ORIGINAL CONTRIBUTION

Remissions in Maternal Depressionand Child PsychopathologyA STAR*D-Child ReportMyrna M. Weissman, PhDDaniel J. Pilowsky, MD, MPHPriya J. Wickramaratne, PhDArdesheer Talati, PhDStephen R. Wisniewski, PhDMaurizio Fava, MDCarroll W. Hughes, PhDJudy Garber, PhDErin Malloy, MDCheryl A. King, PhDGabrielle Cerda, MDA. Bela Sood, MDJonathan E. Alpert, MD, PhDMadhukar H. Trivedi, MDA. John Rush, MDfor the STAR*D-Child Team

PARENTAL DEPRESSION IS AMONG

the most consistent and well-replicated risk factors for child-hood anxiety and disruptive be-

havior disorders and for majordepressive disorder, with more than a2- to 3-fold increased risk in offspringof depressed parents compared with con-trols. These offspring problems often be-gin before puberty, continue into ado-lescence and adulthood, and can betransmitted to the next generation.1 Thelong-term morbidity includes im-paired social and occupational function-ing and increased risk of medical prob-lems as the offspring mature.2 Treatmentof these childhood disorders is contro-versial and is based on a limited num-ber of controlled clinical trials.3 In con-trast, for adults there is considerableevidence for the efficacy of a variety ofpharmacotherapies and psychothera-

pies for depression. Although early on-set major depression is highly familialand has a strong genetic component,4-6

environmental factors, such as dis-

Author Affiliations are listed at the end of this article.Corresponding Author: Myrna M. Weissman, PhD,College of Physicians and Surgeons, Columbia Uni-versity, New York State Psychiatric Institute, Unit 24,1051 Riverside Dr, New York, NY 10032 (mmw3@columbia.edu).

Context Children of depressed parents have high rates of anxiety, disruptive, and de-pressive disorders that begin early, often continue into adulthood, and are impairing.

Objective To determine whether effective treatment with medication of women withmajor depression is associated with reduction of symptoms and diagnoses in their chil-dren.

Design Assessments of children whose depressed mothers were being treated withmedication as part of the multicenter Sequenced Treatment Alternatives to RelieveDepression (STAR*D) trial conducted (between December 16, 2001 and April 24, 2004)in broadly representative primary and psychiatric outpatient practices. Children wereassessed by a team of evaluators not involved in maternal treatment and unaware ofmaternal outcomes. Study is ongoing with cases followed at 3-month intervals.

Setting and Patients One hundred fifty-one mother-child pairs in 8 primary careand 11 psychiatric outpatient clinics across 7 regional centers in the United States. Chil-dren were aged 7 to 17 years.

Main Outcome Measures Child diagnoses based on the Kiddie Schedule for Af-fective Disorders and Schizophrenia; child symptoms based on the Child Behavior Check-list; child functioning based on the Child Global Assessment Scale in mothers whosedepression with treatment remitted with a score of 7 or lower or whose depressiondid not remit with a score higher than 7 on the Hamilton Rating Scale for Depression.

Results Remission of maternal depression after 3 months of medication treatment wassignificantly associated with reductions in the children’s diagnoses and symptoms. Therewas an overall 11% decrease in rates of diagnoses in children of mothers whose depres-sion remitted compared with an approximate 8% increase in rates of diagnoses in chil-dren of mothers whose depression did not. This rate difference remained statistically sig-nificantaftercontrollingfor thechild’sageandsex,andpossibleconfoundingfactors (P=.01).Of the children with a diagnosis at baseline, remission was reported in 33% of those whosemothers’ depression remitted compared with only a 12% remission rate among childrenof mothers whose depression did not remit. All children of mothers whose depressionremitted after treatment and who themselves had no baseline diagnosis for depressionremained free of psychiatric diagnoses at 3 months, whereas 17% of the children whosemothers remained depressed acquired a diagnosis. Findings were similar using child symp-toms as an outcome. Greater level of maternal response was associated with fewer cur-rent diagnoses and symptoms in the children, and a maternal response of at least 50%was required to detect an improvement in the child.

Conclusions Remission of maternal depression has a positive effect on both mothersand their children, whereas mothers who remain depressed may increase the rates of theirchildren’s disorders. These findings support the importance of vigorous treatment for de-pressed mothers in primary care or psychiatric clinics and suggest the utility of evaluatingthe children, especially children whose mothers continue to be depressed.JAMA. 2006;295:1389-1398 www.jama.com

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, March 22/29, 2006—Vol 295, No. 12 1389

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

rupted parent-child attachment and poorparent-child bonding, may mediate theimpact of parental depression on chil-dren’s symptoms.7,8

Only a few studies of children of de-pressed parents have suggested somebenefit for children of reducing paren-tal symptoms, but none of those pub-lished have directly treated parental de-pression in a definitive large sample.9,10

One ongoing study of 129 high-risk off-spring, aged 7 to 17 years, found thatremission of depression in parents af-ter 4 months of various treatments wasassociated with significant reductionsin children’s depressive, internalizing,and externalizing symptoms (J.G., PhD,oral communication, September 2005).

The Sequenced Treatment Alterna-tives to Relieve Depression (STAR*D)trial provided a unique opportunity tostudy the children of ambulatory de-pressed mothers who were being treatedand followed up as part of the studyprotocol.11-13 Our study was restrictedto mothers because the rate of depres-sion is higher in women than it is inmen, particularly women in the child-rearing ages and because mothers aremore likely than fathers to bring theirchildren in for assessments. We previ-ously reported the demographic andclinical characteristics of the mother-child pairs before the commencementof maternal treatment.14 Our focusherein is on the symptomatic and be-havioral functioning of the children as-sessed 3 months after the initiation oftreatment of maternal depression by ateam of evaluators not involved in ma-ternal treatment and unaware of ma-ternal outcomes. These children will befollowed up periodically for a year af-ter maternal depression remission or for2 years should the mother remain de-pressed. We hypothesized that reduc-tion of maternal depression would beassociated with reductions in currentpsychopathological symptoms and dis-orders in their offspring.

METHODSThe sample consisted of 151 motherswho were enrolled (recruited betweenDecember 16, 2001 amd April 24, 2004)

in STAR*D, a multisite US study de-signed to determine the comparativeeffectiveness and acceptability of dif-ferent treatment options for a broadlyrepresentative group of outpatientswith nonpsychotic major depressivedisorder. The rationale, methods, anddesign of the trial have been detailedelsewhere.11-13 Clinical sites includedprimary care and outpatient psychiat-ric care settings serving public- or pri-vate-sector patients. Participants wereadults, aged 18 to 75 years, with non-psychotic major depressive disorder(baseline score on the 17-item Hamil-ton Rating Scale for Depression15,16

[HRSD] �14) and without a lifetime di-agnosis of bipolar disorder, schizophre-nia, or schizoaffective disorders. Pa-tients with concurrent medical andpsychiatric conditions, except as notedabove, were included unless a medicalcondition contraindicated one of thestudy medications. The STAR*D trialoffered 5 treatment levels delivered se-quentially. At the outset, all study par-ticipants were initially treated with cita-lopram (level 1 treatment). Those notremitting with or intolerant of citalo-pram could receive subsequent treat-ment steps provided in an equipose ran-domized design described elsewhere.17

Because the STAR*D-Child studywas an ancillary study and requiredseparate scientific review, it was initi-ated about a year after the adult seg-ment began. Seven of the 14 regionalcenters involved in the trial partici-pated in the trial, based on willingnessto participate, the presence of a sub-stantial number of women in the child-rearing ages, and the availability of cli-nicians experienced at evaluatingchildren. Participating women aged 25to 60 years were screened to ascertainwhether they had children who met eli-gibility criteria. Eligible children had tobe 7 through 17 years of age and liv-ing with their mothers (or in case ofmarital separation or divorce, livingwith her at least 50% of the time). Al-though participating children did notreceive treatment, those who were re-ceiving treatment elsewhere were notexcluded from the study. If a mother

had more than one child aged 7 through17 years, one child was selected usinga table of random numbers. Study is on-going with cases followed at 3-monthintervals. The STAR*D-Child proto-col was reviewed and approved by theinstitutional review boards at each par-ticipating site. Written informed con-sent was obtained for both mothers andchildren.

Maternal Assessments

Mothers received a comprehensive bat-tery of assessments as part of the adultportion of the study,11-13 which in-cluded baseline demographics, psycho-social and clinical features, and diag-nostic and symptomatic status overtime. Information on race and ethnic-ity was collected as part of the demo-graphic assessment, and mothers se-lected their response from a list ofprovided categories.

The mother’s initial diagnosis was es-tablished by clinical interview and con-firmed using a symptom checklist basedon the Diagnostic and Statistical Manualof Mental Disorders, Fourth Edition(DSM-IV).13 The severity of depres-sive symptoms was estimated using theHRSD.15,16 Maternal remission was de-fined as an HRSD score of 7 or less, andresponse was defined as a 50% or greaterreduction of the baseline HRSD score.Remission and response were also as-sessed by self-report, using the 16-item Quick Inventory of DepressiveSymptomatology-Self Report.18-21 Be-cause the findings between the 2 screen-ing tools were similar, we report onlythe latter (Data for the Quick Inven-tory of Depressive Symptomatology-Self Report is available on request). Cli-nicians assessing maternal remissionwere independent of the clinicians as-sessing child outcomes.

Child Assessments

Children’s psychiatric disorders at base-line and the 3-month evaluation wereestablished by direct interview of moth-ers and children using the KiddieSchedule for Affective Disorders andSchizophrenia—Present and LifetimeVersion,22 a widely used valid and re-

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

1390 JAMA, March 22/29, 2006—Vol 295, No. 12 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

liable diagnostic assessment that gen-erates DSM-IV diagnoses. To reduceparticipants’ burden, we selected sec-tions of the Kiddie Schedule for Disor-ders and Schizophrenia that target dis-orders (affective, anxiety, and disruptivebehavior disorder) known to be highlyprevalent among children of de-pressed parents.23,24

The parent version of the Child Be-havior Checklist (CBCL)25 was used toassess children’s symptoms, and the cli-nician-rated Child Global AssessmentScale (C-GAS)26 was used to assess childglobal functioning. The CBCL was ad-ministered to mothers to assess childsocial functioning in 3 domains—activities, socialization, and school func-tioning—and to assess symptoms as-sociated with behavioral and emotionaldisorders. These symptoms are classi-fied as internalizing (predominantlysymptoms associated with anxiety andmood disorders), and externalizing(predominantly symptoms associatedwith disruptive behavior) disorders.Scores are presented as T scores rang-ing from 0 to 100, for which highercompetency scores indicate superior so-cial functioning (scores �30 are con-sidered in the clinically impaired range),and higher psychopathology scores in-dicate a greater number or severity ofsymptoms (scores �70 are consid-ered indicative of clinical impairment).

Overall child functioning was as-sessed using the C-GAS, by the sameclinical interviewer that assessed theKiddie Schedule for Disorders andSchizophrenia. The scale ranges from0 to 100; scores higher than 90 are in-dicative of superior functioning, andscores lower than 70 indicate clinicalimpairment.

Statistical Analyses

Two main outcome measures were ana-lyzed: change in overall rates of chil-dren’s diagnoses from baseline to 3months and change in CBCL-scoresfrom baseline to 3 months.

Changes in rates of child diagnosesfrom baseline to 3 months as a func-tion of mother’s remission and subse-quently mother’s level of response

were analyzed using a repeated mea-sures analysis with binary responsedata, using generalized estimatingequation (GEE) methods.27 A linearprobability model with an identity linkfunction (rather than a logit-link func-tion) was used to model interactionson the additive scale28 and to model adose-response function using rates(rather than odds) as the outcomemeasure because we considered riskdifferences to be a more relevant mea-sure than odds ratios in our study. Theoutcome measures in these modelswere the rates of diagnoses at each ofthe 2 time points, whereas the inde-pendent variables were time, mother’sremission or response level, and a timeby mother’s remission or responselevel interaction term. Analyses wereadjusted for age and sex of child,severity of maternal baseline symp-toms, annual household income,mother’s treatment setting (primary vspsychiatric outpatient care), and treat-ment status of child during 3-monthfollow-up. Household income wasselected a priori as the primary markerfor socioeconomic status; however,the results were unchanged when themother’s educational level or whenboth income and educational levelwere included in the model. The inter-action term of time � mother’s remis-sion status was included to formallytest whether changes in rates of child-hood diagnoses differed significantlybetween mothers whose depressionremitted and mothers whose depres-sion remained. In the analysis ofchild’s diagnosis as a function ofmother’s response level, the interac-tion term of time � mother’s responselevel was included to formally testwhether the change in rate of diagno-sis over time varied with the mother’sresponse level.

The relationship between maternalremission and changes in child’s out-come at 3 months, for children with andwithout diagnoses, was analyzed us-ing separate logistic regression analy-ses with the outcome measure being re-mission for children with a baselinediagnosis and incidence or relapse for

children without a baseline diagnosis,respectively. Analyses were adjusted forage and sex of the child, as well as thecontrol variables listed above. Trendsin rates of child diagnoses by mother’sresponse level in children with a base-line diagnosis and in rates of inci-dence or relapse in children without abaseline diagnoses were examined sepa-rately using the Cochran-Armitage testfor trend.29 Low event rates precludedfitting regression models adjusting forpotential confounders, such as age andsex of child, using generalized linearmodels with an identity-link function,to estimate parameters for adjustedtrends.

Analyses of CBCL change scores wereconducted using linear regressionanalysis. Specifically, the relation ofmother’s remission status to change inCBCL score was modeled so that thechange score was treated as the depen-dent variable, with mother’s remis-sion status as a dichotomous indepen-dent variable, and with the baselinevalue of the CBCL score and the moth-er’s baseline HRDS as covariates. Wealso included the child’s age and sex inthe model to control for potential con-founding. Change scores rather than the3-month score was used as the out-come measure for ease of interpreta-tion. It has been shown that infer-ences resulting from this analysis arevirtually identical no matter which ofthese outcome measures is used.30 In ad-dition to the covariates previouslynoted, the regression analysis was re-peated to include annual household in-come, mother’s treatment setting (pri-mary vs psychiatric outpatient care),and treatment status of child during the3-month follow-up period in order toinvestigate the further potential con-founding effects of these variables. Allstatistical analyses were conducted us-ing SAS statistical software version 9.0(SAS Institute Inc, Cary, NC). P�.05was considered statistically significant.

Six of the 114 mothers who re-ceived both baseline and 3-monthfollow-up assessments were missingfollow-up HRSD scores. Scores forthese mothers were imputed from the

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, March 22/29, 2006—Vol 295, No. 12 1391

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

Quick Inventory of Depressive Symp-tomatology-Self Report, using an item-response theory analysis of the rela-tion between the HRSD and the QuickInventory of Depressive Symptomatol-ogy-Self Report scales, as used by theSTAR*D study.19,31 Sensitivity analy-ses confirmed that the findings did notvary based on inclusion or exclusion ofthese 6 mothers.

RESULTSEight hundred twenty-four women aged25 through 60 years were recruited at8 primary care and 11 psychiatric out-patient clinics across the 7 participat-

ing regional centers (FIGURE 1). Eighthundred eight (98%) of 824 womenwere screened to ascertain whether theyhad at least 1 child aged 7 through 17years; only 177 (22%) of 808 had chil-dren in that age range; 151 (85%) of 177eligible mother-child pairs consentedto participate in the child study. Onehundred fourteen (75%) of 151 of themother-child pairs who received base-line assessments remained in the studyat the time of their child’s 3-month as-sessment. Mothers who dropped outwere not significantly different at base-line from mothers who remained in thestudy on demographics, or clinical char-acteristics, except that mother-childpairs were more likely to drop out ofthe study if the participating child wasmale than if the child was female (70%vs 30%, P=.01). No differences werefound among their children on base-line diagnoses, current or lifetime, se-verity of internalizing and externaliz-ing symptoms, or functioning (dataavailable on request).

Remission of Maternal Depression

Of the mothers who received follow-upassessments, 38/114 (33%) met remis-sion criteria before the 3-month fol-low-up assessment. The average (SD)time to remission was 55 (40) days. Theoverall response rate was 47% (54/114). Of the 38 mothers whose depres-sion remitted, 35 (92%) did so while tak-ing citalopram only (level 1 treatment).Twomothers’medicationswere switchedto extended release venlafaxine-XR, andone mother received a combination regi-men of citalopram and bupropion, priorto remission. TABLE 1 and TABLE 2 sum-marize the baseline characteristics ofmothers and their children. Motherswhose depression remained were finan-cially poorer, more often receiving pub-lic assistance, and less likely to hold a col-lege degree. They also had more severebaseline depression and comorbid anxi-ety but did not differ on age of onset ornumber of major depressive episodes.There were no significant differences onany of the child demographics or base-line clinical characteristics by mothers’remission status.

Figure 1. Study Flow Diagram

151 Mother-Child Pairs Consented and ReceivedBaseline Assessment

824 Women Aged 25 to 60 y Recruited at 7Sites Participating in STAR∗ D-Child

37 Pairs Did Not Complete3-mo Assessment

114 Pairs Completed Child’s 3-mo Assessment

177 Had Children Aged 7 to 17 y

808 Screened

Table 1. Baseline Characteristics of Depressed Mothers by Remission*

Maternal CharacteristicsRemission

(n = 38)No Remission

(n = 76)P

Value‡

Race, No. (%)African American 11 (29) 34 (45)

White 16 (42) 29 (39).17

Hispanic 7 (18) 11 (14)

Other 4 (11) 2 (6)

Education, No. (%)�High school 3 (8) 11 (14)

High school �college 18 (48) 53 (70) .003

�College 17 (44) 12 (16)

Annual household income, No. (%), $�15 000 4 (11) 22 (31)

15 000-40 000 13 (36) 31 (43) .01

�40 000 19 (53) 19 (26)

Receiving public assistance 5 (13) 27 (36) .01

Marital status, No. (%)Currently married 5 (13) 12 (16)

Separated, divorced, or widowed 31 (82) 58 (76) .79

Never married 2 (5) 6 (8)

Age, mean (SD), y 38.6 (7.1) 37.8 (6.4) .59

Clinical history, mean (SD)Baseline Hamilton Rating Scale

for Depression score22.8 (4.1) 25.3 (5.3) .01

Age at first onset, y 23 (12.3) 19 (9.3) .13

No. of major depressive episodes 4 (9.0) 9 (17.8) .19

Comorbid Axis I diagnoses, No. (%)†Generalized anxiety disorder 5 (13) 23 (31) .04

Obsessive-compulsive disorder 3 (8) 19 (25) .03

Panic disorder 1 (2) 10 (13) .09§

Posttraumatic stress disorder 4 (10) 16 (21) .17

Social phobia 17 (45) 34 (45) .95

Substance use disorder 3 (8) 10 (13) .41

Treatment settingPrimary care 19 (50) 50 (65)

.10Psychiatric 19 (50) 26 (34)

*The numbers in each category may vary due to missing data.†Diagnoses based on the Psychiatric Diagnostic Screening Questionnaire.32

‡P Values are based on �2 tests for categorical variables and t tests for continuous variables.§Fisher exact test.

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

1392 JAMA, March 22/29, 2006—Vol 295, No. 12 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

Relation of Maternal Remissionto Changes in Rates ofChild DiagnosesAs shown in TABLE 3, there was an over-all 11% decrease in rates of diagnoses(from 35% [12/34] to 24% [8/34]) inchildren of remitted mothers vs an 8%increase (from 35% [25/71] to 43% [30/71]) in children of mothers with con-tinuing depression. After controlling forthe child’s age and sex, the change inrates was statistically significant for chil-dren of mothers whose depression re-mitted (12.3% decrease; 95% confi-dence interval [CI], 0.08%-23.8%;P = .03) but not for children whosemother’s depression remained (6.5% in-crease; 95% CI, −2.5% to 15.4%;P=.15). Formal tests to determine if theabove rates of changes in children’s di-agnoses varied with mothers’ remis-sion status were statistically signifi-cant (P=.02), and remained significantafter further adjusting for maternal de-pression severity at baseline, maternaltreatment setting, annual household in-come, and child treatment status dur-ing the 3-month follow-up interval(P=.01).

The relation of maternal remissionto child outcomes was also examinedseparately among children with andwithout a diagnosis at baseline. Thirty-seven offspring had psychiatric diag-noses at baseline. Of those whose moth-er’s depression remitted, one third

(4/12) of the children’s own diagnoseshad remitted, whereas only 12% (3/25) of the children of women whose de-pression remained lost their diagno-sis, although this difference was notstatistically significant (P=.21; Table 3).

Sixty-eight children had no psychi-atric disorder at baseline. Of these chil-dren, all remained free of psychiatricdisorders at the 3-month follow-up ifthe maternal depression remitted,whereas 17% (8/46) of children ofmothers who remained depressed had

an onset or relapse over this period(P=.05; Table 3). The higher rates ofonset at 3 months among children ofmothers whose depression remainedwere not associated with variation inlifetime rates of child disorders (P=.85).

FIGURE 2 shows the pattern of changeacross the 3 months, for specific childdisorders as well as for any disorder, bymaternal remission status. Childrenwhose mothers’ depression had remit-ted showed a decrease in the rates ofdepressive (18% [6/34] to 9% [3/34])

Table 2. Children by Maternal Depression Remission*

Child CharacteristicsRemission

(n = 38)No Remission

(n = 76)P

Value‡

Sex, No. (%)Female 17 (45) 44 (58)

.14Male 21 (55) 32 (42)

Age, mean (SD), y 12.2 (2.6) 11.4 (2.7) .19

Clinical history, No. (%)Lifetime DSM-IV diagnoses

Any depressive disorder 9 (24) 15 (20) .61

Any anxiety disorder 6 (18) 17 (23) .55

Any disruptive behavior disorder 10 (30) 19 (26) .64

Current DSM-IV diagnoses

Any depressive disorder 7 (18) 5 (7) .07

Any anxiety disorder 5 (13) 13 (17) .61

Any disruptive behavior disorder 7 (19) 15 (20) .89

Past treatment for emotional problems 12 (35) 16 (21) .13

Treatment for emotional problemsduring STAR*D-Child

2 (6) 10 (13) .23§

Global functioning at baseline, mean (SD)† 68.7 (13) 68.5 (11) .95Abbreviation: DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition.*Numbers may vary in each category due to missing data.†Based on the Child-Global Assessment Scale. Higher scores indicate greater global functioning.‡P values are based on �2 tests for categorical variables and t tests for continuous variables.§Fisher exact test.

Table 3. Relation Between Maternal Remission Status and Change in Any Child Diagnoses, Baseline to 3 Months*

MaternalRemission Status

No. (%)Adjusted Rate

Difference, % (95% CI)†P

Value‡Baseline 3 Months

Rates of Current Child Diagnoses at Baseline and at 3 Months by Maternal Remission Status

Remitted (n = 34) 12 (35) 8 (24) −12.3 (0.08 to 23.8) .02

Unremitted (n = 71) 25 (35) 30 (43) �6.5 (−2.5 to 15.4)

Maternal Remission Status by Child’s Diagnostic Status at 3 Months in Children With a Baseline Diagnosis

Present Absent

Remitted (n = 12) 8 (67) 4 (33) .21§

Unremitted (n = 25) 22 (88) 3 (12)

Maternal Remission Status by Child’s Diagnostic Status at 3 Months, in Children Without a Baseline Diagnosis

Remitted (n = 22) 0 22 (100) .05 �

Unremitted (n = 46) 8 (17) 38 (83)*Numbers may vary in each category due to missing data.†Adjusting for child age and sex.‡P value for interaction term testing for equality of rate differences by maternal remission§By �2 test.�P value based on Fisher exact test.

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, March 22/29, 2006—Vol 295, No. 12 1393

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

and disruptive behavior disorders (18%[6/34] to 12% [4/34]) but with nochange in anxiety disorders (4/34). Incontrast, among children of motherswho did not remit, there was an in-crease in the rates of depressive (7%[5/71] to 11% [8/71]), anxiety (17%[13/71] to 25% [18/71]) and disrup-tive behavior (20% [15/71] to 24% [17/71]) disorders. Although the smallsample size precluded statistical analy-sis on each disorder individually, thechanges in rates of diagnoses betweenchildren of mothers whose depressiondid or did not remitwere significantly

different for the class of internalizing(including depressive or anxiety dis-orders; P=.03) but not externalizing(disruptive behavior) disorders.

To ascertain whether maternaldepression status biased reports ofchildren’s psychopathology, we com-pared Kiddie Schedule for Disordersand Schizophrenia assessments ofeach mother and child (data availableon request). There were no clinicallyor statistically significant differencesbetween maternal and child reports ofdepressive or anxiety symptoms. Inthe case of disruptive behavior disor-

ders, mothers tended to report moresymptoms than their children, whichis consistent with previous findingsthat parents are more likely to reportbehavior and conduct problems intheir children than the children them-selves.33,34 The difference betweenmaternal and child reports did notvary significantly by maternal remis-sion status.

Relation of Maternal Remissionto Change in Child Symptoms

Changes in severity of children’s inter-nalizing and externalizing symptomsover the 3-month period were also ex-amined using changes in CBCL scores.After controlling for the child’s age andsex and adjusting for baseline severityof child and maternal symptoms, therewas a significantly larger decrease in in-ternalizing (adjusted mean score dif-ference, 8.6; P�.001), externalizing(6.6; P=.004), and total (8.7; P� .001)symptoms among children of motherswho had a remission from major de-pressive disorder over the 3-month pe-riod than among children of motherswhose major depressive disorder didnot remit (TABLE 4). The above asso-ciation between maternal remission andchild symptoms remained significant af-ter further adjusting for potential con-founders including maternal socioeco-nomic status, maternal treatmentsetting, occurrence of stressful lifeevents within the assessment interval,and the presence of a father in thehousehold.

There were no significant changes inchild functioning over the same pe-riod by maternal remission status, re-gardless of whether child functioningwas assessed using the maternal-ratedCBCL or the clinician-rated C-GAS(data available on request).

Relation of Level of MaternalResponse to Change in ChildDiagnoses and Symptoms

To quantify the magnitude of mater-nal improvement necessary to detect anappreciable improvement in the child,changes in child symptoms and diag-noses over the 3-month period were as-

Figure 2. Relation Between Maternal Remission Status and Change in Child’s SpecificDiagnoses (Baseline to 3 Months)

50

30

40

20

10

0

% o

f Chi

ldre

n W

ith A

ny D

iagn

osis

Children of Mothers With Remission(n = 38)

AnyDisorder

DepressiveDisorders

AnxietyDisorders

DisruptiveBehaviorDisorders

Baseline

3 Months

Children of Mothers Without Remission(n = 76)

AnyDisorder

DepressiveDisorders

AnxietyDisorders

DisruptiveBehaviorDisorders

Depressive disorders include major depressive disorder, dysthymia, depressive disorder not otherwisespecified, adjustment disorder with depressed mood, and with mixed anxiety and depressed mood. Anxietydisorders include specific phobia, separation anxiety disorder, social phobia, generalized anxiety disorder; obsessive-compulsive disorder; and anxiety disorder not otherwise specified. Disruptive behavior disorders include at-tention deficit hyperactivity disorder, oppositional defiant disorder, and conduct disorder.

Table 4. Relation Between Maternal Remission Status and Change in Child Symptoms(Baseline to 3 Months)

Child Outcomes Baseline

Change in Mean Score (Baseline to 3 Months)by Maternal Remission Status

Remitted† Unremitted‡Adjusted MeanDifference (CI)§

PValue

CBCL psychopathologyT scores*

Internalizing 56.5 (11) −11.6 (11) −0.5 (12) −8.6 (−4.8 to −12.4) �.001

Externalizing 55.5 (12) −9.9 (9) −0.4 (11) −6.6 (−2.4 to −10.8) .004

Total 56.1 (11) −12.5 (11) −0.75 (13) −8.7 (−4.2 to −13.2) �.001Abbreviation: CBCL, Child Behavior Checklist.*The scoring range is from 0 to 100. Lower scores indicate superior functioning. Scores higher than 70 indicate clinical

impairment.†Hamilton Rating Scale for Depression, a 17-item questionnaire, score of 7 or lower closest evaluation prior to child’s

3-month assessment.‡Hamilton Rating Scale for Depression score higher than 7 at closest evaluation prior to child’s 3-month assessment.§Multiple linear regression, with CBCL score at the 3-month assessment as the outcome variable, maternal remission

as a dichotomous independent variable, the child’s baseline symptom severity (CBCL score) and maternal baselineHamilton Rating Scale for Depression score, as covariates, and the child’s age and sex as control variables.

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

1394 JAMA, March 22/29, 2006—Vol 295, No. 12 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

sessed against the percentage change inmaternal depressive symptoms. Mater-nal response was classified in 1 of 5 lev-els, based on the percent reduction inbaseline HRSD scores: �0% (12 moth-ers), 0% to 24% (16 mothers), 25% to49% (33 mothers), and 50% to 74% (27mothers), 75% to 100% (26 mothers).As shown in FIGURE 3A, the change inrates of child diagnoses over the 3months was inversely related to themagnitude of the mother’s responselevel. However, at least 50% maternalresponse was required to discern im-provement in the children. Specifi-cally, children of mothers with 50% orgreater reduction in depression sever-ity had lower overall rates of diag-noses at the 3-month assessment (4%if maternal response was between 50%and 75%; 9% if maternal response ex-ceeded 75%). In contrast, reduction inmaternal depression less than 50% wasassociated with an increase in the ratesof child diagnoses (13% if maternal re-sponse was �25%, and 18% if themothers got worse). After controllingfor the child’s age and sex, a signifi-cant linear relation was found be-tween maternal response level andchange in rates of child diagnoses(P=.04). When examined separately inchildren with and without a diagnosisat baseline, the level of maternal re-

sponse was associated with onset or re-lapse of a child disorder (P=.002) butnot with the child’s remission.

Similar results were obtained on theChild Behavior Checklist for internal-izing (Figure 3B) and externalizing(Figure 3C) symptoms. Greater mater-nal response was associated with agreater decrease in both internalizing(adjusted mean difference, −3.4;P�.001) and externalizing (−3.7;P�.001) symptoms in the children,when tested in a linear regression modelcontrolling for child age, sex, and se-verity of symptoms at baseline, as wellas maternal baseline HRSD and house-hold income.

Effects of Treatment in Children

We did not exclude children receivingtreatment previously or after the base-line assessment. Only 28 children had re-ceived treatment for an emotional prob-lem prior their mothers’ participation inthe trial. The children who improved bytheir 3-month evaluation did not differsignificantly from those who did not onlifetime diagnoses or treatment history,indicating that differential rates of im-provement were unlikely attributable tolifetime course of illness.

We also examined whether the 12children who received outpatient treat-ment during the 3 months differed from

those who did not. There were no dif-ferences in severity of internalizing orexternalizing symptoms at baseline be-tween children who received treat-ment during the 3 months and thosewho did not. Children receiving treat-ment compared with those who did nothad a greater number of DSM-IV diag-noses at baseline (P=.01), most strik-ingly, disruptive behavior disorders(58% vs 15%, P=.002). There was noassociation, however, between childtreatment and maternal remission(P=.70), and the relation of maternalremission to child improvement wassustained after controlling for whetherthe child received any treatment dur-ing the 3-month follow-up interval(data available on request).

COMMENTMany of the children of depressedmothers coming to STAR*D-Child wereacutely symptomatic.14 Over a third hada current psychiatric disorder includ-ing anxiety (16%), depressive (10%),or disruptive behavior disorders (22%);almost half had a past psychiatric dis-order. These high rates are consistentwith findings from numerous studiesof children of depressed parents.35-37

The findings reported herein sug-gest that remission of maternal depres-sion over 3 months is statistically sig-

Figure 3. Relation Between Maternal Response Level and Change in Child Diagnoses and Symptoms

20

–15

0

–5

–10

5

10

15

–20

Maternal Response Level, %

Per

cent

age

Cha

nge

in R

ates

of D

iagn

osis

Change in Child DiagnosesA

<0(n = 12)

0-24(n = 16)

25-49(n = 33)

50-74(n = 27)

≥75(n = 26)

10

–15

0

–5

–10

5

–20

Maternal Response Level, %

Cha

nge

in S

ever

ity o

fIn

tern

aliz

ing

Sym

ptom

s, M

ean

Change in Child InternalizingSymptoms

B

<0(n = 12)

0-24(n = 16)

25-49(n = 33)

50-74(n = 27)

≥75(n = 26)

Cha

nge

in S

ever

ity o

fE

xter

naliz

ing

Sym

ptom

s, M

ean

10

–15

0

–5

–10

5

–20

Maternal Response Level, %

Change in Child ExternalizingSymptoms

C

<0(n = 12)

0-24(n = 16)

25-49(n = 33)

50-74(n = 27)

≥75(n = 26)

Diagnoses were based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Error bars represent 95% confidence intervals. Response levelsreflect the percent reduction of maternal HAM-D between the baseline and 3-month evaluations.

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, March 22/29, 2006—Vol 295, No. 12 1395

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

nificantly associated with reduction inchildren’s current symptoms and diag-noses after controlling for the child’s ageand sex, baseline symptoms, socioeco-nomic status (annual household in-come), as well as severity of maternaldepression at baseline, mother’s treat-ment setting, and the child’s treat-ment status over the 3-month follow-up. Mothers in this study were treatedwith antidepressant medication, but itis likely that the findings reportedherein would apply to any effectivetreatment of depression.

At least 50% maternal response totreatment was required to detect any re-duction in child diagnoses and symp-toms, and children of mothers who re-sponded less than 50% showed anincrease in diagnoses at 3 months. Thisfinding is consistent with the 50% re-sponse threshold routinely used in thedepression literature including theSTAR*D adult study.31 There were nosignificant changes in children’s func-tioning at 3 months, but this lag is con-sistent with previous literature suggest-ing that a reduction in psychiatricsymptoms often precedes improve-ment in functioning.38,39

Because our design was not experi-mental, we cannot demonstrate cau-sality. Furthermore, children’s improve-ment may have had a positive impacton mothers (reverse causation). How-ever, because the duration of the cur-rent maternal depressive episode atbaseline was correlated with the num-ber of children’s internalizing andexternalizing symptoms at baseline(Cynthia Ewell-Foster, PhD, et al, un-published data, December 2005), andthe extent of children’s improvementfollowing maternal remission de-pended on the magnitude of improve-ment in their mothers, reverse causa-tion is not likely to fully account for theassociation between maternal remis-sion and child improvement. It is morelikely that maternal remission trig-gered improvement or prevented de-terioration in the children and that thischange in the children had further im-pact on the mothers. Thus, maternal re-mission seems to have initiated a vir-

tuous cycle, wherein mothers andchildren positively influenced eachother.

These findings need to be consid-ered within the context of the remis-sion rate, the low rate of women withchildren in the overall STAR*D study,and the study limitations. The remis-sion rate at 3 months in this sample was33%, and average time to remission was55 days. If 50% response is consid-ered, these rates increased to 47%.These rates are similar to those re-ported in the overall STAR*D study31

and are higher than those found in ef-ficacy studies among patients withchronic depression.40 These remissionrates should be viewed against the back-ground of mothers participating in thisstudy. They were moderately to se-verely depressed at baseline, with an av-erage of 6 previous episodes and meanonset at age 20 years. Although morethan a third of their children had a cur-rent psychiatric disorder at baseline andmore than half had a lifetime history,some improvement was observed in thechildren in a relatively short time, ie,3 months, and this occurred, in mostcases, without the children receiving di-rect treatment. Even more interestingwas the possible preventive impact ofthe intervention. Whereas children ofunremitting mothers deteriorated (ie,acquired more symptoms and diag-noses) during the 3-month follow-upinterval, none of the children of moth-ers whose depression had remitted hadany onset or recurrence of a psychiat-ric disorder.

Only 22% of participating womenwho were aged 25 through 60 years hadchildren aged 7 to 17 years. The lowproportion of mothers among womenseeking treatment suggests that de-pressed mothers compared with womenwithout children might be less likely toseek and come for treatment. This find-ing is consistent with previous reportsthat depressed low-income women (alarge proportion of the sample in thisstudy) do not use community care avail-able to them, even if it is free.10,41 With-out outreach, child care, transporta-tion, and flexible schedules, these

women are not likely to receive appro-priate treatment for their depression.

This study has limitations. The useof a single antidepressant in an open-trial design without a placebo controldid not allow us to rule out that ma-ternal remission was due to nonspe-cific treatment effects or whether therelation of maternal remission to chil-dren’s outcomes may have been differ-ent if another medication or psycho-therapy had been used. A placebowould have provided information onthe specificity of the treatment butwould have limited generalizability be-cause more severely ill patients mightnot have participated.

Child assessors knew that partici-pating mothers were depressed. Thus,they were not blind to their initial di-agnosis. However, assessors of childoutcomes in the present study were un-aware of maternal responses to treat-ment, and they were not involved in themother’s treatment. As mentionedabove, we cannot rule out the possibil-ity that reverse causation (ie, changesin children’s psychopathology leadingto reductions in maternal depressivesymptoms) contributed to the associa-tion between maternal depression andchild remission. It also is possible thatsome third variable not examined in thepresent study contributed to clinicalchanges in both the mother and thechild (eg, change in levels of stress, fi-nancial strain). Although analyses wereadjusted for the presence of a father inthe house, we were unable to accountfor the impact of the fathers’ psychiat-ric state on their children because fa-thers were not directly assessed in thestudy.

Finally, maternal bias in reportingchildren’s symptoms may have influ-enced the CBCL data, which were basedsolely on maternal reports. Three pointsare worth noting with regard to CBCLscores. First, the duration of the cur-rent maternal depressive episode, butnot the severity of this episode, was as-sociated with the CBCL scores at base-line. Were depression-related bias to in-fluence these scores, one would expectan association between maternal de-

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

1396 JAMA, March 22/29, 2006—Vol 295, No. 12 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

pression severity and maternal re-ports. Second, findings obtained us-ing CBCL scores were similar to thoseobtained using the Kiddie Schedule forDisorders and Schizophrenia, which areunlikely to be biased by maternal per-ception, as separate examinations ofKiddie Schedule for Disorders andSchizophrenia symptoms reported bymother and child revealed similar ratesof depressive and anxiety symptoms. Fi-nally, the lack of functional improve-ment in the children on the clinician-determined C-GAS were mirrored by asimilar lack of improvement reportedby the mother on the functional do-mains of the CBCL.

To our knowledge, this is the firstpublished study to document prospec-tively the relation between remission ofa mother’s depression and her child’sclinical state. These findings are in-triguing because they suggest that anenvironmental influence (ie, the im-pact of maternal depression remis-sion) had a measurable impact on thechild’s psychopathology. Recent stud-ies show that the environmental mayincrease the onset of depressive disor-ders in genetically vulnerable adults andchildren.42,43 Our studies suggest thata reduction in stress associated with ma-ternal remission may reverse the long-standing symptoms in children who arelikely to be genetically vulnerable, al-though we have not genotyped the chil-dren in the study.

From a clinical vantage point, ourfindings suggest that vigorous treat-ment of depressed mothers to achieveremission is associated with positiveoutcomes in their children as well,whereas failure to treat depressed moth-ers may increase the burden of illnessin their children. At a time when thereare many questions about the appro-priate and safe treatment of psychiat-ric disorders in children, these find-ings suggest that it is important toprovide vigorous treatment to moth-ers if they are depressed.

Author Affiliations: Department of Psychiatry, Co-lumbia University and the New York State PsychiatricInstitute, New York (Drs Weissman, Pilowsky, andWickramaratne) and Columbia University, New York(Dr Talati); Department of Psychiatry, University of

Texas Southwestern Medical Center, Dallas (DrsHughes, Rush, and Trivedi); Department of Psychol-ogy and Human Development, Vanderbilt Univer-sity, Nashville, Tenn (Dr Garber); Department of Psy-chiatry, University of North Carolina, Chapel Hill (DrMalloy); Department of Psychiatry, University of Michi-gan, Ann Arbor (Dr King); Department of Psychiatry,University of California at San Diego (Dr Cerda); De-partment of Psychiatry, Virginia Commonwealth Uni-versity, Richmond (Dr Bela Sood); Department of Psy-chiatry, Harvard University, Boston, Mass (Drs Alpertand Fava); and Department of Epidemiology, Univer-sity of Pittsburgh, Pittsburgh, Pa (Dr Wisniewski).Author Contributions: Dr Weissman had full accessto all of the data in the study and takes responsibilityfor the integrity of the data and the accuracy of thedata analysis.Study concept and design: Weissman, Pilowsky,Wickramaratne, Wisniewski, Fava, Garber, Trivedi,Rush.Acquisition of data: Weissman, Pilowsky, Hughes,Garber, Malloy, King, Cerda, Sood, Alpert, Trivedi.Analysis and interpretation of data: Weissman,Wickramaratne, Talati, Fava, Rush.Drafting of the manuscript: Weissman, Talati, Rush.Critical revision of the manuscript for important in-tellectual content: Weissman, Pilowsky, Wickrama-ratne, Talati, Wisniewski, Fava, Hughes, Garber,Malloy, King, Cerda, Sood, Alpert, Trivedi, Rush.Statistical analysis: Wickramaratne, Talati, Fava.Obtained funding: Weissman.Administrative, technical, or material support: Pilowsky,Hughes, Malloy, King, Trivedi, Rush.Study supervision: Weissman, Pilowsky, Wisniewski,Hughes, Garber, King, Cerda, Alpert, Trivedi, Rush.Financial Disclosures: Dr Weissman has received grantsupport from Eli Lilly and GlaxoSmithKline and hasserved as a scientific advisor to Eli Lilly. Dr Fava hasreceived research support from Abbott Laboratories,Lichtwer Pharma GmbH, Lorex Pharmaceuticals; re-ceived honoraria from Bayer AG, Compellis, JanssenPharmaceutica, Knoll Pharmaceutical Co, Lundbeck,Dov Pharmaceuticals, Biovail Pharmaceuticals Inc,BrainCells, Grunenthal GmBH, Sepracor, and Somer-set Pharmaceuticals; and received both research sup-port and honoraria from Aspect Medical Systems, As-traZeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly &Co, Forest Pharmaceutials, GlaxoSmithKline, J&J Phar-maceuticals, Novartis, Organon Inc, Pharmavite, PfizerInc, Roche, Sanofi/Synthelabo, Solvay Pharmaceuti-cals, and Wyeth-Ayerst Laboratories. Dr Hughes hasreceived support from Healthcare Technology Sys-tems Inc and GlaxoSmithKline. Dr Sood serves on thespeaker’s bureau of AstraZeneca. Dr Trivedi has servedas a consultant or has advised for Bristol-Myers Squibb,Eli Lilly, Forest Pharmaceuticals, Wyeth-Ayerst labo-ratories, Sepracor, Johnson & Johnson, Cyberonics, andPfizer; received lecture fees from Bristol-Myers Squibb,Eli Lilly, Forest Pharmaceuticals, Wyeth-Ayerst Labo-ratories, and Cyberonics; received grant support fromthe National Institutes of Health, National Institute ofMental Health, Bristol-Myers Squibb, Cephalon Inc,Predix, Pfizer/Parexel, and Corcept Therapeutics Inc.Dr Rush has received speaking honoraria from and pro-vided scientific consultation to Forest Laboratories.None of the other authors reported disclosures.Funding/Support: This study was supported by grantsR01MH063852 (Dr Weissman, principal investigator)and N01 MH90003 (Dr Rush, principal investigator)from the National Institute of Mental Health. Dr Gar-ber was supported in part by an Independent ScientistAward K02 MH66249 from the National Institute ofMental Health. We appreciate the support of ForestLaboratories for providing citalopram at no cost.Role of the Sponsor: Funding sources played no rolein the design and conduct of the study; collection, man-agement, analysis, and interpretation of the data; orpreparation, review, or approval of the manuscript.

STAR*D Regional Centers: Massachusetts GeneralHospital, Boston, Andrew Nierenberg, MD; Univer-sity of California, San Diego, Sid Zisook, MD; Univer-sity of Michigan, Ann Arbor, Elizabeth Young, MD;University of North Carolina, Chapel Hill, BradleyGaynes, MD, MPH; University of Texas Southwest-ern Medical Center, Dallas, Mustafa Husain, MD;Vanderbilt University, Nashville, Tenn, Steven Hol-lon, PhD; and Virginia Commonwealth University, Rich-mond, Susan Kornstein, MD.Disclaimer: The content of this publication does notnecessarily reflect the views or policies of the Depart-ment of Health and Human Services nor does men-tion of trade names, commercial products, or orga-nizations imply endorsement by the US government.Acknowledgment: We thank Huaiyu Yang, MD, andMichele Candrian, MS, Boston, Mass; Rachel N. Swan,MS, Mi Wu, BS, and Liz Ezell, MA, Nashville, Tenn;Julie Linker, PhD, Richmond, Va; Graham J. Emslie,MD, Jeanne Rintelmann, BA, and Laurie Macleod, RN,CCRC, Dallas, Tex; Elizabeth Santana, marriage andfamily therapy intern, San Diego; Ann Brewster, PhD,Joanne DeVeaugh-Geiss, MA, LPA, and Renee White,MSW, Chapel Hill, NC; Melissa Webster, MSW, andBrooke Tracey, BS, Ann Arbor, Mich; and Diane War-den, PhD, and Kathy Shores-Wilson, PhD, at the Na-tional Coordinating Center.

REFERENCES

1. Weissman MM, Wickramaratne P, Nomura Y, et al.Families at high and low risk for depression: a3-generation study. Arch Gen Psychiatry. 2005;62:29-36.2. Weissman MM, Wickramaratne P, Nomura Y,Warner M, Pilowsky D, Verdeli H. Offspring of de-pressed parents: 20 years later. Am J Psychiatry. Inpress.3. Weller EB, Tucker S, Weller RA. The selective sero-tonin reuptake inhibitors controversy in the treat-ment of depression in children. Curr Psychiatry Rep.2005;7:87-90.4. Kendler KS, Gardner CO, Prescott CA. Toward acomprehensive developmental model for major de-pression in women. Am J Psychiatry. 2002;159:1133-1145.5. Levinson DF, Zubenko GS, Crowe RR, et al. Ge-netics of recurrent early-onset depression (GenRED):design and preliminary clinical characteristics ofa repository sample for genetic linkage studies. Am JMed Genet B Neuropsychiatr Genet. 2003;119:118-130.6. Holmans P, Zubenko GS, Crowe RR, et al. Ge-nomewide significant linkage to recurrent, early-onset major depressive disorder on chromosome 15q.Am J Hum Genet. 2004;74:1154-1167.7. Cicchetti D, Rogosch FA, Toth SL. Maternal de-pressive disorder and contextual risk: contributions tothe development of attachment insecurity and be-havior problems in toddlerhood. Dev Psychopathol.1998;10:283-300.8. Lyons-Ruth K, Bronfman E, Parsons E. Atypical at-tachment in infancy and early childhood among chil-dren at developmental risk, IV: maternal frightened,frightening, or atypical behavior and disorganized in-fant attachment patterns. Monogr Soc Res Child Dev.1999;64:67-96.9. Beardslee WR, Gladstone TR, Wright EJ, CooperAB. A family-based approach to the prevention of de-pressive symptoms in children at risk: evidenceof parental and child change. Pediatrics. 2003;112:e119-e131.10. Verdeli H, Ferro T, Wickramaratne P, GreenwaldS, Blanco C, Weissman MM. Treatment of depressedmothers of depressed children: pilot study of feasibility.Depress Anxiety. 2004;19:51-58.11. Fava M, Rush AJ, Trivedi MH, et al. Backgroundand rationale for the sequenced treatment alterna-

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, March 22/29, 2006—Vol 295, No. 12 1397

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

tives to relieve depression (STAR*D) study. PsychiatrClin North Am. 2003;26:457-494.12. Rush AJ, Trivedi M, Fava M. Depression, IV:STAR*D treatment trial for depression. Am J Psychiatry.2003;160:237.13. Rush AJ, Fava M, Wisniewski SR, et al. Se-quenced treatment alternatives to relieve depression(STAR*D): rationale and design. Control Clin Trials.2004;25:119-142.14. Pilowsky DJ, Wickramaratne P, Rush AJ, et al. Chil-dren of currently depressed mothers: a STAR*AncillaryStudy. J Clin Psychiatry. 2006;67:126-136.15. Hamilton M. A rating scale for depression. J Neu-rol Neurosurg Psychiatry. 1960;23:56-62.16. Hamilton M. Development of a rating scale forprimary depressive illness. Br J Soc Clin Psychol. 1967;6:278-296.17. Lavori PW, Rush AJ, Wisniewski SR, et al. Strength-ening clinical effectiveness trials: equipoise-stratifiedrandomization. Biol Psychiatry. 2001;50:792-801.18. Rush AJ, Gullion CM, Basco MR, Jarrett RB, TrivediMH. The Inventory of Depressive Symptomatology(IDS): psychometric properties. Psychol Med. 1996;26:477-486.19. Rush AJ, Trivedi MH, Ibrahim HM. The 16-itemQuick Inventory of Depressive Symptomatology(QUIDS), clinician rantings (QIDS-C), and self-report(QUIDS-SR): psychometric evaluation patients withchronic major depression [published correction ap-pears in Biol Psychiarty 2003;54:585]. Biol Psychiatry.2003;54:573-583.20. Rush AJ, Bernstein IH, Trivedi MH, et al. Anevaluation of the Quick Inventory of DepressiveSymptomatology and the Hamilton Rating Scalefor Depression: a STAR*D report. Biol Psychiatry. Inpress.21. Trivedi MH, Rush AJ, Ibrahim HM, et al. The In-ventory of Depressive Symptomatology, Clinician Rat-ing (IDS-C) and Self-Report (IDS-SR), and the QuickInventory of Depressive Symptomatology, ClinicianRating (QIDS-C) and Self-Report (QIDS-SR) in pub-lic sector patients with mood disorders: a psychomet-ric evaluation. Psychol Med. 2004;34:73-82.

22. Kaufman J, Birmaher B, Brent D, et al. Schedulefor Affective Disorders and Schizophrenia for SchoolAge Children Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. J Am Acad ChildAdolesc Psychiatry. 1997;36:980-988.23. Hammen C, Burge D, Burney E, Adrian C. Lon-gitudinal study of diagnoses in children of women withunipolar and bipolar affective disorder. Arch GenPsychiatry. 1990;47:1112-1117.24. Weissman MM, Warner V, Wickramaratne P,Moreau D, Olfson M. Offspring of depressed par-ents: 10 years later. Arch Gen Psychiatry. 1997;54:932-940.25. Achenbach TM. Manual for the Child BehaviorChecklist/4-18 and 1991 Profile. Burlington: Univer-sity of Vermont Dept of Psychiatry; 1991.26. Shaffer D, Gould MS, Brasic J, et al. A children’sglobal assessment scale (CGAS). Arch Gen Psychiatry.1983;40:228-231.27. Diggle PJ, Liang KY, Zeger SL. Analysis of Lon-gitudinal Data. Oxford, England: Clarendon Press;1994.28. Agresti A. Categorical Data Analysis. 2nd ed. NewYork, NY: John Wiley & Sons; 2002:120.29. Cochran WG. Some methods for strengtheningthe common Chi -Square tests. Biometrics. 1954;10:417-451.30. Fleiss JL. The Design and Analysis of ClinicalExperiments. New York, NY: John Wiley & Sons; 1986.31. Trivedi MH, Rush AJ, Wisniewski SR, et al; for theSTAR*D Study Team. Outcomes with citalopram fordepression using measurement-based care in STAR*D:implications for clinical research and practice. Am JPsychiatry. 2006;163:28-40.32. Rush AJ, Zimmerman M, Wisniewski SR, et al. Co-morbid psychiatric disorders in depressed outpa-tients: demographic and clinical features. J AffectDisord. 2005;87:43-55.33. Hodges K, Gordon Y, Lennon MP. Parent-childagreement on symptoms assessed via a clinical re-search interview for children: the Child AssessmentSchedule (CAS). J Child Psychol Psychiatry. 1990;31:427-436.

34. Edelbrock C, Costello AJ, Dulcan MK, et al. Parent-child agreement on child psychiatric symptomsassessed via structured interview. J Child PsycholPsychiatry. 1986;27:181-190.35. Gotlib IH, Goodman SH. Children of parents withdepression. In: Silverman WK, Ollendick TH, eds. De-velopmental Issues in the Clinical Treatment of Chil-dren and Adolescents. Boston, Mass: Allyn & Bacon;1999:415-432.36. Downey G, Coyne JC. Children of depressedparents: an integrative review. Psychol Bull. 1990;108:50-76.37. Klein DN, Lewinsohn PM, Seeley JR, Rohde P. Afamily study of major depressive disorder in a com-munity sample of adolescents. Arch Gen Psychiatry.2001;58:13-20.38. Puig-Antich J, Lukens E, Davies M, Goetz D, Bren-nan-Quattrock J, Todak G. Psychosocial functioningin prepubertal major depressive disorders, II: interper-sonal relationships after sustained recovery from af-fective episode. Arch Gen Psychiatry. 1985;42:511-517.39. Mintz J, Mintz LI, Arruda MJ, Hwang SS. Treat-ments of depression and the functional capacity towork [published correction in Arch Gen Psychiatry1993;50:241]. Arch Gen Psychiatry. 1992;49:761-768.40. Koran LM, Gelenberg AJ, Kornstein SG, et al. Ser-traline versus imipramine to prevent relapse in chronicdepression. J Affect Disord. 2001;65:27-36.41. Miranda J, Siddique J, Belin TR, Kohn-Wood LP.Depression prevalence in disadvantaged young blackwomen African and Caribbean immigrants com-pared to US-born African Americans. Soc PsychiatryPsychiatr Epidemiol. 2005;40:253-258.42. Caspi A, Sugden K, Moffitt TE, et al. Influenceof life stress on depression: moderation by a polymor-phism in the 5-HTT gene. Science. 2003;301:386-389.43. Kim-Cohen J, Moffitt TE, Taylor A, et al. Mater-nal depression and children’s antisocial behavior: na-ture and nurture effects. Arch Gen Psychiatry. 2005;62:173-181.

MATERNAL DEPRESSION AND PSYCHOPATHOLOGY IN CHILDREN

1398 JAMA, March 22/29, 2006—Vol 295, No. 12 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

lieve that it is essential for all clinical researchers to pro-vide a complete disclosure of all financial interests that arerelevant to a study.2 Although the term conflict of interest iswidely used, it may entail different meanings. Margolis3 dis-tinguishes between conflicting interests and conflicts of in-terest. The former occur in any situation for which com-peting considerations are presumed to be legitimate. Conflictsof interest, on the other hand, are characterized by an in-dividual occupying dual roles, such as being a researcherand holding a significant financial interest in an area re-lated to the research one is involved in.

At the time of submission of our manuscript to JAMA, weall reached an individual determination of whether our fi-nancial interests, which we report elsewhere,4 did or did notpose a relevant potential conflict of interest with this re-port. Our individual decisions were based on the recogni-tion that the Sequenced Treatment Alternatives to RelieveDepression (STAR*D)–Child study was not designed or pow-ered to demonstrate the efficacy of one antidepressant agentover others or of pharmacological strategies compared witheither nonpharmacological treatments (including psycho-therapy) or favorable environmental and temporal factorsnot related to clinical care. This point is made in the ar-ticle, which provides important observations about moth-ers’ depressive remission and child health but which ap-propriately refrains from going beyond the data to test aparticular agent or modality of treatment.

However, we recognize that JAMA policy requires disclo-sure of any potential conflicts of interest, and in retrospectwe regret that we did not disclose all of our financial inter-ests over the last 5 years and for the foreseeable future. Theyare included herein.

Myrna M. Weissman, PhDweissman@childpsych.columbia.eduCollege of Physicians and SurgeonsColumbia UniversityNew York, NYJonathan E. Alpert, MD, PhDMaurizio Fava, MDMassachusetts General HospitalBoston, MassMadhukar H. Trivedi, MDA. John Rush, MDUniversity of Texas Southwestern Medical CenterDallas

Financial Disclosures: Dr Weissman reported having received grant support fromEli Lilly and GlaxoSmithKline and having served as a scientific advisor to Eli Lilly.Dr Alpert reported having received research support from Aspect Medical Sys-tems, Eli Lilly, Forest Pharmaceuticals, and Pfizer; having received research sup-port and honoraria from Organon Inc; and having received research support fromand having served as a consultant for Pharmavite LLC and Pamlab LLC. Dr Favareported having received research support from Abbott Laboratories, LichtwerPharma GmbH, Lorex Pharmaceuticals; having received honoraria from EPIX Phar-maceuticals, Bayer AG, Compellis, Cypress Pharmaceuticals, Janssen Pharmaceu-tica, Knoll Pharmaceutical, Lundbeck, Dov Pharmaceuticals, Biovail Pharmaceuti-cals, BrainCells Inc, Cypress Pharmaceuticals, Fabre-Kramer Pharmaceuticals,Grunenthal GmBH, MedAvante, Sepracor, and Somerset Pharmaceuticals; and hav-ing received both research support and honoraria from Aspect Medical Systems,

Astra-Zeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, Forest Pharmaceuticals,GlaxoSmithKline, Johnson & Johnson Pharmaceuticals, Novartis, Organon, Pam-lab, Pharmavite, Pfizer, Roche, Sanofi/Synthelabo, Solvay Pharmaceuticals, andWyeth-Ayerst Laboratories. Dr Trivedi reported having provided scientific consul-tation for or having served on the advisory boards of Bristol-Myers Squibb, Cy-beronics, Eli Lilly, Forest Pharmaceuticals, Johnson & Johnson, Pfizer, Sepracor, andWyeth-Ayerst Laboratories; having served on the speaker bureaus for Bristol-Myers Squibb, Cyberonics, Eli Lilly, Forest Pharmaceuticals, and Wyeth-Ayerst Labo-ratories; and having received research support from Bristol-Myers Squibb, Cepha-lon, Corcept Therapeutics, Eli Lilly, GlaxoSmithKline, Janssen Pharmaceutica, NationalInstitute of Mental Health, National Alliance for Research in Schizophrenia andDepression, Pfizer, Predix Pharmaceuticals, and Wyeth-Ayerst Laboratories. Dr Rushreported having provided scientific consultation to or served on advisory boardsof Advanced Neuronetic Systems, Bristol-Myers Squibb, Cyberonics, Eli Lilly, For-est Pharmaceuticals, GlaxoSmithKline, Health Technology Systems, Merck, Neu-ronetics, Ono Pharmaceuticals, Organon Pharmaceuticals, Pfizer, the Urban In-stitute, and Wyeth-Ayerst Laboratories; having received royalties from GuilfordPress and Health Technology Systems and research and grant support from theRobert Wood Johnson Foundation, the National Institute of Mental Health, andthe Stanley Foundation; having served on the speaker bureaus of Cyberonics, EliLilly, Forest Pharmaceuticals, GlaxoSmithKline, and Merck; and owning stock inPfizer.

1. Permutt T, Hebel JR. Simultaneous-equation estimation in a clinical trial of theeffect of smoking on birth weight. Biometrics. 1989;45:619-622.2. Fava GA. Conflict of interest and special interest groups: the making of a counterculture. Psychother Psychosom. 2001;70:1-5.3. Margolis J. Conflict of interest and conflicting interests. In: Beauchamp TL, BowieNE, eds. Ethical Theory and Business. Englewood Cliffs, NJ: Prentice-Hall; 1979.4. American Psychiatric Association. Continuing Medical Education Policy on FullDisclosure. 2005. Available at: http://www.psych.org/edu/ann_mtgs/am/05/programbk/a8pgXIXdisclosureindex03302005.pdf. Accessibility verified May 10,2006.

CORRECTION

Incomplete Financial Disclosure: In the Original Contribution entitled “Remis-sions in Maternal Depression and Child Psychopathology: A STAR*D-Child Re-port” published in the March 22/29, 2006, issue of JAMA (2006;295:1389-1398), financial disclosures were omitted. The amended disclosure statement follows:Dr Weissman reported having received grant support from Eli Lilly andGlaxoSmithKline and having served as a scientific advisor to Eli Lilly. Dr Alpert re-ported having received research support from Aspect Medical Systems, Eli Lilly,Forest Pharmaceuticals, and Pfizer; having received research support and hono-raria from Organon Inc; and having received research support from and havingserved as a consultant for Pharmavite LLC and Pamlab LLC. Dr Fava reported hav-ing received research support from Abbott Laboratories, Lichtwer Pharma GmbH,Lorex Pharmaceuticals; having received honoraria from EPIX Pharmaceuticals, BayerAG, Compellis, Cypress Pharmaceuticals, Janssen Pharmaceutica, Knoll Pharma-ceutical, Lundbeck, Dov Pharmaceuticals, Biovail Pharmaceuticals, BrainCells Inc,Cypress Pharmaceuticals, Fabre-Kramer Pharmaceuticals, Grunenthal GmBH,MedAvante, Sepracor, and Somerset Pharmaceuticals; and having received bothresearch support and honoraria from Aspect Medical Systems, Astra-Zeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, Forest Pharmaceuticals, GlaxoSmithKline, Johnson& Johnson Pharmaceuticals, Novartis, Organon, Pamlab, Pharmavite, Pfizer, Roche,Sanofi/Synthelabo, Solvay Pharmaceuticals, and Wyeth-Ayerst Laboratories. DrTrivedi reported having provided scientific consultation for or having served onthe advisory boards of Bristol-Myers Squibb, Cyberonics, Eli Lilly, Forest Pharma-ceuticals, Johnson & Johnson, Pfizer, Sepracor, and Wyeth-Ayerst Laboratories;having served on the speaker bureaus for Bristol-Myers Squibb, Cyberonics, Eli Lilly,Forest Pharmaceuticals, and Wyeth-Ayerst Laboratories; and having received re-search support from Bristol-Myers Squibb, Cephalon, Corcept Therapeutics, Eli Lilly,GlaxoSmithKline, Janssen Pharmaceutica, National Institute of Mental Health, Na-tional Alliance for Research in Schizophrenia and Depression, Pfizer, Predix Phar-maceuticals, and Wyeth-Ayerst Laboratories. Dr Rush reported having providedscientific consultation to or served on advisory boards of Advanced NeuroneticSystems, Bristol-Myers Squibb, Cyberonics, Eli Lilly, Forest Pharmaceuticals,GlaxoSmithKline, Health Technology Systems, Merck, Neuronetics, Ono Pharma-ceuticals, Organon Pharmaceuticals, Pfizer, the Urban Institute, and Wyeth-Ayerst Laboratories; having received royalties from Guilford Press and Health Tech-nology Systems and research and grant support from the Robert Wood JohnsonFoundation, the National Institute of Mental Health, and the Stanley Foundation;having served on the speaker bureaus of Cyberonics, Eli Lilly, Forest Pharmaceu-ticals, GlaxoSmithKline, and Merck; and owning stock in Pfizer.

LETTERS

1234 JAMA, September 13, 2006—Vol 296, No. 10 (Reprinted) ©2006 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013