: 95-118.

25. Turk J. Impact of cystic fibrosis on familyfunctioning. Pediatrics. 1967;34:67-71.

26. Mellins CA. Families affected by pediatricacquired immunodeficiency syndrome. J DevBehav Pediatr. 1994; 1 5 (suppl) :S54-S60.

27. Koocher GP, O'Malley JE. The DamoclesSyndrome: Psychological Consequences ofSurviving Childhood Cancer. New York, NY:McGraw-Hill; 1981.

28. Patterson JM, Leonard BJ. Caregiving andchildren. In: Kahana E, Biegel DE, WykleML, eds. Family Caregiving across the Life-span. Thousand Oaks, Calif: Sage Publica-tions; 1994: 133-158.

29. Lauler RH, Nakielny W, Wright NA. Psycho-logical implications of cystic fibrosis. CanMed Assoc J. 1966;94: 1043-1048.

30. Kellam SG, Ensminger ME, Turner RJ. Fam-ily structure and the mental health of children.Arch Gen Psychiatry. 1977;34:1012-1022.

31. McLoyd VC. The impact of economic hard-ship on black families and children: psycho-logical distress, parenting, and socioeconomicdevelopment. Child Dev. 1990;61 :311-346.

32. Bird HR, Gould MS, Yager T, Staghezza B,Canino G. Risk factors for maladjustment inPuerto Rican children. JAm Acad Child Ado-lesc Psychiatry. 1989;28:847-850.

33. Brown GW, Harris TO. The Social Origins ofDepression: A Study of Psychiatric Disorderin Women. London, England: Tavistock Pub-lications; 1978.

34. Costello EJ, Costello AJ, Edelbrock C, et al.Psychiatric disorders in pediatric primarycare: prevalence and risk factors. Arch GenPsychiatry. 1988;45:1107-1116.

35. Dohrenwend BP, Dohrenwend BS. SocialStatus and Psychological Disorder. NewYork, NY: Wiley; 1969.

36. Offord DR, Boyle MH, Racine YA, et al. Out-come, prognosis, and risk in a longitudinalfollow-up study. J Am Acad Child AdolescPsychiatry. 1992;31 :916-923.

37. Reinherz HZ, Giaconia RM, Lefkowitz ES,Pakiz B, Frost AK. Prevalence of psychiatricdisorders in a community population of olderadolescents. JAm Acad Child Adolesc Psychi-atry. 1993;32:369-377.

38. Schoenbach VJ, Kaplan BH, Grimson RC,Wagner EH. Use of a symptom scale to studythe prevalence of a depressive syndrome inyoung adolescents. Am J Epidemiol.1982;1 16:791-800.

39. Offord DR, Boyle MH, Szatmari P, et al.Ontario child health study: II. six-month preva-lence of disorder and rates of service utilization.Arch Gen Psychiatry. 1987;44:832-836.

40. Costello EJ, Angold A, Burns BJ, et al. TheGreat Smoky Mountains Study of Youth:goals, designs, methods, and the prevalence ofDSM-III-R disorders. Arch Gen Psychiatry.1996;53:1129-1136.

41. Angold A, Prendergast M, Cox A, HarringtonR, Simonoff E, Rutter M. The Child and Ado-lescent Psychiatric Assessment (CAPA). Psy-chol Med. 1995;25:739-753.

42. Burns BJ, Angold A, Magruder-Habib K,Costello EJ, Patrick MKS. The Child andAdolescent Services Assessment (CASA): Ver-sion 3.0. Durham, NC: Duke University Med-ical Center; 1992.

43. Patrick M, Angold A, Burns B, Costello EJ.The Child and Adolescent Burden Assess-ment (CABA). Department of Psychiatry andBehavioral Sciences, Developmental Epi-

demiology Program; Durham, NC: Duke Uni-versity Medical Center; 1992.

44. Messer SC, Angold A, Costello EJ, Burns BJ.The Child and Adolescent Burden Assess-ment (CABA): measuring the family impactof emotional and behavioral problems. Int JMethods Psychiatr Res 1996;6:261-284.

45. Angold A, Costello EJ. A test-retest reliabilitystudy of child-reported psychiatric symptomsand diagnoses using the Child and AdolescentPsychiatric Assessment (CAPA-C). PsycholMed. 1995;25:755-762.

46. Farmer EMZ, Angold A, Burns BJ, Costello EJ.Reliability of self-reported service use: test-retest consistency of children's responses to theChild and Adolescent Services Assessment(CASA). J Child Fam Stud. 1994;3:307-325.

47. Ascher BH, Farmer EMZ, Burns BJ, AngoldA. The Child and Adolescent ServicesAssessment (CASA): description and psycho-metrics. JEmot Behav Dis. 1996;4:12-20.

48. Costello EJ, Angold A, Burns BJ, Erkanli A,Stangl DK, Tweed DL. The Great SmokyMountains Study of Youth: functional impair-ment and severe emotional disturbance. ArchGen Psychiatry. 1996;53:1137-1143.

49. Burns BJ, Costello EJ, Angold A, et al. Chil-dren's mental health service use across ser-vice sectors. Health Aff 1995;14:147-159.

50. Gotlib IH, Avison WR. Children at risk forpsychopathology. In: Costello CG, ed. BasicIssues in Psychopathology. New York, NY:Guilford Press; 1993: 271-3 19.

51. Garmezy N, Masten AS. Chronic adversities.In: Rutter M, Taylor E, Hersov L, eds. Childand Adolescent Psychiatry: ModernApproaches. Oxford, England: Blackwell Sci-entific Publications; 1994: 191-208.

52. Patterson GR. Depression and Aggression inFamily Interaction. Hillsdale, NJ: Erlbaum;1990.

53. Rutter M. Parental mental disorder as a psy-chiatric risk factor. In: Frances AJ, Hales RE,eds. Review ofPsychiatry, Vol 7. Washington,DC: American Psychiatric Press; 1988:647-663.

80 American Journal of Public Health January 1998, Vol. 88, No. 1