Journal of Affective Disorders 156 (2014) 126–133

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research report

The use of symptom dimensions to investigate the longitudinal effects of life events on depressive and anxiety symptomatology Klaas J. Wardenaar a,n, Tineke van Veen c, Erik J. Giltay c, Frans G. Zitman c, Brenda W.J.H. Penninx b,c,d a University of Groningen, University Medical Center Groningen, Interdisciplinary Center Psychopathology and Emotion regulation (CC-72), PO Box 30.001, 9700 RB Groningen, The Netherlands b University of Groningen, University Medical Center Groningen, Department of Psychiatry, Groningen, The Netherlands c Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands d Department of Psychiatry, VU University Medical Center, Amsterdam, The Netherlands

art ic l e i nf o

a b s t r a c t

Article history: Received 2 October 2013 Received in revised form 6 December 2013 Accepted 7 December 2013 Available online 16 December 2013

Background: Findings on the association between life events and depression have been quite inconsistent. This could be due to the heterogeneity of traditionally used depression outcomes. The aim of this study was to investigate whether specific symptom dimensions can be used as an alternative to detect more specific life event effects. Methods: Participants with/without psychiatric diagnoses were included (n ¼2252). Dimensions of the tripartite model (General Distress [GD], Anhedonic Depression [AD] and Anxious Arousal [AA]) were assessed at baseline, 1-year and 2-year follow-up. Life events occurring between measurements were assessed retrospectively. Longitudinal associations between life events and dimensional scores were analysed with Linear Mixed Models. Results: Occurrence of negative life events was associated with increasing GD and AA, and less with AD. Positive life events were associated with decreasing GD and AD, but not with AA. The association between negative life events and AD was larger in the absence of previous psychiatric problems, lending support to a dimension-specific ‘kindling’ effect. Also, the negative association between negative life events and GD was stronger in those with high neuroticism. Multivariable analyses with individual life events showed that a few strong independent effects remained for each dimension. Limitations: Life event reports were retrospective; only three outcome dimensions were used. Conclusions: These results show that the effects of life events and modifying factors depend, to an extent, on the symptom domain that is considered as outcome, illustrating the need to account for symptom heterogeneity in etiological life event research. & 2013 Elsevier B.V. All rights reserved.

Keywords: Dimensions Depression Tripartite Life events General distress

1. Introduction There is evidence for a relationship between the occurrence of life events and the onset, recurrence and persistence of depression (e.g. Kessler, 1997; Mazure, 1998; Hammen, 2005). Early casecontrol studies showed that adverse life events were more common in depressed patients than in controls (e.g. Paykel et al., 1969; Brown and Harris, 1978). Later, life events were also shown to be associated with first onset of depression and with recurrence in lifetime depressed patients (e.g. Mazure 1998; Ormel et al., 2001). In addition, adverse life events have been shown to be associated with more comorbidity (Paykel, 2003) and longer time to remission (Spinhoven et al., 2011). Also, some depression

n

Corresponding author. Tel.: þ 31 50 36 14610; fax: þ31 50 36 19722. E-mail address: [email protected] (K.J. Wardenaar).

0165-0327/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2013.12.004

subtypes have been found to be more sensitive to the effects of life events: for instance, severe melancholic depression was found to be more sensitive to minor life events than non-melancholic depression (e.g. Harkness and Monroe, 2006). Interestingly, the effects of stressful life events have been found to diminish with increasing numbers of prior episodes (Kendler et al., 2000; Monroe and Harkness, 2005). This is in line with the broadly held ‘kindling’ hypothesis (Post, 1992), which states that the relative contribution of exogenous factors to depression becomes smaller with each depression recurrence. Often replicated, this phenomenon has clarified the dynamic association between life stress/ negative life events and depression recurrence over time (Monroe and Harkness, 2005). Although there is quite some support for the abovementioned associations and phenomena there also is inconsistency in the literature as a whole. There are, for instance, studies that report no or very small associations between life events and depression (e.g.

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

Spinhoven et al., 2010), no or little support for kindling (e.g. Roca et al., 2013), and no associations between life events and depression subtypes (Sun et al., 2012). This could be due to design variations across studies (e.g. in samples, measures and definitions; Mazure, 1998; Hammen, 2005) or methodological issues that affect all life event assessments to some extent (e.g. reporting bias/‘stress generation’, Hammen, 2005; mediating factors, e.g. Kessler, 1997; Michl et al., 2013). Interestingly, much less is known about the role of positive life events. Although these events are hypothesized to decrease depressive severity (e.g. Brown et al., 1988), this effect is under investigated and in little available studies have been found to be modified by other factors. For instance, Oldehinkel et al. (2000) found that positive life events led to more improvement of depression in those with high neuroticism, suggesting a role of person-specific vulnerability and/or reactivity to external stimuli. Apart from the literature's quite narrow focus on negative events, an issue that has received relatively little attention in life event research, but is very important for the replicability and interpretability in any strain of depression research, is the definition of the used outcomes. The traditional depression construct is very heterogeneous (Widiger and Clark, 2000). Two patients that meet DSM-criteria for a depressive episode or have similar scores on a depression severity scale can be very different in terms of their symptom profiles, leading to a situation where traditional psychopathology outcomes show a lot of internal variation. Although more homogeneous subtypes of depression (e.g. melancholic/atypical) have been proposed as a possible solution, these subtypes have limited validity and are quite heterogeneous themselves (Stewart et al., 2007). As a result, most previous research has only been able to detect general effects of life events. This is unfortunate since studies that have employed more specific definitions of life event types and symptom outcomes, have shown relatively consistent patterns of specific associations. On the one hand, life events that involved social loss (e.g. ‘death of a loved one’ and ‘romantic loss’) were found to be associated with increased crying and arousal. On the other hand, life events that were characterized by failure and prolonged stress (e.g. winter) were associated with increased feelings of fatigue and pessimism (Keller and Nesse, 2005). These results were corroborated by a second study (Keller and Nesse, 2006) and further extended in a third (Keller et al., 2007). In this large study, social loss was found to be associated with higher levels of sadness, anhedonia, appetite loss and guilt. Failure and chronic stressors were associated with increased fatigue and insomnia, but less with sadness and anhedonia. This work suggest that the effects of life events vary across different depressive symptoms instead of being syndrome-broad. The abovementioned studies used individual depressive symptoms as outcomes. Symptom dimensions could help to further investigate the specific effects of life events. Dimensions cover distinct symptom domains and follow a severity-continuum from healthy to severely pathological (Goldberg, 2000). As such, dimensions have the advantage that they can be measured reliably with psychometric scales (in contrast to assessment of individual symptoms) and can be used to assess continuous change. Compared to DSM-categories, dimensions also have advantages: they are more homogeneous, circumvent comorbidity (Widiger and Clark, 2000), better represent continuous variations observed in real life (Goldberg, 2000) and provide more statistical power/ sensitivity to change (MacCallum et al., 2002). The tripartite model is a well-known dimensional model of depressive and anxiety symptomatology and assumes three underlying dimensions (Clark and Watson, 1991). General Distress (GD) covers symptoms of psychological distress (e.g. feeling guilty and worry), common to depression and anxiety. Anhedonic Depression (AD) covers symptoms of decreased positive affect and energy,

127

specific to depression. Anxious Arousal (AA) covers symptoms of somatic hyperarousal, specific to anxiety. The model has been well-validated psychometrically (Watson et al., 1995; Keogh and Reidy, 2000) and its external validity has been confirmed by associations with biological mechanisms, such as the Hypothalamo–Pituitary–Adrenal-axis (Wardenaar et al., 2011), metabolic factors (Luppino et al., 2011) and gene sets (van Veen et al., 2012), and with clinical prognosis (Wardenaar et al., 2012). Only few studies have looked at the associations between life events and the tripartite dimensions or comparable constructs/ scales. Several cross-sectional studies have shown that negative life events are associated with increased negative affect/GD and positive life events with increased positive affect/decreased AD (Reich and Zautra, 1981; Zautra and Reich, 1983; Suh et al., 1996). One adolescent study found that negative life events combined with high GD was combined with depression and anxiety; whereas negative life event occurrence and high AD (low positive affect) was specifically associated with depression, in line with the tripartite model (Fox et al., 2010). Another cross-sectional study found several general and specific associations between retrospectively assessed life events and the tripartite dimensions (van Veen et al., 2013). Prospective studies have employed experience sampling methods to measure affective responses to daily hassles, and have also found effects of negatively experienced events on negative affect (e.g. Suls, 1998; Gable et al., 2000; Moberly and Watkins, 2008; Peeters et al., 2003). The latter studies looked at day-to-day effects of events, rated as stressful by respondents, and provide insight in daily life emotional responsivity. The prospective design of the above work could be extended to investigate the more extreme effects of major life events on an epidemiological scale of months/years. In addition, a broader range of dimensions could be included (GD, AD and AA) and both healthy subjects and patients could be included. Importantly, such a design would enable the investigation of the actual added value of dimensions in life event research. This could be done by comparing the effects that are captured by the dimensions with the effects that are captured by traditional DSM-based coursetrajectory classifications. The current study was aimed to investigate the associations between, on the one hand, negative and positive life events, and on the other hand, longitudinal change on the tripartite model dimensions. To this end, symptom-dimension scores and life event reports were collected in a large group of subjects with and without psychiatric diagnoses (n ¼2252) at three consecutive measurements (baseline, 1-year and 2-year follow-up). Associations between life events and change on the dimensions were analysed with Linear Mixed Models (LMM). These analyses were adjusted for DSM-based course trajectories to evaluate the actual added value of the dimensions to captured unique life event effects. Also, analyses were performed to investigate the modifying effect of the presence/absence of a lifetime diagnosis on the effect of negative life events on the dimensions, in line with the kindling-hypothesis. In addition, analyses were done to investigate the potential modifying effect of baseline neuroticism on the association between positive life events and the dimensions.

2. Methods 2.1. Participants Participants came from the Netherlands Study of Depression and Anxiety (NESDA), a large longitudinal study to investigate the course of depressive and anxiety disorders (N ¼2981), who were recruited from community, primary care and specialized mental health care organizations. At baseline, the mean age was 41.9

128

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

years (range 18–65), there were 1002 men and 1979 women, and 2329 participants had a lifetime diagnosis of major depressive disorder (MDD) and/or an anxiety disorder. Six-hundred-fifty-two participants had no lifetime psychiatric diagnosis. Exclusion criteria were not being fluent in Dutch and a primary diagnosis of psychotic, obsessive-compulsive, bipolar or severe addiction disorder because these latter low prevalence disorders could strongly affect the observed course trajectories in NESDA. Detailed objectives and rationales of NESDA can be found elsewhere (Penninx et al., 2008). The Ethical Review Boards of all participating universities approved the research protocol. All participants signed informed consent. All participants had a baseline assessment (T0) consisting of a face-to-face structured psychiatric interview by a trained researchassistant, self-report questionnaires, biological measurements and a blood-draw. After 1 year (T1), all participants completed the selfreport questionnaires and returned these by mail. Two years after baseline (T2), participants were assessed again in a face-to face session, similar to the one at baseline. Dimensional scores were collected at T0, T1 and T2. Participants were included if they provided dimensional scores for all these time-points and information about life events for at least one of the two covered years. In total, 2252 participants (75.5%) provided sufficient data. Included participants were older (t¼  6.1; po0.001), had more years of education (t¼  7.4; po0.001), and were more often female (χ2 ¼ 4.7; p¼ 0.03) than excluded participants.

2.2. Instruments 2.2.1. Dimensions: MASQ-D30 To measure the tripartite dimensions, a 30-item adaptation of the Mood and Anxiety Symptoms Questionnaire: the MASQ-D30 (Wardenaar et al., 2010) was used (original MASQ: Watson et al., 1995). On the MASQ-D30, participants are asked to rate to what extent in the past week they have experienced ‘feelings, sensations, problems and experiences that people sometimes have’ on a 5-point scale, with 1 being ‘“not at all’ and 5 being ‘extremely’. The three 10-item subscales are: ‘GD’, ‘AD’ and ‘AA’. The AD items are reverse-keyed and are rescored before subscale computation. The MASQ-D30 scales have been shown to have adequate psychometric characteristics (Wardenaar et al., 2010).

2.2.2. Life events To assess the occurrence of negative life events between respectively T0 and T1 and T1 and T2, the List of Threatening Events Questionnaire (LTE-Q; Brugha et al., 1985) was retrospectively administered at T1 and T2. The LTE-Q has been shown to be reliable (Brugha and Cragg, 1990). Examples of the assessed negative life events were ‘Serious financial troubles’ and ‘Death of a first-degree relative’. Seven positive life-events were also assessed at T1 and T2 with an additional list (e.g. ‘Making new friends’ and ‘Going on vacation’). The positive life events consisted of the positive opposites of the assessed negative life events and some additional common events that are generally perceived as positive (‘Completing an education’, ‘Going on vacation’). For some negative life events, no positive opposites could be meaningfully defined (e.g. ‘Contact with police or justice’). Therefore, the list of positive events was limited to seven items. For each event, participants were asked to indicate if it happened in the period before assessment and – if yes – when the event occurred or started (in case of long-lasting events). The complete lists of individual life events is displayed in Table 3. Dichotomous working variables were computed for the occurrence of any negative life event (1/0) and the occurrence of any positive life event (1/0).

Table 1 Baseline descriptive characteristics of the used study samples. N

2252

Mean Age at baseline (SD) Number of women (%) Mean years of education (SD)

42.7 (13.1) 1519 (67.5%) 12.4 (3.2)

MASQ-D30 scores, mean (SD) General Distress Anhedonic Depression Anxious Arousal

19.7 (8.4) 33.2 (9.6) 15.5 (5.8)

Baseline DSM-IV diagnoses No disorder Depressive disorders Anxiety disorders Depression and Anxiety

1240 222 415 375

(55.1%) (9.9%) (18.4%) (16.7%)

Lifetime DSM-IV diagnoses No disorder Depressive disorders Anxiety disorders Depression and Anxiety

536 438 274 1004

(23.8%) (19.4%) (12.2%) (44.6%)

Course-trajectory groups Stable Healthy Stable Chronic Unstable course NEO-FFI Neuroticism, mean (SD)

949 431 872 35.5

(42.1%) (19.1%) (38.7%) (9.5)

Depression-specific characteristics Number of episodes, mean (SD) Time since first depression onset, mean (SD) Time since last depression onset, mean (SD)

4.4 (9.0)a 13.9 (11.9)b 3.6 (6.5)c

MASQ-D30¼ Dutch short adaptation of the Mood and Anxiety Symptoms Questionnaire; DSM-IV ¼Diagnostic and Statistical Manual of Mental Disorders Fourth Edition; and NEO-FFI ¼Neuroticism-Extraversion-Openness Five-Factor-Inventory. a b c

Based on available data from 1311 depression cases (58.2%). Based on available data from 1434 depression cases (63.7%). Based on available data from 1422 depression cases (63.1%).

2.2.3. Course-trajectory variables Generic DSM-based course-trajectories were determined based on two data-sources: a diagnostic interview and a life chart interviews (LCI). At T0 and T2, the presence of DSM-IV diagnoses (MDD, Dysthymia, Panic disorder, Social Phobia, GAD, and Agoraphobia) was established with the Composite Interview Diagnostic Instrument (CIDI, WHO version 2.1). The organic exclusion rules were used and diagnoses were hierarchy-free. If at T2 participants got a diagnosis, the LCI was administered to assess the course of the disorder between T0 and T2. The presence of symptoms was evaluated for each month during follow-up. To aid memory, a calendar was used with (Lyketsos et al., 1994). Participants rated the symptom-severity for each symptomatic month on a 5-point scale (no/minimal, mild, moderate, severe, and very severe). Symptomatology was only considered to be present if at least mildly severe. Remission was considered present after at least 3 consecutive months without symptoms. The CIDI and LCI data were combined to define three ‘DSM course-trajectory groups’: (a) the stable healthy group (no depressive and/or anxiety disorder between T0 and T2), (b) the stable chronic group (persistent depressive and/or anxiety disorder between T0 and T2), and (c) the unstable course group (onset of a new disorder/remission from a disorder/remission and recurrence of a disorder).

2.2.4. Additional measures The baseline CIDI was used to determine whether participants had a lifetime diagnosis of depression and/or anxiety before the start of the follow-up period. A dichotomous variable was constructed (1: lifetime diagnosis/0: no lifetime diagnosis) to use in the analyses of effect-modification by prior diagnosis (see below).

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

129

Table 2 Multivariate longitudinal Linear Mixed Models analyses of the association between life event occurrence between measurements and change on the tripartite model dimensions. N ¼ 2252

General Distress Beta (95% CI)

Negative life event (yes/no)

Model 1 Model 2

Positive life event (yes/no)

Model 1 Model 2

0.15 (0.10 0.19)nnn 0.17 (0.10–0.24)nnn  0.16 (  0.09–(  0.23))nnn  0.26 (  0.15–(  0.37))nnn

Anhedonic Depression Beta (95% CI)

Anxious Arousal Beta (95% CI)

0.06 (0.02–0.11)n 0.05 (  0.03–0.12)

0.10 (0.05–0.14)nnn 0.16 (0.09–0.23)nnn

 0.23 (  0.15–(  0.30))nnn  0.23 (  0.11–(  0.34))nnn

 0.14 (  0.07–(  0.21))nnn  0.11 (0.00–(  0.22))

95% CI ¼ 95% Confidence Interval; Data based on Linear Mixed Models analyses with an unstructured covariance matrix Model 1: adjusted for age, gender and MASQ scalescore at baseline (T0). Model 2: additional variables: stable chronic (1/0); stable healthy course (1/0) unstable course (1/0), and four interactions between the course variables and life event variables. n

po 0.025. p o 0.001.

nnn

Table 3 Multivariate longitudinal Linear Mixed Models analyses of the association between individual life event occurrence and change on the tripartite model dimensions. N ¼ 2252

# of occurrences

Negative life events Illness/injury/victimization Illness/injury/victimization in family Death of first degree relative Death of first degree relative or friend Separation of partner Ending of a close friendship A serious problem in friendship/relation Unemployment/job search without result Getting fired from job Serious financial problems Contact with police or justice Something valuable/money was stolen or lost

189 615 205 558 173 320 341 223 107 220 94 174

Positive life events Recovery from a serious illness (in family) Meeting a new partner Making new friends Getting a new job or an important promotion Completing an education Being better off financially Going on vacation

378 241 1191 682 322 387 1869

General Distress

Anhedonic Depression

0.22nnn 0.08n 0.05 0.05 0.17nn 0.09n 0.09n 0.05  0.06 0.32nnn  0.07 0.05

0.18nnn 0.06 0.05 0.02 0.17nn 0.04 0.03 0.12n  0.07 0.15nn  0.10 0.00

 0.01  0.05  0.06n  0.05  0.02  0.02 0.00

 0.09n  0.12n  0.14nnn  0.07n  0.05  0.09n  0.04

Anxious Arousal

0.20nnn 0.04 0.01 0.05 0.03 0.06 0.09n 0.01 0.12 0.10n 0.07 0.01  0.02 0.05  0.06n  0.06 0.00 0.01  0.02

Data are standardized beta coefficients based on Linear Mixed Models analyses with an unstructured covariance matrix. Only the first occurrence of each event for each subject was included. All analyses were adjusted for age, gender and MASQ scale-score at T0. Associations that remained significant after Bonferroni correction (po[0.05/19¼0.0026]) are printed bold and underlined; asterisks indicate uncorrected p-values. n

po 0.05. p o0.01. p o 0.001.

nn

nnn

In addition, the number of previous episodes, the age at first onset and the age at last onset were assessed with the CIDI. The Neuroticism-Extraversion-Openness-Five-Factor-Inventory (NEOFFI; Costa and McCrae, 1992) was used to assess baseline neuroticism.

2.3. Study design We used a longitudinal study design to enable investigation of the association between life event occurrence and change on different symptom dimensions (see Fig. 1). Life events that occurred between T0, T1 and T2 were associated with simultaneous change in dimensional scores between T0, T1 and T2. We used LMM analyses to account for repeated measurements. As a test of the extent to which the dimensions captured change in symptomatology on top of the change that was already captured by the DSM-based course trajectories, variation in the dimensions that was also explained by DSM-based course-trajectories was covaried out.

Fig. 1. The used 3-wave longitudinal research design. MASQ-D30 ¼30-item adaptation of the Mood and Anxiety Symptoms Questionnaire.

2.4. Statistical analyses Variables were standardized to enable effect-size comparisons across different event types and dimensions. Each LMM was conducted with a MASQ-D30 scale as dependent variable and

130

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

positive life event occurrence (1/0) and negative life event occurrence (1/0) as independent variables. Time was used as a repeated measure factor and each LMM was conducted with an unstructured covariance matrix. The baseline (T0) value of the MASQ-D30 dimension under investigation was added as a fixed factor in all models. All analyses were adjusted for covariates. Model 1 included age and gender as covariates. Model 2 additionally included the DSM course trajectory group variables (as dummies) and their interactions with negative and positive life event occurrence (e.g ‘stable chronic  positive life event occurrence’) to covary out all variation in the dependent variable explained by course trajectory membership. To investigate if the effects of negative life events depended on the presence/absence of a prior diagnosis, model 2 was rerun with an interaction between negative life event occurrence (1: yes/0: no) and the presence of a CIDI-determined lifetime depressive or anxiety disorder before T0 (1: lifetime diagnosis/0: no lifetime diagnosis). If this interaction was found to be significant for a certain outcome, additional analyses were conducted in the lifetime diagnosis group to gain more insight in potential factors influencing the effect. The model was run including an interaction between negative life event occurrence (1: yes/0: no) and (a) the number of prior episodes (tertiles: 1, 2–3, and 4 3 episodes), (b) the number of years since first depression onset (quintiles: 0–3 years, 4–7 years, 8–13 years, 14–23 years, and Z24 years), or (c) the number of years since last depression onset (quartiles: 0 years (current), 1 year, 2–5 years, and Z6 years). To investigate whether the effects of positive life events depended on levels of neuroticism, an interaction between positive life event occurrence and NEO-FFI neuroticism quintiles (1:12–27, 2:28–34, 3:35–39, 4:40–44, and 5:45–60) was added to model 2. Finally, to gain more insight in the contribution of individual life event types while adjusting for the co-occurrence of multiple events, multivariable LMM analyses were run with all 19 individual life events as predictor variables. All analyses were done with SPSS 20.0. To restrict false positive rates due to multiple testing, Bonferroni-adjusted p-values (p0 ) were used to asses statistical significance (p0 ¼pn[# tests per outcome]). A p0 o0.05 was taken to indicate statistical significance.

3. Results 3.1. Demographic and psychiatric characteristics The sample characteristics are listed in Table 1. There were 1519 (67.5%) women and the mean age at baseline was 42.7 years (SD ¼13.1). Nine-hundred-forty-nine (42.1%) participants had a stable healthy course, 431 (19.1%) had a stable chronic course, and 872 (38.7%) had an unstable course. At baseline, AD and AA were moderately correlated (r ¼0.48). GD was strongly correlated with AD (r ¼0.68) and AA (r ¼0.62). 3.2. The effects of positive and negative life events The results of the LMM analyses of the association between the occurrence of any negative or positive life event and the MASQD30 scores are shown in Table 2. In model 1, negative life event occurrence was associated strongest with increased GD, AA and weaker with AD. Positive life event occurrence was associated most strongly with AD, followed by GD and AA. In model 2, the association of negative life event occurrence with AD was no longer significant and the association with AA became stronger. The association of positive life event occurrence with AA was no longer significant and the association with GD became stronger. These results indicated that GD is affected by negative and positive

life events, AD is affected mainly by positive life events and AA is affected mainly by negative life events. Interestingly, these effects reflect variation in response to life events that is not also captured by DSM-based course trajectories. 3.3. The modifying effect of prior psychiatric diagnosis When an interaction between negative life event occurrence and prior lifetime diagnosis was added to model 2, this interaction was only significant for AD (beta ¼  0.15; 95% CI:  0.03  (  0.28); p¼ 0.02), indicating a smaller effect in the presence of a previous diagnosis (when interaction term¼1). The beta-coefficient for the effect of negative life events in the absence of a previous diagnosis was 0.19 (95% CI:0.05  0.33). The beta-coefficient of the combined effect of negative life events on AD in the presence of previous diagnoses was much smaller (beta ¼[0.19  0.15 ¼]0.04). This finding supports the hypothesis that the influence of exogenous factors on symptomatology diminishes after the first onset of psychiatric problems. However, in the current analyses this effect turned out to be specific to AD; the interaction effect was nonsignificant for GD (p ¼0.13) and AA (p ¼0.36). For the outcome AD, additional analyses were conducted to gain more insight in the modifying role of prior psychopathology. Within the lifetime depression/anxiety group, there were no significant interactions of negative life event occurrence with the number of prior events (p ¼54), the time since first onset (p ¼0.43) and the time since last onset (p¼ 0.82), suggesting that amongst lifetime patients, the magnitude of the life event effects did not depend upon onset, recency and number of prior episodes. 3.4. The modifying effect of baseline neuroticism When an interaction between positive life event occurrence and neuroticism (quintiles) was added to model 2, this was only significant for the outcome GD (beta ¼  0.07; 95% CI:  0.01  ( 0.13), where the effect indicated that the negative effect of positive life events on GD became stronger with each neuroticism quintile. Stratified analyses in each of the quintiles showed that the negative effect was indeed stronger for higher neuroticism, especially in the two highest quintiles. There was no sign of a dose–response relation (Quintile 1 [n ¼509]: beta ¼  0.14 [p ¼ 0.33]; Quintile 2 [n ¼472] beta ¼  0.02 [p ¼0.88]; Quintile 3 [n ¼ 414]: beta ¼  0.18 [p ¼0.11]; Quintile 4 [n ¼ 449]: beta¼  0.54 [p o0.001]; and Quintile 5 [n ¼403]: beta ¼ 0.27 [p ¼0.07]. 3.5. The effects of individual life events Multivariable associations of the individual life events with GD, AD and AA are shown in Table 3. To adjust for the fact that – by default – some events occur much more frequently within the same subject (e.g. ‘Going on vacation’, ‘Making new friends’) than others (‘Illness/injury/victimization’), for each subject only the first occurrence of each event was included in these analyses. GD was clearly most strongly associated with three negative life events: ‘Illness/ injury/victimization’ (beta¼0.22), ‘Separation of partner’ (beta¼0.17), and ‘Serious financial problems’ (beta¼ 0.32). Associations with other negative life events were smaller and nonsignificant after correction for multiple testing (p40.05/19). There were no significant associations with positive life events. AD showed its strongest associations with the same negative life events as GD, although only two associations were significant: ‘Illness/ injury/victimization’ (beta¼ 0.18) and ‘Separation of partner’ (beta¼0.17). AD showed several associations with positive life events, but after correction for multiple testing, only one significant association remained with ‘Making new friends’ (beta¼  0.14). When analyses were rerun with all occurrences for each individual

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

life event instead of only the first occurrence (results not shown), the more frequent life events ‘Making new friends’ and ‘Going on vacation’ had larger and significant effects (respectively: beta¼ 0.16 and beta¼  0.13). AA only showed a significant association with ‘Illness/injury/victimization’ (beta¼ 0.20) and was not associated with any positive life events. Taken together, these results show that when the co-occurrence of individual life events are taken into account, a few individual key-events seem to show strong associations with GD, AD and AA.

4. Discussion The current longitudinal study investigated the association between life events and longitudinal change on the symptom dimensions of the tripartite model. The results showed that different types of life events have different patterns of association with the tripartite dimensions. Negative life events were most strongly associated with increases in GD and AA. Positive life events were associated with decreases in GD and AD over time. Negative life events were found to have only a very small effect on AD. These associations were barely affected by adjustment for DSM-based course trajectories, indicating that the dimensions captured life event induced variation that is not captured by trajectory group membership as defined by more traditional measures (CIDI and LSI). Building on previous reports in the literature, the potential modifying effect of previous psychiatric diagnosis (depression and/ or anxiety) on the longitudinal association between life events and symptomatology was investigated. It was found that negative life events had a much larger effect on AD in the absence of a prior diagnosis than in the presence of a prior diagnosis. This effect was only seen for AD and not significant for either GD or AA. This could partly explain the initial observation that the association between negative life events and AD seemed weak compared to the other domains. Also following previous work, the modifying effect of neuroticism on the association between positive life events and symptomatology was investigated. These analyses showed that the negative (decreasing) effect of positive life events on GD was considerably stronger in those with high levels of baseline neuroticism. No effect was seen on AD and AA, suggesting that modification was domain-specific, but also possibly because AD and AA are less strongly correlated with neuroticism than GD. Nevertheless, the results suggest that previous modification results (Oldehinkel et al., 2000) may have been driven mostly by the GD-part of depressive symptomatology. Analyses including all individual life events were conducted to investigate their independent effects, while accounting for clustering/co-occurrence of positive and negative events. These analyses revealed only a few strong independent effects. The event ‘Illness/ injury/victimization’ had an effect on all dimensions, ‘Separation from partner’ had an effect on GD and AD, and ‘Serious financial troubles’ only had a pronounced effect on GD. Individual effects of positive life events were only seen for AD (‘Making new friends’). These findings indicate that individual life events vary in the extent to which they independently affect different symptom domains. In addition, the limited number of strong independent life event effects suggests that the observed effects of negative life events as a composite class are explained by combinations (i.e. additive/interactive) of several life event effects rather than by the effect of single events. The current results have several interesting implications. The initial finding with the aggregated event classes (negative and positive) that negative events are associated most strongly with GD and AA, and less with AD, was in line with the previous research on negative and/or positive effect in healthy subjects (e.g.

131

Reich and Zautra, 1981; Suh et al., 1996; Suls, 1998). However, when the modifying effect of prior diagnosis was considered, the effect of negative life events on AD turned out to be substantial in those without a prior diagnosis of depression and/or anxiety. In addition, positive life event occurrence was also associated with GD and not exclusive to AD. These analyses thus suggest that the effects of positive and negative life events are not as exclusive to either GD or AD as would be hypothesized on the basis of previous work in healthy students. This could be explained by the fact that although GD and AD represent separate constructs with supposedly distinct underlying etiological factors, they were quite strongly correlated in the current sample (r ¼0.48) leading to limited differentiation between negative and positive life event effects in the current findings. In healthy samples this intercorrelation is considerably lower (r ¼0.26, Wardenaar et al., 2010), which could explain why in healthy samples more differentiation has been found between the dimensions' external correlates (e.g. Reich and Zautra, 1981). Still, the current results also yielded support for differentiation between GD and AD: (1) when looking at independent effects of individual life events, there were only significant positive life event effects on AD and not on GD and AA. (2) The modifying effect of prior diagnosis was specific to AD. (3) The modifying effect of neuroticism was specific to GD. Taken together, these results indicate that the way in which life events affect symptomatology depends on the symptom dimension that is considered. Change in AA was primarily associated with negative life events, although the associations were smaller than for GD. Analyses of the individual events showed that AA was only associated with ‘Illness/injury/victimization’. Although not previously investigated in a similar fashion, the finding that negative life events have an effect on hyperarousal, which is hypothesized to be specific to anxiety (Clark and Watson, 1991; Mineka et al., 1998), is in line with the idea that negative life events play a role in the onset of anxiety, and panic disorders in particular (Klauke et al., 2010). Also, these findings could reflect the known association between somatic health and the somatic/vegetative type of symptomatology that is operationalized in the AA domain (Luppino et al., 2011). The finding that the effects of life events depend on the presence of lifetime diagnoses is in line with the previous work (e.g. Kendler et al., 2000) on the kindling-effect, which entails that exogenous factors (e.g. life events) play an important role in the first onset of depression but, due to sensitization, less in the onset of recurrent depression. The current results support this hypothesis and add two points of interest. First, kindling was only found for AD, suggesting that it might be symptom- rather than syndrome-specific. Second, the current findings extend previous DSM-based work by showing that kindling applies to dimensional conceptualizations of psychopathology as well. The current findings illustrate the usefulness of dimensions to detect reactivity of mental state to external triggers. Even within groups of patients with a supposedly stable course (e.g. chronic over 2 years), notable variations could be seen on the dimensions, both within and between participants. This indicates that often used DSM-based course-qualifications (e.g. ‘chronic depression’ and ‘full remission’) only tell part of the story. The current epidemiological observations expand on previous work on a smaller time-scale, which found that emotional responsivity to particular daily hassles was also captured very effectively with repeated dimensional assessments (Suls, 1998; Gable et al., 2000; Peeters et al., 2003). The latter concept of emotional reactivity as a characteristic that varies across individuals can be translated to the current epidemiological setting. Variations in emotional reactivity could explain why life events sometimes do and sometimes do not lead to onset/recurrence of

132

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

depression (Kessler, 1997; Mazure, 1998; Hammen, 2005). In addition, many mechanisms have been described that might act on this process, including kindling, coping mechanisms (Billings and Moos, 1981; Kraaij et al., 2003), the amount and quality of social support (Cohen and Willis, 1985), genetic predisposition (Wichers et al., 2007) and early life adversity (Heim and Nemeroff, 2001). The current results add to this by showing that the effects of life events depend on the particular type of symptom domain (GD, AD or AA) and modifying factors, suggesting even more complex underlying dynamics. Although the current study had several strong characteristics, including a large sample-size, a longitudinal approach and systematic data collection, there are also limitations. First, findings apply to a group of healthy persons and psychiatric outpatients and cannot be generalized to other populations. Second, only three dimensions were used to describe depressive and anxiety symptomatology; whereas in reality other relevant symptom (sub) dimensions could exist (den Hollander-Gijsman et al., 2012). Third, only a limited set of effect-modifiers was considered; the roles of other protective (coping and social support) and susceptibility factors (e.g. genetic predisposition) could not be tested. Fourth, retrospective reporting of life events could be biased (Hammen, 2005). Fifth, systematic differences between included and excluded participants could have led to some selection bias. In conclusion, the dimensions of the tripartite model can be used to capture the specific effects of negative and positive life events and to gain interesting insights in some of the modifiers that act on these effects. Moreover, the observed effects transcended the traditional DSM-based course-trajectory distinctions, suggesting that using dimensions on top of DSM-based course definitions could help to better understand the complex mechanisms that underlie the link between life events and psychopathology.

Role of funding source The funding sources had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication.

Conflict of interest All authors declare that they have no conflicts of interest.

Acknowledgements The infrastructure for the NESDA study (www.nesda.nl) is funded through the Geestkracht program of the Netherlands Organisation for Health Research and Development (Zon-Mw, Grant no. 10-000-1002) and is supported by participating universities and mental health care organizations (VU University Medical Center, GGZ inGeest, Arkin, Leiden University Medical Center, GGZ Rivierduinen, University Medical Center Groningen, Lentis, GGZ Friesland, GGZ Drenthe, Scientific Institute for Quality of Healthcare (IQ healthcare), Netherlands Institute for Health Services Research (NIVEL) and Netherlands Institute of Mental Health and Addiction (Trimbos Institute).

References Billings, A.G., Moos, R.H., 1981. The role of coping responses and social resources in attenuating the stress of life events. J. Behav. Med. 4, 139–157. Brown, G.W., Harris, T.O., 1978. Social Origins of Depression. Free Press, London. Brown, G.W., Adler, Z., Bifulco, A., 1988. Life events, difficulties and recovery from chronic depression. Br. J. Psychiatry 152, 487–498. Brugha, T., Bebbington, P., Tennant, C., Hurry, J., 1985. The list of threatening experiences: a subset of 12 life event categories with considerable long-term contextual threat. Psychol. Med. 15, 189–194. Brugha, T.S., Cragg, D., 1990. The list of threatening experiences: the reliability and validity of a brief life events questionnaire. Acta Psychiatr. Scand. 82, 77–81. Clark, L.A., Watson, D., 1991. Tripartite model of anxiety and depression: psychometric evidence and taxonomic implications. J. Abnorm. Psychol. 100, 316–336. Cohen, S., Wills, T.A., 1985. Stress, social support, and the buffering hypothesis. Psychol. Bull. 98, 310–357. Costa, P.T., McCrae, R.R., 1992. Normal personality assessment in clinical practice: the NEO personality inventory. Psychol. Assess. 4, 5–13.

den Hollander-Gijsman, M.E., Wardenaar, K.J., de Beurs, E., van der Wee, N.J., Mooijaart, A., van Buuren, S., Zitman, F.G., 2012. Distinguishing symptom dimensions of depression and anxiety: an integrative approach. J. Affect. Disord. 136, 693–701. Fox, J.K., Halpern, L.F., Ryan, J.L., Lowe, K.A., 2010. Stressful life events and the tripartite model: relations to anxiety and depression in adolescent females. J. Adolesc. 33, 43–54. Gable, S.L., Reis, H.T., Elliot, A.J., 2000. Behavioral activation and inhibition in everyday life. J. Pers. Soc. Psychol. 78, 1135–1149. Goldberg, D., 2000. Plato versus Aristotle: categorical and dimensional models for common mental disorders. Compr. Psychiatry 41, 8–13. Hammen, C., 2005. Stress and depression. Annu. Rev. Clin. Psychol. 1, 293–319. Harkness, K.L., Monroe, S.M., 2006. Severe melancholic depression is more vulnerable than non-melancholic depression to minor precipitating life events. J. Affect. Disord. 91, 257–263. Heim, C., Nemeroff, C.B., 2001. The role of childhood trauma in the neurobiology of mood and anxiety disorders: preclinical and clinical studies. Biol. Psychiatry 49, 1023–1039. Keller, M.C., Neale, M.C., Kendler, K.S., 2007. Association of different adverse life events with distinct patterns of depressive symptoms. Am. J. Psychiatry 164, 1521–1529. Keller, M.C., Nesse, R.M., 2005. Is low mood an adaptation? Evidence for subtypes with symptoms that match precipitants. J. Affect. Disord. 86, 27–35. Keller, M.C., Nesse, R.M., 2006. The evolutionary significance of depressive symptoms: different adverse situations lead to different depressive symptom patterns. J. Pers. Soc. Psychol. 91, 316–330. Kendler, K.S., Thornton, L.M., Gardner, C.O., 2000. Stressful life events and previous episodes in the etiology of major depression in women: an evaluation of the “kindling” hypothesis. Am. J. Psychiatry 157, 1243–1251. Keogh, E., Reidy, J., 2000. Exploring the factor structure of the mood and anxiety symptom questionnaire (MASQ). J. Pers. Assess. 74, 106–125. Kessler, R.C., 1997. The effects of stressful life events on depression. Annu. Rev. Psychol. 48, 191–214. Klauke, B., Deckert, J., Reif, A., Pauli, P., Domschke, K., 2010. Life events in panic disorder-an update on “candidate stressors”. Depress. Anxiety 27, 716–730. Kraaij, V., Garnefski, N., de Wilde, E.J., Dijkstra, A., Gebhardt, W., Maes, S., ter Doest, L., 2003. Negative life events and depressive symptoms in late adolescence: bonding and cognitive coping as vulnerability factors? J. Youth. Adolesc. 32, 185–193. Luppino, F.S., van Reedt Dortland, A.K., Wardenaar, K.J., Bouvy, P.F., Giltay, E.J., Zitman, F.G., Penninx, B.W., 2011. Symptom dimensions of depression and anxiety and the metabolic syndrome. Psychosom. Med. 73, 257–264. Lyketsos, C.G., Nestadt, G., Cwi, J., Heithoff, K., Eaton, W.W., 1994. The life chart interview – a standardized method to describe the course of psychopathology. Int. J. Methods. Psychiatr. Res. 4, 143–155. MacCallum, R.C., Zhang, S.B., Preacher, K.J., Rucker, D.D., 2002. On the practice of dichotomization of quantitative variables. Psychol. Methods 7, 19–40. Mazure, C.M., 1998. Life stressors as risk factors in depression. Clin. Psychol.: Sci. Pract. 5, 291–313. Michl, L.C., McLaughlin, K.A., Shepherd, K., Nolen-Hoeksema, S., 2013. Rumination as a mechanism linking stressful life events to symptoms of depression and anxiety: longitudinal evidence in early adolescents and adults. J. Abnorm. Psychol. 122, 339–352. Mineka, S., Watson, D., Clark, L.A., 1998. Comorbidity of anxiety and unipolar mood disorders. Annu. Rev. Psychol. 49, 377–412. Moberly, N.J., Watkins, E.R., 2008. Ruminative self-focus and negative affect an experience sampling study. J. Abnorm. Psychol. 117, 314–323. Monroe, S.M., Harkness, K.L., 2005. Life stress, the “kindling” hypothesis, and the recurrence of depression: Considerations from a life stress perspective. Psychol. Rev. 112 (2), 417–445. Oldehinkel, A.J., Ormel, J., Neeleman, J., 2000. Predictors of time to remission from depression in primary care patients: do some people benefit more from positive life change than others? J. Abnorm. Psychol. 109, 299–307. Ormel, J., Oldehinkel, A.J., Brilman, E.I., 2001. The interplay and etiological continuity of neuroticism, difficulties, and life events in the etiology of major and subsyndromal, first and recurrent depressive episodes in later life. Am. J. Psychiatry 158, 885–891. Paykel, E.S., Myers, J.K., Dienelt, M.N., Klerman, G.L., Lindenthal, J.J., Pepper, M.I., 1969. Life events and depression: a controlled study. Arch. Gen. Psychiatry. 21, 753–760. Paykel, E.S., 2003. Life events and affective disorders. Acta Psychiatr. Scand. 108 (suppl.), 61–66. Peeters, F., Nicolson, N.A., Berkhof, J., Delespaul, P., deVries, M., 2003. Effects of daily events on mood states in major depressive disorder. J. Abnorm. Psychol. 112, 203–211. Penninx, B.W., Beekman, A.T., Smit, J.H., Zitman, F.G., Nolen, W.A., Spinhoven, P., Cuijpers, P., De Jong, P.J., Van Marwijk, H.W., Assendelft, W.J., Van Der Meer, K., Verhaak, P., Wensing, M., De Graaf, R., Hoogendijk, W.J., Ormel, J., Van Dyck, R., 2008. The Netherlands Study of Depression and Anxiety (NESDA): rationale, objectives and methods. Int. J. Methods Psychiatr. Res. 17, 121–140. Post, R.M., 1992. Transduction of psychosocial stress into the neurobiology of recurrent affective disorder. Am. J. Psychiatry 149, 999–1010. Reich, J.W., Zautra, A., 1981. Life events and personal causation: Some relationships with satisfaction and distress. J. Pers. Soc. Psychol. 41, 1002–1012. Roca, M., Gili, M., Garcia-Campayo, J., Armengol, S., Bauza, N., García-Toro, M., 2013. Stressful life events severity in patients with first and recurrent depressive episodes. Soc. Psychiatry Psychiatr. Epidemiol. 48, 1963–1969.

K.J. Wardenaar et al. / Journal of Affective Disorders 156 (2014) 126–133

Spinhoven, P., Elzinga, B.M., Hovens, J.G., Roelofs, K., van Oppen, P., Zitman, F.G., Penninx, B.W., 2011. Positive and negative life events and personality traits in predicting course of depression and anxiety. Acta Psychiatr. Scand. 124, 462–473. Spinhoven, P., Elzinga, B.M., Hovens, J.G., Roelofs, K., Zitman, F.G., van Oppen, P., Penninx, B.W., 2010. The specificity of childhood adversities and negative life events across the life span to anxiety and depressive disorders. J. Affect. Disord. 126, 103–112. Stewart, J.W., McGrath, P.J., Quitkin, F.M., Klein, D.F., 2007. Atypical depression: current status and relevance to melancholia. Acta. Psychiatr. Scand. 115 (suppl.), 58–71. Suh, E., Diener, E., Fujita, F., 1996. Events and subjective well-being: only recent events matter. J. Pers. Soc. Psychol. 70, 1091–1102. Suls, J., 1998. Emotional reactivity to everyday problems, affective inertia, and neuroticism. Pers. Soc. Psychol. Bull. 24, 127. Sun, N., Li, Y., Cai, Y., Chen, J., Shen, Y., Sun, J., Zhang, Z., Zhang, J., Wang, L., Guo, L., Yang, L., Qiang, L., Yang, Y., Wang, G., Du, B., Xia, J., Rong, H., Gan, Z., Hu, B., Pan, J., Li, C., Sun, S., Han, W., Xiao, X., Dai, L., Jin, G., Zhang, Y., Sun, L., Chen, Y., Zhao, H., Dang, Y., Shi, S., Kendler, K.S., Flint, J., Zhang, K.A., 2012. Comparison of melancholic and nonmelancholic recurrent major depression in Han Chinese women. Depress. Anxiety 29, 4–9. van Veen, T., Goeman, J.J., Monajemi, R., Wardenaar, K.J., Hartman, C.A., Snieder, H., Nolte, I.M., Penninx, B.W., Zitman, F.G, 2012. Different gene sets contribute to different symptom dimensions of depression and anxiety. Am. J. Med. Genet. B: Neuropsychiatric Genet. 159, 519–528.

133

van Veen, T., Wardenaar, K.J., Carlier, I.V., Spinhoven, P., Penninx, B.W., Zitman, F.G., 2013. Are childhood and adult life adversities differentially associated with specific symptom dimensions of depression and anxiety? Testing the tripartite model. J. Affect. Disord. 146, 238–245. Wardenaar, K.J., van Veen, T., Giltay, E.J., de Beurs, E., Penninx, B.W., Zitman, F.G., 2010. Development and validation of a 30-item short adaptation of the Mood and Anxiety Symptoms Questionnaire (MASQ). Psychiatry Res. 179, 101–106. Wardenaar, K.J., Vreeburg, S.A., van Veen, T., Giltay, E.J., Veen, G., Penninx, B.W., Zitman, F.G., 2011. Dimensions of depression and anxiety and the hypothalamopituitary-adrenal axis. Biol. Psychiatry 69, 366–373. Wardenaar, K.J., Giltay, E.J., van Veen, T., Zitman, F.G., Penninx, B.W., 2012. Symptom dimensions as predictors of the two-year course of depressive and anxiety disorders. J. Affect. Disord. 136 (3), 1198–1203. Watson, D., Weber, K., Assenheimer, J.S., Clark, L.A., Strauss, M.E., McCormick, R.A., 1995. Testing a tripartite model: I. Evaluating the convergent and discriminant validity of anxiety and depression symptom scales. J. Abnorm. Psychol. 104, 3–14. Wichers, M., Myin-Germeys, I., Jacobs, N., Peeters, F., Kenis, G., Derom, C., Vlietinck, R., Delespaul, P., van Os, J., 2007. Genetic risk of depression and stress-induced negative effect in daily life. Br. J. Psychiatry 191, 218–223. Widiger, T.A., Clark, L.A., 2000. Toward DSM-V and the classification of psychopathology. Psychol. Bull. 126, 946–963. Zautra, A.J., Reich, J.W., 1983. Life events and perceptions of life quality: developments in a two-factor approach. J. Community Psychol. 11, 121–132.

The use of symptom dimensions to investigate the longitudinal effects of life events on depressive and anxiety symptomatology.

Findings on the association between life events and depression have been quite inconsistent. This could be due to the heterogeneity of traditionally u...
362KB Sizes 0 Downloads 0 Views