children of depressed parents: an integrative review · pdf filechildren of depressed parents:...

Psychological Bulletinl990,VoI.108,No. 1,50-76

Copyright 1990 by the American Psychological Association, Inc.0033-2909/90/S00.75

Children of Depressed Parents: An Integrative ReviewGeraldine DowneyUniversity of Denver

James C. CoyneUniversity of Michigan

This article reviews the various literatures on the adjustment of children of depressed parents,difficulties in parenting and parent-child interaction in these families, and contextual factors thatmay play a role in child adjustment and parent depression. First, issues arising from the recurrent,episodic, heterogeneous nature of depression are discussed. Second, studies on the adjustment ofchildren with a depressed parent are summarized. Early studies that used depressed parents as con-trols for schizophrenic parents found equivalent risk for child disturbance. Subsequent studies usingbetter-defined samples of depressed parents found that these children were at risk for a full range ofadjustment problems and at specific risk for clinical depression. Third, the parenting difficulties ofdepressed parents are described and explanatory models of child adjustment problems are outlined.Contextual factors, particularly marital distress, remain viable alternative explanations for bothchild and parenting problems. Fourth, important gaps in the literature are identified, and a consis-tent, if unintentional, "mother-bashing" quality in the existing literature is noted. Given the limita-tions in knowledge, large-scale, long-term, longitudinal studies would be premature at this time.

Depression has traditionally been viewed as a problem of theindividual. Psychologists now know that this view is limited.Research over the past two decades has revealed the interdepen-dence between depressed persons and their social context(Coyne, 1990). Parents' depression and their child's adjustmentis a case in point.

Depression is a highly prevalent disorder, especially amongwomen of child-bearing age (Weissman, 1987). At any giventime, approximately 8% of mothers are clinically depressed(Weissman, Leaf, & Bruce, 1987). The rate increases to 12% inmothers who have recently given birth (O'Hara, 1986). Conse-quently, large numbers of children are exposed to parental de-pression. Understanding the impact of parental depression onchildren is, therefore, a matter of great social as well as theoreti-cal significance.

Initially, the children of depressed persons served as controlsin high-risk research on the offspring of schizophrenic parents.This led to the serendipitous finding that children of depressedparents were just as disturbed as children of schizophrenic par-ents. A review of early research on children of depressed parentsled Beardslee, Bemporad, Keller, and Klerman (1983) to con-clude that these children were at considerable risk for a fullrange of psychological difficulties, including depressive symp-tomatology (see also Orvaschel, 1983). They qualified their con-clusion by noting that few studies had a control group or usedcriterion-based child diagnoses, and called for investigationsthat remedied these limitations. They also urged researchers toexamine how parental depression is linked with child malad-justment and to search for child and family characteristics thatmight modify this link.

We thank Barbara Bettes, Niall Bolger, John Richters, Abigail Stew-art, and the reviewers for helpful comments on earlier versions of thisarticle.

Correspondence concerning this article should be addressed to Geral-dine Downey, Psychology Department, University of Denver, Denver,Colorado 80208.

The study of children of depressed parents has made consid-erable progress toward answering these calls, as the set of studiesin a recent special issue of Developmental Psychology demon-strates (Dodge, 1990). Recent research on the adjustment ofthese children shows admirable sophistication in its use of ad-vances in child assessment and nosology. Researchers have alsoconsidered why these children are at risk. Efforts to address thisquestion have incorporated methodological and theoretical de-velopments in the study of family interaction, human attach-ment, stress and coping, and cognitive and interactional aspectsof psychopathology.

The hypothesis that is receiving the most research attentionis that children's problems result directly from living with a de-pressed parent, especially a depressed mother. However, sup-port can also be found for the hypothesis that the associationbetween parental depression and child problems is spurious,that is, that both are caused by preexisting conditions such asmarital turmoil or family stress. This hypothesis suggests thatto understand children's adjustment problems, the broader in-terpersonal processes operating in the lives of depressed parentsmust be considered.

A rival to both these hypotheses is that the connection be-tween parent and child problems has a genetic basis. However,twin and adoption studies show that genetic factors can at bestpartially account for the problems of children with a depressedparent (Allen, 1976; Cadoret, O'Gorman, Heywood, &Troughton, 1985; von Knorring, Cloninger, Bohman, & Sig-vardsson, 1983). The more appropriate question is, how do ge-netic and other biological vulnerabilities combine with contex-tual factors to influence child adjustment? Relevant methodolo-gies are now available to tackle this question. Yet, there has beenlittle effort to integrate biological and contextual risk factors,reflecting the view that these factors offer competing rather thancomplementary explanations.

The heightened interest in children of depressed parents overthe past 5 years has resulted in an empirical base that is consid-erably stronger than what was available when earlier reviews

50

CHILDREN OF DEPRESSED PARENTS 51

were undertaken (Beardslee et al., 1983; Burbach & Borduin,1986; Forehand, McCombs, & Brody, 1987; Orvaschel, 1983).We believe that an integrative review is needed to ensure thatfuture study designs are informed by what is known about chil-dren of depressed parents and, more generally, about affectivedisorders and their interpersonal context. Toward this end, wereview and evaluate recent progress in the relevant literatures.

First, we summarize the research documenting the adjust-ment problems in children of depressed parents. We begin withan overview of the high-risk studies of schizophrenia and turnto the more recent studies specifically on children of depressedparents. Second, we evaluate the research that seeks to explainthe adjustment problems of these children. We pay particularattention to research on the parenting behavior of depressedpersons because of the centrality of this transmission model inthe literature. We then consider evidence for the alternativemodel, which holds that processes in the broader interpersonalcontext may account for both parental depression and child ad-justment. We end by drawing some tentative conclusions andidentifying an agenda for future research.

Despite considerable progress, it is becoming apparent thatthe resolution of basic questions about children of depressedparents will be more difficult than it first appeared. Before turn-ing to the studies of these children, we briefly consider someissues that contribute to the intractability of the task.

Methodological Concerns

An Emerging View of Depression and Its Social Context

Traditionally, it has been assumed that depression occurs ina single episode that usually resolves without enduring impair-ment. The emerging view is that depression is heterogeneousand highly variable in course, but is basically a recurrent, epi-sodic disorder with varying degrees of residual difficulties(Akiskal, 1982; Clayton, 1983). Persons who become clinicallydepressed can expect 5-6 episodes in their lifetime (Zis &Goodwin, 1979). Although most episodes resolve in 6-9months, 20% continue for at least 2 years. About 40% of personsseeking treatment for major depression have "double depres-sion," in that they suffer from a more enduring dysthymia aswell as major depression (Keller & Shapiro, 1982). In addition,anxiety disorders and personality disturbance often co-occurwith major depression (Black, Bell, Hulbert, & Nasrallah, 1988;Merikangas, Prusoff, & Weissman, 1988).

Depression is frequently accompanied by interpersonaldifficulties, particularly in close relationships. Depressedwomen have a high rate of marital conflict (Weissman & Paykel,1974), persisting for up to 4 years after an episode (Rounsaville,Prusoff, & Weissman, 1980), and divorce is common (Briscoe& Smith, 1973; see Coyne, 1990, for a review). To make mattersmore complex, depressed persons tend to marry persons with apsychiatric illness or a family history of psychopathology (Meri-kangas & Spiker, 1982). When depressed persons have psychiat-rically disturbed spouses, their own symptoms are more severe,and marital and family disturbance is more likely (Merikangas,Weissman, Prusoff, & John, 1988). Overall, persons vulnerableto depression may marry persons who contribute both to theirdepression and to their difficulties as parents (Quinton, Rutter,& Liddle, 1984). More generally, depressed persons tend to live

in adverse circumstances that may precede, co-occur with, andpersist beyond their depression, providing a plausible alterna-tive explanation for the difficulties that they and their childrenexperience (Brown & Harris, 1978).

This emerging view of depression and its social context hasimportant implications for the study of children of depressedparents. First, it implies that children chosen simply because aparent meets the criteria for major depression will vary greatlyin their exposure to parental depression, to other forms of pa-rental psychopathology, to marital conflict, and to adverse livingconditions. Treating these children as a homogeneous groupmay impede progress in understanding the processes linking pa-rental depression and child adjustment. Second, it implies theneed to develop complex models and research designs to sortthe effects of the various influences on child adjustment andthe need for a cautionary approach to the use of conventionalcontrol and matching strategies (Cicchetti & Aber, 1986;Walker, Downey, & Nightingale, 1989).

Selection of Depressed ParentsDepression has been construed both as a continuum of psy-

chological distress and as a discrete diagnostic entity. Its corre-lates vary with its definition (e.g., Breslau & Davis, 1986; Lew-insohn, Hoberman, & Rosenbaum, 1988). Coyne and Gotlib(1983) reviewed problems that arise when depression is identi-fied from elevated scores on self-report measures rather thanfrom structured diagnostic interviews. They noted that the rela-tion between self-report questionnaire responses and diagnosesof depression based on clinical interviews is equivocal; thatmany persons from community samples who are identified asdepressed from questionnaire responses do not meet the diag-nostic criteria for clinical depression; and that some of the de-pression that is identified in this way is mild and transient. For-tunately, most of the recent studies of adjustment in children ofdepressed parents have used diagnosis as the criteria for paren-tal depression. By contrast, research on parenting by depressedpersons has generally used self-reported depression scores ob-tained from community samples. Thus, there may be markeddiscontinuities between this research and research on the ad-justment problems of children with a depressed parent. Consis-tent with this suggestion, Forehand et al. (1987) concluded thatprior research showed a stronger relation between parental de-pression and child adjustment when depression was identifiedthrough diagnostic interviews than when depression was identi-fied through questionnaire responses.

In the two decades of research on children of depressed per-sons, diagnostic practices and nosological terminology havechanged in ways that can prove confusing. Parents who werediagnosed as manic-depressive solely because of the severity oftheir depression would now be considered unipolar, rather thanbipolar, and the term neurotic depression does not map well onto current diagnostic schemes. As a result, earlier findings thatthe children of neurotically depressed parents showed the high-est level of disturbance are difficult to interpret (Sameroff,Seifer, & Zax, 1982). Many of the parents considered schizo-phrenic in early high-risk studies would not meet Diagnosticand Statistical Manual of Mental Disorders, third edition(DSM-III), criteria for the disturbance, and significant num-bers have been rediagnosed as bipolar disordered (Erlenmeyer-

52 GERALDINE DOWNEY AND JAMES C. COYNE

Kimling & Cornblatt, 1987; Weintraub, 1987; Wynne, Cole, &Perkins, 1987). Thus, early findings that the children of de-pressed and schizophrenic parents show similar difficulties mayreflect the presence of parents in the schizophrenic sample whowould now be considered affectively disordered.

Structured diagnostic interviews such as the Schedule of Affec-tive Disorders and Schizophrenia (SADS) and rigorously definedcriteria such as the Research Diagnostic Criteria (RDC) now allowgreater consistency in the selection of depressed parents. Yet eventhe use of diagnoses derived from the same structured interviewsdoes not ensure that samples of depressed persons are homoge-neous. An important source of heterogeneity is whether depressionco-occurs with a personality disorder. DSM-III and the revisedDSM-IH incorporate a multiaxial diagnostic system that allowsresearchers to code both major disorders such as depression (AxisI) and personality disorders (Axis II). A recent study of consecutivehospital admissions found that 35% of patients with major depres-sion also had a personality disorder (Shea, Glass, Pilkonis, Wat-kins, & Docherty, 1987).

Personality disorders that co-occur with major depression arelikely to have a complex effect on children. Aside from thedifficulties posed by living with a parent who is borderline, pas-sive-aggressive, avoidant, or dependent, Axis II diagnoses areassociated with an earlier onset of depression, poorer recovery,more life stress, and more frequent suicide attempts, all ofwhich may have a negative effect on children (Black et al., 1988;Phofl, Stangl, & Zimmerman, 1984). In fact, parental personal-ity disorder may have a stronger impact on child adjustmentthan parental depression (Rutter & Quinton, 1984).

Characteristics of depressed people also vary by treatmentsetting (Paykel, Klerman, & Prusoff, 1970). How they vary haschanged in the time from the earliest studies of children of de-pressed parents to the present. Antidepressant medication ismore widely used and effective than in the past. Hospitaliza-tions are less frequent and of shorter duration. Treatment hasshifted to primary medical care, and more antidepressants arenow prescribed by primary care physicians than by psychia-trists. For these and other reasons, samples of depressed personsselected from hospitals and clinics where research is conductedwill differ from samples selected from these settings in the pastand from the larger population of depressed persons who aretreated in the community or who go untreated. At the outset ofour review, we should caution that most of the better-designedstudies of children with a depressed parent have used hospitaland clinic samples. Thus, their findings pertain to persons whoare more seriously disturbed than the typical depressed person.For example, in the University of California, Los Angeles(UCLA), Family Stress Project, depressed mothers had a meanof 15.5 lifetime episodes and a mean of 2.1 hospitalizations fordepression (Hammen, Gordon, et al., 1987b). It is importantnot to generalize uncritically from studies of more disturbed,treated people to the population of depressed persons.

Control Groups, Matching, and Statistical ControlsSimple comparisons between children of depressed and non-

depressed parents may suffice for documenting the problemsshown by children of depressed parents. However, if one's goalis explanation, they do not suffice. We have already noted sev-eral ways in which families that include a depressed parent

differ from other families (e.g., high levels of marital conflictand adversity). These differences may provide alternative expla-nations for the association between parental depression andchild problems. These differences may also require substantialqualifications as to when or under what circumstances such anassociation emerges. Conventional wisdom holds that the prob-lem of co-occurring risk factors can be addressed in study de-signs by including appropriate control groups that are matchedwith the target group on factors that are likely to influence ad-justment. Reflecting this view, several high-risk studies ofschizophrenia included children of depressed patients as a com-parison group to test the hypothesis that the problems of chil-dren with schizophrenic parents are due to schizophrenia per seand not to psychiatric disturbance or its correlates (e.g., stress,marital discord). Children of depressed parents were viewed asthe ideal control group because depression and schizophreniaare considered to be equally severe but genetically distinct dis-orders (Kendler, Gruenberg, & Strauss, 1981;Rosenthal, 1970).

Surprisingly few recent studies of children with depressedparents include a nondepressed psychiatric control group. Evenamong those that do, the apparent explanatory power that sucha control group affords may prove illusory, as illustrated by arecent study comparing children of bipolar-disordered parentswith children of nondepressed, psychiatric control parents(Klein, Depue, & Slater, 1985). The bipolar index parents wereselected from an inpatient unit, but the psychiatric control par-ents were drawn from outpatient clinics. The bipolar index par-ents had a mean of 7.7 hospitalizations, whereas the psychiatriccontrol parents had a mean of 0.3 hospitalizations. We cannottell whether the high rate of affective disturbance in the off-spring of bipolar index parents was due to their parent's disor-der or to their parent's repeated hospitalization and its corre-lates. In general, patients with a nonaffective diagnosis willdiffer from depressed patients on several background and clini-cal variables. This limits the extent to which adequate matchingcan be achieved and, thus, the conclusions that can be drawnjustifiably from comparisons.

Most applications of matching across index and controlgroups are fairly mechanical and include efforts to ensure thatthe groups are comparable on demographic variables such age,sex, social class, number of children, and family structure. In-vestigators may be unable to implement more complex match-ing strategies. They seldom have (a) a large enough sample fromwhich to draw matched depressed parents and controls, (b)sufficient knowledge of the characteristics of their patient popu-lations or of the variables most crucial for matching purposes,or (c) an understanding of how matching or selecting on onevariable affects other relevant variables (Walker, Downey, &Nightingale, 1989).

Many factors that affect child outcomes are not amenable tomatching because they cannot be assessed until a study is un-derway. These include marital distress, life events, and parents'current level of psychological distress. Studies that have consid-ered these potential confounding variables have treated them ascovariates (e.g., Beardslee, Schultz, & Selman, 1987; Hammen,Adrian, et al., 1987; Weintraub & Neale, 1984). However, thehigh-risk literature provides several examples of the wisdom oftesting for interactions between risk factors in order to identifythe circumstances under which parental psychopathology andchild disturbance are linked (Walker, Downey, & Nightingale,


1989). For example, Walker, Downey, and Bergman (1989)found that parental psychiatric status and child maltreatmentinteracted such that the behavior problems of children with aschizophrenic parent increased over time only if the childrenalso experienced maltreatment.

Although testing for interactions between parental depres-sion and other contextual factors is clearly warranted, the co-occurrence of risk factors puts practical limits on such tests. Forexample, marital distress is common in families with a de-pressed parent. Testing how marital distress and parental de-pression interact to affect child adjustment requires findingenough maritally distressed families that do not include a de-pressed parent and enough maritally nondistressed familieswith a depressed parent. In practice, this may be a difficult task,and adopting this strategy introduces new problems of interpre-tation. Groups constructed in this way are unrepresentative ofthe larger populations of families of maritally distressed and de-pressed persons. Furthermore, maritally distressed persons whoare not currently depressed are at risk for depression in the nearfuture and are likely to have a history of depression (Lewinsohnetal., 1988).

SummaryWe have identified several aspects of depression and its con-

text that pose difficulties for drawing firm conclusions abouthow the disorder relates to problems in offspring. One mightstart with the suggestion that the study of the effects of depres-sion on offspring is best done by using well-described, homoge-neous samples of depressed parents, appropriate matched con-trol groups, and statistical controls and tests for interactions toexamine possible mediating and moderating influences onthese children. The barriers to implementing such a designshould be becoming apparent, and it should not be surprisingthat few studies come close to meeting this ideal. At present,probably too little is known to make the most effective use ofmatching and statistical controls, but this does not mean re-searchers ought to abandon trying. This judgment may seemtoo pessimistic, but it should be noted that other authors areraising similar doubts about conventional ways of disentanglingcausal influences in the naturalistic environment and are cau-tioning that an uncritical application of these approaches canbe misleading (Kessler, 1987).

Adjustment in Children of Depressed ParentsStudies have investigated the relation between parental de-

pression and a wide range of child outcomes. In this review, wefocus on three sets of outcomes: (a) measures of psychologicalfunctioning, (b) clinical disorders, and (c) potential markers ofvulnerability to depression. We begin with a discussion of high-risk studies of schizophrenia that included children of de-pressed parents as a control group. We then discuss more recentstudies that specifically target children of depressed parents.

High-Risk Studies of SchizophreniaTable 1 summarizes some features of the seven high-risk

studies that included a depressed parent control group (Cohler,Grunebaum, Weiss, Hartman, & Gallant, 1977; Fisher, Kokes,

Harder, & Jones, 1980; Garmezy & Devine, 1984; Goodman,1987; Sameroff, Barocas, & Seifer, 1984; Weintraub & Neale,1984; Worland, Janes, Anthony, McGinnis, & Cass, 1984).Other studies also included depressed parents in their psychiat-ric control group, but early reports from these studies com-bined depressed and nondepressed psychiatric controls (Erlen-meyer-Kimling & Cornblatt, 1987; McNeil & Kaij, 1987).

Beyond consisting of currently or formerly hospitalized pa-tients, the seven groups of depressed parents lacked uniformity.Some studies combined parents with neurotic and psychotic de-pression, whereas others included only parents with psychoticdepression. Three studies were restricted to intact families andfour studies included only families with a depressed mother.The distinction between unipolar and bipolar parental disor-ders was made in just some reports from two studies (Fisher,Schwartzman, Harder, & Kokes, 1984; Weintraub, Winters, &Neale, 1986). All seven studies had a longitudinal componentin their design, but the extent of follow-up varied considerablyand the studies of school-aged children have not yet reportedextensively on their follow-up data.

Parental diagnostic procedures varied across studies from themost reliable diagnostic tools available to hospital chart re-views. Initially, most of the high-risk studies did not use DSM-III or RDC criteria to assess parental diagnosis. Child assess-ment procedures also differed across studies, but all studies in-cluded an assessment of the child's social adjustment. The threestudies of preschool children used some combination of motherand observer reports of child adjustment. The four studies ofschool-aged children included teacher or peer ratings of socialand academic competence. With one exception, the initial childassessments did not use diagnostic procedures because suitableprocedures and criteria were not well developed when the stud-ies began. Moreover, schizophrenia, the outcome of interest,rarely emerges in childhood. The one study that included a di-agnostic assessment began much more recently than the others(Goodman, 1987).

Despite inconsistencies across measures and samples, theschool-aged children of affectively disturbed and schizophrenicparents showed similar deficits in comparison with matched orrandom control children. Effects specifically associated with ei-ther diagnosis were strikingly absent. In some studies, on somemeasures, particularly teacher ratings of academic and socialcompetence, children of schizophrenic and depressed parentsresembled children of normal control parents.

The story is more complex for studies begun with preschoolchildren because the findings from these longitudinal studies var-ied with the children's developmental stage. For example, Cohlerand his colleagues found that the 3-year-old children of depressedand schizophrenic mothers showed similar cognitive deficits, butat age 9 years children of depressed mothers were more impairedthan children of schizophrenic mothers (Cohler et al., 1977;Grunebaum, Cohler, Kaufman, & Gallant, 1978). Sameroff et al.(1984) reported that children of depressed mothers had moreproblems immediately after birth than children of schizophrenicor control mothers, but subsequently resembled control childrenin mental and motor development. Although depressed motherscontinued to report the highest level of child deviance through age4 years, this appears to reflect the mothers' functional impairmentrather than their specific diagnosis. Goodman (1987) also foundsocial and cognitive deficits and adjustment problems in the off-


Table 1High!Risk Studies of Schizophrenia With a Depression Control Group

Study Sample Child ageParent diagnostic

criteriaChild adjustment

data

Emory University Project(Goodman, 1987)

Massachusetts Mental HealthCenter Project (Cohler,Grunebaum, Weiss,Hartman, & Gallant,1977; Gamer, Gallant,Grunebaum, & Cohler,1977)

Minnesota High!Risk Study(Rolf, 1972; Garmezy&Devine, 1984; Rolf &Garmezy 1974)

Rochester Child and FamilyStudy (Fisher etal., 1984;Baldwin, Cole, & Baldwin,1982; Wynne, Cole, &Perkins 1987)

Rochester LongitudinalStudy (Sameroff, Barocas,&Seifer, 1984;Samero«f,Seifer,&Zax, 1982;Sameroff, Seifer, Zax, &Barocas, 1987)

St. Louis High!Risk Study(Worland, Janes, Anthony,McGinnis,&Cass, 1984)

Stony Brook High!RiskStudy (Weintraub &Neale, 1984; Weintraub,1987)

Depressed = 36, schizophrenic = 71,control = 38 (single, low!incomeBlack mothers from treatmentfacility)

Depressed = 12 (psychotic),schizophrenic = 29, control = 57(mothers from intact familiesparticipating inposthospitalization program)

Depressed = 27 (neurotic),schizophrenic = 31, externalizingdisorder =31, internalizingdisorder = 27 (matched andrandomized controls; hospitalrecords, mothers)

Psychotic depressed = 38,nonpsychotic depressed = 39,schizophrenic = 20, nonpsychoticnondepressed = 28 (boys fromintact, middle! to upper!SESfamilies; no normal controlgroup; records of previouslyhospitalized patients)

Depressed = 58 (neurotic),schizophrenic = 29, personalitydisordered = 40, control = 57(mothers; sample obtained bycrossing maternity cases withpsychiatric register)

Depressed = 60 (12), schizo!phrenic = 100 (11), medicalcontrol = 78 (20), control = 130(50), (hospital inpatients; intactfamilies)

Depressed = 91, schizophrenic =153, matched and random = 443,controls = 153, (hospitalinpatients)

Longitudinal study: 0!5 years; DSM!III, based on Observer!rated social3 assessments, 1 year apart hospital records competence,

developmental andDSM!IIIdiagnosticassessments

IQ and attentionaltasks

Longitudinal study: 3 years,5!6 years, 8!12 years

9!12 years

4, 7, or 10 years

Longitudinal study: birth, 3,12, 30, & 48 months

6!20 years

6!16 years

DSM!II, based onpsychiatrist'sratings ofsymptoms

DSM!II, based on Teacher and peerhospital records ratings of social

competence

DSM!III, based on Teacher and peerstructuredinterview(present stateinterview; Wing,Cooper, &Sartorius, 1976).

DSM!II, based onstructuredinterview (CAPPS;Endicott &Spitzer 1972).

DSM!II, based onhospital records.Recent reportsused DSM!III.

DSM!II, based onstructuredinterview (CAPPS;Endicott &Spitzer, 1972)

ratings ofcompetence

Standardizeddevelopmentalassessments;maternal andobserver ratings;at 4 years onlymaternal ratingsof child obtained

Clinical ratings ofpsychopathologybased on IQ,visual!motor, andprojective tests;teacher!ratedcompetence

Teacher and peerratings ofcompetence,attentionalmeasures, and IQ

Note. DSM = Diagnostic and Statistical Manual of Mental Disorders, II = second edition, III = third edition; SES = socioeconomic status, CAPPS =Current and Past Psychopathology Scales. Numbers in parentheses are " for recent reports.

spring of both schizophrenic and depressed parents. The groupthat appeared most impaired depended on what was measuredand when it was measured. Rates of DSM!III disorders were lowin children of both schizophrenic and depressed parents at all threeassessments. However, rates of diagnosis (usually developmentaldelays) for these children doubled from 5% to 10% between thefirst and final assessment, whereas diagnoses were notably absentin the control children.

Despite the inconsistencies across time and measures, thesestudies suggest that, like school!aged children, infants and pre!school children with a depressed parent are more impaired than

control children but resemble young children with a schizo!phrenic parent.

Similarity in children of depressed and schizophrenic parents.How should the similarity in adjustment between children ofdepressed and schizophrenic parents be interpreted? One possi!bility is that the specific parental diagnosis is less relevant tochild adjustment than other dimensions of parental psychopa!thology (Watt, 1984). Indeed, the studies that investigated chro!nicity and severity of impairment have found that these factorsare stronger predictors than diagnosis of child adjustment(Harder, Kokes, Fisher, & Strauss, 1980; Sameroff et al., 1984;


Wynne et al., 1987). We might have reached a different conclu-sion if the studies had included diagnostic assessments or mea-sures that were more germane to depression and its risk factors.

Lack of specificity may also reflect the widespread use ofDSM, second edition (DSM-II), criteria in ascertaining paren-tal diagnosis. In several high-risk studies, rediagnosis usingDSM-IH criteria has led to sizable decreases in the schizo-phrenic parent sample and corresponding increases in the bipo-lar parent sample (e.g., Weintraub, 1987; Wynne et al., 1987).The Stony Brook High-Risk Study found some evidence thatthe children of DSM-II schizophrenic parents who also metthe DSM-IH criteria for schizophrenia were more maladjustedthan children of parents who were rediagnosed as affectivelydisordered (Weintraub, 1987). When other high-risk research-ers complete their rediagnosis of parents by means of DSM-IHcriteria, children of schizophrenic parents may yet appear moreimpaired than children of affectively disturbed parents.

Even if children of schizophrenic and depressed parents con-tinue to show similar levels of adjustment following rediagnosiswith DSM-IH criteria, two issues warrant consideration. First,similar adjustment patterns in children of schizophrenic anddepressed parents need not imply that the underlying processesare similar for both groups. Illustrating this point, Emery,Weintraub, and Neale (1982) found that although children ofschizophrenic and depressed parents showed similar levels ofmaladjustment, marital discord explained the adjustmentproblems in children with a depressed parent but not in chil-dren of schizophrenic parents.

Second, the prognostic value of childhood adjustment prob-lems may not be the same for the offspring of depressed andschizophrenic parents. For example, there is suggestive evidencethat low social competence is a precursor of schizophrenia butnot of depression. In a follow-back study of school records, chil-dren who later developed schizophrenia showed low social com-petence, but the social competence of children who became de-pressed resembled that of controls (Lewine, Watt, Prentsky, &Fryer, 1980).

Follow-up diagnostic studies. The children in most of thehigh-risk studies have now entered late adolescence and adult-hood, and diagnostic assessments of them are underway. Thesedata will permit a prospective test of the validity of potentialrisk indicators identified in childhood. The rediagnosis of manyschizophrenic parents as bipolar disordered and the increasinginterest in children of depressed parents has resulted in morecareful and systematic reports about children of depressed par-ents. Thus, these studies have the potential to further under-standing of the long-term effects of having a depressed parent.

In the studies that have reported follow-up results, distur-bance rates were higher in the offspring of both schizophrenicand affectively disordered parents than in normal control chil-dren. Children of schizophrenic parents showed the highestoverall rates of any disorder (Erlenmeyer-Kimling & Cornblatt,1987; Weintraub, 1987). Only the Stony Brook High-Riskstudy has reported rates of specific disorders (Weintraub, 1987).Children of affectively disturbed parents showed high rates ofaffective disorders, whereas children of schizophrenic parentsshowed high rates of borderline and personality disorders(Weintraub, 1987). Although this finding awaits replication, itsuggests that "depression breeds true" and anticipates the re-

sults of studies with control groups of medically ill parents tobe discussed later.

Lessons from high-risk research. Beyond substantive find-ings, the high-risk studies of schizophrenia have provided valu-able lessons about the pitfalls of carrying out this type of re-search. First, they highlight the difficulties in interpreting find-ings from studies in which psychiatric groups are not welldescribed. Second, they illustrate the problems involved in fol-lowing high-risk samples longitudinally (Lewine, Watt, &Grubb, 1984). These include differential rates of attritionacross study groups and increases in heterogeneity over time insamples originally selected to be homogeneous. Families withthe highest levels of psychosocial risk are the most likely to dropout. Families with a disturbed parent that were initially selectedto be intact show higher rates of marital disruption than controlfamilies. These processes can seriously compromise research-ers' ability to distinguish developmental changes in childrenfrom changes that are due to the changing composition of thesample.

Although the results of the long-term, follow-up studies of thehigh-risk samples are beginning to emerge, investigators havenot yet examined whether early maladjustment predicts laterdiagnostic outcomes. Instead, they treat each wave as a separatecross-sectional assessment. Thus, the prognostic value of theintermediate risk markers identified during childhood remainunclear.

Clearly, these long-term, longitudinal, high-risk studies haveproved difficult to implement successfully. Nonetheless, it is im-portant to acknowledge the sophistication of some aspects of thestudy designs and the breadth of domains of child and familyfunctioning assessed. The St. Louis High-Risk Study is particu-larly impressive (Worland et al., 1984). This study incorporatedboth medical and psychiatric control groups and obtained nat-uralistic observations of family interactions at home, clinicalinterviews, and child cognitive and psychomotor assessments.Without an understanding of the methodological and logisticaldifficulties that these studies encountered, subsequent studieson children of depressed parents are in danger of succumbingto similar problems. We now turn to these more recent studies.

Recent Studies of Children of Depressed Parents

Psychological functioning. Early, uncontrolled studies of chil-dren with a depressed parent provided suggestive evidence thatthese children were at risk for a wide range of problems in psy-chological functioning (Beardslee et al., 1983). More recentstudies that included control groups have confirmed this assess-ment. These studies are summarized in Table 2. All of the stud-ies used standardized diagnostic criteria (e.g., DSM-IH orRDC) to identify depressed parents. Parents were usually re-cruited from inpatient or outpatient treatment facilities, andstudies varied in whether they combined parents with mild andmajor depression or included only parents with major depres-sion. Samples also varied in the age composition of the children,family structure, and sex of the depressed parent. Althoughmeasures of psychological functioning differed across studies,most used valid and reliable instruments (e.g., Child BehaviorChecklist; Achenbach & Edelbrock, 1983).

School-aged children of depressed parents generally showhigher levels of both externalizing and internalizing symptoms


than normal control children. This finding emerges in reportsby teachers (Lee & Gotlib, 1989a, 1989b; Richters & Pelligrini,1989), by parents (Billings & Moos, 1983; Breslau, Davis, &Prabucki, 1988; Lee & Gotlib, 1989a, 1989b; Richters & Pelli-grini, 1989), and by the children themselves (Breslau et al.,1988; Hirsch, Moos, & Reischl, 1985). Other indicators of mal-adjustment in these children include higher levels of treatmentfor psychiatric disturbance (Hammen, Gordon, et al., 1987b;Klein, Clark, Dansky, & Margolis, 1988; Klein et al., 1985; Or-vaschel, Welsh-Allis, & Weijai, 1988; Weissman, 1988), higherlevels of functional impairment (Beardslee et al., 1987; Kleinet al., 1988; Lee & Gotlib, 1989a; Weissman, Gammon, et al.,1987), and a higher proportion of children scoring in the clini-cal range of symptom checklists (Lee & Gotlib, 1989a). Thesechildren also have deficits in social (Hammen, Gordon, et al.,1987b; Richters & Pelligrini, 1989) and academic (Billings &Moos, 1983; Hammen, Gordon, etal., 1987b; Weissman, Gam-mon, et al., 1987) competence that are not due to intellectuallimitations. Finally, they are in poorer physical health than chil-dren of control parents (Billings & Moos, 1983; Weissman,Gammon, et al., 1987).

Little is known about the impact of parental depression onthe adjustment of infants and toddlers beyond the knowledgegained from the high-risk studies of schizophrenia. In thesestudies, children of depressed parents were more poorly ad-justed than normal control children, but were similar to chil-dren of schizophrenic parents (Cohler et al., 1977; Goodman,1987; Sameroff et al., 1984). The only additional study of ad-justment in young children with a depressed parent reportedthat these children showed symptoms of depression and antiso-cial behavior (Gaensbauer, Harmon, Cytryn, & McKnew, 1984;Zahn-Waxier et al., 1984,1988). These symptoms were evidentduring the children's second year of life and continued to char-acterize the children at age 6 (Zahn-Waxier et al., 1988). Thesample consisted of 7 male offspring of bipolar parents, 5 ofwhom had a unipolar-disordered spouse. The small size anduniqueness of the sample limits the generalizability of thesefindings. Nonetheless, the study shows that it is possible to iden-tify depressive tendencies in the offspring of affectively dis-turbed parents early in life, and that these problems continue.

Although this body of studies confirms that children of de-pressed parents are at heightened risk for general adjustmentproblems even in early childhood, issues of specificity persist. Inaccord with findings from the schizophrenia high-risk studies,three studies reported that children of medically or nonde-pressed psychiatrically ill parents were indistinguishable fromchildren with a depressed parent (Hammen, Gordon, et al.,1987b; Hirsch et al., 1985; Lee & Gotlib, 1989a, 1989b). Theremaining studies included only a normal control group. Thislack of specificity suggests that general measures of child malad-justment do not reveal the distinct consequences of parentaldepression. However, a failure to reject the null hypothesis isdifficult to interpret, for two reasons. First, the small samplesused in these studies has meant that their power to detect groupdifferences is low. Second, the studies provide little informationabout the medical or nondepressed psychiatric parent groups.For example, Hirsch et al. (1985) included a medical controlgroup of women with arthritis, but did not assess the psychiatrichistory of these women. Lee and Gotlib (1989a, 1989b) con-cluded that nonsignificant differences in adjustment between

children of depressed and nondepressed psychiatric outpatientsimplied that parental psychiatric status is more important forchild adjustment than is depression. Yet Lee and Gotlib alsoreported that children of depressed parents were twice as likelyas children of nondepressed psychiatric patients to score in theclinical range of the adjustment measure. Here, as elsewherein considerations of nonspecificity, null findings are difficult tointerpret.

Clinical diagnoses. Only one of the studies reviewed byBeardslee et al. (1983) used diagnostic criteria and included acontrol group (Welner, Welner, McCrary, & Leonard, 1977). Inthis study, 7% of children with a depressed parent met the diag-nostic criteria for adult depression and 25% showed consider-able depressive symptomatology, whereas none of the controlchildren were depressed. The findings from nine additionalstudies now confirm that children of depressed parents are in-deed at heightened risk for affective diagnoses. These studieswere based on independent samples of school-aged and youngadult offspring, included a control group, and used DSM-IHor RDC diagnostic criteria based on structured interviews withthe child and, with two exceptions, the parent. This emphasison diagnosis was possible because of recent developments innosology and structured interviews with children (Carlson &Garber, 1986). The use of direct interviews with children is animportant improvement, given concerns about parents' abilityto provide reliable reports about children's depressive symp-tomatology. Comparisons of mothers' and children's reportsshow that they generally agree about externalizing symptoms,but that mothers' estimates of their children's depressive symp-toms are considerably lower than the children's own estimates(Breslau et al., 1988; Weissman, Wickramaratne, etal., 1987).

Table 2 gives details of the nine studies, and Table 3 summarizestheir findings. Five studies focused primarily on parents with aunipolar disorder; three focused on bipolar-disordered parents.Only one study included both types of disorder (Hammen, Gor-don, et al., 1987b). As with the high-risk studies of schizophrenia,studies varied considerably in sample composition. Some studiesincluded only depressed mothers and some included only childrenin late adolescence and young adulthood.

In each of the nine studies, children of depressed parentswere more likely to receive a diagnosis than control children.However, affective disturbance was the only diagnosis that chil-dren of depressed parents received significantly more often thancontrols across the studies (see Table 3). The rate of any affectivedisorder was three times higher in children of unipolar-disor-dered parents and 1.75 times higher in children of bipolar-disor-dered parents than it was in control children. Children of uni-polar-disordered parents were at particularly high risk for majordepressive disorders (MDD); their rate of MDD was six timesthat of control children. Children of bipolar-disordered parentsshowed significantly higher rates of MDD in just one of thethree studies that reported this disorder. The children in thisstudy were older (15-23 years) than children in the other twostudies, suggesting that serious depressive disorders may have alater onset in children of bipolar-disordered parents.

Unlike depression, full-blown mania was generally absent inthe offspring of either unipolar- or bipolar-disordered parents.However, one study found significant evidence of cyclothymiain adolescent and young adult offspring of bipolar-disorderedparents (Klein et al., 1985). The cases of manic disorders (bipo-


lar 2, hypomania, cyclomania) found in all nine samples wererestricted to the offspring of depressed parents. The low level ofmanic or bipolar cases may reflect the age span covered by thestudies. Manic episodes are thought to occur rarely in prepubes-cent children (Depue & Munroe, 1978), and parent and childreports of manic symptoms are unreliable (Edelbrock, Costello,Dulcan, Kalas, & Conover, 1985).

There were no consistently reliable differences between chil-dren of depressed and nondepressed parents in rates of any spe-cific nonaffective disorder. Nonetheless, most other disordersoccurred more often in the children of depressed parents thanthey did in control children. Children of depressed parentsshowed significantly higher rates of conduct disorder in threestudies (Beardsleeetal., 1987; Hammen, Gordon, etal., 1987b;Klein et al., 1988); in one study, they showed significantlyhigher rates of attentional deficit disorder (Orvaschel et al.,1988); and in three studies, they showed significantly higherrates of substance abuse (Beardslee et al., 1987; Hammen, Gor-don, et al., 1987b; Weissman, 1988). Some studies also reportedthat children of depressed parents were more likely to receivemultiple diagnoses than were control children (Decina et al.,1983; Klein et al., 1988; Orvaschel et al., 1988; Weissman,Gammon, et al., 1987).

It is unclear whether the high rates of depression in childrenof unipolar-depressed parents are a specific consequence of pa-rental depression, because only two studies included a high-riskcontrol group. Both studies found significantly higher rates ofaffective disorders in children of depressed parents than in chil-dren of medically ill or psychiatric control parents (Hammen,Gordon, et al., 1987b; Klein et al., 1988). By contrast, one ofthe studies reported that these two groups of children were in-distinguishable on a measure of general adjustment (Hammen,Gordon, et al., 1987). When considered together with the resultsfrom the Stony Brook High-Risk Study (Weintraub, 1987),these findings suggest that clinical depression may be a specificconsequence of parental clinical depression. Thus, future stud-ies that aim to identify distinct effects of parental depressionshould not limit their child assessments to general adjustmentmeasures.

Risk factors. Unlike the schizophrenia high-risk studies,identifying premorbid risk markers was not the initial goal ofthe high-risk studies of depression. However, several promisinglines of inquiry are now underway that draw on current theoriesand empirical investigations of vulnerability to depression inadulthood. Putative risk factors being investigated includeaffect regulation, the handling of separation and failure, attribu-tional style, and negative self-views.

Problems with separation from a loved one have long beenimplicated as a risk factor in adult depression (Bowlby, 1980).Such difficulties may be exacerbated in persons who developinsecure attachments to their caretaker in infancy (Cummings& Cicchetti, in press). Parent-child attachment has been stud-ied most extensively in the National Institute of Mental Health(NIMH) program of research on offspring of affectively dis-turbed parents. Gaensbauer et al. (1984) found that, at 18months, six of the seven children with a bipolar parent wereinsecurely attached to their mother, whereas only one of thecontrol children was not securely attached. High rates of inse-cure attachment were found also in a larger sample of motherswith bipolar disorder (79%) or major depression (47%), espe-

cially when the disorder was severe (Radke-Yarrow, Cummings,Kuczynski, & Chapman, 1985). The rate of insecure attach-ment in both the control and minor depression groups resem-bled the general population rate (25%-30%). Insecure attach-ments were characterized primarily by avoidance or disorga-nized, disoriented behavior previously found in severely abusedchildren (Crittenden, 1988). It is not yet known whether suchpatterns of attachment predict later depression as distinct fromother forms of maladjustment.

The regulation of emotion and social interaction is also beingexplored as a potential risk factor for affective disturbance inboth NIMH samples. When observed interacting with a peer atage 30 months and again at age 6 years, the seven children ofbipolar-disordered parents showed problems with empathy andunusual responses to conflict (Zahn-Waxier, Cummings, Mc-Knew, & Radke-Yarrow, 1984; Zahn-Waxier et al., 1988). Inassessing these findings, Zahn-Waxier et al. (1984) suggestedthat "emotional disregulation, accompanied by poor social re-lationships evidenced in an impoverished ability to give to oth-ers, illustrates one process by which depression may develop"(p. 240). Confirmatory evidence of these difficulties comes fromthe larger NIMH study of children of unipolar- and bipolar-dis-ordered and control mothers (Zahn-Waxier, Kochanska, Krup-nick, & McKnew, 1990). Children with affectively disturbedmothers responded to protective tests with exaggerated andcomplex representations of interpersonal conflict and distress.

On the basis of a related line of research, Klein and his col-leagues also argue that subdromal problems with affect regula-tion (dysthymia or cyclothymia) in childhood index risk forlater affective disturbance (Klein et al., 1985, 1988). Evidenceof elevated levels of subdromal cyclothymia in the offspring ofbipolar-disordered parents and of dysthymia in the offspring ofunipolar-disordered parents are consistent with this argument.However, as with the NIMH research, the prognostic value ofthese findings and their specificity to parental affective distur-bance have yet to be established.

Only the UCLA Family Stress Project has investigated thelinks among parental affective disturbance, risk factors, andchild diagnostic outcomes. In an ambitious program of re-search, Hammen and her colleagues have tested cognitive theo-ries of vulnerability to depression in high-risk children (Ham-men, 1988; Hammen, Adrian, & Hiroto, 1988; Jaenicke et al.,1987). Specifically, they examined whether a negative self-con-cept and a negative attributional style (i.e., a tendency to makeglobal, internal, stable attributions about negative events; Peter-son & Seligman, 1984) were more characteristic of childrenwith an affectively disturbed parent than of control children andwhether these cognitions predicted subsequent changes in childadjustment. Children of unipolar- and bipolar-disordered par-ents had a significantly more negative self-concept and a morenegative attributional style than children of normal parents. Anegative self-concept predicted increases in both affective andnonaffective diagnoses over a 6-month period. Contrary to the-oretical expectations, a negative attributional style predicted in-creases only in nonaffective diagnoses; it did not predict depres-sion. These results confirm the need to go beyond identifyingheightened levels of theoretically relevant variables in childrenof depressed parents in order to ask what these findings implyfor child adjustment.

These programs of research on risk factors for later depres-


Table 2Studies of Children With a Depressed Parent

Study

Beardslee, Schultz,&Selman(1987); Kelleretal. (1986)

Billings & Moos(1983, 1985)

Breslau, Davis, &Prabucki(1988)

Connors,Himmelhoch,Goyette, Ulrich,& Neil (1979)

Cytryn, McKnew,Bartko, Lamour,& Hamovit(1982)

Decina et al.(1983)

Gershon et al.(1985)

Hammen, Adrian,etal. (1987);Hammen,Gordon, et al.(1987b)

Hirsch, Moos, &Reischl(1985)

Klein, Clark,Dansky, &Margolis(1988)

KJein, Depue, &Slater (1985)

Sample

Depression = 108 (mainlyunipolar), control = 64(research and treatmentfacilities andcommunity facilities)

Unipolar = 218, con-trol = 236, (inpatientand outpatientfacilities)

Unipolar = 55 (major),control = 278 (mothersfrom communitysample)

Bipolar = 16, unipolar =43 (affective disordersclinic)

Depressed = 19 (major),control = 21(hospitalized atNational Institute ofMental Health [NIMHclinic])

Bipolar =31, control =18 (outpatient clinic)

Bipolar = 29, control =37 (NIMH affectivedisorders clinic)

Unipolar = 19 (major),bipolar = 16, medicalcontrol = 18, control =39 (middle SESmothers in treatmentfor depression; medicalcontrols had chronicarthritis or diabetes.)

Unipolar = 16, medicalcontrol = 16, control =16(unipolarsfromtreatment facilities;medical controls hadarthritis.)

Unipolar = 47 (major),medical control = 33,control = 38 (mainlyconsecutive hospitaladmissions)

Bipolar = 37, nonaffectivepsychiatric control =22 (bipolars recruitedfrom inpatients,controls fromoutpatients.)

Child age

11-19 years

Up to 20 years

8-23 years

1-18 years

5- 15 years

7- 14 years

6- 17 years

8- 16 years

12-18 years

14-22 years

15-21 years

Parent diagnosticcriteria

RDC, based onSADS-L

RDC, based onhospital records

DSM-lII, based onDiagnosticInterviewSchedule

Feighner, Robins,Guze,Woodruff, &Munoz(1972)criteria

RDC, based onSADS-L

RDC, based onSADS-L, RDCinterview

RDC, based onSADS-L

RDC, based onSADS-L; BeckDepressionInventory;family straininterview

RDC, based onhospital records

RDC, based onSADS-L, RDCinterview guide,& hospitalrecords

DSM-III, based onSADS-L

Child data

DSM-III, based on DiagnosticInterview for Children andAdolescents interviews withchild and mother; RochesterAdaptive BehaviorInventory; social cognitivemeasures

Parents completedquestionnaires about child'smental and physical healthand functioning (Health &Daily Living Form). Processrepeated 1 year later.

DSM-III symptom scales,based on DiagnosticInterview Schedule forChildren interview withchild and DiagnosticInterview Schedule forChildren — Parent Versioninterview with mother

Both parents completed theConnors ParentQuestionnaire to assessbehavioral disturbance.

DSM-III, based oninformation from parents,child, and school; used tocomplete Psychiatric &Affective Rating Scales

DSM-III, based onsemistructured interviewwith parents and evaluationof child using Mental HealthAssessment Form and GAS;IQ; Rorschach

DSM-III, based on K-SADSinterviews with mother andchild

DSM-III, based on K-SADS;mother's CBCL; child socialcognition and depression;Connors's Teacher RatingScale

Children completed theHopkins (Psychiatric)Symptom Checklist

DSM-III, based on SADS-Linterview with child; lifeactivities interview

DSM-III, based on SADS-Linterview with child


Table 2 (continued)

Study

Lee&Gotlib(1989a, 1989b)

Orvaschel, Welsh-Allis, & Weijai(1988)

Welner, Welner,McCrary, &Leonard (1977)

Weissman,Gammon et al.,(1987);Weissman(1988)

NIMH studies(Gaensbauer,Harmon,Cytryn, &McKnew 1984;Zahn-Waxler,Cummings,McKnew, &Radke- Yarrow,1984)

Radke- Yarrow,Cummings,Kuczynski, &Chapman(1985)

Richters &Pelligrini(1989)

Sample

Depressed = 16,psychiatric control =10, medical control =8, control = 27(outpatient facilities)

Unipolar = 6 1 (major),control = 46 (researchtreatment facility)

Unipolar = 75 (major),control = 152 (middleSES, inpatients)

Unipolar = 125 (major),control = 95 (researchtreatment facility)

Bipolar = 7, control = 20(bipolar parent was inNIMH treatmentprogram; male childrenfrom intact families)

Bipolar = 14, unipolar =42 (major), unipolar =12 (minor), control =3 1 (community sampleof mothers)

Depression = 26 (inepisode), depression =

Child age

7-13 years

6- 17 years

6-16 years

6-23 years

Longitudinal study:2-3 years, 6years

2-3 years

9 years (study islongitudinal)

Parent diagnosticcriteria

DSM-III, based onSADS

DSM-III, based onSADS-L

Feighner et al.(1972), basedon intakeinformation

RDC, based onSADS-L



RDC, based onSADS-L

Child data

Child Assessment Schedule;CBCL (parent and teacherversions)

DSM-III, based on K-SADSinterviews with mother andchild; IQ; teacher CBCL

Mother and child interviewsabout developmental history,adjustment, and symptomsofpsychopathology

DSM-III, based on K-SADSinterviews with mother andchild; IQ; GAS; schoolperformance; treatmenthistory

Attachment during strangesituation at 12 and 18months; observer ratings ofchild behavior duringinteractions with peers andadults at 30 months and at 6years

Attachment during strangesituation procedure

Parent and teacher forms ofthe CBCL

27 (in remission),control = 25(community sample ofmothers)

Note. RDC = Research Diagnostic Criteria, SADS = Schedule of Affective Disorders and Schizophrenia, L = Lifetime; DSM-III = Diagnostic andStatistical Manual of Mental Disorders, third edition; SES = socioeconomic status; GAS = Global Assessment Scale; CBCL = Child BehaviorChecklist. K-SADS = Kiddie Schedule for Affective Disorders and Schizophrenia.

sion provide several promising lines of inquiry. Testable theo-ries relating childhood risk factors with adult depression areclearly needed because clinical depression does not usuallyemerge until adulthood. We must remember, however, that ademonstration of elevated levels of a potential risk factor is aninsufficient test of a theory. One must also show that the puta-tive risk factor predicts later psychopathology.

Summary: Adjustment of ChildrenWith a Depressed Parent

Children of depressed parents show heightened rates of gen-eral problems in adjustment, putative markers of risk for de-pression, and clinical depression. Difficulties are found even ininfancy. They emerge in peer, teacher, and observer reports aswell as in self- and parent reports. The general adjustment prob-lems shown by children of depressed parents include social andacademic difficulties at school and internalizing and externaliz-ing behavior problems. In showing these problems, they resem-

ble other children who experience the stress and disruption thataccompany serious parental psychiatric or medical illness.

We do not yet know whether the putative risk markers fordepression are a unique consequence of parental depression be-cause most studies did not include the appropriate controlgroups. Existing results question the specific prognostic valueof some of the risk markers and caution against inferring riskfor depression from elevated rates of proposed risk factors.

There is some suggestion that depression, and especially ma-jor depression, is a more specific consequence of having a uni-polar-depressed parent. Depression is the only diagnosable dis-order for which children of depressed parents show significantlyheightened risk. The risk for bipolar disorder in children witha bipolar-disordered parent is not yet established because fewstudies have investigated the offspring of bipolar-disorderedparents and because manic episodes have a later onset than de-pressive episodes.

Several caveats are in order. First, there is little agreementbetween mothers and children about children's affective symp-


Table 3Diagnostic Status of Children of Affectively Disordered and Control Parents(Percentage of Sample in Each Diagnostic Category)

Study Age NAny AW-///

diagnosis

Affective

Any MajorNonaffective

only

Control

Beardslee, Schultz, & Selman(1987)

Cytryn, McKnew, Bartko,Lamour, & Hamovit(1982)

Decinaetal. (1983)Gershonetal.(1985)Hammen, Gordon, et al.

(1987b)Klein, Depue, & Slater

(1985)Klein, Clark, Dansky, &

Margolis(1988)Orvaschel, Welsh-Allis, &

Weijai(1988)Weissman(1988)

11-19

5-157-146-17

8-16

15-21

14-22

6-176-23

64

211837

39

22

38

4697

—

—551

29

18

24

1557

23

230

38

17

5

0

4—

2

8—14

9

5

0

013

—

—513

12

13

14

11—

Unipolar

Beardslee etal. (1987)Cytryn etal. (1982)Decinaetal. (1983)Gershon etal. (1985)Hammen etal. (1987)Klein etal. (1985)Klein etal. (1988)Orvaschel etal. (1988)Weissman(1988)

11-195-157-146-178-16

15-2114-226-176-23

10819

——19

—4761

125

————74—514176

3870——74—3221—

2623——47—

91528

————0—1920—

Bipolar

Beardslee etal. (1987) 11-19Cytryn etal. (1982) 5-15Decinaetal. (1983) 7-14Gershon etal. (1985) 6-17Hammen etal. (1987) 8-16Klein etal. (1985) 15-21Klein etal. (1988) 14-22Orvaschal etal. (1988) 6-17Weissman(1988) 6-23

31291637

52729243

26416738

10253

2631255

Note. DSM-III = Diagnostic and Statistical Manual of Mental Disorders, third edition.

toms, and the diagnostic rules used in the face of inconsistenciesare never explicit. Second, most studies compared only childrenof depressed and normal parents. Thus, conclusions about thespecific risk for major depression associated with parental de-pression are based on only two studies that also included a med-ical control group (Hammen, Gordon, et al., 1987b; Klein etal., 1988). Third, the studies focused primarily on the preadultyears. Because the study of childhood depression is just begin-ning, it is not yet known whether what is being observed is aprecursor of adult depression or a short-term stress reactionwith limited developmental implications.

Depression is a heterogenous episodic phenomenon. Childadjustment is likely to reflect the type of depression, the degreeof family stress and marital discord experienced, and the par-ent's level of social impairment as distinct from affective symp-

tomatology. In addition, child adjustment may respond tochanges in the parent's symptomatology. With a few notableexceptions, researchers have ignored the heterogeneity of affec-tive disturbance. For example, only two studies have comparedthe adjustment of children of unipolar- and bipolar-disorderedparents (Conners, Himmelhoch, Goyette, Ulrich, & Neil, 1979;Hammen, Gordon, et al., 1987b). Consistent with recent find-ings from the high-risk studies of schizophrenia (Fisher et al.,1984; Weintraub, 1987), these studies depict children of unipo-lar-disordered parents as somewhat less socially competent andmore clinically disturbed than children of bipolar-disorderedparents. A possible explanation for these differences is that thecourse of illness may be chronic for a higher proportion of uni-polar-disordered parents than bipolar-disordered parents.

The effect on children of whether a parent is currently in an


episode has also been neglected. However, available evidencesuggests that child adjustment does not fluctuate as parents movein and out of episodes (Billings & Moos, 1985; Richters & Pelli-grini, 1989). Perhaps the chronic impairment and family stressthat often accompany parental depression have more serious im-plications for child development than the acute impairment asso-ciated with acute depressive episodes (Richters, 1987).

In summary, parental depression places children at consider-able risk for adjustment problems in a variety of domains, in-cluding depression. The question that we must now address ishow and under what circumstances parental depression islinked with child disturbance.

Explaining the Link Between Parental Depressionand Child Adjustment

In the introduction we identified three possible, but nonexclu-sive, explanations for maladjustment in children of depressedparents: genetic factors, parenting by a depressed person, andcharacteristics of the depressed parent's broader interpersonalcontext that may provide an alternative explanation for both theirchildren's problems and their own depression. Our methodologi-cal review suggested a further possibility. The impairment of de-pressed persons aside from diagnosis (e.g., personality disorder,enduring distress) may explain their children's problems.

The genetic hypothesis is implicit in studies of children witha psychiatrically disturbed parent. However, direct tests of thishypothesis are difficult because trait markers of genetic liabilityhave not yet been identified. Twin and adoption studies provideindirect tests of this hypothesis. As we have noted, the resultsof these studies confirm that these children's problems cannotbe attributed solely to genetic factors. In fact, the results of onemajor adoption study of children of affectively disturbed par-ents led the investigator to doubt that these children's risk foradult affective disturbance is discernible during childhood (Ca-doret, 1983). The adoptive parents of children who became de-pressed as adults reported that these children were not particu-larly symptomatic during childhood. Cadoret (1983) has sug-gested that affective disturbance in young children living withan affectively disturbed parent is due to stress and that "if thereis a genetic diathesis, it may not manifest itself early in life"(p. 65). However, this conclusion may be more applicable tochildren of bipolar-disordered than of unipolar-disordered par-ents (Cadoret et al., 1985).

Of the three social transmission models outlined, the modelthat has received the most attention proposes that there is some-thing uniquely depressing about the parenting behavior of a de-pressed person, or more specifically, a depressed mother. Thismodel has the virtue of simplicity, but, as we shall see, it is be-coming clear that the processes linking parent and child depres-sion are considerably more complex.

Parenting Behavior of Depressed Persons

Research on the parenting practices of depressed parents re-flects the growing interest in the interpersonal context of depres-sion. Coyne (1976a) showed that depressed persons evoke nega-tive reactions in strangers following brief interactions. Re-searchers have since documented several characteristics ofdepressed persons' behavior during social encounters that may

explain this response, although these differences are not alwaysspecific to depression (Coyne, Birchill, & Stiles, in press; Youn-gren & Lewinsohn, 1980). First, depressed persons' rate of be-havior is reduced. In social situations they speak less often andwith less intensity; they gaze at their partner less frequently; andthey respond more slowly. Second, they are more hostile andirritable, especially in intimate relationships. Third, they actmore depressed. This is evident in their sad and anxious facialexpression and posture, and in the content of their speech.Overall, these behaviors are unlikely to foster positive relation-ships with significant others.

This may be especially true of the parent-child relationship.Parenting is a particularly complex form of social interaction.The sustained effortful behavior that it involves is likely toprove difficult for depressed parents, especially when their chil-dren are young and exaggerated affective tone and a high toler-ance for aversive behavior are required. When interacting withher young child, the typical mother gears her behavior towardfostering sustained positive interaction and expresses a highlevel of positive affect. Hostility and aggression are seldom seen.Mothers maintain the interaction by responding quickly andcontingently with exaggerated facial expression and vocal into-nation (Cohn & Tronick, 1987).

Depressed mothers have been described in the clinical litera-ture as experiencing difficulties in the parenting role that reflectthe symptoms of their disorder (Burbach & Borduin, 1986). Forexample, Weissman and Paykel (1974) observed that "at thesimplest level, the helplessness and hostility which are associ-ated with acute depression interfere with the ability to be awarm and consistent mother" (p. 121). Fisher et al. (1980) alsosuggested that "depressed patients may display high degrees ofnon-acknowledgement such that they do not interact meaning-fully with the child" (p. 354). In general, depressed mothersview the role of parent less positively than do control mothers.They experience negativity toward the demands of parenthoodand feelings of rejection and hostility toward their child (Col-letta, 1983; Davenport, Zahn-Waxier, Adland, & Mayfield,1984; Webster-Stratton & Hammond, 1988; Weissman & Pay-kel, 1974). They also perceive themselves to be less competentand adequate than other parents (Davenport et al., 1984; Flem-ing, Ruble, Flett, & Shaul, 1988; Rutter, 1966; Webster-Stratton& Hammond, 1988; Weissman & Paykel, 1974).

Evidence that depressed mothers also show difficulties duringinteractions with their children comes from observational stud-ies of clinically depressed mothers and of mothers with elevateddepressive symptomatology. The behaviors that characterizedepressed persons' interactions their children resemble the be-haviors that characterize depressed persons' interactions withadults. These studies are summarized in Table 4. The studiesare based mainly on observations of mothers during short struc-tured or semistructured interactions with their infant or tod-dler, although some studies have focused on school-aged chil-dren. None of the studies observed depressed fathers. Most ofthe studies identified depressed mothers from elevated depres-sive symptomatology, but these studies and studies of clinicallydepressed mothers typically yield similar results.

The high-risk studies of schizophrenia provide the earliestsystematic evidence about the parenting behavior of depressedmothers. Cohler et al. (1977) observed that psychotic mothers,whether depressed or schizophrenic, had more difficulty inter-


Table 4Studies of Parenting by Depressed Persons

Study Sample Child age Parent diagnostic criteria Observation situation

Bettes(1988)

Cohn, Matias, Tronick,Connell, & Lyons-Ruth (1986)

Cohn, Campbell, Matias,& Hopkins (1990)

Field (1984)

Field etal. (1985)

Field etal.( 1988); Field,Healy, Goldstein, &Guthertz(1990)

Fleming, Ruble, Flett, &Shaul(1988)

Hops etal. (1987)

Livingood, Daen, &Smith (1983)

Lyons-Ruth, Zoll,Connell, &Grunebaum(1986)

Panaccione & Wahler(1986)

NIMH Study(Davenport, Zahn-Waxler, Adland, &Mayfield 1984)

Breznitz & Sherman(1987)

Kochanska, Kuczynski,Radke-Yarrow, &Welsh (1987)

UCLA Family StressStudy (Gordon, et al.,l989;Hammen,Gordon, etal., 1987a)

Unipolar = 10, control = 26(community sample ofmothers)

Unipolar = 13 (communitysample of low SESmothers with multipleother risk factors; nocontrol group)

Unipolar = 24, control = 12(community sample)

Unipolar = 12, control = 12(mothers identified bysocial workers asdepressed postpartum)

Unipolar = 12, control = 12(mothers prenatallyidentified as depressed)

Unipolar = 40, control = 34(low SES communitysample of mothers)

#=41 (max.)(primiparous, middleSES mothers from intactfamilies fromcommunity; correlationalstudy)

Unipolar = 27, control = 27(mothers from intactfamilies; outpatienttreatment facilities)

Unipolar = 25, control = 25(community sample ofmothers from intact,middle-SES families)

N = 56 (correlational studyof mothers from samesource as Cohn et al.,1986)

N = 33 (correlational studyof mothers in Head Startprogram)

Bipolar = 7, control = 20

3-4 months

6-7 months

2 months

3 months

3-5 months

3-6 months

Longitudinal study:3 days and 1, 3and 16 months

3-16 years

2 days

1 year

M = 4.3 years

Longitudinal study:12,15.&18months

BDI > 10 3-15 min face-to-face interactionin the home

CES-D in range for outpatient 6 min face-to-face and 40 mindepressives naturalistic interaction

Unipolar = 14, control =18 3 years(mothers)

Unipolar = 37, bipolar = 1-4 years17, control = 33

Unipolar = 19 (major),bipolar = 16

8-16 years

RDC based on SADS

BDI>15

BDI > 11

BDI > 12

10-item mood scale

RDC based on SADS

BDI > 10

CES-D

BDI

DSM-III based on SADS-L

RDC based on SADS-L

RDC based on SADS-L

RDC based on SADS-L

3 min face-to-face interaction

3 min face-to-face interactionduring mother's simulation oflooking normal and of lookingdepressed

10 min face-to-face free play

3 min face-to-face interactionwith mother and with astranger

At 3 days, feeding situation; at 1,3, & 16 months, naturalisticinteraction for 15-20 min.

Naturalistic in-home interactionof all family members

15-min feeding session

40 min naturalistic in-homeinteraction

120 min naturalistic in-homeinteraction in four 30-minsessions

Behavior of mothers and childduring strange situationprocedure

Interaction in unstressful andmildly stressful situation

90 min naturalistic interactionwith control episodes targeted

5 min interaction duringachievement and duringconflict resolution task

Note. BD! = Beck Depression Inventory; CES-D = Center for Epidemiotogieal Studies-Depression; SES = socioeconomic status; RDC = ResearchDiagnostic Criteria; SADS = Schedule of Affective Disorders and Schizophrenia, L = Lifetime; DSM-III = Diagnostic and Statistical Manual ofMental Disorders, third edition.

acting with their children than did control mothers and con-fused their own needs with their children's needs. Sameroffetal. (1982) described depressed mothers as being less spontane-ous, less vocal, less positive, and more distant than controlmothers when playing with their 4-month-old children. How-ever, these differences did not emerge consistently during later

observations, and depressed mothers resembled schizophrenicmothers on many measures. Although these studies show thatdepressed mothers differ from control mothers in ways that mayaffect children's social development, the results indicate that thedeficits may not be specific to depression. Unfortunately, recentresearch has not considered this possibility carefully.


The dampening effect of depression on social behavior is evi-dent in the picture of the depressed mother emerging from therecent cross-sectional studies (see Table 4, Bettes, 1988; Cohn,Campbell, Matias, & Hopkins, 1990; Cohn, Matias, Tronick,Connell, & Lyons-Ruth, 1986; Davenport et al., 1984; Field,1984; Field, Healy, Goldstein, & Gutherz, 1990; Field et al.,1988; Field et al., 1985; Fleming et al., 1988; Livingood, Daen,& Smith, 1983; Radke-Yarrow, Richters, & Wilson, in press).During interactions with their young children, depressed moth-ers showed lower rates of behavior and of affective expression.Specifically, they expressed little positive affect, perhaps re-flecting their anhedonia. They also responded more slowly, lesscontingently, and less consistently to the children.

These findings support the view that depression may affectparenting by reducing the effort that parents put into interact-ing with their child. This may be due to the reduced energylevels and self-absorption that are symptomatic of depression.Studies of the speech of depressed mothers have provided moredetailed evidence of reduced effort in parenting. Clinically de-pressed mothers speak less often to their 3-year-old children andrespond more slowly to their children's speech (Breznitz &Sherman, 1987). Mothers with mild depressive symptoms alsorespond more slowly to their 4-month-old infants (Bettes,1988). Moreover, the length of their pauses and utterances aremore variable than those of control mothers, and they are alsoless likely to use the exaggerated intonation typical of caretak-ers' speech with infants. These findings led Bettes (1988) to sug-gest that depression impedes mothers' ability to imbue theirspeech with the affective signals thought to play an importantrole in the socialization of affect modulation in children.

The dampening effect of depression on effortful interactionis also evident in the strategies used by clinically depressed orbipolar-disordered mothers to resolve control episodes withtheir children (Kochanska, Kuczynski, Radke-Yarrow, &Welsh, 1987). Depressed mothers choose strategies that requireless cognitive effort more frequently than do control mothers.These include enforcing obedience unilaterally or withdrawingwhen faced with child resistance. Control mothers, by contrast,are more likely to negotiate a solution with their child. The for-mer approach maximizes short-term compliance by the child,whereas the latter approach maximizes long-term compliancegoals (Kuczynski, 1984). Thus, by opting for less cognitivelyeffortful conflict resolution strategies, depressed mothers maysocialize their children to resolve interpersonal conflict throughcoercion or withdrawal.

The hostility and irritability that are symptomatic of depres-sion and that characterize depressed adults' interactions withtheir spouses also emerge in their interactions with their chil-dren. Studies have reported that depressed mothers are moreirritable toward their infants than are control mothers (Cohnet al., 1986, 1990; Field, 1984; Lyons-Ruth, Zoll, Connell, &Grunebaum, 1986; Radke-Yarrow et al., in press), and there issome suggestion that maternal irritability and hostility in-creases under stress (Cohn et al., 1986; Field et al., 1990; Lyons-Ruth et al., 1986; Rutter, 1990).

Hostility is more pronounced in the interactions of depressedmothers with school-aged children than with younger children,perhaps reflecting the more active role of older children duringinteractions. Clinically depressed mothers from the UCLAFamily Stress Project were more irritable and critical and less

positive and task focused than other mothers during a conflictresolution task with their child (Gordon et al., 1989; Hammen,Gordon, et al., 1987a). Mothers' depressive symptomatologyalso predicted hostility toward children in a nonclinical sample.Panaccione and Wahler (1986) found a strong association be-tween mother's depressive symptoms and hostile child-directedbehavior, including shouting and slapping, even when the adver-sity of child behavior was partialed out.

Surprisingly, observational studies have paid very little atten-tion to depressed mothers' sad affect during interactions withtheir children. In a key exception, the Oregon Research Insti-tute group (Biglan et al., 1985; Biglan, Rothlind, Hops, & Sher-man, 1989; Hops et al., 1987) reported that depressed mothersshowed high levels of sadness during interactions with husbandsand children. Expressions of anger by other family memberswere particularly likely to evoke these indications of sadness.Mothers' displays of sadness functioned, in turn, to suppressthe hostility of other family members. Thus, by observing thecontrol that mothers' distress exerted over other family mem-bers' behavior, children may learn the coercive value of display-ing distress. The failure of other investigators to study mothers'sadness largely reflects a reliance on coding schemes developedfor purposes other than to study depressed parents and theirchildren. Future observational studies need to be better in-formed by existing knowledge about depression and to tailortheir coding schemes accordingly, as the Oregon Research Insti-tute group has done.

In sum, these observational studies show that, in at least twoimportant respects, the behavior of depressed mothers duringinteractions with their children resembles that of depressed per-sons during interactions with other adults. First, their behaviorand affective expression is constricted and their speech is flat ascompared with other mothers. They respond less positively, lessfrequently, and less quickly to their children's efforts to engagetheir attention. Second, they show heightened levels of child-directed hostility and negativity, and their attempts to controlchild behavior are marked by coercion rather than by negotia-tion. More studies are needed before any conclusions can bedrawn about mothers' displays of sad affect during family inter-action.

Although existing studies provide a rich description of howdepressed mothers differ from normal control mothers, this de-scription is not a basis for concluding that a unique causal rela-tion exists between parents' depression and their children's de-pression and maladjustment that is mediated-through depres-sion-related parenting difficulties. This conclusion is prematureuntil we can show that (a) the difficulties of depressed parentsare a unique consequence of depression rather than a sharedconsequence of other aspects of parental dysfunction or of stres-sors in the broader social environment; (b) specific parentingbehaviors in depressed parents are associated with child depres-sion and adjustment problems; and (c) the direction of causalitygoes from depressed parent to child. We review the evidence foreach of these conditions.

Specificity of parenting difficulties of depressed persons. Webegin with a consideration of some possible explanations for theparenting difficulties of depressed persons. First, these difficul-ties may be a specific consequence of depression. In this casethey should not be evident in parents with other disorders or inparents who are experiencing stress but not clinical depression.


Second, the difficulties may be a final common pathway throughwhich different sources of stress on mothers, including depres-sion, affect child adjustment. For example, these difficultiesmight be a consequence of depression in a nonmaritally distressedmother and a consequence of marital discord in a nondepressedmother. Third, they may be attributable to a correlate of depres-sion such as marital discord, chronic stress, or personality disorder.In this case, parenting difficulties should be restricted to those de-pressed mothers who also show the true causal factor (e.g., maritaldiscord), and should occur also in nondepressed mothers whoshow that factor. Of course these models may work differently forthe two types of parenting problems identified: hostility and con-stricted behavior and affect.

The relative merits of these models are difficult to establish.The samples used in most observational studies of depressedparents are not large enough to use statistical controls to ex-plore alternative explanations, and most studies included onlynormal comparison groups. Nonetheless, several pieces of evi-dence converge to suggest that the parenting difficulties of de-pressed mothers are not unique. First, the high-risk studies ofschizophrenia found similar difficulties in the parenting of de-pressed and schizophrenic mothers. In fact, when interactingwith their child, schizophrenic mothers have been found toshow more of such characteristically depressive behaviors aswithdrawal and detachment than do depressed mothers (Good-man & Brumley, 1990). Second, there is considerable overlapbetween the parenting difficulties of clinically depressed moth-ers and mothers with mild depressive symptomatology. Third,parenting difficulties similar to those reported in depressedmothers have been found in mothers coping with several differ-ent stressors, including premature infants (Crnic, Greenberg,Ragozin, Robinson, & Basham, 1983), divorce (Hetherington,Cox, & Cox, 1982), and poverty (McLoyd, 1989). For example,following divorce, custodial mothers often become self-in-volved, erratic, uncommunicative, inconsistent, and punitive indealing with their children (Hetherington, Stanley-Hagan, &Anderson, 1989).

These findings suggest that the parenting difficulties observedin depressed mothers may be common to mothers who are dis-tressed because of stressors that originate within the family(e.g., marital conflict), outside the family (e.g., neighborhood,economic conditions), or within the mother herself (e.g., physi-cal illness, psychiatric illness) (cf. Patterson, 1982; Wahler &Dumas, 1989). Chronic stressors can produce chronic or inter-mittent depressive symptoms or general distress without the de-velopment of clinical depression (Breslau & Davis, 1986). Theprolonged distress and dysfunction shared by clinically de-pressed mothers and nonclinically depressed mothers understress may have more consequences for their parenting thanother aspects of the syndrome of depression. Consistent withthis view, Hammen and her colleagues found that chronic stressand mother's current depressive symptomatology were morepredictive than affective diagnosis of the mother's critical andunproductive behavior during interactions with her child (Gor-don et al., 1989; Hammen, Gordon, et al., 1987a). Moreover,the usual differences in parenting between depressed and nor-mal control mothers did not emerge when depressed motherswho were poor, Black, and single were compared with controlsliving in similar circumstances (Goodman & Brumley, 1990),

further suggesting that stress has a stronger impact than diagno-sis on parenting.

In sum, there is still no evidence that the parenting difficultiesof depressed mothers are a specific consequence of depression.The task is now to determine whether the various groups ofmothers who show similar parenting difficulties share a com-mon correlate (e.g., chronic stress) or whether these difficultiesare a common pathway through which different stressors affectchild adjustment.

Causal relation between depressed persons' parenting behav-ior and child adjustment. Diagnostic studies of children with adepressed parent have concentrated on children aged 6 yearsand older, whereas studies of parenting by depressed personshave concentrated on infants and toddlers. Thus, it is not yetknown whether the parenting difficulties of depressed personscan account for depression or other adjustment problems intheir offspring. The consequences of the interactional difficul-ties that children of depressed mothers show in infancy are un-clear. They may be transitory, disappearing when mothers'symptoms subside. Or they may indicate a vulnerability to laterdepression that results from the depressed mothers' parentingdifficulties or from some constitutional vulnerability. Unfortu-nately, as we have noted, there is still little known about theprognostic value of putative markers of vulnerability to depres-sion. Nonetheless, there is evidence linking parents' negativityand hostility toward their children with children's negative self-concept (Jaenicke et al., 1987) and attachment difficulties(Radke-Yarrow et al., in press).

Indirect evidence of an association between negative, hostileparenting behavior and adjustment problems in children witha depressed parent comes from several sources. Retrospectivereports of harsh, unfair disciplinary practices during childhoodhave been found to account for the relation between parents'depression and depression in their adult children (Holmes &Robins, 1988). Child abuse has been found to be more prognos-tic than parental depression of child psychopathology (Kashani,Shekim, Burk, & Beck, 1987). A history of harsh, punitive, andinconsistent parenting is common in clinically depressed chil-dren (see Burbach & Borduin, 1986, for review), and abusedchildren show high levels of depressive symptoms (Kazdin,Moser, Colbus, & Bell, 1985).

Direct evidence of a link between hostile, negative parenting bydepressed mothers and child maladjustment comes from two ob-servational studies (Hammen, Gordon, et al., 1987a; Radke-Yar-row et al., in press). In both cases, however, the mother's negativebehavior was reciprocated by the child. This reciprocal patternmakes it difficult to interpret Radke-Yarrow et al.'s finding thatthis interaction pattern predicted maladjustment in the index chil-dren 4 years later (Radke-Yarrow et al., in press). The child's mal-adjustment might have been caused by the mother's behavior or itmight be a continuation of preexisting child problems that elicitnegative responses from mothers (Bugental, Blue, & Cruzcosa,1989; Bugental & Shennum, 1984). Hammen, Burge, and Stans-bury (1990) used longitudinal analyses to disentangle these causalprocesses and found that mother's difficulties did contribute tosubsequent changes in child adjustment net of the child's initialadjustment level. However, the analyses revealed a much morecomplex reciprocal process, with both the child and the mothercontributing to each other's current difficulties and to the child'sfuture difficulties.


Thus, negative, hostile parenting may be one source of theadjustment problems in children of depressed parents. Thisparenting style has been identified as a risk factor for severalother forms of maladjustment and especially for conduct disor-ders (Patterson, 1982). Whereas it may provide a sufficient ex-planation for the externalizing symptoms of children with a de-pressed parent, it may not explain their higher rates of majordepression. Perhaps this parenting style elicits clinical depres-sion in children only in combination with a parental style thatis behaviorally and affectively nonresponsive and that is charac-terized by high rates of dysphoric affect. Although researchershave speculated about the impact of diminished rates of mater-nal behavior and constricted affect on children, this questionhas not yet been investigated.

Direction of causality in mother-child interaction. We havenoted that the reciprocal aversiveness of interactions betweendepressed mothers and their children poses problems for assess-ing the contribution of the depressed mother's behavior to herschool-aged children's maladjustment. Difficulties with disen-tangling the causal direction of mother-child interaction arepresent even in infancy. Reciprocity between depressed moth-ers and their infants is evident in the first months of the infant'slife. Several of the observational studies summarized in Table 4have reported that the behavior of children of depressed moth-ers resembles their mothers' behavior. Compared with controlchildren, children of depressed mothers direct their behaviortoward their mother less frequently, smile and express happi-ness less often, and are more irritable and fussy (Cohn et al.,1986, 1990; Field, 1984; Field et al., 1985, 1988; Lyons-Ruthet al., 1986). Two recent studies undertook time series analysesto disentangle the causal direction of interactions betweenmothers and their infants (Cohn et al., 1990; Field et al., 1990).Both studies found that regardless of mother's clinical status,mothers and infants contributed equally to maintaining the ob-served interaction pattern. Neither the mother nor the infanttook the lead in directing the course of the interaction.

Although these studies provide a snapshot of the structureand content of mother-child interaction, they do not reveal theprocess by which this structure came about. In their efforts tounderstand this process, Field and her colleagues took as theirstarting point the finding that when nondepressed mothers sim-ulated depression, their infants responded with distress and per-sistent efforts to reinstate the mother's normal mood (Cohn &Tronick, 1983). When this study was replicated with childrenof depressed mothers, the behavior of these children changedlittle when their mother simulated depression (Field, 1984).This lack of response suggests that the children may haveadapted to their mothers' depression and are no longer upset byit. Alternatively, their lack of distress may reflect a passive cop-ing style that generalizes to other situations.

Field et al.'s (1988) finding that 3-month-old infants of de-pressed mothers behaved similarly during interactions withtheir mother and with a female stranger supports this latter al-ternative. This depressed interaction style may develop veryearly in life in response to the depressed mother's behavior, ormay be present from birth. There is some evidence to supportthe latter possibility. Infants of depressed mothers have beenfound to show limited responsivity to stimulation, depressedactivity levels after delivery (Field et al., 1985; Sameroff et al.,1982), and more perinatal problems (Weissman et al., 1986).

Whatever the origin of the child's interactional pattern, it isnoteworthy that it generalizes to strangers early in life and alsoevokes a negative response from strangers (Field et al., 1988).This points to the need to consider parent-child interaction asan interpersonal system and to explore more fully the child'scontribution to its maintenance.

Summary. Maternal depression is associated with difficultiesin parenting. However, the difficulties that have been identifiedare not specific to depressed mothers. They also occur in moth-ers experiencing high stress and distress. There is some sugges-tion that these factors may have a more direct impact on parent-ing than does clinical depression. This raises the possibility thatsome of the parenting difficulties of depressed mothers are dueto a correlate of depression (e.g., chronic stress, marital discord)or that a variety of stressors can result in similar parentingdifficulties.

The implications of these parenting difficulties for the disor-ders that emerge in children of depressed parents in later child-hood and adolescence are not yet known. There is an associa-tion between negative, hostile parenting and child maladjust-ment, but the direction of causality is unclear. As early as age 3months, children reciprocate their depressed mother's negativeand less-effortful interaction behavior and evoke negative re-sponses from strangers (Field et al., 1988). Although the originsof these children's problematic style of interaction are not yetknown, it is clear that as a group they are more challenging toparents than are control children.

This collection of evidence challenges the view that the par-enting behavior of depressed mothers is the only route by whichdepression and general maladjustment is transmitted intergen-erationally. It argues instead for moving from a unidirectionalmaternal behavior -»• child maladjustment model to look at thebroader interpersonal context of maternal depression and at thechild's contribution to the interpersonal context.

Interpersonal Context of Depression

We noted earlier several ways in which the interpersonal contextof families with a depressed parent differs from that of controlfamilies. Marital discord and family stress are more common infamilies with a depressed parent. Social impairments aside fromdiagnosis are also more characteristic of depressed persons and oftheir spouses. Just as these co-occurring risk factors may explainthe parenting difficulties of depressed women, they may also ex-plain the adjustment difficulties of their children.

The similarity between children of depressed parents andchildren of medically ill or nonaffectively disturbed parentsprovides indirect support for this proposal. More direct supportcomes from studies showing that child maladjustment also de-pends on nondiagnostic aspects of parental impairment (i.e.,chronicity and severity of the illness, current level of distressand adaptive functioning) and on chronic stress and marital dis-cord. This is true for children's general adjustment problems(Billings & Moos, 1983; Hammen, Adrian, et al., 1987; Kelleret al., 1986) and for clinical disorders (Hammen, Adrian, et al.,1987; Keller et al., 1986). The effect of these co-occurring riskfactors on general adjustment problems is powerfully illustratedby Billings and Moos (1983). They found that in families witha depressed parent where stress was low and support was highand the illness was not severe, only 10% of the children were


disturbed as compared with 3% in control families and 25% inall families with a depressed parent. In light of our argumentthat parental depression may place children at specific risk forclinical depression, it is noteworthy that when Hammen,Adrian, et al. (1987) controlled for chronic stress and distress,rates of clinical depression in children of unipolar-depressedmothers were unaffected.

Marital conflict. Adjustment problems in children, especiallyconduct disturbance, increase following divorce (Emery, 1982;Hetherington et al., 1989) and respond to parental conflict(Long & Forehand, 1987; Long, Forehand, Fauber, & Brody,1987). The high rates of conflict in the marriages of depressedpersons suggest that marital discord may be a particularly via-ble alternative explanation for their children's adjustment prob-lems. In fact, studies have shown that marital distress plays animportant role in accounting for these children's general adjust-ment problems (Connersetal., 1979; Emery etal., 1982; Kelleret al., 1986; Kuyler, Rosenthal, Igel, Dunner, & Fieve, 1980) andfor current and lifetime diagnoses (Keller et al., 1986). Usingfamilies from the Stony Brook High-Risk Project, Emery et al.(1982) conducted the strongest test yet of the role of maritaldistress in the adjustment of children with a depressed parent.As noted earlier, marital discord accounted for the relation be-tween parental depression and child competence, and this effectwas specific to depression. These results led Emery et al. to con-clude that

the relationship between having a unipolar depressed or a bipolarparent and children's disturbed behavior is largely accounted forby concomitant marital discord. Where there is no marital discordpresent in families with an affectively disturbed parent, the risk forproblematic school behavior is similar to that found in controls,(p. 226)

This set of findings provides some empirical corroborationfor Rutter's (1966) speculations on the relation between prob-lematic child behavior and marital discord. He suggested that

children may be adversely affected by disturbed relationships be-tween the two parents. Mental disorder, especially longstandingmental disorder, is often associated with hostility and discord be-tween husband and wife. Quarreling and bickering may increaseand the atmosphere in the house may reflect rising tension. Thismay sometimes have more impact on the child than the mentalillness itself, (p. 110)

Whereas marital conflict may explain the general adjustmentproblems of children with a depressed parent, it is not yetknown whether it can explain their high rates of clinical depres-sion. Two studies using different designs provide converging evi-dence that the processes underlying depression in these childrenmay differ from those underlying the externalizing problemsthat contribute heavily to general adjustment scores. Hops andhis colleagues investigated interaction patterns in maritally dis-cordant and nondiscordant families that included a clinicallydepressed mother, and in normal control families (Biglan et al.,1985; Friedman, 1984; Hops et al., 1987; Hops, Sherman, &Biglan, in press). Hostile, irritable behavior was most commonin the maritally distressed families. By contrast, families with adepressed mother showed high rates of dysphoric affect and lowrates of positive affect irrespective of marital distress. Childrenof depressed mothers in maritally distressed families were moreirritable than children in nonmaritally distressed families, but

both groups of children showed similar levels of depressive be-havior. Fendrich, Werner, and Weissman (1990) used epidemio-logical data to compare the relative impact of parental depres-sion and family disharmony on children. Whereas family dis-cord was most strongly linked with conduct disorders, parentaldepression was most strongly linked with child depression.Thus, marital discord may account for externalizing problemsbut not for clinical depression.

Linking marital discord, parental depression, and child ad-justment. Several causal models of the association betweenmarital discord, parental depression, and child maladjustmentare possible (see Figure 1). First, marital discord may mediatethe link between parental depression and child maladjustment(Model 1). This implies that parental depression affects childadjustment primarily by increasing children's risk of exposureto marital discord. Second, parental depression may mediatethe relation between marital discord and child maladjustment(Model 2). This implies that depression has a direct impact onchild adjustment and that marital discord affects child adjust-ment primarily by increasing children's exposure to parentaldepression. Third, the relation between parental depression andchild maladjustment may be spurious, with marital discordcausing both parental depression and child maladjustment(Model 3). Fourth, the relation between marital discord andchild maladjustment may be spurious, with parental depressioncausing both marital discord and child maladjustment (Model4). Of course, these models are for heuristic purposes and donot incorporate the complex reciprocal processes that probablyoperate.

In evaluating these models, we need to consider two strandsof evidence. First, we must consider the research just reviewedon the relative impact of marital discord and parental depres-sion on child adjustment. This research demonstrates the needto distinguish between externalizing problems and depressionin children in evaluating the efficacy of the models we have out-lined. It appears that marital discord directly increases chil-dren's risk for externalizing problems, whereas parental depres-sion has its primary impact on children's risk for depression.

Second, we must consider studies of the relation betweenmarital discord and parental depression. Research on the inter-personal context of depression in adults provides some evidencethat marital discord plays a causal role in the onset and mainte-nance of adult depression (see Coyne, 1990, for review; Epstein,Baldwin, & Bishop, 1983; Miller et al., 1986). Patients withmarital problems are less responsive to psychotherapy (Court-ney, 1984; Jacobson, Schmaling, & Holtzworth-Munroe, 1987)and antidepressant medication (Rounsaville et al., 1980), andare more likely to relapse (Hooley & Teasdale, 1989). Whenfamily functioning does not improve, patients experiencelonger depressive episodes (Keitner, Miller, Epstein, Bishop, &Fruzzetti, 1987). Furthermore, spousal criticism at the time of adepressed person's admission to a hospital predicts subsequentrelapse (Hooley, Orley, & Teasdale, 1986; Hooley & Teasdale,1989; Vaughn & Leff, 1976). Thus, marital conflict may affectthe course of depression, and continued disharmony may impedeits progress, limit response to treatment, and be associated withhigh rates of relapse. In addition, people who become divorcedare at heightened risk for increased depressive symptomatology(Menaghan & Lieberman, 1986) and for clinical depression (Heth-erington et al., 1989). Finally, many spouses of depressed persons


Model 1

Model 3

ParentalDeoresslon

^ *k MaritalDiscord

ChildProblems

Mpdel 4

MaritalDiscord « fe*— • w

ParentalDeoression d fer Child

Problems

Model 5

Figure 1. Alternative models linking marital discord, parental depression, and child problems.

have adjustment problems that predate their marriages (Farter &Hadzi-Pavolic, 1984; Quinton et al., 1984) and that may play arole in their spouse's depression (Coyne, 1990).

There is evidence that depression is also a source of stress onthe marital relationship. Approximately 40% of the adults liv-ing with a currently depressed persons are so distressed thatthey meet standardized criteria for referral (Coyne et al., 1987).Most of the elevated distress associated with living with a cur-rently depressed person is due to the burden of coping with theillness.

The research we have reviewed offers some support for allfour models, provided that we distinguish between children'sexternalizing problems and their depression. First, marital dis-tress contributes directly to children's externalizing problems,and by inducing and maintaining parental depression, indi-rectly increases their risk for clinical depression. Consistentwith Model 3, the direct relation between marital discord andchild depression may be spurious, and may be explained by

their joint relation with parental depression. Second, parentaldepression directly increases children's risk for depression and,by increasing marital strife, indirectly increases their risk forexternalizing problems. Consistent with Model 4, the direct re-lation between parental depression and children's externalizingproblems may be spurious and may be explained by their jointrelation with marital discord. Model 5, which combines Models1-4, illustrates these observations. Because few studies have ex-amined the broader interpersonal context of these children'slives, the conclusions that Model 5 incorporates are tentativeand require further study.

Children's contribution to the interpersonal processes of de-pression. Our discussion so far has treated child maladjustmentas an outcome of depression and marital distress. However, evi-dence from multiple sources and across multiple situationsshows that children of depressed parents pose a greater parent-ing challenge than the average child. Problems emerge early inthese children, and they may be more difficult from birth than


other children. As with spouses, the feedback available frominteractions with children may further validate the feelings ofineptitude, worthlessness, and rejection that characterize de-pressed persons, thus maintaining their depression.

Although the impact of children's behavior on parents' de-pression has not been systematically investigated, there is somereason to believe that child disturbance contributes to the con-tinuation of maternal depression. First, there is evidence of areciprocal relation between maternal depression and child mal-adjustment (Hammen et al., 1990). Second, declines in moth-ers' depression parallel the successful treatment of their chil-dren's behavior problems (Forehand, Wells, & Griest, 1980;Patterson, 1982). Third, parents of less-disturbed children aremore likely to recover from their depression within a year,whereas parent's recovery status does not predict changes inchild adjustment (Billings & Moos, 1985). Thus, a comprehen-sive model of the interpersonal context of depression must in-corporate the child's contribution to the interpersonal context.

Efforts to understand child characteristics that might modifyrisk for psychopathology have typically focused on identifyingsources of resilience in children. Several studies have identifiedsuperior social and social cognitive skills as characteristic of re-silient children (Beardslee et al., 1987; Beardslee & Podorefsky,1988; Downey & Walker, 1989; Kaufman, Grunebaum, &Cohler, 1979;Pelligrinietal., 1986; Radke-Yarrow & Sherman,in press). A frequent conclusion is that social skills help ensurethat children receive positive attention from adults other thantheir depressed parent. However, socially skilled children mayalso be less likely to confirm their depressed parent's feeling ofineptitude and rejection, and thus to elicit the negative re-sponses that are common in depressed parents.

A recent study of resilient children participating in theNIMH project provides some support for this suggestion(Radke-Yarrow & Sherman, in press). Resilient children (i.e.,socially and academically competent children without psychiat-ric problems) were characterized by high intelligence and socialcharm, and by holding a special place in their families becauseof some trait that their parents valued. Radke-Yarrow and Sher-man argued that for these reasons they received all the warmthand resources that their parents could provide.

Studies linking temperament with resilience in children(Garmezy, 1983) have suggested further ways in which childcharacteristics may elicit or maintain particular parental be-haviors. Perhaps as Crockenberg (1981) has found for childrenin general, temperamentally easygoing children are more im-pervious to their depressed mother's behavior than are moreirritable children. Imperviousness to mothers' negative behav-ior may impede the development of the reciprocal pattern ofnegativity often seen in the interactions of depressed mothersand their children. Bugental and her colleagues have docu-mented such a pattern in families with a child at high risk forbeing abused (Bugental et al., 1989). Compared with their low-risk siblings, children at high risk for abuse were behaviorallymore "difficult" and evoked a more negative reaction in theirmothers, who reported feeling little control over their "difficult"child's behavior. The high-risk children also evoked more nega-tive reactions from strangers who felt threatened by them, butnot from strangers who felt in control of the situation. Similarprocesses may operate when depressed mothers are faced witha difficult child, given their perceptions of being incompetent

parents. The coercive behavior that these children evoke in theirmothers may further exacerbate existing child difficulties.

Commentary, Integration, and a Look AheadResearch on the children of depressed parents has now

achieved considerable momentum, with a great increase in thepast 5 years in the quantity and quality of available studies. Theaccumulation of better-designed studies has yielded some con-sistent findings. It is clear that depression in parents is associ-ated with problems of adjustment and diagnosable disorders,particularly depressive disorders, in their children. However, thespecificity and origin of this association is not definitively estab-lished. The evidence at this point supports the tentative conclu-sion that an increase in diagnosable depression is specific to thechildren of depressed parents, but that more general problemsin adjustment are shared with children of mothers with a physi-cal illness or a psychiatric condition other than depression andof mothers who are otherwise under stress. In particular, mari-tal discord is a viable alternative explanation for the general ad-justment difficulties of children with a depressed parent.

The behavior observed between depressed mothers and theirchildren has continuities with the behavior of depressed moth-ers toward other adults. Depressed mothers emit lower rates ofbehavior and show constricted affect; they adopt less-effortfulcontrol strategies; and they show considerable hostility and neg-ativity. Here, too, there are unresolved issues of specificity. Un-fortunately, the studies that established the heightened risk fordepression among the children of depressed parents have usu-ally involved a more seriously disturbed sample of mothers orhave focused on children who are older than the children in thestudies of parenting by depressed mothers. Thus, it is not yetpossible to draw conclusions about possible links between par-enting and this apparently specific risk.

With recent progress in the study of children of depressedparents, it is becoming more obvious that adequate explanatorymodels must incorporate considerable complexity (see Rutter,1990, for a similar conclusion). We have reviewed evidence sug-gesting reciprocal influences in interactions between depressedmothers and their children. Furthermore, the interpersonal con-text of depression—that is, conditions such as marital discordand family stress that precede, precipitate, or co-occur with ma-ternal depression—is clearly implicated. The difficulties of de-pressed mothers and the adjustment problems of their childrenmay be contingent on such conditions. At least some of the asso-ciation between depression and both parenting problems andchild maladjustment may be spurious, with the same interper-sonal factors (e.g., marital discord) that produce depression inmothers also producing parenting and child problems. Whereasthis conclusion may be true of children's externalizing prob-lems, it does not appear to be as true for their depression.

With the recent accumulation of studies, the requirementsfor making a meaningful contribution to the literature have be-come more stringent. There is now a need for more specifichypotheses, measures tailored to these hypotheses, and atten-tion to the heterogeneity of depressed persons and their spousesand life circumstances, as well as their children and their vari-ous adjustment problems. Recent studies are meeting many ofthese requirements. Nonetheless, our review has revealed someimportant gaps in the factors being considered and the methods


being used. These gaps limit the generalizability of the resultsthat are being obtained, as well as the ability of researchers tointerpret the patterns of association that are found. We nowidentify the areas most in need of attention.

Patient characteristics. Most depressed persons are eithertreated in the community or fail to obtain treatment (Weissman& Myers, 1980). The processes that bring a depressed person toa tertiary facility where rigorous research is being conductedare not well understood, but the resulting treatment populationis biased in terms of severity, chronicity, and treatment history.Many of the studies that we have reviewed report exceptionallyhigh rates of past episodes and hospitalizations among the de-pressed mothers. There is a pressing need for basic data on theadjustment of children with clinically depressed persons whosedisturbance is less severe. In the absence of such data, we shouldbe cautious in our generalizations from the children of the terti-ary care and inpatient populations that are currently beingstudied.

Whereas most studies of the problems among children of de-pressed parents now use interview-based diagnosis for identify-ing depressed mothers, studies of parenting behavior still relymainly on self-report depression scores. It is important thatstudies of parenting behavior routinely use diagnosis based onstructured interviews, and that the two types of studies becomecomparable in terms of severity of parental disturbance and ageof the children. There is also a need for attention to Axis IIdiagnoses, because personality disorders are common amongdepressed patients. They may be both a direct source of parent-ing problems and increased risk for child disturbance and beassociated with systematic differences in the course and severityof parents' depression.

Investigators should also consider the severity of depressivesymptoms in both depressed and comparison groups (Fisher& Kokes, 1983; Hammen, Adrian, et al, 1987). Prolonged orrecurrent periods of psychological distress may be one of thecrucial commonalities among mothers who are depressed,mothers with nonaffective psychiatric disorders or physical con-ditions, and mothers under stress. Some investigators have ig-nored the problem of depressive symptoms in comparisongroups. Others have tried to solve it by establishing exclusioncriteria that have the unintended effect of producing unrepre-sentative samples with truncated distributions of depressionscores. Lee and Gotlib (1989a, 1989b), for example, reported amean depression score for their psychiatric comparisons groupthat was not significantly lower than their exclusion score forelevated depressive symptoms. This suggests selective samplingfrom a population with a high prevalence of depressive symp-tomatology.

Resources and positive characteristics of depressed mothersthat might reduce the risk to their children have been totallyneglected. Despite a growing interest in protective factors andevidence that the contextual factors that contribute to childmaladjustment may also precipitate mothers' depression, de-pressed mothers are still studied almost exclusively in terms oftheir negative impact on children. The ability of depressedmothers to continue to provide for their children, even thoughthey are demoralized and highly symptomatic and the childrenare difficult and unrewarding, may be a protective factor in theface of other adversities.

Depressed fathers. Depressed mothers have received consid-

erably more attention than depressed fathers. Although somestudies have found that children of depressed fathers are lessdisturbed than children of depressed mothers (Keller et al.,1986; Klein et al., 1985), other studies find equivalent distur-bance levels (Billings & Moos, 1983; Klein et al., 1988; Weiss-man, Gammon, et al., 1987). Thus, the effects of paternal de-pression are in need of further study. Aside from the intrinsicinterest of families with a depressed father, comparisons withfamilies where the depressed mother is the primary caretakermay elucidate the processes underlying the high levels of distur-bance in children of depressed mothers.

Spouses of depressed mothers. The spouses of depressedmothers remain shadowy figures. We have noted that there issome evidence of selective mating, that husbands are inclinedto be disturbed themselves, and that marital conflict is oftenpresent. These factors contribute to the wife's vulnerability tobecome depressed, the course of her illness, parenting prob-lems, and children's risk for disturbance. A comprehensivemodel of the causes of disturbance in the children of depressedparents will require attention to the many ways in which hus-bands contribute both positively and negatively to their wife'sfunctioning and their child's well-being. Progress in researchers'understanding of the association between maternal depressionand child outcomes will accelerate if they routinely assess thebackground, current functioning, and lifetime psychiatric his-tory of spouses. Furthermore, observational studies shouldmore routinely sample family interaction in which husbandsare present and should examine husbands' interactions withchildren and their effects on interactions between mothers andchildren.

Role of marital turmoil. Marital turmoil appears to haveboth direct and indirect effects on the parenting difficulties ofdepressed persons and on their children's difficulties, and re-mains a plausible rival explanation for some of these difficulties.At this point it is not enough to merely demonstrate that mari-tal turmoil is a factor; more refined hypotheses about the pro-cesses by which this association comes about are needed.Among the possibilities are that parenting and child well-beingare jeopardized by (a) inconsistencies in child rearing, not onlybetween the depressed parent and the spouse but also withineach parent as they become embroiled in marital disputes; (b)lack of support for parenting; and (c) children witnessing andbeing unwitting participants in overt marital conflict. Thesefactors may directly affect child well-being, interact with paren-tal depression, or mediate its effects.

Separation and divorce. Some studies included only intact,two-parent families, whereas others have also included single-parent families but have failed to explore the effects of familyconfiguration. Studies that have examined these effects suggestthat children of depressed parents in separated or divorced fam-ilies are at higher risk than those in intact families (Conners etal., 1979; Kuyler et al., 1980). However, these studies have notsimultaneously considered the level of conflict between the par-ents. Perhaps it is not marital dissolution per se that is thesource of problems, but the overt parental conflict that leads upto it and continues after separation (Long et al., 1987). Whenseparation reduces the children's exposure to parental conflict,it may reduce their risk for maladjustment. Adequately address-ing the impact of marital dissolution on children requires at-tending to both family configuration and marital conflict.


Depressive behavior of parents. The interactional model ofdepression (Coyne, 1976a; Coyne et al., in press) emphasizesthe support-seeking displays of sadness and the inhibiting qual-ity of depressed persons. Clinical writers have also suggestedthat depressed mothers lean on their children, engaging in a"frantic search for support" (Fabian & Donohue, 1956, p. 403).Drake and Price (1975) viewed these children as "a primaryvehicle through which parents meet their dependency needs"(p. 164). Whereas the hostility and irritability of depressed par-ents has received considerable attention, their characteristicallydepressed behavior has been virtually ignored (for exception,see Hops et al., 1987). There needs to be a greater emphasison measures of parenting and parent-child interaction that aretheoretically relevant and tailored to the phenomenon of de-pression. Such measures may reveal the processes underlyingclinical depression in children of depressed parents.

Experience of depressed parents and their children. Ques-tions about the reciprocal influences of depressed parents, theirspouses, and their children might be more readily answered byasking family members directly about what transpires betweenthem. Structured interviews, questionnaires about how eachfamily member coped with critical incidents, and structureddaily diaries all have unrealized potential for uncovering familyprocesses that short-term observations cannot reveal. These ap-proaches are particularly applicable to older children who canserve as reliable informants. Self-report procedures may revealthat children with serious adjustment problems reinforce somedepressed women's sense of guilt and inadequacy. At the sametime, more positive aspects of the family relationships of de-pressed persons may be identified, as the following studyshowed. Williams and Carmichael (1985) found that a lower-class, multiethnic Australian sample of depressed women re-ported a full range of difficulties with their school-aged childrenbut not with their infants. In fact, most of the "mothers werevery protective of their infants and had good relationships withthem, and the infants seemed to be the one source of joy andcomfort in their otherwise sad, drab lives" (p. 86). Positive fea-tures of the family relationships of depressed persons have beenignored in the search for protective factors. Interviews, ques-tionnaires, and daily diaries might be particularly appropriatefor identifying these processes.

Child characteristics. Children with high social skills and in-telligence are at reduced risk for disturbance. Beyond this, littleis known about how child characteristics modify risk for malad-justment. Evidence that mother-child interaction is reciprocaland that some children show difficulties shortly after birth sug-gests a possible influential role for children's reactive character-istics, including temperament. By this, we mean that some chil-dren may prove troublesome for any mother and may be partic-ularly difficult for mothers who have a low tolerance forfrustration and are prone to demoralization. As in other areasof child and adult psychopathology, a comprehensive explana-tory model for disturbance in children of depressed parents willprobably be diathesis-stress in form (Zubin & Spring, 1977),but for now researchers must learn more about child vulnerabil-ity and protective factors. Comparisons within the families ofdepressed parents as well as between families will help to clarifythis issue. We have already noted that comparisons of siblingsare proving useful in elucidating the processes linking maternal

feelings of ineptitude, difficult child behavior, and child abusein high-risk families (Bugental et al., 1989).

Depression in children of depressed parents. The tentativeconclusion that children of depressed parents are at particularrisk for diagnosable depression suggests a need to sharpen thefocus on the nature of their depression. As a start, researchersneed to know about the construct validity and clinical featuresof these diagnoses, and the natural history of the condition thatis being identified. Recent studies have indicated that major de-pressive disorder among hospitalized children is similar to thatamong hospitalized adults in terms of phenomenology, familyhistory, drug-versus-placebo response, and neuroendocrinemarkers (Preskon, Weller, Hughes, Weller, & Bolte, 1987). How-ever, the continuities between milder depressive disorder in chil-dren, particularly nonreferred children identified through ascreening, and the adult disorder are unclear. We have noted thespeculation of one genetics researcher that the depression foundin the young children of affectively disturbed parents is a stressresponse and not an expression of a genetic vulnerability (Ca-doret, 1983). On the other hand, in a recent retrospective studythe proportion of children of depressed parents who remaineddepressed for 2 years was similar to rates found in investigationsof both children referred for psychiatric disorder and adult de-pressed patients (Keller, Beardslee, Lavori, Wunder, & Samuel-son, 1988). This finding may be seen as underscoring the seri-ousness of the child depression identified in the studies that wehave reviewed, although more research is needed.

Another way of addressing questions about the nature of de-pression in children of depressed parents is by examining bio-logical correlates of their depression. One study of children withmajor depressive disorder found that 40% met Research Diag-nostic Criteria for endogenous depression (Chambers, Puig-Antich, Tabrizi, & Davies, 1985). It would be interesting to es-tablish what proportion of children of depressed parents metthese criteria and the degree of concordance with their parents.During a depressive episode, many depressed children appar-ently show some of the same evidence of abnormalities in thehypothalamic-pituitary-adrenal (HPA) axis as do depressedadults, such as that demonstrated with the dexamethasone sup-pression test (DST; Weller & Weller, 1988). Although the DSTis hardly the "gold standard" for depression that some once con-sidered it to be, this test can be an important research tool whenproperly administered and interpreted (Baldessarini & Aranas,1985). Namely, when taken together with an elevated symptomscore or diagnosis based on structured interview, a positive DSTor other neuroendocrine marker offers further assurance of thespecificity of the diagnosis (Weller & Weller, 1988). Findings ofpositive DST results with the depressed children of depressedparents can be used to strengthen arguments for the connectionbetween the disturbance they manifest and risk for depressionin adulthood. The test might also be useful in studying the roleof psychosocial factors such as family interaction in the develop-ment of biological risk for depression. There is some evidencethat affective symptoms in the context of unsupportive relation-ships in early childhood are an antecedent of enduring alter-ations in the HPA axis functions in adults (Breier et al., 1988).To date, support for this intriguing hypothesis has depended onretrospective reports of adults about childhood adversity. Inclu-sion of HPA measures in studies of the children of depressedparents could provide a much more direct test.


Specificity reconsidered. Addressing the issue of specificitythrough the inclusion of appropriate control groups is a way ofclarifying what is unique among the many complex processesoccurring in families with a depressed parent. Yet the issue ofspecificity has sometimes been distorted, with investigators be-coming set on giving substantive interpretations to the findingthat children of depressed parents do not differ from some com-parison group. When one uses small convenience samplesmatched on only a few variables and global measures of childadjustment as criteria, null findings are likely but not particu-larly meaningful. Concluding that similar levels of child out-comes reflect the same underlying processes may therefore beincorrect. For instance, children of mothers recently hospital-ized for an acute medical condition may be as distressed as chil-dren of mothers hospitalized for depression, but in the first casedistress may be due to the mother's hospitalization and in thesecond the distress may decrease following the mother's hospi-talization because of a reduction in aversive interactions.

To be useful, comparison groups should be chosen with a par-ticular rationale, be representative of the population fromwhich they are sampled, and be matched and well described.Measures of child outcome should include both diagnosis andmeasures of general adjustment. Assessments of interveningprocesses should be extensive enough to identify both differ-ences and similarities between families of depressed and com-parison parents. Identification and recruitment of appropriatecomparison groups can be difficult, and speculative interpreta-tion of null results can be hazardous. At this point, one mightargue that too little is known to pick appropriate comparisongroups, other than normal controls. Faced with limited re-sources and the choice between using a small, somewhat arbi-trarily selected comparison group or doubling the size of thedepressed parent group, an investigator might do well to adoptthe second strategy and explore the relations among variableswithin the depressed sample. For example, comparisons of fam-ilies in which a parent is currently in an episode with families inwhich the parent is in remission are particularly relevant givenknowledge of the effects of living with a depressed person onadults (Coyne et al., 1987). Later, carefully selected comparisongroups might be used to test the hypotheses that have been iso-lated and refined in this fashion.

The next step. We have identified a large set of domains thatshould be considered in efforts to understand the connectionbetween depression in parents and child problems. No studycan be expected to address all of them, but cognizant of thiscomplexity, researchers should be cautious in interpreting theresults of any study. Within each of these domains, there needsto be a great deal of sorting and refinement of hypotheses. Expe-rience from studies of the children of schizophrenic parents,some of which we have reviewed, suggests that it is foolhardy toproceed with ambitious longer-term, longitudinal studies with-out first carefully refining one's research questions. Baldwin's(1960, cited in Garmezy & Devine, 1984) earlier commentsabout the need for caution in launching ambitious longitudinalstudies of personality development are equally appropriate tothe current state of research on children of depressed parents:

It should be apparent that a longitudinal study of the effects of one,two, or three variables of childhood experience upon later person-ality is a big investment for relatively small return—quantitativelyspeaking. It behooves us, therefore, to precede such an undertaking

with careful pretesting, study of cross-sectional differences, andcruder retrospective studies to establish the likelihood of majoreffects. A longitudinal study is the last, not the first step in a re-search program. It is an absolutely essential research method if weare to get firm knowledge of psychological change, (p. 27)

Before undertaking long-term, longitudinal studies of childrenof depressed parents, researchers should learn what they can fromthe studies of schizophrenia. Researchers can learn about the logis-tics of longitudinal high-risk studies, and can also learn substan-tively about children of depressed parents. Children in these earlystudies are now reaching adulthood, and some of the studies areundertaking diagnostic assessments. Thus, existing data ostensiblycollected for another purpose can be used to address questionsabout whether children of depressed parents are at specific risk foradult depression, as distinct from childhood depression, and tosort the relevance of some childhood risk factors for adult vulnera-bility to depression.

Implications for the development of an interactional perspec-tive on depression. An interactional perspective on depressionstarts with the assumption that depression takes on the charac-teristics of an emergent interpersonal system, and that an under-standing of depressed persons requires that something abouttheir interpersonal circumstances or ecological niches beknown—that is, the conditions of their everyday lives, the prob-lems that they face and how they cope, and the reciprocal influ-ences between them and the key persons around them (Coyne,1976b, Coyne et al., in press). Despite recent refinements, thisperspective provides a rather sketchy picture of the interper-sonal processes associated with depression. Findings concern-ing the children of depressed parents can serve as an importantimpetus for further refinements. The studies we reviewed high-light the correlates, if not clearly the effects, of depression inone set of close relationships, those between parent and child.The children are indeed distressed and are more likely to bedepressed. However, as with the spouses of depressed persons,explanations of their maladjustment will require a complexmodel (Coyne & DeLongis, 1989; Coyne et al., in press). Stud-ies of interactions between depressed parents and their childrenhave downplayed the effects of the parents' depressive displaysand support seeking, but as in the study of the marital interac-tions of depressed persons, these studies highlight the impor-tance of reciprocal irritability and overt hostility. They also callattention to the negative effects of the less-effortful interactionstrategies used by depressed persons. Most important, the studyof the children of depressed parents reinforces the notion ofhow difficult the close relationships of depressed persons can befor everyone involved, and the complexity of the links to thelarger context.

Concluding RemarksThe high rate of serious psychological problems in children

of depressed parents suggests an important social and publichealth problem. The literatures that we have reviewed also indi-cate that depressed mothers have a variety of difficulties as par-ents. Yet our review points to the need for caution in how theconnection between the problems of these children and theirparents is drawn. In particular, researchers must be sensitiveto the pitfalls of laying too much responsibility on the allegedshortcomings of depressed mothers. Taking an overview of the


diverse literatures that we have considered, we find a distinctand consistent, even if unintentional, "mother-bashing" qualityto much of this body of work. Studies establishing rates of dis-turbance in children have been interpreted in terms of the ad-verse consequences of the difficulties of depressed mothers.Studies finding both adjustment problems in the children andparenting problems in the mothers are taken as proof of thesimple link between the two. What is most troubling is that thedesigns of many studies provide no opportunity to challengeprevailing simplistic notions about the nature of this link. Theysimply fail to include assessments of the contextual factors thatmight produce spurious relations between the parenting pro-vided by depressed persons and their children's difficulties.

As we have noted, depression in a parent may represent onlyone source of the many disadvantages and deprivations facedby children of depressed parents. Furthermore, depression maybe just one of the problems that depressed parents have in deal-ing with these children. Little is known about the compensatorystrengths and resources of these parents, the reactive character-istics of their children, or the broader context of the mother-child relationship. However, what is known suggests that thesefactors should receive more attention. Depressed parents andtheir children are heterogenous groups and need to be describedin terms of their strengths as well as their difficulties, if research-ers are to better understand why and under what circumstancesparental depression is associated with problems in child devel-opment. Clearly, this requires moving beyond the deficit modelimplicit in the inordinate attention given to negative parentalattitudes and behavior and problematic child outcomes. At thispoint, it is matter of social responsibility as well as good theoret-ical and empirical sense to adopt a more contextual view ofdepressed parents and their children.

References

Achenbach, T. M, & Edelbrock, C. S. (1983). Manual for the ChildBehavior Checklist and Revised Behavior Profile. Burlington, VT:University of Vermont, Department of Psychiatry.

Akiskal, H. S. (1982). Factors associated with incomplete recovery inprimary depressive illness. Journal of Clinical Psychiatry, 43, 266-271.

Allen, M. (1976). Twin studies of affective illness. Archives of GeneralPsychiatry, 33, 1476-1478.

Baldessarini, R. J., & Aranas, G. W. (1985). Does the dexamethasonesupression test have a clinical utility? Journal of Clinical Psychiatry,46, 25-30.

Baldwin, A. (1960). The study of child behavior and development. InP. H. Mussen (Ed.), Handbook of research methods in child develop-ment (pp. 3-35). New York: Wiley.

Baldwin, A., Cole, R., & Baldwin, C. (1982). Parent pathology, familyinteraction, and the competence of the child in school. Monographsof the Society for Research in Child Development, 47(5, Serial No.197).

Beardslee, W, Bemporad, J., Keller, M. B., & Klerman, G. L. (1983).Children of parents with a major affective disorder: A review. Ameri-can Journal of Psychiatry, 140, 825-832.

Beardslee, W., & Podorefsky, D. (1988). Resilient adolescents whoseparents have serious affective and other psychiatric disorders: Impor-tance of self-understanding and relationships. American Journal ofPsychiatry, 145, 63-69.

Beardslee, W, Schultz, L., & Selman, R. (1987). Level of social cogni-tive development, adaptive functioning, and DSM-III diagnoses in

adolescent offspring of parents with affective disorders: Implicationsof the development of the capacity for mutuality. Developmental Psy-chology, 25,807-815.

Bettes, B. (1988). Maternal depression and motherese: Temporal andintonational features. Child Development, 59, 1089-1096.

Biglan, A., Hops, H., Sherman, L., Friedman, L., Arthur, J., & Osteen,V. (1985). Problem solving interactions of depressed mothers andtheir spouses. Behavior Therapy, 16, 431 -451.

Biglan, A., Rothlind, J., Hops, H., & Sherman, L. (1989). Impact ofdistressed and aggressive behavior. Journal of Abnormal Psychology,98, 218-228.

Billings, A. G., & Moos, R. H. (1983). Comparison of children of de-pressed and nondepressed parents: A social environmental perspec-tive. Journal of Abnormal Child Psychology, 11, 483-486.

Billings, A. G., & Moos, R. H. (1985). Children of parents with unipolardepression: A controlled one-year follow-up. Journal of AbnormalChild Psychology, 14, 149-166.

Black, D. W, Bell, S., Hulbert, J., & Nasrallah, A. (1988). The impor-tance of Axis II disorders in patients with major depression: A con-trolled study. Journal of Affective Disorders, 14, 115-122.

Bowlby, J. (1980). Attachment and Loss. Volume HI: Loss. New York:Basic Books.

Breslau, N., & Davis, G. (1986). Chronic stress and major depression.Archives of General Psychiatry, 43, 309-314.

Breslau, N., Davis, G., & Prabucki, K. (1988). Depressed mothers asinformants in family history research—Are they accurate? Psychia-try Research, 24, 345-359.

Breznitz, Z., & Sherman, T. (1987). Speech patterning of natural dis-course of well and depressed mothers and their young children. ChildDevelopment, 58, 395-400.

Breier, A., Kelsoe, J. F, Kirwin, P. D., Seller, S. A., Wolkowitz, O. M.,& Pickar, D. (1988). Early parental loss and development of adultpsychopathology. Archives of General Psychiatry, 45, 87-93.

Briscoe, C. W, & Smith, J. B. (1973). Depression and marital turmoil.Archives of General Psychiatry, 29, 811-817.

Brown, G. W, & Harris, T. (1978). Social origins of depression. NewYork: Free Press.

Bugental, D. B., Blue, J., & Cruzcosa, M. (1989). Perceived control overcaregiving outcomes: Implications for child abuse. DevelopmentalPsychology, 25, 532-539.

Bugental, D. B., & Shennum, W. A. (1984). "Difficult" children as elici-tors and targets of adult communication patterns: An attributional-behavioral transactional analysis. Monographs of the Society for Re-search in Child Development, 49(1, Serial No. 205).

Burbach, D. J., & Borduin, C. M. (1986). Parent-child relations and theetiology of depression: A review of methods and findings. ClinicalPsychology Review, 6, 133-153.

Cadoret, R. J. (1983). Discussion. In S. B. Guze, F. J. Earls, & J. E.Varrett (Eds.), Child psychopathology and development (p. 65). NY:Raven Press.

Cadoret, R., O'Gorman, T, Heywood, E., & Troughton, E. (1985). Ge-netic and environmental factors in major depression. Journal ofAffective Disorders, 9, 155-164.

Carlson, V., & Garber, J. (1986). Developmental issues in the classifica-tion of depression in children. In M. Rutter, C. Izard, & P. Read(Eds.), Depression in young people (pp. 399-434). New York: Guil-ford Press.

Chambers, W, Puig-Antich, J., Tabrizi, M. A., & Davies, M. (1985).The assessment of affective disorders in children and adolescents bysemi-structured interview: Test-retest reliability of the Schedule forAffective Disorders and Schizophrenia for school-age children, pres-ent episode version. Archives of General Psychiatry, 42, 696-697.

Cicchetti, D., & Aber, J. L. (1986). Early precursors of later depression:An organizational perspective. In L. Lipsitt & C. Rovee-Collier (Eds.),Advances in infancy (Vol. 4, pp. 87-137). Norwood, NJ: Ablex Press.

Clayton, P. J. (1983). The prevalence and course of the affective disor-


ders. In J. M. Davis & J. W. Maas (Eds.), The affective disorders (pp.193-202). Washington, DC: American Psychiatric Press.

Cohler, B. J., Grunebaum, H. U., Weiss, J. L., Hartman, C. R., & Gal-lant, D. H. (1977). Child care attitudes and adaptation to the mater-nal role among mentally ill and well mothers. American Journal ofOrthopsychiatry, 46, 123-133.

Cohn, J., Campbell, S. B., Matias, R., & Hopkins, J. (1990). Face-to-face interactions of postpartum depressed and nondepressed mother-infant pairs at 2 months. Developmental Psychology, 26, 15-23.

Cohn, J., Matias, R., Tronick, E., Connell, D., & Lyons-Ruth, K.(1986). Face-to-face interactions of depressed mothers and their in-fants. In E. Tronick & T. Field (Eds.), Maternal depression and infantdisturbance (pp. 31-46). San Francisco: Jossey-Bass.

Cohn, J., & Tronick, E. (1983). Three-month-old infants' reaction tosimulated maternal depression. Child Development, 54, 185-190.

Cohn, J., & Tronick, E. (1987). Mother-infant interaction: The se-quence of dyadic states at 3,6, and 9 months. Child Development, 54,185-193.

Colletta, N. D. (1983). At risk for depression: A study of young mothers.Journal of Genetic Psychology, 142, 301-310.

Conners, C. K., Himmelhoch, J., Goyette, C. H., Ulrich, R., & Neil,J. F. (1979). Children of parents with affective illness. Journal of theAcademy of Child Psychiatry, 18, 600-607.

Courtney, R. H. (1984). The effectiveness of social workers in the man-agement of depressed female patients in general practice. Psychologi-cal Medicine, ^(Monograph Suppl. 6).

Coyne, J. C. (1976a). Depression and the response of others. Journal ofAbnormal Psychology, 85, 186-193.

Coyne, J. C. (1976b). Toward an interactional description of depression.Psychiatry, 39, 28-40.

Coyne, J. C. (1990). Interpersonal processes in depression. In G. I.Keitner (Ed.), Depression and families (pp. 31-54). Washington, DC:American Psychiatric Press.

Coyne, J. C., Burchill, S. A. L., & Stiles, W. B. (in press). An interac-tional perspective on depression. In C. R. Snyder & D. O. Forsyth(Eds.), Handbook of social and clinical psychology: The health per-spective. New York: Pergamon.

Coyne, J. C., & DeLongis, A. (1989). Marriages and spouses of de-pressed persons: An intergrative review. Unpublished manuscript.

Coyne, J. C., & Gotlib, I. (1983). The role of cognition in depression:A critical appraisal. Psychological Bulletin, 94, 472-505.

Coyne, J. C., Kessler, R. C., Tal, M., Turnbull, J., Wortman, C. B., &Greden, J. (1987). Living with a depressed person: Burden and psy-chological distress. Journal of Consulting and Clinical Psychology, 55,347-352.

Crittenden, P. (1988). Relationships at risk. In J. Belsky & T. Nezworski(Eds.), Clinical implications of attachment theory (pp. 136-174).Hillsdale, NJ: Erlbaum.

Crnic, K. A., Greenberg, M. T, Ragozin, N., Robinson, N. M., & Bas-ham, R. B. (1983). Effects of stress and social support on mothers ofpremature and full-term infants. Child Development, 54, 209-217.

Crockenberg, S. B. (1981). Infant irritability, mother responsiveness,and social influences on the security of mother-infant attachment.Child Development, 52, 857-865.

Cummings, E. M., & Cicchetti, D. (in press). Toward a transactionalmodel of relations between attachment and depression. In M. T.Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment inthe preschool years: Theory, research, and intervention. Chicago: Uni-versity of Chicago Press.

Cytryn, E. M., McKnew, D., Bartko, J. J., Lamour, M., & Hamovit, J.(1982). Offspring of patients with affective disorders II. Journal of theAmerican Academy of Child Psychiatry, 21, 389-391.

Davenport, Y. B., Zahn-Waxler, C., Adland, M. L., & Mayfield, A.(1984). Early child rearing practices in families with a manic-depres-sive parent. American Journal of Psychiatry, 142, 230-235.

Decina, P., Kestenbaum, C. J., Farber, S., Kron, L., Gargan, M., Sack-

eim, H. A., & Fieve, R. R. (1983). Clinical and psychological assess-ment of children of bipolar probands. American Journal of Psychia-try, 140, 548-553.

Depue, R. A., & Monroe, S. M. (1978). Learned helplessness in theperspective of the depressive disorders: Conceptual and definitionalissues. Journal of Abnormal Psychology, 87, 3-20.

Dodge, K. (Ed.). (1990). Developmental psychopathology in children ofdepressed mothers. Developmental Psychology, 26, 3-6.

Downey, G., & Walker, E. F. (1989). Social cognition and adjustment inchildren at risk for psychopathology. Developmental Psychology, 25,835-845.

Drake, R. E., & Price, J. L. (1975). Depression: Adaptation to disrup-tion and loss. Perspectives in Psychiatric Care, 13, 163-169.

Edelbrock, G., Costello, A. J., Dulcan, M. K., Kalas, R., & Conover,N. C. (1985). Age differences in the reliability of the psychiatric inter-view of the child. Child Development, 56, 265-274.

Emery, R. (1982). Interpersonal conflict and the children of discord anddivorce. Psychological Bulletin, 92, 310-330.

Emery, R., Weintraub, S., & Neale, J. (1982). Effects of marital discordon the school behavior of children of schizophrenic, affective disor-dered, and normal parents. Journal of Abnormal Child Psychology,76,215-225.

Endicott, J., & Spitzer, R. (1972). Current and past psychopathologyscales (CAfPS). Archives of General Psychiatry, 27, 678-687.

Epstein, N. B., Baldwin, I. M., & Bishop, D. S. (1983). The McMasterassessment device. Journal of Marital and Family Therapy, 9, 171-180.

Erlenmeyer-Kimling, L., & Cornblatt, B. (1987). The New York High-Risk Project: A follow-up report. Schizophrenia Bulletin, 13, 451-461.

Fabian, A. A., & Donohue, J. F. (1956). Maternal depression: A chal-lenging child guidance problem. American Journal of Orthopsychia-try, 26, 400-405.

Feighner, J. P., Robins, E., Guze, S. B., Woodruff, R. A. Jr., & Munoz,R. (1972). Diagnostic criteria for use in psychiatric research. Archivesof General Psychiatry, 26, 57-63.

Fendrich, M., Warner, V., & Weissman, M. M. (1990). Family risk fac-tors, parental depression, and psychopathology in offspring. Develop-mental Psychology, 26,40-50.

Field, T. (1984). Early interactions between infants and their postpar-tum depressed mothers. Infant Behavior and Development, 7, 517-522.

Field, T, Healy, B., Goldstein, S., & Guthertz, M. (1990). Behavior-statematching and synchrony in mother-infant interactions of nonde-pressed vs. depressed dyads. Developmental Psychology, 26, 7-14.

Field, T, Healy, B., Goldstein, S., Perry, S., Bendell, D., Schamberg, S.,Zimmerman, E., & Kuhn, G. (1988). Infants of depressed mothersshow "depressed" behavior even with non-depressed adults. ChildDevelopment, 60, 1569-1579.

Field, T, Sandberg, D., Garcia, R., Vega-Lahr, N., Goldstein, S., & Guy,L. (1985). Pregnancy problems, postpartum depression, and earlymother-infant interactions. Developmental Psychology, 21, 1152-1156.

Fisher, L., & Kokes, R. F. (1983). Assessing the relationship betweenparental behavior and child functioning. In H. L. Morrison (Ed.),Children of depressed parents: Risk, identification and intervention(pp. 47-64). New York: Grune & Stratton.

Fisher, L., Kokes, R. F, Harder, D., & Jones, J. E. (1980). Child compe-tence and psychiatric risk: IV. Summary and integration of findings.Journal of Nervous and Mental Disease, 168, 353-355.

Fisher, L., Schwartzman, P., Harder, D., & Kokes, R. (1984). A strategyand methodology for assessing school competence in high-risk chil-dren. In N. Watt, E. J. Anthony, L. Wynne, & J. Rolf (Eds.), Childrenat risk (pp. 355-359). New Vbrk: Cambridge University Press.

Fleming, A., Ruble, D., Flett, G., & Shaul, D. (1988). Postpartum ad-justment in first-time mothers: Relations between mood, maternal


attitudes, and mother-infant interactions. Developmental Psychology,24,11-81.

Forehand, R., McCombs, A., & Brody, G. (1987). The relationship be-tween parental depressive mood states and child functioning. Ad-vances in Behavior Therapy and Research, 9, 1-20.

Forehand, R., Wells, K., & Griest, D. (1980). An examination of thesocial validity of a parent training program. Behavior Therapy, 11,488-502.

Friedman, L. (1984). Family interaction among children of unipolar de-pressed mothers: A naturalistic observation study. Unpublished doc-toral dissertation, University of Oregon, Eugene.

Gaensbauer, J. J., Harmon, R. J., Cytryn, L., & McKnew, D. H. (1984).Social and affective development in infants with a manic-depressiveparent. American Journal of Psychiatry, 141, 223-229.

Gamer, E., Gallant, D., Grunebaum, H., & Cohler, B. J. (1977). Chil-dren of psychotic mothers. Archives of General Psychiatry, 34, 592-597.

Garmezy, N. (1983). Stressors of childhood. In N. Garmezy & M. Rut-ter (Eds.), Stress, coping, and development in children (pp. 43-84).New York: McGraw-Hill.

Garmezy, N., & Devine, V. (1984). Project Competence: The Minne-sota studies of children vulnerable to psychopathology. In N. Watt,E. J. Anthony, L. Wynne, & J. Rolf (Eds.), Children at risk for schizo-phrenia (pp. 289-303). New \brk: Cambridge University Press.

Gershon, E., McKnew, D., Cytryn, L., Hamovit, J., Schreiber, J., Hibbs,E., & Pellegrini, D. (1985). Diagnosis in school-age children of bipo-lar affective disorder parents and normal controls. Journal of AffectiveDisorders, 5,283-291.

Goodman, S. (1987). Emory University Project on children of disturbedparents. Schizophrenia Bulletin, 13, 412-423.

Goodman, S. H., & Brumley, H. E. (1990). Schizophrenic and de-pressed mothers: Relational deficits in parenting. Developmental Psy-chology, 26, 31-39.

Gordon, D., Burge, D., Hammen, C., Adrian, C., Jaenicke, C., & Hi-roto, D. (1989). Observations of interactions of depressed womenwith their children. American Journal of Psychiatry, 146, 50-55.

Grunebaum, H. U., Cohler, B. J., Kaufman, C., & Gallant, D. H.(1978). Children of depressed and schizophrenic mothers. Child Psy-chiatry and Human Development, 8, 219-228.

Hammen, C. (1988). Self cognitions, stressful events, and the predictionof depression in children of depressed mothers. Journal of AbnormalChild Psychology, 16, 347-360.

Hammen, C., Adrian, C., Gordon, D., Burge, D., Jaenicke, C., & Hi-roto, D. (1987). Children of depressed mothers: Maternal strain andsymptom predictors of dysfunction. Journal of Abnormal Psychology,96, 190-198.

Hammen, C., Adrian, C., & Hiroto, D. (1988). A longitudinal test of theattributional vulnerability model in children at risk for depression.British Journal of Clinical Psychology, 27, 37-46.

Hammen, C., Burge, D., & Stansbury, K. (1990). Relationship ofmother and child variables to child outcomes in a high-risk sample:A causal modeling analysis. Developmental Psychology, 26, 24-30.

Hammen, C., Gordon, D., Burge, D., Adrian, C., Jaenicke, C., & Hi-roto, G. (1987a). Communication patterns of mothers with affectivedisorders and their relationship to children's status and social func-tioning. In K. Hahlweg & M. J. Goldstein (Eds.), Understanding ma-jor mental disorder: The contribution of family interaction research(pp. 103-119). New York: Family Process Press.

Hammen, C., Gordon, G., Burge, D., Adrian, C., Jaenicke, C., & Hi-roto, G. (1987b). Maternal affective disorders, illness, and stress: Riskfor children's psychopathology. American Journal of Psychiatry, 144,736-741.

Harder, D. W, Kokes, R. F, Fisher, L., & Strauss, J. (1980). Child com-petence and psychiatric risk IV. Relationships of parent diagnosticclassifications and parent psychopathology severity to child function-ing. Journal of Nervous and Mental Disease, 168, 343-347.

Hetherington, E. M., Cox, M., & Cox, R. (1982). Effects of divorce onparents and children. In M. Lamb (Ed.), Nontraditional families (pp.233-288). Hillsdale, NJ: Erlbaum.

Hetherington, E. M., Stanley-Hagen, M., & Anderson, E. R. (1989).Marital transitions: A child's perspective. American Psychologist, 44,303-312.

Hirsch, B. J., Moos, R. F, & Reischl, T. M. (1985). Psychosocial adjust-ment of adolescent children of a depressed, arthritic, or normal par-ent. Journal of Abnormal Psychology, 94, 154-164.

Holmes, S. J., & Robins, L. N. (1988). The role of parental disciplinarypractices in the development of depression and alcoholism. Psychia-try, 51, 24-36.

Hooley, J. M., Orley, J., & Teasdale, J. D. (1986). Levels of expressedemotion and relapse in depressed patients. British Journal of Psychia-try, 148, 642-647.

Hooley, J., & Teasdale, J. (1989). Predictors of relapse in unipolar de-pressives: Expressed emotion, marital distress, and perceived criti-cism. Journal of Abnormal Psychology, 98, 229-235.

Hops, H., Sherman, L., & Biglan, A. (in press). Maternal depression andmarital discord in children's behaviors: A developmental perspective.In G. R. Patterson (Ed.), Depression and aggression: Two facets offamily interaction. Hillsdale, NJ: Erlbaum.

Hops, H., Biglan, A., Sherman, L., Arthur, J., Friedman, L., & Osteen,R. (1987). Home observations of family interactions of depressedwomen. Journal of Consulting and Clinical Psychology, 55, 341-346.

Jacobson, N. S., Schmaling, K. B., & Holtzworth-Munroe, A. (1987).A component analysis of behavioral marital therapy: Two year follow-up and prediction of relapse. Journal of Marital and Family Therapy,13, 187-196.

Jaenicke, C., Hammen, C., Zupan, B., Hiroto, D., Gordon, D., Adrian,C., & Burge, D. (1987). Cognitive vulnerability in children at risk fordepression. Journal of Abnormal Child Psychology, 15, 559-572.

Kashani, J. H., Shekim, W. Q, Burk, J. P., & Beck, N. C. (1987). Abuseas a predictor of psychopathology in children and adolescents. Jour-nal of Clinical Child Psychology, 16, 43-50.

Kaufman, C., Grunebaum, H., & Cohler, B. (I979). Superkids: Compe-tent children of psychotic mothers. American Journal of Psychiatry,136, 1398-1402.

Kazdin, A. E., Moser, J., Colbus, D., & Bell, R. (1985). Depressivesymptoms among physically abused and psychiatrically disturbedchildren. Journal of Abnormal Psychology, 94, 298-307.

Keitner, G. I., Miller, I. W, Epstein, N. B., Bishop, D. S., & Fruzzetti,A. E. (1987). Family functioning and the course of major depression.Comprehensive Psychiatry, 28, 54-64.

Keller, M., Beardslee, W, Dorer, D., Lavori, P., Samuelson, H., & Kler-man, G. (1986). Impact of severity and chronicity of parental affec-tive illness on adaptive functioning and psychopathology in children.Archives of General Psychiatry, 43, 930-937.

Keller, M., Beardslee, W, Lavori, P., Wunder, J., & Samuelson, H.(1988). Course of major depression in nonreferred adolescents: A ret-rospective study. Journal of Affective Disorders, 15, 235-243.

Keller, M. B., & Shapiro, R. W. (1982). "Double depression": Superim-position of acute depressive episodes on chronic depressive disorders.American Journal of Psychiatry, 139, 438-442.

Kendler, K., Gruenberg, A. M., & Strauss, J. S. (1981). An independentanalysis of the Copenhagen sample of the Danish adoption study ofschizophrenia. Archives of General Psychiatry, 38, 973-990.

Kessler, R. C. (1987). The interplay of research design strategies anddata analysis procedures in evaluating the effects of stress on health.In S. V. Kasl & C. L. Cooper (Eds.), Stress and health: Issues in re-search methodology (pp. 113-140). New "York: Wiley.

Klein, D., Clark, D., Dansky, L., & Margolis, E. T. (1988). Dysthymiain the offspring of parents with primary unipolar affective disorder.Journal of Abnormal Psychology, 94, 115-127.

Klein, D., Depue, R. A., & Slater, J. F. (1985). Cyclothymia in the ado-


lescent offspring of parents with bipolar affective disorder. Journal ofAbnormal Psychology, 94, 115-127.

Kochanska, G., Kuczynski, L., Radke-Yarrow, M., & Welsh, J. D.(1987). Resolutions of control episodes between well and affectivelyill mothers and their young child. Journal of Abnormal Child Psychol-ogy, 75,441-456.

Kuczynski, L. (1984). Socialization goals and mother-child interaction:Strategies for long-term and short-term compliance. DevelopmentalPsychology, 20, 1061-1073.

Kuyler, P. L., Rosenthal, L., Igel, G., Dunner, D. L., & Fieve, R. R.(1980). Psychopathology among children of manic depressive pa-tients. Biological Psychiatry, 15, 589-597.

Lee, C, & Gotlib, I. (1989a). Clinical status and emotional adjustmentof children of depressed mothers. American Journal of Psychiatry,146, 478-483.

Lee, C., & Gotlib, I. (1989b). Maternal depression and child adjust-ment: A longitudinal analysis. Journal of Abnormal Psychology, 98,78-85.

Lewine, R., Watt, N., & Grubb, T. (1984). High-risk-for-schizophreniaresearch: Sampling bias and its implications. In N. Watt, E. J. An-thony, L. Wynne, & J. Rolf (Eds.), Children at risk for schizophrenia(pp. 557-564). New York: Cambridge University Press.

Lewine, R. R., Watt, N. F, Prentsky, R. A., & Fryer, J. H. (1980). Child-hood social competence in functionally disordered psychiatric pa-tients and in normals. Journal of Abnormal Psychology, 89, 132-138.

Lewinsohn, P. M., Hoberman, H. M., & Rosenbaum, M. (1988). Aprospective study of risk factors for unipolar depression. Journal ofAbnormal Psychology, 97, 251-264.

Livingood, A. B., Daen, P., & Smith, B. D. (1983). The depressedmother as a source of stimulation for her infant. Journal of ClinicalPsychology, 39, 369-375.

Long, N., & Forehand, R. (1987). The effects of parental divorce andparental conflict on children: An overview. Developmental and Be-havioral Pediatrics, 8, 292-296.

Long, N., Forehand, R., Fauber, R., & Brody, E. (1987). Self-perceivedand independently observed competence of young adolescents as afunction of parental marital conflict and recent divorce. Journal ofAbnormal Child Psychology, 15, 15-27.

Lyons-Ruth, K., Zoll, D., Connell, D., & Grunebaum, H. U. (1986).The depressed mother and her one-year-old infant. Environment, in-teraction, attachment, and infant development. In E. Tronick & T.Field (Eds.), Maternal depression and infant disturbance (pp. 61 -82).San Francisco: Jossey-Bass.

McLoyd, V. (1989). The declining fortunes of Black children: Psycholog-ical distress, parenting, and socioemotional development in the con-text of economic hardship. Unpublished manuscript, University ofMichigan.

McNeil, T, & Kaij, L. (1987). Swedish High-Risk Study: Sample char-acteristics at age 6. Schizophrenia Bulletin, 13, 372-381.

Menaghan, E., & Lieberman, M. (1986). Changes in depression follow-ing divorce: A panel study. Journal of Marriage and the Family, 48,319-328.

Merikangas, K., Prusoff, B., & Weissman, M. (1988). Parental concor-dance for affective disorders: Psychopathology in offspring. Journalof Affective Disorders, 15, 279-290.

Merikangas, K., & Spiker, D. G. (1982). Assortative mating among in-patients with primary affective disorder. Psychological Medicine, 12,753-764.

Merikangas, K., Weissman, M., Prusoff, B., & John, K. (1988). Assorta-tive mating and affective disorders: Psychopathology in offspring. Psy-chiatry, 51, 48-57.

Miller, P. M., Dean, C., Ingham, J. G., Kreitman, N. B., Sashidharan,S. P., & Surtees, P. G. (1986). The epidemology of life events and long-term difficulties with some reflections on the concept of indepen-dence. British Journal of Psychiatry, 148, 686-696.

O'Hara, M. (1986). Social support, life events, and depression during

pregnancy and the puererium. Archives of General Psychiatry, 43,569-573.

Orvaschel, H. (1983). Maternal depression and child dysfunction. In B.Lahey & A. Kazdin (Eds.), Advances in clinical child psychology (Vol.6, pp. 169-197). New York: Plenum Press.

Orvaschel, H., Welsh-Allis, G., & Weijai, Y. (1988). Psychopathology inchildren of parents with recurrent depression. Journal of AbnormalChild Psychology, 16, 17-28.

Panaccione, V., & Wahler, R. (1986). Child behavior, maternal depres-sion, and social coercion as factors in the quality of child care. Journalof Abnormal Child Psychology, 14, 273-284.

Parker, G., & Hadzi-Pavlovic, D. (1984). Modification of levels of de-pression in mother-bereaved women by parental and marital relation-ships. Psychological Medicine, 14, 125-135.

Patterson, G. (1982). Family coercive process. Eugene, OR: CastiliaPress.

Paykel, E. S., Klerman, G. L., & Prusoff, B. A. (1970). Treatment settingand clinical depression. Archives of General Psychiatry, 36, 1261-1269.

Pelligrini, A. D., Kosisky, S., Nackman, D., Cytryn, L., McKnew, D.,Gershon, E., Hamovit, J., & Cammuso, K. (1986). Personal and so-cial resources in children of patients with bipolar affective disorderand children of normal control subjects. American Journal of Psychi-atry, 143,856-961.

Peterson, C., & Seligman, M. E. P. (1984). Causal explanations as a riskfactor for depression: Theory and evidence. Psychological Review, 91,347-374.

Phofl, B., Stangl, D., & Zimmerman, M. (1984). The implications ofDSM III-R personality disorders for patients with major depression.Journal of Affective Disorders, 7, 309-318.

Preskon, S. H., Weller, E. B., Hughes, C. W, Weller, R. A., & Bolte, K.(1987). Depression in prepubertal children: Dexamethasone nonsup-pression predicts differential response to imiprimine vs. placebo. Psy-chopharmacology Bulletin, 23, 128-133.

Quinton, D., Rutter, M., & Liddle, C. (1984). Institutional rearing, par-enting difficulties, and marital support. Psychological Medicine, 14,107-124.

Radke-Yarrow, M., Cummings, E. M., Kuczynski, L., & Chapman, M.(1985). Patterns of attachment in two- and three-year olds in normalfamilies and families with parental depression. Child Development,36, 884-893.

Radke-Yarrow, M., Richters, J., & Wilson, W. E. (in press). Child devel-opment in a network of relationships. In R. Hinde & J. Stevenson-Hinde (Eds.), Individuals in a network of relationships. Cambridge,England: Cambridge University Press.

Radke-Yarrow, M., & Sherman, T. (in press). Hard growing: Childrenwho survive. In J. Rolf, A. Masten, D. Cicchetti, K. Nuechterlein, &S. Weintraub (Eds.), Risk and protective factors in the development ofPsychopathology. Cambridge, England: Cambridge University Press.

Richters, J. (1987). Chronic versus episodic stress and the adjustmentof high-risk offspring. In K. Hahlweg & M. J. Goldstein (Eds.), Un-derstanding major mental disorder: The contribution of family inter-action research (pp. 74-90). New York: Family Process Press.

Richters, J., & Pellegrini, D. (1989). Depressed mothers' judgementsabout their children: An examination of the depression-distortion hy-pothesis. Child Development, 60, 1068-1075.

Rolf, J. (1972). The social and academic competence of children vulner-able to schizophrenia and other behavior pathologies. Journal of Ab-normal Psychology, 80, 225-243.

Rolf, J., & Garmezy, N. (1974). The school performance of childrenvulnerable to behavior pathology. In D. Ricks, A. Thomas, & M. Roff(Eds.), Life history research in psychopathology (Vol. 3, pp. 87-107).Minneapolis: University of Minnesota Press.

Rosenthal, D. (1970). Genetic theory and abnormal behavior. New \fork:McGraw-Hill.

Rounsaville, B. J., Prusoff, B. A., & Weissman, M. M. (1980). The


course of marital disputes in depressed women: A 48-month follow-up study. Comprehensive Psychiatry, 21, 111-118.

Rutter, M. (1966). Children of sick parents. Oxford, England: OxfordUniversity Press.

Rutter, M. (1990). Commentary: Some focus and process considera-tions regarding the effects of parental depression on children. Devel-opmental Psychology, 26, 60-67.

Rutter, M., & Quinton, D. (1984). Parental psychiatric disturbance:Effects on children. Psychological Medicine, 14, 853-880.

Sameroff, A. J., Barocas, R., & Seifer, R. (1984). The early developmentof children born to mentally ill women. In N. Watt, E. J. Anthony, L.Wynne, & J. Rolf (Eds.), Children at risk for schizophrenia (pp. 482-514). New \brk: Cambridge University Press.

Sameroff, A. J., Seifer, R., & Zax, M. (1982). Early development of chil-dren at risk for emotional disorder. Monographs of the Society forResearch in Child Development, 47(1, Serial No. 199).

Sameroff, A., Seifer, R., Zax, M., & Barocas, R. (1987). Early indicatorsof developmental risk: Rochester Longitudinal Study. SchizophreniaBulletin, 13, 363-394.

Shea, M., Glass, D., Pilkonis, P., Watkins, J., & Docherty, J. (1987).Frequency and implications of personality disorders in a sample ofdepressed outpatients. Journal of Personality Disorders, 1, 27-42.

Vaughn, C. E., & Leff, J. P. (1976). The influence of family and socialfactors on the course of psychiatric illness. British Journal of Psychia-try, 129, 125-137.

Von Knorring, A., Cloninger, R., Bohman, M., & Sigvardsson, S.(1983). An adoption study of depressive disorders and substanceabuse. Archives of General Psychiatry, 40, 943-950.

Wahler, R., & Dumas, J. (1989). Attentional problems in dysfunctionalmother-child interactions: An interbehavioral model. PsychologicalBulletin, 105, 116-130.

Walker, E., Downey, G., & Bergman, A. (1989). The effects of parentalpsychopathology and maltreatment on child behavior: A test of thediathesis-stress model. Child Development, 60, 15-24.

Walker, E., Downey, G., & Nightingale, N. (1989). The nonorthogonalnature of risk for psychopathology. Journal of Primary Prevention, 9,143-163.

Watt, N. (1984). In a nutshell: The first two decades of high-risk re-search in schizophrenia. In N. Watt, E. J. Anthony, L. Wynne, & J.Rolf (Eds.), Children at risk for schizophrenia (pp. 572-596). NewYork: Cambridge University Press.

Webster-Stratton, C., & Hammond, M. (1988). Maternal depressionand its relationship to life stress, perceptions of child behavior prob-lems, parenting behaviors, and child conduct problems. Journal ofAbnormal Child Psychology, 16, 299-315.

Weintraub, S. (1987). Risk factors in schizophrenia: The Stony BrookHigh-Risk Project. Schizophrenia Bulletin, 13, 439-449.

Weintraub, S., & Neale, J. (1984). Social behavior of children at risk forschizophrenia. In N. Watt, E. J. Anthony, L. Wynne, & J. Rolf (Eds.),Children at risk for schizophrenia (pp. 279-286). New York: Cam-bridge University Press.

Weintraub, S., Winters, K. C., & Neale, J. M. (1986). Competence andvulnerability in children with an affectively disordered parent. In M.Rutter, C. E. Izard, & P. B. Read (Eds.), Depression in young people(pp. 205-220). New York: Guilford Press.

Weissman, M. (1987). Advances in psychiatric epidemiology: Rates andrisks for major depression. American Journal of Public Health, 77,445-451.

Weissman, M. (1988). Psychopathology in the children of depressedparents: Direct interview studies. In D. L. Dunner, E. S. Gershon, &J. E. Barrett (Eds.), Relatives at risk for mental disorder (pp. 143-159). New York: Raven Press.

Weissman, M., Gammon, D., John, K., Merikangas, K., Warner, V.,

Prusoff, B., & Sholomskas, D. (1987). Children of depressed parents.Archives of General Psychiatry, 44, 747-753.

Weissman, M., John, K., Merikangas, K., Prusoff, B., Wickramaratne,P., Gammon, D., Argold, A., & Waxner, V. (1986). Depressed parentsand their children. American Journal of Diseases of Childhood, 140,801-805.

Weissman, M., Leaf, P., & Bruce, M. L. (1987). Single parent women.Social Psychiatry, 22, 29-36.

Weissman, M., & Meyers, J. (1980). Psychiatric disorders in a U.S. com-munity: The application of research diagnostic criteria to a resur-veyed community sample. Acta Psychiatricia Scandanavica, 62, 99-111.

Weissman, M., & Paykel, E. S. (1974). The depressed woman: A studyof social relationships. Chicago, IL: University of Chicago Press.

Weissman, M., Wickramaratne, P., Warner, V, John, K., Prusoff, B.,Merikangas, K., & Gammon, G. D. (1987). Assessing psychiatric dis-orders in children. Archives of General Psychiatry, 44, 747-753.

Weller, E. B., & Weller, A. A. (1988). Neuroendocrine changes in affec-tively ill children and adolescents. Endocrinology ofNeuropsychiatricDisorders, 17,41-54.

Welner, Z., Welner, A., McCrary, M. D., & Leonard, M. A. (1977). Psy-chopathology in children of inpatients with depression: A controlledstudy. Journal of Nervous and Mental Disease, 164, 408-413.

Williams, H. E., & Carmichael, A. (1985). Depression in mothers in amulti-ethnic urban industrial municipality in Melbourne: Aetiologi-cal factors and effects on infants and preschool children. Journal ofChild Psychology and Psychiatry and Allied Disciplines, 26,277-288.

Wing, J. K., Cooper, J. E., & Sartorius, N. (1974). Measurement andclassification of psychiatric symptoms. London: Cambridge Univer-sity Press.

Worland, J., Janes, C., Anthony, E. J., McGinnis, M., & Cass, L. (1984).St. Louis Risk Research Project: Comprehensive progress report ofexperimental studies. In N. Watt, E. J. Anthony, L. Wynne, & J. Rolf(Eds.), Children at risk for schizophrenia (pp. 105-147). New York:Cambridge University Press.

Wynne, L., Cole, R., & Perkins, P. (1987). University of Rochester Childand Family Study: Risk research in progress. Schizophrenia Bulletin,13, 403-476.

Youngren, M. A., & Lewinsohn, P. M. (1980). The functional relation-ship between depression and interpersonal behavior. Journal of Ab-normal Psychology, 89, 333-341.

Zahn-Waxler, C., Cummings, E. M., McKnew, D., & Radke-Yarrow,M. (1984). Altruism, aggression, and social interactions in youngchildren with a manic-depressive parent. Child Development, 55,112-122.

Zahn-Waxler, C., Kochanska, G., Krupnick, J., & McKnew, D. (1990).Patterns of guilt in children of depressed and well mothers. Develop-mental Psychology, 26, 51-59.

Zahn-Waxler, C., Mayfield, A., Radke-Yarrow, M., McKnew, D., Cy-tryn, L., & Davenport, Y. (1988). A follow-up investigation of off-spring of parents with bipolar disorder. American Journal of Psychia-try, 145, 506-509.

Zahn-Waxler, C., McKnew, D., Cummings, E. M., Davenport, Y. B., &Radke-Yarrow, M. (1984). Problem behaviors in peer interaction ofyoung children with a manic-depressive parent. American Journal ofPsychiatry, 141, 236-240.

Zis, A. P., & Goodwin, F. K. (1979). Major affective disorder as a recur-rent illness: A critical review. Archives of General Psychiatry, 36, 835-839.

Zubin, J., & Spring, B. (1977). Vulnerability—a new view of schizo-phrenia. Journal of Abnormal Psychology, 86, 103-126.

Received December 5,1988Revision received November 13, 1989

Accepted November 30, 1989 •

children of depressed parents: an integrative review · pdf filechildren of depressed parents:...

Documents