Journal of Affective Disorders 105 (2008) 185–193www.elsevier.com/locate/jad

Research report

Stressful life events and depressive problems in early adolescent boysand girls: The influence of parental depression, temperament

and family environment☆

Esther M.C. Bouma a,⁎, Johan Ormel a, Frank C. Verhulst b, Albertine J. Oldehinkel a,b

a Department of Psychiatry and School of Behavioural and Cognitive Neurosciences, University Medical Centre Groningen,University of Groningen, The Netherlands

b Department of Child and Adolescent Psychiatry, Erasmus Medical Centre, Rotterdam, The Netherlands

Received 1 December 2006; received in revised form 9 May 2007; accepted 10 May 2007Available online 14 June 2007

Abstract

Background: Stressful life events increase the probability of depressive problems in early adolescence. Several genetic andenvironmental risk factors may change individual sensitivity to the depressogenic effect of these events. We examined modificationby parental depression and gender, and mediation of the former by temperament and family environment.Methods: Data were collected as part of a longitudinal cohort study of (pre)adolescents (n=2127). During the first assessment waveat approximately age 11, we assessed parental depression, family functioning, perceived parenting behaviours, and temperamentalfrustration and fearfulness. At the second wave, about two and a half years later, stressful life events between the first and secondassessment were assessed. Depressive problems were measured at both waves.Results: Adolescents with parents who had a (lifetime) depressive episode were more sensitive to the depressogenic effect ofstressful events than adolescents without depressed parents. Furthermore, girls are more sensitive to these effects than boys. Themodifying effect of parental depression was not mediated by temperament, family functioning and perceived parenting.Limitations: Life events were assessed without consideration of contextual information. Depressive problems were measured byquestionnaires that did not directly represent DSM-IV criteria. The measure of parental depression was unspecific regardingseverity and timing of depressive episodes.Conclusion: The results suggest that gender and parental depression are associated with increased sensitivity to depression afterexperiencing stressful life events during adolescence.© 2007 Elsevier B.V. All rights reserved.

Keywords: Adolescent depressive problems; Stressful life events; Parental depression; Gender; Mediated modification

☆ E.M.C. Bouma analysed the data and wrote the paper supervisedby A.J. Oldehinkel. J. Ormel and F.C. Verhulst gave valuablesuggestions in the final stages of the paper. All authors contributedto and have approved the final manuscript.⁎ Corresponding author. Department of Psychiatry, University

Medical Centre Groningen, P.O. Box 30.000, 9700 RB Groningen,The Netherlands. Tel.: +31 50 3612774; fax: +31 50 361 9722.

E-mail address: E.M.C.Bouma@med.umcg.nl (E.M.C. Bouma).

0165-0327/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.jad.2007.05.007

1. Introduction

Depressive problems in adolescence can be persistentand recurrent (Angst et al., 2000) and represent apotential cause for psychopathology and poor psycho-social adjustments in adult life (Goodyer et al., 2000).

Fig. 1. Schematic model of hypothesised associations.

186 E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

The experience of stressful life events is a well-established risk factor for the development of depressiveproblems in adolescence and adulthood (Goodyer et al.,2000; Kendler et al., 1995; Kessler, 1997; Larson andHam, 1993; Ormel et al., 2001; Rutter, 2000; Silberget al., 2001). Whether or not stressful life events triggerdepressive problems depends on adolescents' sensitivityto stress, which is influenced by a variety of genetic andenvironmental risk factors. We propose that parentaldepression is one of the risk factors that may amplify theassociation between stressful life events and depressiveproblems in adolescence.

Depressive problems run in families (e.g. Goodmanand Gotlib, 1999), which may be due to transmittedvulnerability genes as well as to offspring exposure to anadverse and non-supportive family environment. Fromtwin, adoption and family studies we know thatdepression is moderately heritable (e.g. Kendler andPrescott, 1999; Sullivan et al., 2000; Kendler et al.,1999), and that genetic factors may modify theassociation between life events and depression (Kendleret al., 1995; Silberg et al., 2001). Consistent with this,recent molecular genetic work suggests that a functionalpolymorphism in the promotor region of the serotonintransporter gene moderates the depressogenic effects ofstressful life events (Caspi et al., 2003; Eley et al., 2004;Jacobs et al., 2006; Zammit and Owen, 2006). Based onthese findings we hypothesised that parental depressionwould amplify the effect of stressful life events onadolescents' depressive problems.

Furthermore, individual differences in sensitivity tostressful life events may be explained by differences innegative affectivity. Negative affectivity, a temperamentdimension encompassing frustration and fearfulness,reflects the tendency to experience negative emotionswhen confronted with environmental challenges (Roth-bart et al., 2000). Several studies (e.g. Lonigan et al.,2003), including work from our own group (Oldehinkelet al., 2006) have revealed associations betweennegative affectivity and depressive problems in adoles-cence. Negative affectivity and depressive problemsshare genetic risk factors (e.g. Kendler et al., 2006).Ormel et al. (2005) showed that the effect of parentalpsychopathology on offspring psychopathology waspartly mediated by preadolescent temperament. Con-sidering this, we expected that (part of) the postulatedmodification of the depressogenic effect of stressful lifeevents by parental depression would be mediated bytemperamental frustration and fearfulness.

The modifying effect of parental depression couldalso be mediated by family environmental factors. Lackof family closeness, poor communication, absence of

supportive relationships, parental rejection and littleemotional warmth have all been found to be prevalent infamilies with depressed parents, and are associated withan increased risk of emotional and behavioural problemsin offspring (e.g. Pilowksy et al., 2006; Meares et al.,2000; Davies and Windle, 1997; Beardslee et al., 1996).The stress-buffering model (Cohen and Wills, 1985)proposes that, when faced with negative experiences,individuals with greater support from families andfriends are less likely to become depressed. Conversely,poor family circumstances may increase the impact ofstressful life events. Hence, family environmentalfactors could also mediate (part of) the postulatedmodification of the depressogenic effect of stressful lifeevents on adolescents' depressive problems.

Prevalence of depressive problems, number and kindof stressful events, and sensitivity to stress, are alldifferent for males and females (e.g. Brown and Harris,1978; Sherrill et al., 1997). The transition intoadolescence seems to be the starting point for an increasein depressive problems in girls (Angold et al., 1998),probably set off because of a variety of gender specifichormonal and social developments (Ge et al., 2001;Larson and Ham, 1993; Cyranowski et al., 2000; Silberget al., 2001). Therefore, associations between stressfullife events and adolescents' depressive problems cannotbe studied without taking into account the possibility ofgender differences in each of the effects examined.

The aim of this study was to explore if parentaldepression and gender modified the sensitivity to thedepressogenic effect of stressful events in a largepopulation-based sample of early adolescents. Inaddition, we investigated if the effect of parentaldepression was mediated by temperamental frustrationand fearfulness, and by family environmental factors,more specifically, perceived parental emotional warmthand rejection and family functioning. The hypothesisedassociations are presented in Fig. 1.

187E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

2. Methods

2.1. Sample

TRAILS: The TRacking Adolescents' IndividualLives Survey is a large prospective population studyof Dutch adolescents who are measured biennially untilat least 25 years of age. The present study involves datafrom the first and second assessment wave of TRAILS,which ran from, respectively, March 2001 to July 2002,and September 2003 to December 2004. At the firstwave, 2230 children were enrolled in the study of which2149 children participated in the second wave. Themean age at T1 was 11.09 (SD=0.56), and 50.8% weregirls. The mean age at T2 was 13.6 (SD=0.53), and51.2% were girls. For more details see De Winter et al.(2005).

2.2. Procedures

Measures used in this study come from the TRAILSFamily Interview held with one of the parents (usuallythe mother, 95.6%) at the first assessment wave (T1) andself-report questionnaires filled out by parents andadolescents at both the first and second assessment wave(T1 and T2).

Depressive problems. Mental health problems wereassessed by the Child Behavioral Checklist (CBCL)parent report form (Achenbach, 1991b; Verhulst andAchenbach, 1995). The CBCL contains a list of 120behavioural and emotional problems, which parents canrate as 0 = not true, 1 = somewhat or sometimes true, or2 = very or often true in the past 6 months. In addition tothe CBCL we administered the self-report version of thisquestionnaire, the Youth Self-Report (YSR, Achenbach,1991a). We used the CBCL/YSR Depressive Problemsscale (Achenbach et al., 2003), which contains of 13items (Cronbach's α T1 CBCL=0.68, T2 CBCL=0.73,T1 YSR=0.77, T2 YSR=0.77) covering depressedmood, anhedonia, loss of energy, feelings of worthless-ness and guilt, suicidal ideation, sleep problems andeating problems. The mean of the standardised parentand adolescent scores was used as a measure ofdepressive problems in this study since a combinedscore is less sensitive to context and perspective(Kraemer et al., 2003). When information of oneinformant was missing or unreliable (YSR: T1 n=34,T2 n=36, CBCL: T1 n=153, T2 n=205) the compositescore was based on only one informant.

Stressful life events were assessed at T2 by aquestionnaire containing 36 events, of which 25 werestressful. Examples of these events are parental divorce

or death, severe illness/death of family members, seriousquarrel with a friend, romantic break-up and victimiza-tion. The items had a yes/no format to indicate whetheror not the event had occurred in the last 2 years. Of the2127 adolescents included in our analysis, 78.2%experienced at least one stressful life event in thisperiod. The severity of an event was rated on a four-point scale ranging from 0 = not unpleasant to 3 = veryunpleasant. Of all the events included 39% werecategorised as very unpleasant, 28% as unpleasant and33% as mildly unpleasant. Events that were relativelyoften rated as very unpleasant were; serious illness,death of a dear one, parental divorce, and conflicts withand between family members. The other events weremore or less equally divided across the severitycategories. The scores of all 25 events were summedto create a total severity score. Because this total scorewas highly skewed and we did not expect meaningfuldifferences between low and average levels of stressregarding their association with depressive problems,we dichotomised this variable into low/average stress(score ≤6: 81.4% of the participants) and high stress(score N6: 18.6% of the participants).

Lifetime parental depression was assessed by meansof the TRAILS Family History Interview (FHI). A briefdescription of the main symptoms of depression wasgiven, followed by a series of questions to assess(lifetime) prevalence, treatment, and presence during thepast year. Depressions characterized by professionaltreatment, medication use, or episodes in the last yearwere given extra weight, because of their greater(assumed) severity and/or impact on the adolescents.For each parent, parental depression was scored as 0 =no depression, 1 = depression without treatment/medication and last episode more than a year ago, and2 = depression with treatment/medication or episode inthe last year. In our dataset, 14.6% males and 26.8%females had experienced a depressive episode. Thesenumbers are comparable to the DSM-IV lifetime rates(15.4% males and 27.4% females) obtained by CIDI-interviews in NEMESIS, a large population study (Bijlet al., 1998) and in the European Study of Epidemiologyof Mental Disorders (2004). We summed the data ofboth parents which resulted in a parental depressionscore that ranged from 0 to 4. Missing values (fathern=53, mother n=65) were imputed by group means.

Temperament was assessed at T1 by the parentversion of the short form of the Early AdolescentTemperament Questionnaire Revised (EATQ-R, Hart-man, 2000, Putnam et al., 2001). Fearfulness (five items,Cronbach's α=0.63) denotes worrying and unpleasantaffect related to the anticipation of distress. Frustration

188 E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

(five items, Cronbach's α=0.74) is indicative ofnegative affect related to interruption of ongoing taskor goal blocking.

Parenting. At T1, perceived parenting behaviourwere assessed with the EMBU-C (Markus et al., 2003),the child version of the EMBU (a Swedish acronym forMy Memories of Upbringing, developed by Perris et al.,1980). In the present study, we used the scales Rejectionand Emotional Warmth. The Rejection scale contains of12 items with Cronbach's α=0.84 for fathers and 0.83for mothers. The Emotional Warmth scale contains 18items with Cronbach's α=0.91 for both parents. Theassociations for father and mother were high, both forrejection (r=0.68, pb0.001) as well as for emotionalwarmth (r=0.79, pb0.001). Therefore we combinedthem into a single measure. When information of oneinformant was missing or unreliable (EmotionalWarmth: father n=70, mother n=30, Rejection: fathern=72, mother n=31) the composite score was basedon only one informant.

Family functioningwas assessed at T1, by a modifiedversion of the General Functioning Scale of theMcMaster Family Assessment Device (FAD, Epsteinet al., 1983). This scale assesses six dimensions of familyfunctioning; communications, problem solving, affec-tive responsiveness, affective involvement, roles andbehaviour control. The modified scale consists of 12items with a Cronbach's α=0.85. Parents could rate theiragreement on a 4-point scale; 1 = totally disagree, 2 =disagree, 3 = agree, 4 = totally agree.

2.3. Statistical methods

All continuous variables were standardised to a meanof 0 and a standard deviation of 1 to obtain internallycomparable regression coefficients. Interaction termswere created by multiplying the standardised scores. Ap-value smaller than 0.05 was considered statisticallysignificant. First, means of and correlation betweenvariables were calculated and gender differences inmeans and percentages were analysed by t-tests and χ2-tests respectively. After that, we examined modificationof the relationship between stressful events anddepressive problems by parental depression and gender.In case of a significant interaction of parental depressionand stressful life events, we subsequently examinedmediation of this effect by fearfulness, frustration,rejection, emotional warmth and family function.

Main and interaction effects of stressful life events,parental depression and gender on adolescent depressiveproblems were investigated by means of a backwardstepwise regression analysis with T2 depressive pro-

blems as dependent variable. The initial model includesall main and interaction effects of stressful life events,parental depression and gender. First, we examined thethree-way interaction between stressful life events,parental depression and gender (a significant interactionwould imply that the mediation analyses had to bestratified by gender). If not significant, this interactionwas dropped from the model. In the second step, non-significant two-way interactions were dropped from themodel. The last step examined the main effects. Modelterms could only be dropped if the variables were notincluded in higher-order terms. To examine to whatextent the effect of stressful life events might be due topre-event depressive problems, we repeated the finalanalysis adjusting for T1 depressive problems, byincluding this variable in the model.

If parental depression significantly modifies theeffects of stressful life events on adolescent depressiveproblems, this effect may be mediated by temperamentor family-related risk factors. Mediation of an associ-ation between a determinant and an outcome can occuronly if the mediating variable is associated with bothfactors, follows the determinant in time and precedes theoutcome (Baron and Kenny, 1986). In our case, we didnot investigate simple mediation, but mediated modifi-cation (that is mediation of the modifying effect ofparental depression on the association between stressfullife events and depressive problems). Consequently, ourmediators should still be significantly associated withparental depression, but rather than being associatedwith depressive problems, their interaction with stressfullife events should be significant. A factor that mediatesthe effect of parental depression, should be related tosensitivity to stressful life events itself too, that is,increase the effect of life events on depressive problems.Putative mediators were only included in subsequentanalyses when both associations were found.

To asses the amount of mediation, the (direct)interaction effect of parental depression and stressfulevents was compared with the interaction effect afterincluding the (main and interaction) effects of themediator in the model. The larger the relative reductionin the regression coefficient in the interaction betweenparental depression and stressful life events, the morethis effect was mediated by the variable under study.

3. Results

3.1. Descriptive statistics

Mean scores or percentages of the variables used areshown in Table 1. For descriptive purposes, we

Table 1Descriptive statistics

Boys Girls Gender difference

n Mean (SD) n Mean (SD) t or χ df p

Dep. problems T1 1027 3.13 (2.50) 1080 3.18 (2.43) −4.63 210 0.76Dep. problems T2 1033 2.47 (2.30) 1075 3.17 (2.90) −6.16 2087 a b0.001Stress (0,1) 1039 14.8% 1088 22.2% 18.88 1 b0.001Parental depression 1007 0.80 (1.18) 1060 0.89 (1.17) 0.16 2065 0.86Fearfulness 930 2.34 (0.70) 976 2.49 (0.75) −4.48 1903 a 0.88Frustration 931 2.83 (0.68) 976 2.74 (0.63) 3.18 1881 a b0.001Emotional warmth 1027 3.17 (0.51) 1082 3.26 (0.49) −4.42 2090 a b0.001Rejection 1027 1.51 (0.32) 1081 1.45 (0.29) 5.03 2053 a b0.001Family functioning 957 2.49 (0.16) 1002 2.50 (0.10) 0.15 1957 0.15

a Degrees of freedom not equal to n−1 due to correction for unequal variances.

189E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

presented the mean of the unstandardised total scores ofparent and self-reported adolescent depressive problemsrather than the mean of the standardised scores.

3.2. Bivariate associations

Correlations between the variables in the study weregenerally low to moderate (Table 2). In both sexes,depressive problems at T2 were significantly associatedwith all other variables.

3.3. Modification of the stress-depressive problemsassociation by parental depression and gender

The three-way interactions of stressful events,parental depression and gender were not significant(B=−0.09, p=0.36), indicating that the interaction ofstressful life events and parental depression wascomparable for boys and girls, and that mediationanalysis did not have to be stratified by gender. The two-

Table 2Bivariate associations

1. 2. 3. 4.

1. Stress .05 a .16a,b .31a

2. Parental depression .03 a .21 b .173. Depr. problems T1 .14a,b .19 b .524. Depr. problems T2 .19a,b .18 b .60 b

5. Rejection .12a,b .04 .32 b .226. Emotional warmth − .03 − .02 − .16 b − .127. Fearfulness .14a,b .16 b .27 b .208. Frustration .11a,b .13 b .25 .239. Family functioning .04 a .07 a .05 .09

Note: Girls' correlations are printed above the diagonal; boys' correlations ba Point biserial correlations for associations between a continuous and a db Significant association at pb0.05.

way interaction of parental depression and gender wasnot significant either (B=0.02, p=0.61).

Parental depression and gender both significantlymoderated the association between stressful life eventsand depressive problems. Sensitivity to stress, asindicated by the interaction variable between stressfullife events and parental depression, was larger inadolescents whose parents have a history of depressionthan in adolescents whose parents have no such history.Furthermore, the significant interaction between stress-ful life events and gender indicated that girls are moresensitive to this effect than boys. Coefficients of the finalmodel are shown in Table 3. These results are visualisedin Fig. 2 in which the estimated depressive problems forboys and girls are shown, both in absence and presenceof parental depression and stressful life events.

We performed a posthoc-analyses to examine theeffect of stressful life events in absence of parentaldepression by adding parental depression as unstan-dardised variable to the model (B=0.32, p=b0.001).

5. 6. 7. 8. 9.

,b .10a,b .04 a .09a,b .05 a .04 a

b .01 − .04 .09 b .10 b .11 b

b .37 b − .22 b .31 b .22 b .11 b

.28 b − .16 b .19 b .12 b .06 b

b − .33 b .26 b .24 b .03b − .31 b − .05 .02 .004b .23 b − .05 .43 b .05b .26 b − .001 .43 b − .01b .03 .01 − .01 − .04

elow the diagonal.ichotomous variable.

Table 3Effect modification by parental depression and gender

Predictors B t p

Stress (0 = low, 1 = high) 0.39 5.48 b0.001Parental depression 0.12 5.90 b0.001Gender (0 = boys, 1 = girls) 0.12 3.06 0.003Stress⁎parental depression 0.11 2.47 0.014Stress⁎gender 0.27 2.85 0.004

Dependent variable: depressive problems at T2.Note1: R2=0.113 (S.E.=0.81).

190 E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

This effect was stronger for girls than for boys (B=0.27,p=0.004).

Adjusted for depressive problems at T1 (i.e. beforethe stressful events) the main effects of gender andparental depression were still significant, as well as theinteraction between stressful life events and gender(B=0.27, p=0.01). The interaction between stressfullife events and parental depression was no longersignificant but showed a trend in the expected direction(B=0.07, p=0.074).

3.4. Analysis of mediation

Because the interaction between parental depressionand stressful events did not show gender differences,mediation analyses were not stratified by gender. One ofthe prerequisites of mediation was that the mediator hadto be associated with parental depression. Onlyfrustration, fearfulness and family functioning met thisfirst criterion (see Table 2) however, none of thesefactors met the second prerequisite and were notincluded in further analyses. Interesting to note is thatrejection did meet the second prerequisite for mediation;the interaction between rejection and stressful life eventswas significantly associated with depressive problems(B=0.126, p=0.003).

Fig. 2. Estimated problem

4. Discussion

The first goal of this study was to explore if therelationship between stressful life events and depressiveproblems in adolescence was modified by parentaldepression and gender in a large population sample ofDutch early adolescents. Secondly, we wanted toinvestigate if the effect of parental depression wasmediated by temperamental frustration and fearfulness,and by family environmental factors. Consistent with ourexpectations and other studies (e.g. Goodyer et al., 2000;Larson and Ham, 1993), stressful life events wereassociated with depressive problems in both boys andgirls, but considerably stronger in girls than in boys.Furthermore,we found that adolescentswhose parents hada history of depression reportedmore depressive problemsafter the occurrence of stressful life events than adoles-cents whose parents did not have such history. This effectof parental depression was not mediated by temperamentor family environment factors included in this study.

The finding that parental depression was associatedwith a greater offspring sensitivity to stressful life eventsis in line with results reported by Silberg et al. (2001),despite differences in study design (twin versuspopulation sample, parent versus child report, onlygirls versus both genders, age differences). This adds tothe evidence that familial risk of depression is likely tobe expressed by increased stress-sensitivity. A differ-ence between Silberg et al.'s and our study is that theydid not find an association between stressful life eventsand depression in the absence of parental emotionaldisorders, while we also found an effect in the group ofadolescents whose parents have never had depressiveproblems. Our findings converge with molecular geneticevidence (e.g. Caspi et al., 2003) that genetic poly-morphisms contribute to individual differences insensitivity to stressful experiences.

s for boys and girls.

191E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

Children from parents who have a history ofdepression may be at risk for developing emotionalproblems not only because of a genetically predisposedsensitivity to stress but also because of the possibilitythat they experience more stressful events than childrenfrom parents without such history. In our sample,however, parental depression was not significantlyassociated with the number of stressful life events,which could be due to the relatively large proportion ofparents who have a history of depression but had nodepressive problems during the study period.

The association between stressful events and depres-sive problems was much stronger for girls than for boys,both with and without adjusting for pre-existingdepressive problems. This could indicate that girlsbecome more sensitive to the depressogenic effects ofstressful life events during the transition from childhoodto adolescence as suggested several times (e.g. Cyra-nowski et al., 2000; Silberg et al., 2001). It should benoted, however, that the gender difference in depressiveproblems during early adolescence, was due to adecrease of depressive problems in boys, rather thanan increase in girls, compared to preadolescent levels.The lack of increase in depressive problems in girls inour sample might suggest that their sensitivity to eventsdid not change, while boys became less sensitive. Datafrom subsequent waves will make it possible to explorethis further.

To the best of our knowledge, no one has analysedmediated modification of the stress-depressive problemsassociation by parental depression. There are only threestudies that examined mediation of the (main) effects ofparental psychopathology on offspring psychopatholo-gy. Davies and Windle (1997) found that family discordwas a modest mediator of the effect of maternaldepressive symptoms on middle-adolescent' depressivesymptoms. Ormel et al. (2005), using the same sampleas described in the present study, found that one third ofthe effect of familial loading of parental psychopathol-ogy on internalising and externalising problem behav-iour in offspring was mediated by offspringtemperament. Recently, Burt et al. (2006), reportedthat parenting and family environmental factors partiallymediated the association between parental psychopa-thology and offspring psychopathology in late adoles-cence. Although these studies involved mediation of themain effect of parental psychopathology on offspringproblem behaviour instead of mediated modification,they seem to provide indirect support that the effect ofparental depression on adolescents' sensitivity tostressful events is mediated by temperament and familyenvironment. Yet we did not find evidence for such

mediation. Tentatively, we propose a number of factorsthat could have contributed to these results.

There were no significant associations between theparental depression measure and the family-relatedputative mediators emotional warmth, parental rejec-tion, and poor family functioning. This could be due tothe fact that our parental depression measure yielded ahighly heterogeneous group, regarding both severity ofthe depressive problems and their timing. Mild or longremitted depressive problems are not very likely toaffect current parenting and family functioning. Ourresults suggest that the association between parentaldepression and parenting may be less pronounced oreven absent in population samples. Indeed, recentprospective studies found no stress-buffering effects offamily support (e.g. Burton et al., 2004).

The absence of associations could also signifymethodological weakness of the parental depressionmeasure, which was based on retrospective report.However, this is contradicted by the finding that theparental depression measure was strong enough tomoderate the association between stressful events andoffspring depressive problems. Although we did notfind evidence for family-related factors, it is stillpossible that parental depression influenced the impactof stressful life events through other family difficultieswhich were not included in this model.

Temperamental frustration and fearfulness did notmediate the interaction between parental depression andstressful life events either. Tentatively, negative affectivityreflects individual differences in reactivity to daily hasslesbetter than reactivity to more severe and rare life events(Zammit and Owen, 2006). Frustration and fearfulnessmay be more relevant with regard to minor provocations,annoyances, and threats than to the bigger facts of life.

Our study has important assets: it was based on alarge prospective population sample of pre- and earlyadolescents and it used multiple informants whichdecreased the risk of inflated associations. The prospec-tive design allowed us to adjust for depressive problemsalready present before the stressful events occurred, andrevealed that the interaction of gender and life eventsremained significant and that the interaction of parentaldepression and life events still showed a trend in theexpected direction.

There are also limitations. First of all, life events wereassessed retrospectively and based on self-reports insteadof interview ratings taking into account contextualinformation (Brown and Harris, 1978). Individuals withdepressive problems tend to over-report number as well asseverity of stressful life events ful events (Brewin et al.,1993). Second, depressive problems were based on

192 E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

questionnaire data. The CBCL/YSR Depressive Problemsscale was not developed to asses depressive problemsaccording to DSM-IV criteria, but constructed on the basisof expert ratings of the original, empirically derived, CBCLand YSR scale items. Consequently, the items do notrepresent one-to-one counterparts with all DSM-IV criteria.Third, as mentioned above, parental depression wasassessed retrospectively. Despite these limitations, thisstudy provides an excellent starting point for futureresearch of the mechanisms by which parental depressionmay be associated with increased stress-sensitivity inchildren of depressed parents.

To conclude, offspring of depressed parents have ahigher stress-sensitivity than offspring of non-depressedparents. This highlights the importance of monitoringchildren from high-risk families, particularly when theyare exposed to high amounts of life stress. Furthermore,early adolescent girls have an increased stress-sensitivitycompared to boys. It is important to extend our knowledgeabout this widely observed gender difference, to providethe best, possibly differential, care for adolescent boys andgirls with depressive problems. More research is neededon mechanisms by which parental depression influencesoffspring's increased stress-sensitivity.

Acknowledgements

This research is part of the TRacking Adolescents'Individual Lives Survey (TRAILS). Participating centresof TRAILS include variousDepartments of theUniversityof Groningen, the Erasmus Medical Centre of Rotterdam,the University of Nijmegen, the Trimbos Institute, and theUniversity of Utrecht, The Netherlands. TRAILS isfinancially supported by grants from the NetherlandsOrganization for Scientific Research (GB-MW 940-38-011, GB-MAG 480-01-006, ZonMw 100.001.001; NWO175.010.2003.005) and the Department of Justice(WODC), and by the participating centres. The fundingsources had no involvement in the study design;collection and interpretation of the data; writing thereport, and decision to submit the paper for publication.

References

Achenbach, T.M., 1991a. Manual for the Child Behavior Checklist/4-18and 1991 Profile. University of Vermont, Vermont, Burlington VT.

Achenbach, T.M., 1991b. Manual for the Youth Self-Report and 1991Profile. University of Vermont, Vermont, Burlington VT.

Achenbach, T.M., Dumenci, L., Rescola, L.A., 2003. DSM-orientedand empirically based approaches to constructing scales from thesame items pools. Journal of Clinical Child and AdolescentPsychology 32, 328–340.

Angold, A., Costello, E.J., Worthman, C.M., 1998. Puberty anddepression: the roles of age, pubertal status and pubertal timing.Psychological Medicine 28, 51–61.

Angst, J., Kasper, S., Weiller, E., 2000. Recurrent brief depression: afrequent syndrome in clinical practice. International Journal ofPsychiatry in Clinical Practice 4, 195–199.

Baron, R.M., Kenny, D.A., 1986. The moderator mediator variabledistinction in social psychological-research — conceptual, strate-gic, and statistical considerations. Journal of Personality and SocialPsychology 51, 1173–1182.

Beardslee, W.R., Keller, M.B., Seifer, R., Lavori, P.W., Staley, J.,Podorefsky, D., Shera, D., 1996. Prediction of adolescent affectivedisorder: effects of prior parental affective disorders and childpsychopathology. Journal of the American Academy of Child andAdolescent Psychiatry 35, 279–288.

Bijl, R.V., Ravelli, A., Van Zessen, G., 1998. Prevalence of psychiatricdisorder in the general population: results of the NetherlandsMental Health Survey and Incidence Study (NEMESIS). SocialPsychiatry and Psychiatric Epidemiology 33, 587–595.

Brewin, C.R., Andrews, B., Gotlib, I.H., 1993. Psychopathology andearly experience: a reappraisal of retrospective reports. Psycho-logical Bulletin 11, 82–98.

Brown, G.W., Harris, O.T., 1978. The Social Origins of Depression: AStudy of Psychiatric Disorder in Women. Travistock, London.

Burt, K.B., Van Dulmen, H.M., Calivati, J., Egeland, B., Sroufe, L.A.,Forman, D.R., Appleyard, K., Carlson, E.A., 2006. Mediatinglinks between maternal depression and offspring psychopathology:the importance of independent data. Journal of Child Psychologyand Psychiatry 46, 490–499.

Burton, E., Stice, E., Seeley, J.R., 2004. A prospective test of the stress-buffering model of depression in adolescent girls: no support onceagain. Journal of Consulting and Clinical Psychology 72, 689–697.

Caspi, A., Sugden, K., Moffit, T.E., Taylor, A., Craig, I.W., Harrington,H., McClay, J., Mill, J., Martin, J., Braithwaite, A., Poulton, R.,2003. Influence of life stress on depression: moderation by apolymorphism in the 5-HTT gene. Science 301, 386–389.

Cohen, S., Wills, I.A., 1985. Stress, social support, and the bufferinghypothesis. Psychological Bulletin 98, 310–357.

Cyranowski, J.M., Frank, E., Young, E., Shear, M.K., 2000.Adolescent onset of the gender difference in lifetime rates ofmajor depression — a theoretical model. Archives of GeneralPsychiatry 57, 21–27.

Davies, P.T., Windle, M., 1997. Gender-specific pathways betweenmaternal depressive symptoms, family discord, and adolescentadjustment. Developmental Psychology 33, 657–668.

De Winter, A., Oldehinkel, A.J., Veenstra, R., Brunnekreef, J.A.,Verhulst, F.C., Ormel, J., 2005. Evaluation of non-response bias inmental health determinants and outcomes in a large sample of pre-adolescents. European Journal of Epidemiology 20, 173–181.

Epstein, N.B., Baldwin, L.M., Bishop, D.S., 1983. The McMaster familyassessment device. Journal ofMarital and Family Therapy 9, 171–180.

European Study of Epidemiology of Mental Disorders (ESEMeDproject), 2004. Prevalence of mental disorders in Europe: resultsfrom the ESEMeD project vol. 109 (supplement 420).

Eley, T.C., Sugden, K., Corsico, A., Gregory, A.M., Sham, P., McGuffin,P., Plomin, P., Craig, I.W., 2004. Gene–environment interactionanalysis of serotonin system markers with adolescent depression.Journal of American Child and Adolescent Psychiatry 43, 298–306.

Ge, X.J., Conger, R.D., Elder, G.H., 2001. Pubertal transition, stressfullife events, and the emergence of gender differences in adolescentdepressive symptoms. Developmental Psychology 37, 404–417.

193E.M.C. Bouma et al. / Journal of Affective Disorders 105 (2008) 185–193

Goodman, S.H., Gotlib, I.H., 1999. Risk for psychopathology in thechildren of depressed mothers: a developmental model for under-standing mechanisms of transmission. Psychological Reviews 106,458–490.

Goodyer, I.M., Herbert, J., Tamplin, A., Altham, P.M.E., 2000. First-episode major depression in adolescents — affective, cognitiveand endocrine characteristics of risk status and predictors of onset.British Journal of Psychiatry 176, 142–149.

Hartman, C.A., 2000. Nederlandse vertaling van de Early AdolescentTemperament Questionnaire. Intern rapport, DisciplinegroepPsychiatrie. Universiteit van Groningen, Nederland.

Jacobs, N., Gunter, K., Peeters, F., Derom, K., Vlietinck, R., Van Os,J., 2006. Stress-related negative affectivity and genetically alteredserotonin transporter function: evidence of synergism in shapingrisk of depression. Archives of General Psychiatry 63, 989–996.

Kendler, K.S., Prescott, C.A., 1999. A population-based twin study oflifetime major depression in men and women. Archives of GeneralPsychiatry 56, 39–44.

Kendler, K.S., Kessler, R.C., Walters, E.E., MacLean, C., Neale, M.C.,Heath, A.C., Eaves, L.J., 1995. Stressful life events, geneticliability, and onset of an episode of major depression in women.American Journal of Psychiatry 152, 833–842.

Kendler, K.S., Karkowski, L.M., Prescott, C.A., 1999. Causalrelationship between stressful life events and the onset of majordepression. American Journal of Psychiatry 156, 837–841.

Kendler, K.S., Gatz, M., Gardner, C.O., Pedersen, N.L., 2006. ASwedish twin study of lifetime major depression. AmericanJournal of Psychiatry 163, 109–114.

Kessler, R.C., 1997. The effects of stressful life events on depression.Annual Review of Psychology 48, 191–214.

Kraemer, H.C., Measelle, J.R., Ablow, J.C., Essex, M.J., Boyce, W.T.,Kupfer, D.J., 2003. A new approach to integrating data frommultiple informants in psychiatric assessment and research: mixingand matching contexts and perspectives. American Journal ofPsychiatry 160, 1566–1577.

Larson, R., Ham, M., 1993. Stress and storm and stress in earlyadolescence — the relationship of negative events with dysphoricaffect. Developmental Psychology 29, 130–140.

Lonigan, C.J., Phillips, Beth. M., Hoe, E.S., 2003. Relations ofpositive and negative affectivity to anxiety and depression inchildren: evidence from a latent variable longitudinal study.Journal of Consulting and Clinical Psychology 71, 465–481.

Markus, M.T., Lindhout, I.E., Boer, F., Hoogendijk, T.H.G., Arrindell,W.A., 2003. Factors of perceived parental rearing styles: theEMBU-C examined in a sample of Dutch primary school children.Personality and Individual Differences 34, 503–519.

Meares, P.A., Firminger, K.B., Oyserman, D., Moybray, C.T., 2000.Parenting among mothers with a serious mental illness. AmericanJournal of Orthopsychiatry 7, 296–315.

Oldehinkel, A.J., Veenstra, R., Ormel, J., De Winter, A.F., Verhulst,F.C., 2006. Temperament, parenting, and depressive symptoms in a

population sample of preadolescents. Journal of Child Psychologyand Psychiatry 47, 684–695.

Ormel, J., Oldehinkel, A.J., Brilman, E.I., 2001. The interplay andetiological continuity of neuroticism, difficulties, and life events inthe etiology of major and subsyndromal, first and recurrentdepressive episodes in later life. American Journal of Psychiatry158, 885–891.

Ormel, J., Oldehinkel, A.J., Ferdinand, R.F., Hartman, C.A., DeWinter, A.F., Veenstra, R., Vollebergh, W., Minderaa, R.B.,Buitelaar, R.K., Verhulst, F.C., 2005. Internalizing and external-izing problems in adolescence: general and dimension specificeffects of familial loadings and preadolescent temperament traits.Psychological Medicine 35, 1825.

Perris, C., Jacobsson, L., Lindstrom, H., Knorring, L.V., Perris, H.,1980. Development of a new inventory for assessing memories ofparental rearing behavior. Acta Psychiatrica Scandinavica 61,265–274.

Pilowksy, D.J., Wikramaratne, M.P.H., Nomura, Y., Weissman, M.M.,2006. Family discord, parental depression, and psychopathology inoffspring: 20-year follow-up. Journal of the American Academy ofChild and Adolescent Psychiatry 45, 452–460.

Putnam, S.P., Ellis, L.K., Rothbart, M.K., 2001. The structure oftemperament from infancy through adolescence. In: Eliasz, A.,Angleitner, A. (Eds.), Advances/Proceedings in Research onTemperament. Pabst Scientist Publisher, Germany, pp. 165–182.

Rothbart, M.K., Ahadi, S.A., Evans, D.E., 2000. Temperament andpersonality: origins and outcomes. Journal of Personality andSocial Psychology 78, 122–135.

Rutter, M., 2000. Negative life events and family negativity:accomplishments and challenges. In: Harris, T. (Ed.), WhereInner and Outer Worlds Meet: Essays in Honour of George W.Brown. Tayler and Francis, London, pp. 129–149.

Sherrill, J.T., Anderson, B.A., Franke, E., Renolds, C.F., Tu, X.M.,Patterson, D., Ritenour, A., Kupfer, D.J., 1997. Is stress morelikely to provoke depressive episodes in women than in men?Depression and Anxiety 6, 95–105.

Silberg, J., Rutter, M., Neale, M., Eaves, L., 2001. Genetic moderationof environmental risk for depression and anxiety in adolescentgirls. British Journal of Psychiatry 179, 116–121.

Sullivan, P.F., Neale, M.C., Kendler, K.S., 2000. Genetic epidemiol-ogy of major depression: Review and meta-analysis. AmericanJournal of Psychiatry 157, 1552–1562.

Verhulst, F.C., Achenbach, T.M., 1995. Empirically based assessmentand taxonomy of psychopathology: cross-cultural applications. Areview. European Child and Adolescent Psychiatry 4, 61–76.

Zammit, S., Owen, M.J., 2006. Stressful life events, 5-HTT genotypeand risk of depression. British Journal of Psychiatry 188, 199–201.