Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2012 Jun 19.
Published in final edited form as: Int J Cogn Ther. 2010 Dec 1;3(4):368–379. doi: 10.1521/ijct.2010.3.4.368

Specificity of Stress Generation: A Comparison of Adolescents with Depressive, Anxiety, and Comorbid Diagnoses

Nicole P Connolly 1, Nicole K Eberhart 2,4, Constance L Hammen 1, Patricia A Brennan 3
PMCID: PMC3378060  NIHMSID: NIHMS382002  PMID: 22724042

Abstract

Individuals with a history of depression experience more stress that is dependent in part on their own actions. However, it is unclear whether stress generation is a unique feature of depression, or a universal process that is also present in other types of psychopathology, such as anxiety disorders. The current study addressed this issue by comparing adolescents with a history of “pure” (i.e., non-comorbid) depressive disorders, pure anxiety disorders, comorbid depression and anxiety, and no disorder, on their levels of dependent and independent stress. Results indicated that adolescents with pure depression experienced more dependent stress than adolescents with pure anxiety, and adolescents with any internalizing diagnosis experienced more dependent stress than controls. Further, adolescents with comorbid depression and anxiety reported the highest levels of stress generation. The results suggest that while stress generation may be more strongly associated with depression than anxiety in adolescence, it is not unique to depression.

Keywords: stress generation, depression, anxiety, comorbidity

Decades of research have provided strong evidence that stress is associated with vulnerability to depression (e.g., Paykel, 2003). More recently, a growing body of research has focused on the opposite direction of causality, examining depressed individuals' tendency to contribute to the occurrence of stressors in their lives, a process labeled stress generation (Hammen, 1991). In particular, Hammen (1991) found that women with a history of major depressive disorder tend to engage in situations that create increased levels of acute stressors in their lives, particularly those with interpersonal content, as compared to women with bipolar and medical diagnoses. While this association between self-generated stressors and depression has been replicated in samples of both adolescents and adults (e.g., Chun, Cronkite, & Moos, 2004; Cui & Vaillant, 1997; Davila, Hammen, Burge, Paley, & Daley, 1995; Harkness & Luther, 2001; Harkness, Monroe, Simons, & Thase, 1999; see Hammen, 2005, for a review), little research has examined whether this process is a unique feature of depressive disorders, or a more universal process that is also present in other types of psychopathology, particularly anxiety disorders.

Anxiety disorders are of particular interest given the high degree of overlap between anxiety and depressive diagnoses. Estimates of comorbidity between anxiety disorders and depressive disorders, particularly among children and adolescents, range from 16% to 75% (Angold & Costello, 1993; Brady & Kendall, 1992) depending on the sample and the methods used to assign diagnoses. Because of these high rates of co-occurrence, researchers have examined numerous similarities and differences between anxiety and depressive disorders, in an attempt to explain both the high rates of comorbidity and the distinct symptom profiles that characterize the different disorders. However, surprisingly little research has examined whether stress generation is a common factor in the two families of disorders, or perhaps a specific facet of depression that helps distinguish between the disorders.

Only two recent studies have directly addressed whether the stress generation hypothesis could account for some of the differences between depressive and anxiety disorders. In a daily diary study of college students, Joiner, Wingate, Gencoz, & Gencoz (2005) found some evidence that increased occurrence of stressors was specifically associated with depressed mood. They found that depressive symptoms predicted increases in the total number of life events reported, whereas anxiety symptoms did not predict changes in reports of stressors among young adults. Uhrlass and Gibb (2007) also examined specificity of the stress generation hypothesis in a college student sample, utilizing weekly assessments. Similar to Joiner et al. (2005), Uhrlass and Gibb found that higher levels of depressive symptoms predicted a higher rate of reporting negative life events throughout the course of the study compared to anxious symptoms.

While these two studies suggest a stronger link between depression and stressful events than between anxiety and stressful events, interpretation of the results is limited by several factors in the respective studies. First, both studies relied on self-report measures of stressors, which may artificially restrict the range of life experiences reported. Further, the studies presented a difference in total reported stressors but did not distinguish between stressors that are dependent in part on the actions of the individual, versus stressors that are more independent of the individual’s actions. It is unclear whether the difference reflects a true increase in the creation of stressful situations among individuals with a higher level of depressed mood, or whether it mainly reflects a reporting bias in which current symptoms are associated with higher self-reports of stress. Additionally, both studies relied upon questionnaire-based assessments of depressed and anxious mood, also limiting the reach of their findings.

The present study seeks to address these limitations by examining patterns of stress generation in a high-risk sample of over 800 Australian youth, who were examined at age 15. Specifically, semi-structured life stress and diagnostic interviews were used to examine the levels of stress generation experienced by individuals with lifetime histories of depressive disorders, anxiety disorders, and comorbid depressive and anxiety disorders, as compared to healthy controls. The main hypothesis was that adolescents with a history of only depressive disorders would report higher levels of recent stressful life events that are at least in part dependent on their own actions (i.e., stress generation), as compared to adolescents with only anxiety disorders. This effect was expected to be particularly salient for interpersonal stressors, given that stress generation findings have been strongest for relationship stress (e.g., Hammen, 1991; Daley et al., 1997). However, individuals with a history of only depression or anxiety diagnoses were not expected to differ in the levels of independent (i.e., fateful) stress they had experienced, in line with previous research (e.g., Hammen, 1991). Further, based on previous evidence that stress generation is more elevated among depressed women with comorbid diagnoses (Daley et al., 1997; Harkness & Luther, 2001), it was hypothesized that adolescents with comorbid depression and anxiety diagnoses would experience more dependent stress than individuals with “pure” depression or anxiety alone. Finally, it was hypothesized that individuals with any internalizing diagnosis would engage in elevated levels of stress generation in comparison to healthy controls.

Method

Participants

Youth were assessed as part of an ongoing longitudinal study of adolescents and their families in Queensland, Australia. The sample was drawn from participants in the Mater University Study of Pregnancy (MUSP), which examined a birth cohort of 7,775 children born at the Mater Misericordiae Mothers’ Hospital in Brisbane between 1981 and 1984. The current study examined a subset of these participants at age 15.

A sample of 815 adolescents, 412 males (50.6%) and 403 females (49.4%), was selected by investigators Patricia Brennan (Emory University) and Constance Hammen (UCLA) for follow-up at age 15 in a study of children at risk due to maternal depression. The Delusion-States Symptom Inventory (DSSI; Bedford, Foulds, & Sheffield, 1976), a measure of anxious and depressive symptomatology administered to mothers during the initial phases of the original study, was used to select mothers with varying levels of severity and chronicity of depressive symptoms (including no depression). Maternal depression status was verified by diagnostic interview (SCID). Approximately 45% of the current sample of mothers had a diagnosis of major depressive disorder or dysthymic disorder during the child’s lifetime.

At the age 15 wave of the study, the mean age of the adolescents was 14.9 years old (SD = .37). According to the mothers’ reports of ethnicity, 729 (89.3%) of the adolescents were Caucasian, 8 (1.0%) were Asian, 6 (0.7%) were Maori or Aborigine, and 51 (6.2%) were of mixed descent. Overall, the families in this sample represented a lower middle class to lower class SES. Of the mothers, 575 (70.4%) reported completing grade 10 or less.

Procedure

Interviewers conducted 3.5-hour in-home interviews for both the mothers and the target adolescents. After obtaining consent/assent from the participants, trained clinicians administered semi-structured diagnostic interviews and semi-structured chronic strain and stressful life event interviews privately to both the adolescents and their parents. Participants then completed a series of self-report questionnaires on laptop computers.

Measures

Diagnostic Evaluations. The

Schedule for Affective Disorders and Schizophrenia in School-Aged Children (K-SADS-E; Orvaschel, 1995) was administered during the age 15 assessment to establish current and lifetime diagnoses of Axis I psychiatric disorders for the youth. The K-SADS-E is a widely used semi-structured interview for diagnosing current presence or past history of Axis I psychiatric disorders for children and adolescents. Trained clinicians administered the interview separately to both the adolescents and their mothers. Diagnoses were “best estimates” based on both sources of information and any additional information available.

To determine interrater reliabilities in the current sample, 75 of the KSADS-E interview tapes were randomly selected for rating by a second clinician. Overall, the weighted kappa for depressive disorders was .82 for current disorders and .73 for past disorders. The weighted kappa for anxiety disorders was .76 for current disorders and .79 for past disorders.

Depression Severity

Depressive symptoms, which were examined in secondary analyses, were measured using the Beck Depression Inventory (BDI; Beck, 1987). This widely-used measure has been shown to be reliable and valid in a number of samples (e.g., Beck, Steer, & Garbin, 1988). Cronbach's alpha was .86 in the present sample.

Life Stress Interview

The UCLA Life Stress Interview (e.g. Hammen, Ellicott, Gitlin, & Jamison, 1989; Hammen 1991) was used to assess episodic (i.e., acute) stressors participants had experienced in the past 12 months. This semi-structured interview is based on contextual threat methods of stress assessment proposed by Brown & Harris (1978), in which the context surrounding the event is taken into consideration when evaluating the event’s objective impact. Interviewers probed the adolescents for details of stressful events across several domains. Afterward, a team of independent raters blind to the subject’s diagnostic status reviewed each event to determine the level of impact and the degree to which the event was dependent upon actions by the subject. The level of impact was rated on a 5-point scale, with 1 representing little to no impact on the individual and 5 representing extreme adversity. The level of dependence was also rated on a 5-point scale, with 1 representing an event that was completely independent of the subject’s actions (e.g. a family member passing away), 3 indicating partial dependence on the subject, and 5 indicating complete dependence on the subject’s behavior(e.g., damaged the car while driving under the influence). Dependency was dichotomized such that scores of 3 or higher indicate a dependent event and scores below 3 indicate independence. Raters also assessed whether each event was interpersonal in nature. These categorizations were used to sum life event ratings in a given domain in order to create composites of total dependent (i.e., self-generated) non-interpersonal and interpersonal stress.

A group of 89 cases were selected at random and both Australian and US teams independently rated episodic life events to determine interrater reliability. Both severity and independence ratings showed high interrater reliability with severity ratings having an ICC of .92 and independence ratings having an ICC of .89.

Results

Four groups were formed based on diagnostic history through age 15. The depressive disorder group consisted of 61 adolescents (7.5%) who reported a lifetime history of major depressive disorder and/or dysthymic disorder and no other disorders. The anxiety disorder group consisted of 61 youth (7.5%) who reported a lifetime history of generalized anxiety disorder, obsessive compulsive disorder, separation anxiety disorder, posttraumatic stress disorder, panic disorder, social phobia, or specific phobia and no other disorders. The comorbid group consisted of 20 adolescents (2.5%) who met criteria for a lifetime history of both a depressive disorder and an anxiety disorder but no other disorders. The control group consisted of 568 adolescents (69.7%) who had no history of psychological diagnoses. Adolescents who had a lifetime history of attention deficit/hyperactivity disorder, oppositional defiant disorder, substance abuse disorder, or eating disorder (N = 105, 12.9% of the sample) were excluded from the analyses to control for potential confounding effects of other disorders or comorbidities.

To examine group differences in stress generation while controlling for potential confounding variables, a series of hierarchical regression analyses were conducted. The different stress variables (e.g., independent stress, dependent stress) were examined as dependent variables in separate regression equations. On the first step of each regression analysis, dummy-coded variables for gender and maternal history of depressive disorders were entered as controls, given their significant correlations with the various measures examined. On the second step of the regression analyses, contrast codes for the four diagnostic groups were entered as predictors. The four diagnostic groups were examined simultaneously in order to conduct an omnibus test of diagnostic group differences in each type of stress examined, while the orthogonal contrast codes enabled testing of specific planned comparisons of interest. The codes compared levels of stress experienced by 1) the “pure” depression versus the “pure” anxiety only group, 2) the control group versus the average across the other three groups with internalizing disorders, and 3) the comorbid group versus the average across the “pure” depression and anxiety groups. Table 1 presents group means for each type of stress.

Table 1.

Descriptive Statistics for Measures of Stress Across the Four Diagnostic Groups

Diagnostic Group
Variable Control (N = 568) Depressed (N = 61) Anxious (N = 61) Comorbid (N = 20) Total
Independent Stress M = 2.86 (SD = 2.59) M = 3.91 (SD = 3.07) M = 3.14 (SD = 2.79) M = 4.21 (SD = 4.71) M = 3.04 (SD = 2.78)
Dependent Stress M = 2.56 (SD = 2.37) M = 3.55 (SD = 2.92) M = 2.49 (SD = 2.30) M = 5.00 (SD = 4.29) M = 2.70 (SD = 2.52)
 Non-Interpersonal Stress M = 1.02 (SD = 1.55) M = 1.42 (SD = 1.89) M = 0.98 (SD = 1.45) M = 2.00 (SD = 2.02) M = 1.09 (SD = 1.61)
 Interpersonal Stress M = 1.72 (SD = 1.95) M = 2.59 (SD = 2.36) M = 1.70 (SD = 2.04) M = 3.23 (SD = 2.97) M = 1.86(SD = 2.07)
Open in a new tab

Independent Stress

First, adolescents in the pure depressed, pure anxious, comorbid, and control groups were compared on their levels of total independent stress (i.e., the sum of severity ratings for stressful life events judged to be independent of the individual's actions). Results are summarized in Table 2. There was a small, statistically significant predictive relationship between diagnostic group status and level of independent stress, F(3,704) = 2.93, p = 0.033. When looking at differences between diagnostic groups, adolescents in the control group reported significantly less stress from independent life events than adolescents across the three diagnostic groups, p = 0.011. Although the pure depressed group appeared to have experienced higher levels of independent stress than the pure anxious group, no statistically significant difference among these diagnostic groups was detected, ns.

Table 2.

Prediction of Age 15 Stressors from Demographic Factors and Diagnostic Group

β t p
DV: Total Independent Stress
Step 1 – Demographics <0.001***
 Gender 0.17 4.50 <0.001***
 Maternal Depression History 0.08 2.18 0.029*
Step 2 – Diagnostic Group 0.033*
 Depression vs. Anxiety 0.04 1.15 0.252
 Control vs. Internalizing Diagnosis −0.11 −2.55 0.011*
 Comorbid vs. Pure Disorder 0.01 0.22 0.829
DV: Total Dependent Stress
Step 1 – Demographics <0.001***
 Gender 0.15 3.94 <0.001***
 Maternal Depression History 0.05 1.31 0.189
Step 2 – Diagnostic Group <0.001***
 Depression vs. Anxiety 0.08 2.19 0.029*
 Control vs. Internalizing Diagnosis −0.16 −3.73 <0.001***
 Comorbid vs. Pure Disorder 0.13 3.25 0.001**
DV: Interpersonal Dependent Stress
Step 1 – Demographics <0.001***
 Gender 0.17 4.54 <0.001***
 Maternal Depression History 0.04 1.04 0.300
Step 2 – Diagnostic Group 0.001**
 Depression vs. Anxiety 0.06 1.49 0.137
 Control vs. Internalizing Diagnosis −0.13 −3.14 0.002**
 Comorbid vs. Pure Disorder 0.13 3.18 0.002**
DV: Non-Interpersonal Dependent Stress
Step 1 – Demographics 0.488
 Gender 0.03 0.78 0.438
 Maternal Depression History 0.03 0.90 0.367
Step 2 – Diagnostic Group 0.039*
 Depression vs. Anxiety 0.07 1.80 0.072†
 Control vs. Internalizing Diagnosis −0.10 −2.25 0.025*
 Comorbid vs. Pure Disorder 0.06 1.38 0.170
Open in a new tab

Note.

†

p < .10,

*

p < 05,

**

p < .01, p< .001.

Dependent Stress

To examine stress generation hypotheses, the groups were then compared on their levels of total dependent stress (i.e., the sum of severity ratings for events judged to be at least partially a result of the subject’s own actions). Results indicated that diagnostic group membership significantly predicted level of total dependent stress, F(3,704) = 7.24, p < 0.001. Adolescents with a pure depressive disorder reported more stress from dependent events than adolescents with a pure anxiety disorder, p = 0.029. Adolescents in the control group reported less dependent stress than adolescents in the internalizing disorder groups, p < 0.001. Adolescents in the comorbid group reported more dependent stress than adolescents in either of the pure internalizing disorder groups, p = 0.001.

Dependent, interpersonal stress

To explore whether levels of certain types of dependent stressful events differ across diagnostic groups, dependent events were classified as either being interpersonal or non-interpersonal in nature. Results suggested that there were significant differences between groups in total dependent interpersonal stress, F(3,704) = 5.32, p = 0.001. Adolescents with comorbid depressive and anxiety disorders had higher levels of dependent interpersonal stress than adolescents with pure anxiety or depressive disorders, p = 0.002. Although the pure depressed group appeared to have experienced more dependent interpersonal stress than the pure anxious group, these group differences were not statistically significant. However, there was evidence that adolescents with any internalizing disorder reported more dependent interpersonal stress than adolescents with no diagnosis, p = 0.002.

Dependent, non-interpersonal stress

Groups were also compared on their total severity ratings across events classified as dependent on the individual’s actions and non-interpersonal in nature. Overall, diagnostic status significantly predicted level of dependent, non-interpersonal stress, F(3,704) = 2.80, p = 0.039. Adolescents in the control group reported less dependent, non-interpersonal stress than adolescents in the three diagnostic groups, p = 0.025. Adolescents in the depressive disorder group demonstrated a trend toward reporting higher stress from dependent non-interpersonal events as compared to individuals in the anxiety disorder group, p = 0.072.

Post-hoc tests

Additional analyses were conducted in order to examine potential alternative explanations for the finding of greater dependent and specifically dependent interpersonal stress in the comorbid versus internalizing groups. Given that the depressed group reported greater dependent stress than the anxious group, we examined whether the higher level of stress experienced by the depressed group differed from the comorbid group. Analysis of covariance (ANCOVA) controlling for maternal depression and gender confirmed that the comorbid group experienced marginally greater dependent stress, F(1,77) = 3.21, p = 0.077, and in particular, significantly greater dependent interpersonal stress, F(1,77) = 5.02, p = 0.028, as compared to the depressed group. Further, the depressed and comorbid groups did not differ in their levels of depression severity at the time of the assessment, F(1,76) = 0.98, p = 0.324.

Gender Differences

Because rates of stress generation have been demonstrated to significantly differ between genders, interactions between gender and diagnostic group were entered on a subsequent step in each of the analyses to determine if gender may moderate the relationship between diagnostic group and reported stress. Overall, no significant interaction effect was found between diagnostic group and gender in the prediction of total independent stress, p > 0.050, ns. However, gender did appear to significantly moderate the relationship between diagnostic group and total dependent stress, p < 0.001. The interaction effect is pictured in Figure 1. Overall, it appeared that the interaction effect was primarily driven by the difference in total dependent stress between males and females in the comorbid anxiety and depressive disorder group, with females with a comorbid disorder reporting significantly more dependent stress than males with comorbid disorders. Similar patterns were noted when dependent stress was broken down into interpersonal dependent stress, p = 0.001, and non-interpersonal dependent stress, p = 0.036.

Figure 1.

Figure 1

Open in a new tab

Effect of gender and diagnostic group on total dependent stress.

Discussion

The primary goal of the present study was to explore whether adolescents with a history of a depressive disorder would differ from adolescents with a history of an anxiety disorder in the amount of stress that they experience, particularly stressors that they play a role in generating. Overall, results suggested that adolescents with a history of a “pure” (i.e., non-comorbid) depressive disorder reported significantly more dependent stress (i.e., events in which their own actions played a significant role in causing the stressor) at age 15 than adolescents with pure anxiety disorders. This is consistent with Hammen’s (1991) theory of the role of stress generation in depressive disorders, namely that depressed individuals tend to contribute to the occurrence of stressors in their own lives, which in turn contribute to continued depression.

Adolescents in the comorbid depressive and anxiety disorder group did not significantly differ from adolescents with pure depressive or anxiety disorders in their levels of total independent stress that is outside their control. However, adolescents with comorbid internalizing disorders did display significantly more dependent stress generation than adolescents in the pure disorder groups, particularly the depression only group, and this result did not appear to be driven by a difference in depression severity alone. Adolescents with comorbid diagnoses were particularly likely to contribute to stress in their relationships with others. These results replicate Daley et al.'s (1997) finding that young women with depression comorbid with a non-depressive disorder engage in higher levels of interpersonal stress generation than individuals with depression alone, and Harkness and Luther's (2001) finding that depressed women who simultaneously had comorbid diagnoses of both dysthymia and anxiety experienced higher levels of dependent stress than women who had depression alone or only comorbid dysthymia or anxiety The current study expands upon this earlier research on adult women by providing evidence that comorbid individuals' greater susceptibility to stress generation emerges by age 15. Further, while Daley et al. did not distinguish between different comorbid diagnoses, and Harkness and Luther only found increased stress generation risk for depressed individuals who simultaneously had both dysthymic and anxiety disorders as comorbid diagnoses, the current study provides evidence that a tendency to contribute to dependent stressors may occur in comorbid anxiety without additional comorbid diagnoses.

The study also expanded upon previous research on clinical risk factors for stress generation by examining gender differences in the relationship between diagnosis and stress generation. The study found preliminary evidence that comorbid diagnoses may be associated with higher levels of stress generation among adolescent girls as compared to boys. These results expand upon previous evidence that girls have higher rates of stress generation than boys (e.g., Shih, Eberhart, Hammen, & Brennan, 2006; and see Shih & Auerbach, this issue). However, the current study's results should be viewed as tentative, as the comorbid group consisted of just 6 males and 14 females. Future research should examine whether this result holds up in a larger sample of comorbid youth.

Overall, findings suggest that while anxiety and depressive disorders do share a number of similar features and may have some common genetic underpinnings (e.g., see Middeldorp, Cath, Van Dyck, Boomsma, 2005, for a review), there are also differences in the role of stress in these different classes of disorders. Overall, adolescents with a history of pure depression experienced higher levels of stress that is dependent in part on their own actions, as compared to adolescents with pure anxiety, suggesting that stress generation processes may be specific to depression. These results are consistent with Hammen's (1991) initial finding of greater stress generation in women with depression diagnoses versus bipolar disorder or medical illness, as well as considerable research that has conceptualized stress generation as a facet of depression (see Hammen, 2005, for a review). However, the current study also found that individuals with any internalizing diagnosis experience more dependent stress than healthy controls, and moreover, that stress generation in depression appears to be aggravated by the presence of a comorbid anxiety diagnosis. This suggests that models of stress generation originally developed to explain the development of depression (e.g. Hammen, 1991) may have some applicability to anxiety disorders. The current study's findings also refine and expand upon prior findings suggesting that stress generation is more strongly associated with daily and weekly assessments of depressive symptoms than anxiety symptoms among college students (Joiner, et al., 2005; Uhrlass & Gibb, 2007) by extending the results to clinically significant diagnoses and stressors occurring over a longer time period, and distinguishing between different types of stressors.

There are a number of potential mechanisms through which depression and anxiety may influence stress generation. One possible mechanism may be through the individual’s interpersonal style. There is evidence that both depression and anxiety are characterized by a maladaptive interpersonal style, particularly an insecure attachment style (e.g., Eberhart & Hammen, 2006; Eng, Heimberg, Hart, Schneier, & Liebowitz; 2001; Safford, Alloy, Crossfield, Morocco, & Wang, 2004), and further, that insecure attachment and other maladaptive approaches to relationships are in turn associated with increased stress generation (e.g., Eberhart & Hammen, 2009; Hankin, Kassel, & Abela, 2005; Shahar & Priel, 2003). Thus, a common maladaptive interpersonal style may be one mechanism through which both depression and anxiety influence stress generation. Along similar lines, a common cognitive style could drive stress generation in the disorders. There is ample evidence that maladaptive cognitions are associated with both depression (e.g., Alloy et al., 2006; Hankin, Abramson, & Siler, 2001; Iacoviello, Alloy, Abramson, Whitehouse, & Hogan, 2006) and anxiety (e.g., Reardon & Williams, 2007; Riskind, Williams, Gessner, Chrosniak, & Cortina, 2000), and recent research has found that cognitive vulnerability factors that are associated with anxiety and depression are also predictive of stress generation (e.g., Riskind, Black, & Shahar, 2010; Safford, Alloy, Abramson, & Crossfield, 2007; Shih, Abela, & Starrs, 2009). Thus, it is also possible that maladaptive ways of thinking drive stress generation in both disorders.

Negative affect, or more broadly general distress, may be a key pathway through which both depressive and anxiety diagnoses contribute to stress generation. Indeed, the tripartite model of depression and anxiety (Clark & Watson, 1991) posits that the overlap between depression and anxiety may be explained by a shared general distress factor. Along these lines, it is notable that the highest levels of stress generation were found among adolescents with comorbid depression and anxiety diagnoses. It is possible that this group had the highest levels of dependent stressors in part because they had the highest levels of distress. Further, adolescents with more than one diagnosis may also have chronic symptomatology and impairment that contributes to dependent stressors. Future research should examine these and other mechanisms through which depression, anxiety, and their co-occurrence influence the stressors that individuals experience.

It should be noted that while differences between the depressed and anxious groups in interpersonal and non-interpersonal subtypes of dependent stress were not statistically significant, they were supported by the pattern of the means or marginal p-values, which tentatively suggested greater levels of these types of stress among the depressed group. Large variances in the stress variables were apparent, and future studies with larger samples of adolescents with depression and anxiety disorders may be needed to provide sufficient power to attain statistical significance.

Several limitations of the sample should also be considered when evaluating the results of this study. First, the current sample was selected to over-sample mothers with a depressive disorder. Maternal history of depressive disorders has been controlled in analyses to account for this bias in the sample selection and to statistically control for potential confounding effects of maternal depression. However, caution should be taken in extending findings from this sample to other non-selected samples, and further research is needed to determine whether similar patterns hold in unselected community samples of adolescents. Further, while the study is based on a large sample of over 800 individuals, the various diagnostic categories examined had relatively low frequencies, which may have reduced the available power to uncover significant effects. In addition, while the study's largely Caucasian, middle to lower class sample is representative of the Australian population it was drawn from, the results may not generalize to more ethnically diverse populations.

Finally, while the study utilized semi-structured interviews to measure life stress and diagnoses, there were some limitations to these assessments. It should be noted that reporting of stressors was retrospective, and thus may be subject to memory biases. Moreover, the design of the study did not allow us to ascertain whether adolescents' diagnoses preceded the stressful life events they reported. Future research should prospectively examine whether the different diagnoses predict future patterns of stress generation. Further, due to the small number of individuals with internalizing disorder diagnoses, the study examined lifetime diagnoses, combining individuals with current and past disorders in its analyses. Thus, it is unclear whether the pattern of results would be replicated if the study examined individuals with only current or only past diagnoses. Further research is needed to establish whether the increased levels of stress generation found among individuals with pure depression versus pure anxiety, internalizing disorders versus controls, and comorbid diagnoses versus pure diagnoses, also occur prospectively, and if they are equally applicable to current versus past internalizing diagnoses.

Despite these limitations, the current study represents an important step toward understanding patterns of stress generation among adolescents with depression and anxiety diagnoses.

Acknowledgments

The research reported in this article was supported by a National Institute of Mental Health Grant (R01MH5223901) awarded to Constance Hammen and Patricia Brennan and a National Research Service Award postdoctoral fellowship (5-T32-MH14584) awarded to Nicole Eberhart while at the University of California, Los Angeles. We are grateful to the project staff in Brisbane, Australia, Robyne Le Brocque and interviewers Cherie Dalton, Barbara Mann, Eileen Tone, Sandra Fergusson, Lisa Manning, and Molly Robbins, and to Jake Najman, William Bor. Michael O'Callaghan, Gail Williams, and Margaret Andersen of the University of Queensland, Australia

References

  1. Alloy LB, Abramson LY, Whitehouse WG, Hogan ME, Panzarella C, Rose DT. Prospective incidence of first onsets and recurrences of depression in individuals at high and low cognitive risk for depression. Journal of Abnormal Psychology. 2006;115:145–145. doi: 10.1037/0021-843X.115.1.145. [DOI] [PubMed] [Google Scholar]
  2. Angold A, Costello EJ. Depressive comorbidity in children and adolescents: Empirical, theoretical, and methodological issues. American Journal of Psychiatry. 1993;150(12):1779–1791. doi: 10.1176/ajp.150.12.1779. [DOI] [PubMed] [Google Scholar]
  3. Bedford A, Foulds GA, Sheffield BF. A new personal disturbance scale (DSSI/sAD) British Journal of Social & Clinical Psychology. 1976;15(4):387–394. doi: 10.1111/j.2044-8260.1976.tb00050.x. [DOI] [PubMed] [Google Scholar]
  4. Brady EU, Kendall PC. Comorbidity of anxiety and depression in children and adolescents. Psychological Bulletin. 1992;111(2):244–255. doi: 10.1037/0033-2909.111.2.244. [DOI] [PubMed] [Google Scholar]
  5. Chun C, Cronkite RC, Moos RH. Stress generation in depressed patients and community controls. Journal of Social and Clinical Psychology. 2004;23:390–412. [Google Scholar]
  6. Clark LA, Watson D. Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology Special Issue: Diagnoses, Dimensions, and DSM-IV: The Science of Classification. 1991;100(3):316–336. doi: 10.1037//0021-843x.100.3.316. [DOI] [PubMed] [Google Scholar]
  7. Brown GW, Harris T. Social origins of depression. London: Free Press; 1978. [DOI] [PubMed] [Google Scholar]
  8. Cui XJ, Vaillant GE. Does depression generate negative life events? Journal of Nervous and Mental Disease. 1997;185:145–150. doi: 10.1097/00005053-199703000-00003. [DOI] [PubMed] [Google Scholar]
  9. Daley SE, Hammen C, Burge D, Davila J, Paley B, Lindberg N, Herzberg DS. Predictors of the generation of episodic stress: A longitudinal study of late adolescent women. Journal of Abnormal Psychology. 1997;104:251–259. doi: 10.1037//0021-843x.106.2.251. [DOI] [PubMed] [Google Scholar]
  10. Eberhart NK, Hammen CL. Interpersonal predictors of stress generation. Personality and Social Psychology Bulletin. 2009;35:544–556. doi: 10.1177/0146167208329857. [DOI] [PubMed] [Google Scholar]
  11. Eberhart NK, Hammen CL. Interpersonal predictors of onset of depression during the transition to adulthood. Personal Relationships. 2006;13:195–206. [Google Scholar]
  12. Eng W, Heimberg RG, Hart TA, Schneier FR, Liebowitz MR. Attachment in individuals with social anxiety disorder: The relationship among adult attachment styles, social anxiety, and depression. Emotion. 2001;1(4):365–380. doi: 10.1037/1528-3542.1.4.365. [DOI] [PubMed] [Google Scholar]
  13. Davila J, Hammen C, Burge D, Paley B, Daley SE. Poor interpersonal problem solving as a mechanism of stress generation in depression among adolescent women. Journal of Abnormal Psychology. 1995;104:592–600. doi: 10.1037//0021-843x.104.4.592. [DOI] [PubMed] [Google Scholar]
  14. Hammen C. Stress and depression. Annual Review of Clinical Psychology. 2005;1:293–319. doi: 10.1146/annurev.clinpsy.1.102803.143938. [DOI] [PubMed] [Google Scholar]
  15. Hammen CL. The generation of stress in the course of unipolar depression. Journal of Abnormal Psychology. 1991;100:555–561. doi: 10.1037//0021-843x.100.4.555. [DOI] [PubMed] [Google Scholar]
  16. Hammen C, Ellicott A, Gitlin M, Jamison KR. Sociotropy/autonomy and vulnerability to specific life events in patients with unipolar depression and bipolar disorders. Journal of Abnormal Psychology. 1989;98(2):154–160. doi: 10.1037//0021-843x.98.2.154. [DOI] [PubMed] [Google Scholar]
  17. Hankin BL, Abramson LY, Siler M. A prospective test of the hopelessness theory of depression in adolescence. Cognitive Therapy and Research. 2001;25(5):607–632. [Google Scholar]
  18. Hankin BL, Kassel JD, Abela JRZ. Adult attachment dimensions and specificity of emotional distress symptoms: Prospective investigations of cognitive risk and interpersonal stress generation as mediating mechanisms. Personality and Social Psychology Bulletin. 2005;31:136–151. doi: 10.1177/0146167204271324. [DOI] [PubMed] [Google Scholar]
  19. Harkness KL, Luther J. Clinical risk factors for the generation of life events in major depression. Journal of Abnormal Psychology. 2001;110:564–572. doi: 10.1037//0021-843x.110.4.564. [DOI] [PubMed] [Google Scholar]
  20. Harkness KL, Monroe SM, Simons AD, Thase ME. The generation of life events in recurrent and non-recurrent depression. Psychological Medicine. 1999;29:135–144. doi: 10.1017/s0033291798007752. [DOI] [PubMed] [Google Scholar]
  21. Iacoviello BM, Alloy LB, Abramson LY, Whitehouse WG, Hogan ME. The course of depression in individuals at high and low cognitive risk for depression: A prospective study. Journal of Affective Disorders. 2006;93:61–61. doi: 10.1016/j.jad.2006.02.012. [DOI] [PubMed] [Google Scholar]
  22. Joiner TE, Jr, Wingate LR, Gencoz T, Gencoz F. Stress generation in depression: Three studies on its resilience, possible mechanism, and symptom specificity. Journal of Social & Clinical Psychology. 2005;24(2):236–253. [Google Scholar]
  23. Middeldorp CM, Cath DC, Van Dyck R, Boomsma DI. The co-morbidity of anxiety and depression in the perspective of genetic epidemiology: A review of twin and family studies. Psychological Medicine. 2005;35(5):611–624. doi: 10.1017/s003329170400412x. [DOI] [PubMed] [Google Scholar]
  24. Orvaschel H. Schedule for Affective Disorder and Schizophrenia for School-Age Children Epidemiologic Version 5. Ft. Lauderdale. FL: Center for Psychological Studies, Nova Southeastern University; 1995. [Google Scholar]
  25. Paykel ES. Life events and affective disorders. Acta Psychiatrica Scandinavica. 2003;108:61–66. doi: 10.1034/j.1600-0447.108.s418.13.x. [DOI] [PubMed] [Google Scholar]
  26. Reardon JM, Williams NL. The specificity of cognitive vulnerabilities to emotional disorders: anxiety sensitivity, looming vulnerability and explanatory style. Journal of Anxiety Disorders. 2007;21:625–643. doi: 10.1016/j.janxdis.2006.09.013. [DOI] [PubMed] [Google Scholar]
  27. Riskind JH, Black D, Shahar G. Cognitive vulnerability to anxiety in the stress generation process: Interaction between the looming cognitive style and anxiety sensitivity. Journal of Anxiety Disorders. 2010;24:124–128. doi: 10.1016/j.janxdis.2009.09.007. [DOI] [PubMed] [Google Scholar]
  28. Riskind JH, Williams NL, Gessner T, Chrosniak LD, Cortina J. The Looming Maladaptive Style: anxiety, danger, and schematic processing. Journal of Personality and Social Psychology. 2000;79:837–852. doi: 10.1037//0022-3514.79.5.837. [DOI] [PubMed] [Google Scholar]
  29. Safford SM, Alloy LB, Abramson LY, Crossfield AG. Negative cognitive style as a predictor of negative life events in depression-prone individuals: A test of the stress generation hypothesis. Journal of Affective Disorders. 2007;99:147–154. doi: 10.1016/j.jad.2006.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Safford SM, Alloy LB, Crossfield AG, Morocco AM, Wang JC. The relationship of cognitive style and attachment style to depression and anxiety in young adults. Journal of Cognitive Psychotherapy. 2004;18(1):25–41. [Google Scholar]
  31. Shahar G, Priel B. Active vulnerability, adolescent distress, and the mediating/suppressing role of life events. Personality and Individual Differences. 2003;35:199–218. [Google Scholar]
  32. Shih JH, Abela JRZ, Starrs C. Cognitive and interpersonal predictors of stress generation in children of affectively ill parents. Journal of Abnormal Child Psychology. 2009;37:195–208. doi: 10.1007/s10802-008-9267-z. [DOI] [PubMed] [Google Scholar]
  33. Shih JH, Eberhart NK, Hammen CL, Brennan PA. Differential exposure and reactivity to interpersonal stress predict sex differences in adolescent depression. Journal of Clinical Child and Adolescent Psychology. 2006;35:103–115. doi: 10.1207/s15374424jccp3501_9. [DOI] [PubMed] [Google Scholar]
  34. Uhrlass DJ, Gibb BE. Childhood emotional maltreatment and the stress generation model of depression. Journal of Social & Clinical Psychology. 2007;26(1):119–130. [Google Scholar]

RESOURCES