Professional Documents
Culture Documents
Predictors of The First Onset of A Major Depressive Episode and Changes in Depressive Symptoms Adolescence: Stress and Negative Cognitions
Predictors of The First Onset of A Major Depressive Episode and Changes in Depressive Symptoms Adolescence: Stress and Negative Cognitions
Judy Garber
DePaul University
Vanderbilt University
This 6-year longitudinal study examined stressors (e.g., interpersonal, achievement), negative cognitions
(self-worth, attributions), and their interactions in the prediction of (a) the first onset of a major
depressive episode (MDE), and (b) changes in depressive symptoms in adolescents who varied in risk for
depression. The sample included 240 adolescents who were first evaluated in Grade 6 (M 11.86 years
old; SD 0.57; 54.2% female) and then again annually through Grade 12. Stressful life events and
depressive diagnoses were assessed with interviews; negative cognitions and depressive symptoms were
assessed with self-report questionnaires. Discrete time hazard modeling revealed a significant interaction
between interpersonal stressors and negative cognitions, indicating that first onset of an MDE was
predicted by high negative cognitions in the context of low interpersonal stress, and by high levels of
interpersonal stressors at both high and low levels of negative cognitions. Analyses of achievement
stressors indicated significant main effects of stress, negative cognitions, and risk in the prediction of an
MDE, but no interactions. With regard to the prediction of depressive symptoms, multilevel modeling
revealed a significant interaction between interpersonal stressors and negative cognitions such that
among adolescents with more negative cognitions, higher levels of interpersonal stress predicted higher
levels of depressive symptoms, whereas at low levels of negative cognitions, the relation between
interpersonal stressors and depression was not significant. Risk (i.e., maternal depression history) and sex
did not further moderate these interactions. Implications for intervention are discussed.
Keywords: stress, depression, negative cognitions, adolescents
780
pression over time (e.g., Hankin & Abramson, 2001; Kraaij et al.,
2003). More recent studies using short-term longitudinal designs
have found prospective evidence of the moderating effect of negative cognitions on stress in the prediction of change in depressive
symptoms (Abela & Hankin, 2008). In a review of 18 such studies,
Lakdawalla et al. (2007) reported that the follow-up lengths ranged
from 1 to 60 months (M 8.5 months; median 3.0 months;
mode 12 months), and sample sizes ranged from 34 to 1,507
(M 310.56; median 176; mode 79). The current longitudinal study builds on this literature in the following important
ways by: (a) following a moderate size sample (N 240) of youth
over a longer time intervalsix years, (b) testing the cognitive
vulnerability hypothesis to predict both the first onset of a major
depressive episode (MDE) and changes in depressive symptoms
during adolescence, (c) using interview-based measures of stress
and depressive disorders, (d) examining two types of stressors
(interpersonal, achievement), (e) testing a composite index of
negative cognitions, and (f) exploring the direct and moderating
role of sex and two important features of maternal depression
history and current level of depressive symptoms.
list. Regarding cognitions, Hammen and colleagues measured selfconcept, self-schema, and attributional style, whereas Lewinsohn
et al., measured attributional style and dysfunctional attitudes.
Given the different findings in these earlier investigations, and the
paucity of studies testing the cognitive vulnerability model with
respect to clinical diagnoses in youth despite the frequent call for
such research (e.g., Abela, 2001; Abela & Veronneau-McArdle,
2002), a primary aim of the current prospective study was to test
the cognitive diathesis-stress model using a larger, high-risk sample of same-age youth to predict both the first onset of a major
depressive episode as well as changes in depressive symptoms.
Stressors
Not all types of stressors are related to depression in the same
way and not all stressors are similarly moderated by negative
cognitions to predict depression. Given the heterogeneity in how
stress has been defined and measured in the literature, inconsistent
findings across studies are not surprising (Monroe & Simons,
1991). Investigations of the cognitive diathesis-stress model typically have used a broad index of stress comprised of events in
several domains. A few studies have tested the model in youth with
regard to specific stressors such as peer rejection and victimization
(e.g., Panak & Garber, 1992; Prinstein & Aikins, 2004; Prinstein,
Cheah, & Guyer, 2005), trauma (e.g., Gibb & Alloy, 2006), and
academic failure (e.g., Hilsman & Garber, 1995). These studies
found evidence consistent with the cognitive-stress model with
regard to the particular stressors examined, although they did not
contrast different types of stressors within the same study (McMahon, Grant, Compas, Thurm, & Ey, 2003).
Stressors can be categorized as interpersonal and achievement
events. Interpersonal stressors involve interactions with another
person(s) such as conflict, rejection, and break-ups. Achievement
stressors typically involve failure or disappointment in relation to
a goal. The examination of interpersonal stressors is particularly
relevant to cognitive-interpersonal models of depression (e.g.,
Gotlib & Hammen, 1992; Joiner & Coyne, 1999). The integration
of cognitive and interpersonal perspectives takes into account the
social context in which cognitions develop and the special salience
of interpersonal relationships (Rudolph, Hammen, & Burge, 1997;
Rudolph et al., 2000). Indeed, interpersonal stressors have been
found to have a stronger relation to depression than noninterpersonal stressors (e.g., Rudolph & Hammen, 1999a; Rudolph et al.,
2000).
The most commonly used method to assess stressful life events
is with self-report checklists, which usually yield a total count of
events rather than separate scores for interpersonal or noninterpersonal events. Although some self-report checklists and contextual
threat interviews have been found to have overlapping items and to
be similarly associated with depression (Lewinsohn, Rohde, &
Gau, 2003), life events checklists typically do not adequately
measure severe life events or gather information about the timing
of the events (Duggal et al., 2000; Grant & McMahon, 2005;
McQuaid, Monroe, Roberts, Kupfer, & Frank, 2000). In contrast,
contextual threat interviews allow investigators to make objective
ratings of the impact of a stressful event and to date the onset and
offset of an event (Brown & Harris, 1978). These interviews
facilitate the acquisition of detailed information about contextual
factors surrounding the event, the objective impact of the event on
the participant, the duration of the event, and the possible role of
the individual in generating the event.
Several stress interviews have been developed for use with
children and adolescents (e.g., Hammen, 1991; Hankin, Mermelstein, & Roesch, 2007; Williamson et al., 1998). Few studies,
however, have tested the cognitive-stress interaction in adolescents
using such life stress interviews (e.g., Hammen, 1988; Hammen et
al., 1988). Therefore, another aim of the present study was to
examine the cognitive vulnerability model using objective threat
ratings of interpersonal and achievement events in adolescents.
Cognitive Vulnerability
The present study focused on cognitions that cut across the main
cognitive theories of depression and are considered to be central to
the development of depression (Garber, 2007). In particular, we
operationalized the cognitive vulnerability in terms of perceived
self-worth and attributional style because these constructs are
fundamental to several of the leading psychological theories of
depression including Becks (1967) cognitive theory, Browns
self-esteem vulnerability model (Brown, Bifulco, & Harris, 1987;
Southall & Roberts, 2002), and the Helplessness (Abramson et al.,
1978) and Hopelessness theories (Abramson, Metalsky, & Alloy,
1989; Abramson et al., 1978). The cognitive vulnerability hypothesis common to all of these theories is that individuals with
certain maladaptive thinking patterns (e.g., low self-esteem; negative attributional style) are at increased risk for depression when
they experience negative life events because of how they interpret
and respond to those events.
Previous research has shown that various measures of negative
cognitions in children and adolescents are highly correlated
(Abela, Aydin, & Auerbach, 2006; Adams, Abela, & Hankin,
2004) suggesting that they may represent a common underlying
construct that forms a latent factor (Ginsburg et al., 2009; Gotlib,
Lewinsohn, Seeley, Rohde, & Redner, 1993; although see also
Abela, 2001; Conley, Haines, Hilt, & Metalsky, 2001). For example, in a study of several different cognitive inventories, measures
of self-esteem and attributional style, in particular, were found to
load onto the same factor (Adams et al., 2007). In the current
study, we examined the relations between the measures of selfworth and attributional style across multiple assessments, and
created a composite cognitive vulnerability index similar to that
used in other studies testing the cognitive-stress model in both
adults (Alloy et al., 2006) and children (Turner & Cole, 1994).
Whereas several studies (e.g., Abela, 2001; Hankin et al., 2004)
have explicitly compared different measures of the cognitive vulnerability as prescribed by the various cognitive theories (e.g.,
Becks cognitive model vs. the Hopelessness theory), the current
study used a composite index of the cognitive vulnerability to
address a different set of questions regarding specific stressors,
moderators, and measures of depression.
781
Sex Differences
Studies examining sex differences in cognitive vulnerability and
in the diathesis-stress interaction have yielded inconsistent results.
Whereas some investigations have reported no sex differences in
attributional style (Gladstone, Kaslow, Seeley, & Lewinsohn,
1997; Hankin, Abramson, & Siler, 2001; Thompson, Kaslow,
Weiss, & Nolen-Hoeksema, 1998), other studies have shown that
girls have more negative attributional and inferential styles (Hankin & Abramson, 2002), more negative automatic thoughts (Calvete & Cardenoso, 2005), and lower self-esteem (Allgood-Merten,
Lewinsohn, & Hops, 1990) than boys. One study reported a more
negative attributional style for boys (Gladstone et al., 1997).
Several researchers (e.g., Hankin & Abramson, 2001; Hyde,
Mezulis, & Abramson, 2008; Nolen-Hoeksema & Girgus, 1994;
Rudolph et al., 2000) have suggested that the rise in depression
rates among adolescent girls might be partially due to their experiencing higher levels of stress (Compas, Slavin, Wagner, & Vannatta, 1986; Rudolph, 2002; Rudolph & Hammen, 1999a) or their
being more reactive to stress than boys (Little & Garber, 2004;
Rudolph, 2002; Rudolph & Hammen, 1999a; Seiffge-Krenke &
782
Method
Participants
Participants were 240 mothers and children first assessed when
they were in 6th grade (mean age 11.86, SD .57). The sample
was 54.2% female, 82% Caucasian, 14.7% African American, and
3.3% other (Hispanic, Asian, Native American, or mixed ethnic
background). Families were predominantly working (e.g., nurses
aid, sales clerk) to middle class (e.g., store manager, teacher) with
a mean socioeconomic status (Hollingshead, 1975) of 41.84 (SD
13.25).
Procedures
Parents of 5th grade children from metropolitan public schools
were invited to participate in a study about parents and children. A
brief health history questionnaire comprised of 24 medical conditions (e.g., diabetes, heart disease, depression) and 34 medications
(e.g., Prozac, Elavil, Valium) was sent with a letter describing the
study to over 3,500 families. Of the 1,495 mothers who indicated
an interest in participating, the 587 who endorsed either a history
of depressive symptoms, use of antidepressants, or no history of
psychopathology were interviewed further by telephone. The remaining families were excluded because the mother did not indicate depression or indicated other kinds of psychiatric problems
without depression, or had a serious medical illness (e.g., cancer,
multiple sclerosis). Of the 587 families screened, 238 were excluded because they did not indicate sufficient symptoms to meet
criteria for a depressive disorder (38%), had other psychiatric
disorders that did not also include a depressive disorder (19%), the
mother or the target child had a serious medical condition (14%),
the family no longer was interested (21%), the target child was in
the wrong grade (6%), or the family had moved out of the area
(2%). The remaining 349 mothers who indicated that they had a
history of depression or had no psychiatric problems were interviewed in person using the Structured Clinical Interview for DSM
diagnoses (SCID; Spitzer, Williams, Gibbon, & First, 1990).
Based on the SCID, 149 families then were excluded because the
mother indicated a history of a psychiatric diagnosis that did not
also include a mood disorder or reported a serious medical condition, or the child had a serious and/or chronic medical illness or a
pervasive developmental disorder.
The final sample of 240 families consisted of 185 mothers who
had had a depressive disorder (147 mothers had had diagnoses of
Major Depressive Disorder (MDD); the remaining 38 mothers had
had diagnoses of Dysthymia, Depression NOS, or Adjustment
Disorder with Depressed Mood); 55 mothers were life-time free of
psychopathology (low risk). Among the depressed mothers, the
average number of depressive episodes they had had was 2.05
(SD 1.19); 26.7% (n 65) also had had alcohol and/or drug
abuse or dependence, and 27.5% (n 66) had had an anxiety
disorder.
Mothers depressive symptoms were assessed annually with the
Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock,
& Erlbaugh, 1961), which measures affective, cognitive, behav-
Measures
Adolescent depression diagnosis and symptoms.
The
Schedule for Affective Disorders and Schizophrenia for SchoolAge ChildrenEpidemiological versionPresent and Lifetime (KSADS-PL; Kaufman et al., 1997) is a semistructured clinical
interview from which diagnoses of depressive disorders can be
made. At the first assessment (Grade 6), the K-SADS-PL was
administered to mothers and children. Follow-up interviews were
conducted annually using the Longitudinal Interval Follow-up
Evaluation (LIFE; Keller et al., 1987), which parallels the
K-SADS and assesses disorders since the previous interview.
Mothers and children reported separately about the youths depressive symptoms in the K-SADS-PL and LIFE interviews, and were
combined according to the standard procedures described in the
instructions of these interviews (Ambrosini, 2000). The LIFE
yields a depression score on a six-point scale from 1 (no or one
depressive symptom and no impairment) to 6 (meets criteria for a
major depressive episode with marked impairment). A score of 5
indicates a definite MDE. All interviews were audio-taped. A
second rater who was unaware of the scoring of the primary
interviewer reviewed a random 25% of the interview audiotapes.
Kappa (Cohen, 1960) was 0.81 for depressive disorders.
The Childrens Depression Inventory (CDI; Kovacs, 1981,
1985) contains 27 items measuring cognitive, affective, and behavioral symptoms of depression. Each item lists three statements,
scored 0 through 2, in order of increasing symptom severity. The
CDI has adequate internal consistency, testretest reliability, and
convergent validity with other self-report measures (e.g., Abela,
2001; Cole, Hoffman, Tram, & Maxwell, 2000; Smucker, Craig-
783
784
Missing Data
Data were available across Waves as follows: 207 participants
had complete data on all study measures at the assessment conducted at Wave 1, 181 at Wave 2, 158 at Wave 3, 156 at Wave 4,
and 146 at Wave 5. In the DTHM analyses, participants were
censored when they dropped out. MLM analyses allow for variability in how many times participants are measured. In the current
study, all available data from each participant were used to calculate the growth trajectories. Participants with and without missing
data were compared on study variables using ANOVAs on continuous variables and chi-square analyses for categorical variables.
Those who did not complete the study had significantly higher
levels of achievement stress at Wave 1, F 9.96, p .01 and
were more likely to be male than those who completed the study
[2(1) 11.75, p .01]. No other significant differences were
found at any other time point or on any other measure or demographic characteristic. Participants available data were used in
each of the analyses.
Results
Descriptive Statistics
Descriptive statistics and bivariate correlations for all study
variables were computed. Means and standard deviations for the
entire sample for each study variable are presented in Table 1. Data
are also presented separately for males and females and for lowand high-risk participants. One-way ANOVAs were used to test
for sex and risk differences in study variables.
Compared to low-risk youth, high-risk adolescents had significantly more interpersonal stressors, more negative cognitions,
higher depression scores at each wave, and more achievement
stressors at Wave 6. One-way ANOVAs testing for risk group
differences on monthly stress scores showed that high-risk youth
had higher interpersonal stress levels during 43 out of 70 (61%)
months.2 For achievement stressors, significant differences between high- and low-risk adolescents were found for 2 out of 70
months (2%).
Correlations among the study variables are presented in Table 2.
Interpersonal stressors were moderately correlated over time;
achievement stressors showed low stability across time. The negative cognitions composite variable was moderately stable over
time with correlations ranging from .32 to .66; depressive symptoms also were moderately stable across waves (rs .48 to .60).
At most time points, a more negative cognitive style was significantly correlated with higher levels of depressive symptoms.
785
Table 1
Means, Standard Deviations for the Total Sample, and as a Function of Sex and Risk
Sex
Total Sample
Variable
Wave 1
Interpersonal stress
Achievement stress
Negative Cognitionsa
Depressive symptomsb
Wave 2
Interpersonal stress
Achievement stress
Negative cognitionsa
Depressive symptomsb
Wave 3
Interperpersonal stress
Achievement stress
Negative cognitionsa Cognitionsa
Depressive symptomsb
Wave 4
Interpersonal stress
Achievement stress
Negative cognitionsa Cognitionsa
Depressive symptomsb
Wave 5
Interpersonal stress
Achievement stress
Negative cognitionsa Cognitionsa
Depressive symptomsb
a
Females
Risk
Males
Low
ANOVA
High
Sex
Risk
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
2.18
.18
.00
4.76
2.69
.60
.87
4.45
2.10
.11
.08
4.62
2.76
.35
.82
4.60
2.27
.25
.10
4.93
2.61
.80
.92
4.27
2.24
.07
.32
3.08
3.27
.23
.78
3.10
2.19
.21
.09
5.28
2.50
.67
.88
4.67
.22
3.39
2.49
.25
.01
2.04
9.96
9.72
1.75
.16
.00
4.68
2.21
.54
.86
5.06
1.58
.12
.06
4.43
2.09
.53
.88
4.62
1.95
.21
.10
4.98
2.34
.55
.92
5.56
1.53
.24
.26
2.88
1.88
.86
.82
3.14
1.81
.14
.08
5.27
2.30
.40
.85
5.43
1.60
1.75
1.08
.58
.69
1.64
6.08
8.55
1.71
.16
.00
4.88
2.20
.39
.88
5.22
1.75
.15
.10
5.17
2.27
.41
.89
5.81
1.67
.16
.12
4.51
2.12
.37
.87
4.35
1.84
.18
.30
3.57
2.40
.44
.81
3.64
1.67
.15
.09
5.31
2.14
.38
.89
5.58
.08
.02
2.98
.73
.24
.34
6.91
3.94
1.47
.10
.00
5.17
2.08
.26
.83
5.40
1.56
.13
.04
5.58
2.34
.31
.87
5.64
1.35
.07
.05
4.66
1.73
.17
.79
5.08
1.38
.11
.34
3.10
2.01
.33
.79
3.95
1.49
.10
.11
5.88
2.11
.23
.82
5.65
.57
3.38
.46
1.32
.12
.10
9.80
9.92
1.21
.09
.01
5.72
1.72
.33
.88
6.39
1.21
.09
.06
6.32
1.85
.25
.88
6.81
1.20
.08
.07
4.83
1.55
.41
.88
5.65
1.40
.08
.28
4.15
1.67
.23
.69
5.58
1.15
.09
.08
6.31
1.73
.36
.92
6.60
.00
.06
.98
2.44
.84
.03
5.69
4.23
Adolescents negative cognitions composite scores at study entry were tested as a time-invariant predictor in the DTHM.3 Untransformed variables were used in the DTHM analyses because a
person-centered score could not be calculated based on the single
measurement of the cognitive variable at Wave 1. Moreover,
person-centered scores are not appropriate for use with the depression diagnostic outcome variable as individuals are removed from
the analysis once they experience an onset of an event. Finally,
separate models were tested for each stressor type (interpersonal,
achievement). Model-trimming was performed as follows: threeway interactions with sex or risk were examined first, then twoway interactions, and then main effects.
Post hoc probing of interaction effects followed the guidelines
recommended by Aiken, West, and Reno (1991) and clarified by
Holmbeck (2002). For each significant interaction effect, new
variables were created at one standard deviation (SD) above the
mean and one SD below the mean. These new variables were used
to create new conditional moderator variables, and analyses were
rerun using the conditional moderator variable to obtain simple
slopes of the moderation effects. These simple slopes then were
used in the graphical representations of the interactions.
Results.
The hazard function modeled the probability of
participants experiencing a first onset of a depressive episode
[Depression Rating Scale (DSR) score of 5] during the six years
of the study. Seventeen participants were excluded from the
DTHM analyses because they had a previous or current depressive
3
Monthly measures of cognitions were not available, and adolescents
annual cognitions scores could not be divided into discrete monthly periods.
W1 Interpersonal Stress
W1 Achievement Stress
.03
W1 Negative Cognitions
.03
W1 Depressive Symptoms
DeepDepression
.15
W2 Interpersonal Stress
.36
W2 Achievement Stress
.08
W2 Negative Cognitions
.15
W2 Depressive Symptoms
.10
W3 Interpersonal Stress
.18
W3 Achievement Stress
.05
W3 Negative Cognitions
.15
W3 Depressive Symptoms
.10
W4 Interpersonal Stress
.07
W4 Achievement Stress
.08
W4 Negative Cognitions
.11
W4 Depressive Symptoms
.03
W5 Interpersonal Stress
.08
W5 Achievement Stress
.02
W5 Negative Cognitions
.07
W5 Depressive Symptoms
.16
.01
.08
.14
.02
.02
.11
.02
.08
.04
.06
.05
.14
.20
.07
.05
.12
.17
.05
.52
.18
.20
.07
.06
.58
.48
.33
.60
.14
.16
.05
.07
.55
.46
.26
.44
.00
.01
.07
.06
.44
.34
.22
.42
.01
.03
.12
.09
.32
.28
.23
.49
.01
.12
.00
.30
.12
.03
.02
.06
.07
.08
.08
.10
.01
.02
.03
.07
.00
.05
.16
.05
.08
.13
.01
.00
.03
.15
.07
.07
.01
.55
.10
.13
.05
.07
.66
.52
.48
.48
.02
.10
.07
.02
.54
.43
.40
.46
.09
.08
.02
.01
.43
.44
.41
.49
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
1.
2.
3.
4.
Variable
Table 2
Correlations Among Study Variables
.01
.13
.05
.21
.17
.10
.03
.21
.08
.08
.12
.01
.04
.03
.01
.08
.01
.02
.07
.08
.00
10
12
13
14
.46
.01
.01
.00
.05
.07
.11
.62 .03
.18
.61 .58
.06
.08
.28
.57
.51
.07
.04
.09
.52
.04
.03
.36
.04
.10
.02
.03
.04
.11
11
.49
.12
.04
.61
.52
15
17
18
19
20
.06
.00
.14
.46 .04
.04
16
786
CARTER AND GARBER
787
Table 3
Discrete Time Hazard Model Time Specification Results
Deviance difference in comparison to
TIME
n parameters
Deviance
Previous model
General model
AIC
General
Constant
Linear
Quadratic
Cubic
Fourth order
Fifth order
70
1
2
3
4
5
6
890.65
859.53
858.30
850.46
850.39
843.66
842.33
31.12
1.23
7.84
0.07
6.73
1.33
31.12
32.35
40.19
40.26
46.99
48.32
1044.66
861.53
862.30
856.46
858.39
853.66
854.33
Note.
AIC Akaike Information Criterion. Deviance statistics in bold show significant changes in chi-square.
for the quadratic model was lower than the cubic model. Thus, the
most parsimonious specification for time was the quadratic model,
which included the month and the squared month variables.
The first series of models tested the three-way interactions to
address whether sex or risk moderated the effects of stress and
negative cognitions on the onset of a depressive episode. The
hazard function then was modeled as a function of a set of control
and substantive predictors. Similar analyses were run separately
for each stressor type. The main effects of sex, Wave 1 cognitions,
and mean monthly stress ratings were included as were the twoway interactions between stress and cognitions, stress and sex, and
cognitions and sex, and the three-way interaction among stress, cognitions, and sex. Parallel analyses were conducted to test risk as a
moderator of the cognitive-stress interaction, and included the threeway and two-way interactions among stressors, cognitions, and risk.
The model with both the three- and two-way interactions was
not significant for interpersonal stressors and did not converge for
achievement stressors. The next series of models included the
two-way interactions between stressor type and the negative cognitions composite. The interaction between interpersonal stressors
and negative cognitions was significant (see top of Table 4), over
and above the significant effect of risk. Simple slope analyses of
the interaction (see Figure 1) indicated that the first onset of a
major depressive episode (MDE) was predicted by high negative
cognitions in the context of low interpersonal stress (Odds 1.14,
p .01, CI 1.04 1.25, Hazard .53), as well as by high levels
of interpersonal stressors at both high (Odds .51, p .05, CI
.27.97, Hazard .35) and low levels of negative cognitions
(Odds .53, p .05, CI .30.97, Hazard .34).
The interaction of negative cognitions with achievement stressors was not significant. Therefore, we next tested the most parsimonious model for achievement stressors and found significant
main effects of time, risk, stressors, and negative cognitions in
predicting the hazard function (see bottom of Table 4). High-risk
youth had significantly increased odds of developing depression
over the course of the study. Additionally, the negative cognitive
composite and achievement stressors each significantly predicted
depression onset.4 For every one unit increase in the mean monthly
achievement stress score, the odds of developing a depressive
disorder increased by 1.24. Sex was not a significant predictor or
moderator of the onset of an MDE in either the interpersonal or
achievement stressor models.
788
Table 4
Discrete Time Hazard Model Results: Two-Way Interactions of Stress and Negative Cognitions, and Main Effects of Stress and
Negative Cognitions, Controlling for Risk
Two-way interactions
Interpersonal
Achievement
Stressor type
Odds
CI (95%)
Wald
Hazard
Odds
CI (95%)
Wald
Hazard
Intercept
Risk
Month
Month Month
Stress
Negative Cognitionsa
Stress Neg Cognitions
4.66
3.66
.75
1.00
.65
.04
1.03
1.1012.22
.70.82
1.001.00
.28.97
.971.12
1.001.06
36.82
4.44
47.04
42.41
4.29
1.30
3.88
.79
.43
.50
.34
.51
.51
5.17
2.82
.74
1.00
2.51
1.08
.94
.849.43
.68.81
1.001.00
.867.30
1.021.15
.861.04
48.53
2.83
46.54
41.27
2.83
6.28
1.37
.74
.43
.50
.71
.52
.49
1.0011.82
.68.81
1.001.00
1.071.45
1.011.15
33.87
3.88
46.10
42.18
7.95
5.19
.77
.43
.50
.55
.52
Intercept
Risk
Month
Month Month
Stress
Negative Cognitions
.01
3.45
.74
1.00
1.24
1.08
60
50
40
Depressive
Episode Hazard 30
Function
20
10
0
Low
High
Interpersonal Stressors
The Y-axis includes negative values because the scores were centered
around each individuals mean scores over the course of the study.
789
Table 5
Multi-Level Model Results of the 2-Way Interaction Between Stressor Type and Negative
Cognitions Predicting Depressive Symptoms
Two-way interactions
Interpersonal
Stressor type
Initial status
Level 1
Age
Prior year Depressive Symptoms (CDI)
Stress
Negative Cognitions
Stress Negative Cognitions
Level 2
Risk
Person-Mean Stress
Person-Mean Negative Cognitions
Achievement
t
2.46
.01
.01
.10
2.36
1.81
1.91
4.38
2.48
3.46
2.19
.15
.19
.02
.84
1.06
1.43
4.38
.05
3.14
1.34
.05
.10
.50
.02
.40
.07
1.88
.16
.19
.19
.92
.31
.06
.58
.38
not significant.6 With regard to achievement stress, analyses revealed significant main effects for previous years depression
scores (B .19, p .001) and for negative cognitions (B .84,
p .01); the interaction between achievement stress and negative
cognitions was not significant (B 1.06, SE .79, ns), however.
To allow for more directs comparisons to other studies in the
literature (e.g., Abela & McGirr, 2007; Shih, 2006), we conducted
additional analyses using alternative methods of specifying the
time-varying predictors. Because the choice of centering methods
potentially provides different information about the effects of
time-varying predictors, we also tested grand-mean and personmean centered variables, Time 1-centered variables, and variables
that were not centered at all (Singer & Willett, 2003). The method
followed by Abela and Skitch (2007) in which only stressors, but
not cognitions or depressive symptoms were person-centered also
was tested. Using these alternative methods of variable specification, however, did not yield significant interaction effects.
1.5
Depressive
Symptoms
0.5
-0.5
-1
-1.5
Discussion
Several important findings emerged from this 6-year longitudinal study of predictors of both the first onset of a major depressive
episode (MDE) and changes in depressive symptoms during adolescence. First, evidence consistent with the cognitive-stress interaction model (Abramson et al., 1978, 1989; Beck, 1967; Brown &
Harris, 1978) was found using two different data analytic approaches (i.e., DTHM and MLM) and two different outcomes (i.e.,
onset of an MDE and changes in depressive symptoms). That is,
the relation between interpersonal stressors and depression varied
significantly as a function of level of negative cognitions. The first
onset of an MDE was predicted by either high negative cognitions
in the context of low interpersonal stress or high interpersonal
stress regardless of level of negative cognitions. Under conditions
of both low interpersonal stress and low negative cognitions,
6
Low
High
Interpersonal Stress
790
Stressors
The current investigation tested the cognitive-stress model separately for interpersonal and achievement stressors because not all
types of life events are similarly related to depression (Monroe &
Simons, 1991). Both theory (e.g., Gotlib & Hammen, 1992; Joiner
& Coyne, 1999) and empirical studies have highlighted the special
salience of social stressors (e.g., romantic break-ups) with regard
to depressive disorders (Rudolph et al., 2000) and depressive
symptoms (Rudolph & Hammen, 1999b). Consistent with this
literature, we found that interpersonal stressors, in particular, interacted with negative cognitions to predict both the first onset of
an MDE and changes in depressive symptoms. Interpersonal stressors are more common during adolescence and are associated with
earlier pubertal timing in both males and females (Rudolph, 2008).
Thus, one explanation for the rise in depressive symptoms and
diagnoses in adolescence may be an increase in the experience of
interpersonal stressors during this developmental period, particularly for girls (Natsuaki et al., 2009; Rudolph, 2002). In contrast,
there was a significant main effect of achievement stressors predicting the first onset of an MDE, but no interaction with negative
cognitions. The smaller number of achievement as compared to
interpersonal stressors reported by participants quite possibly resulted in a more restricted range of achievement events, which
could have reduced our power to detect the interaction.
Moreover, stronger support for the cognitive-stress model might
have been found if we had tested the specific vulnerability hypothesis (e.g., Beck, 1983; Coyne & Whiffen, 1995). According to this
perspective, when there is a match between the content of an
individuals cognitive style and the domain of a stressful event
(e.g., interpersonal, achievement), depression is especially likely to
occur. For example, individuals who believe that they have to be
perfect likely will become very distressed when they fail to accomplish a desired goal; persons who highly value being liked and
accepted by others are at increased risk for depression if they
encounter an interpersonal stressor such as rejection.
In addition, certain events that were objectively classified by
independent raters as being interpersonal or achievement-focused
may have been experienced quite differently by some participants.
For example, receiving a bad grade on an exam typically would be
classified as an achievement stressor, but for some youth a bad
grade might be an interpersonal stressor because for them, the
salient event is their parents disapproval. Without assessing the
subjective meaning of events for participants, we cannot be certain
that objectively rated life events reflect participants actual experiences.
791
Sex Differences
Sex differences in levels of study variables or in the relations
among them were not observed in the current study. These results
are consistent with other studies that have not found sex differ-
792
Limitations
Limitations of the current study provide directions for future
research. First, both cognitive vulnerability and depressive symptoms were assessed with self-report measures. In contrast, depressive disorders were diagnosed by clinicians and contextual stress
levels were rated by independent judges based on separate interviews with the adolescent and mother. Thus, although the relation
between negative cognitions and depressive symptoms could have
been inflated due to the use of a common assessment procedure
(i.e., self-report), the link between stressors and depressive disorders is less likely to have been the result of such shared method
variance. Alternative ways of assessing cognitions include
laboratory-based tasks, such as the attentional dot-probe and incidental recall tasks (Garber & Kaminski, 2000), have shown promise for examining cognitive vulnerability in the context of stress
(Jacobs, Reinecke, Gollan, & Kane, 2008).
Second, both a strength and a limitation of the current study was
that participants were recruited so as to oversample offspring of
mothers with histories of depression (high risk). This strategy was
used in order to have greater variability on the measures of interest
(e.g., stress, cognitions, and depression). Interestingly, over and
above the potent predictor of maternal depression, the cognitivestress interaction was still significant indicating that more variance
needed to be explained beyond risk. The results of the current
study, however, may not generalize to a purely community sample
or to offspring of depressed fathers (Connell & Goodman, 2002;
Kane & Garber, 2004).
Third, as with any longitudinal study, some participants were
lost over the six years. Attrition analyses revealed that those who
Clinical Implications
An important clinical implication of the current study is that
interventions that teach youth strategies for coping with negative
life events, particularly interpersonal stressors, may be especially
useful in treating and preventing depression. Indeed, interventions
that emphasize interpersonal communication (Young, Mufson, &
Davies, 2006), social skills (Gillham, Hamilton, Freres, Patton, &
Gallop, 2006; Reinecke, Ryan, & DuBois, 1998), and coping
(Compas et al., 2009) have been found to be efficacious in the
treatment and prevention of depression in youth. Approaches that
integrate both cognitive and interpersonal strategies may be especially effective and therefore should be the focus of future intervention efforts (Garber, 2006; Jaycox, Reivich, Gillham, & Seligman, 1994). Finally, results of the current study highlight the need
for the continued development and dissemination of depression
prevention programs that target cognitive restructuring and coping
with stress, particularly in offspring of depressed parents (e.g.,
Compas et al., 2009; Garber et al., 2009).
References
Abela, J. R. Z. (2001). The hopelessness theory of depression: A test of the
diathesis-stress and causal mediation components in third and seventh
grade children. Journal of Abnormal Child Psychology, 29, 241254.
doi:10.1023/A:1010333815728
Abela, J. R. Z., Aydin, C., & Auerbach, R. P. (2006). Operationalizing the
vulnerability and stress components of the hopelessness theory of
depression: A multi-wave longitudinal study. Behavior Research and
Therapy, 44, 15651583. doi:10.1016/j.brat.2005.11.010
793
794
795
796
Daley, S. E. (2000). Toward an interpersonal life-stress model of depression: The developmental context of stress generation. Development
and Psychopathology, 12, 215234. doi:10.1017/S0954579400002066
Ruscio, J., & Ruscio, A. M. (2000). Informing the continuity controversy:
A taxometric analysis of depression. Journal of Abnormal Psychology,
109, 473 487. doi:10.1037/0021-843X.109.3.473
Sandberg, S., Rutter, M., Pickles, A., McGuinness, D., & Angold, A.
(2001). Do high-threat life events really provoke the onset of psychiatric
disorder in children? Journal of Child Psychology and Psychiatry, 42,
523532. doi:10.1111/1469-7610.00746
Seiffge-Krenke, I., & Stemmler, M. (2002). Factors contributing to gender
differences in depressive symptoms: A test of three developmental
models. Journal of Youth and Adolescence, 31, 405 417. doi:10.1023/
A:1020269918957
Seligman, M. E. P., Peterson, C., Kaslow, N. J., Tanenbaum, R. L., Alloy,
L. B., & Abramson, L. Y. (1984). Attributional style and depressive
symptoms among children. Journal of Abnormal Psychology, 93, 235
238. doi:10.1037/0021-843X.93.2.235
Shih, J. H. (2006). Sex differences in stress generation: An examination of
sociotropy/autonomy, stress, and depressive symptoms. Personality and
Social Psychology Bulletin, 32, 434. doi:10.1177/0146167205282739
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis:
Modeling event change and event occurrence. New York, NY: Oxford
University Press.
Smucker, M. R., Craighead, W. E., Craighead, L. W., & Green, B. J.
(1986). Normative and reliability data for the Childrens Depression
Inventory. Journal of Abnormal Child Psychology, 14, 2539. doi:
10.1007/BF00917219
Southall, D., & Roberts, J. E. (2002). Attributional style and self-esteem in
vulnerability to adolescent depressive symptoms following life stress: A
14-week prospective study. Cognitive Therapy and Research, 26, 563
579. doi:10.1023/A:1020457508320
Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M. B. (1990).
Users Guide for the Structured Clinical Interview for DSMIIIR.
Washington, DC: American Psychiatric Press.
Thompson, M., Kaslow, N. J., Weiss, B., & Nolen-Hoeksema, S. (1998).
Childrens Attributional Style Questionnaire Revised: Psychometric
examination. Psychological Assessment, 10, 166 170. doi:10.1037/
1040-3590.10.2.166
Tompson, M., Pierre, C., Boger, K., McKowen, J., Chan, P., & Freed, R.
(2009). Maternal depression, maternal expressed emotion, and youth
psychopathology. Journal of Abnormal Child Psychology, 38, 105117.
doi:10.1007/s10802-009-9349-6
Turner, J. E., & Cole, D. A. (1994). Development differences in cognitive
diatheses for child depression. Journal of Abnormal Child Psychology,
22, 1532. doi:10.1007/BF02169254
Williamson, D. E., Birmaher, B., Frank, E., Anderson, B. P., Matty, M. K.,
& Kupfer, D. J. (1998). Nature of life events and difficulties in depressed
adolescents. Journal of the American Academy of Child & Adolescent
Psychiatry, 37, 1049 1057. doi:10.1097/00004583-199810000-00015
Young, J. F., Mufson, L., & Davies, M. (2006). Efficacy of interpersonal
psychotherapy-adolescent skills training: An indicated preventive intervention for depression. Journal of Child Psychology and Psychiatry, 47,
1254 1262.
Zuroff, D. C., Mongrain, M., & Santor, D. A. (2004). Conceptualizing and
measuring personality vulnerability to depression: Comment on Coyne
and Whiffen (1995). Psychological Bulletin, 130, 489 511.
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.