Longitudinal associations between rumination and depressive symptoms in a probability sample of adults

Journal of Affective Disorders 260 (2020) 680–686

Contents lists available at ScienceDirect

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

Research paper

Longitudinal associations between rumination and depressive symptoms in a probability sample of adults

T

Mark A. Whismana, , Alta du Ponta,b, Peter Butterworthc ⁎

a

Department of Psychology and Neuroscience, University of Colorado Boulder, United States Institute for Behavioral Genetics, University of Colorado Boulder, United States c Centre for Research on Ageing, Health and Wellbeing, Research School of Population Health, The Australian National University, Australia b

ARTICLE INFO

ABSTRACT

Keywords: Depression Rumination Neuroticism Longitudinal Prospective Probability sample

Background: According to the Response Styles Theory, rumination maintains and exacerbates depression. This study was conducted to examine the bidirectional longitudinal associations between rumination and depressive symptoms in a probability sample of Australian adults, evaluate the degree to which the strength of these longitudinal associations was moderated by gender, and test whether these longitudinal associations remained statistically significant when adjusting for the influence of demographic characteristics and neuroticism. Methods: A probability sample of Australian adults (N = 5891) completed self-report measures of rumination, neuroticism, and depressive symptoms at baseline and rumination and depressive symptoms at a four-year follow-up. Results: Results from regression analyses indicated that rumination predicted residual change in depressive symptoms and depressive symptoms predicted residual change in rumination, suggesting that rumination and depressive symptoms influence one another in a bidirectional, recursive fashion. Gender was not a significant moderator of the longitudinal associations between rumination and depressive symptoms. Analyses including the covariates of age, gender, and neuroticism, a personality trait characterized by heightened negative emotionality, indicated that the bidirectional longitudinal associations between rumination and depressive symptoms were not explained by their shared association with demographic characteristics or neuroticism. Limitations: Within-person analyses involving repeated assessments, shorter time intervals, and assessment of brooding rumination would provide a stronger test of the potential causal association between rumination and depressive symptoms. Conclusions: Study findings suggest that rumination may be both a potential risk factor for and a consequence of depressive symptoms in adults.

1. Introduction A substantial body of research has documented associations between depression and rumination, a pattern of repetitive and passive thinking about symptoms of depression and possible causes and consequences of such symptoms (for a review, see Nolen-Hoeksema et al., 2008). According to the Response Styles Theory (RST; NolenHoeksema, 1991), rumination is a way of responding to distress that is causally associated with the onset, severity, and maintenance of depression. As reviewed elsewhere (Nolen-Hoeksema et al., 2008), longitudinal studies have supported the RST insofar as rumination at baseline predicts increases in depressive symptoms (NolenHoeksema and Morrow, 1991) and the onset of depressive disorders (e.g., Just and Alloy, 1997). For example, in a sample of 1132 adults, ⁎

rumination significantly predicted major depressive disorder diagnostic status one year later, after adjusting for baseline diagnostic status and depressive symptoms (Nolen-Hoeksema, 2000), and a study of undergraduates found that rumination was associated with the number of major depressive episodes experienced over a 2.5-year period (Spasojević and Alloy, 2001). Similar findings have been reported in studies of adolescents (Wilkinson et al., 2013) and older adults (Gan et al., 2015). Although most longitudinal studies have examined the unidirectional association from rumination to depression (i.e., a vulnerability hypothesis), depression may also lead to increased rumination. For example, according to the scar hypothesis (Lewinsohn et al., 1981), depressive episodes create changes in cognition that persist during remission. Consistent with this perspective, one study found that

Corresponding author. E-mail address: [email protected] (M.A. Whisman).

https://doi.org/10.1016/j.jad.2019.09.035 Received 11 May 2019; Received in revised form 15 August 2019; Accepted 8 September 2019 Available online 10 September 2019 0165-0327/ © 2019 Elsevier B.V. All rights reserved.

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al.

currently and formerly dysphoric individuals reported higher levels of rumination than never dysphoric individuals (Roberts et al., 1998). In adolescents, reciprocal relations between rumination and depressive symptoms were found across 4 years (Nolen-Hoeksema et al., 2007), whereas another study found no relation between depressive symptoms at baseline and rumination 6 months later, but did find a significant association between depressive symptoms at 6 months and increased rumination at 12 months (Calvete et al., 2015). In adults, a study of recently discharged, formerly depressed inpatients and gender- and age-matched community controls found that depressive symptoms predicted rumination 5 months later in both samples (Huffziger et al., 2009). Rumination is considered a response to discrepancies between actual and desired states (Martin and Tesser, 1989); thus, desires to be happy when one is depressed may also lead to increased rumination. Consistent with this hypothesis, one study found that trait rumination decreased as depressive symptoms abated (Roberts et al., 1998). Additional evidence for this hypothesis includes research from the affect valuation literature, which examines peoples’ evaluations of certain affective states. For example, one study found that people with major depressive disorder and generalized anxiety disorder had more extreme beliefs than healthy controls about the extent to which they should experience negative affect. Furthermore, beliefs that one should be experiencing less negative affect was associated with trait rumination, even after adjusting for negative affect (Thompson et al., 2016). Although there are some exceptions, including a study in which participants were recruited through random-digit dialing (NolenHoeksema, 2000), most prior studies evaluating the longitudinal association between rumination and depression in non-clinical samples have been based on samples of convenience, including undergraduates (e.g., Grassia and Gibb, 2008; Nolan et al., 1998; Roberts et al., 1998). Similarly, much of the research on rumination and depression is from samples from the United States, although there are studies from other countries such as Germany (Huffziger et al., 2009), China (Gan et al., 2015), and Japan (Sakamoto et al., 2001). Examining the associations between rumination and depression in probability samples of different age groups and in different countries is important to evaluate the generalizability of their association, as there may be cultural or historical influences on rumination and depression that could influence the strength of their longitudinal association (Grossmann and Kross, 2010). Demonstrating longitudinal associations between rumination and depression is consistent with causal hypotheses, but stronger inferences can be made if it is shown that the rumination–depression association is not due to shared association with other variables that could serve as rival explanations for their association (Kenny, 1979). Neuroticism, a personality trait characterized by heightened negative affect (Lahey, 2009), is one important potential rival explanation for the longitudinal associations between rumination and depression. Neurotcism is associated with rumination (e.g., Muris et al., 2005; Nolan et al., 1998; Roberts et al., 1998) and depression (for a meta-analysis, see Kotov et al., 2010). Additionally, given that both rumination and neuroticism have moderate to large genetic correlations with depression (e.g., Fanous et al., 2002; Johnson et al., 2014), and that rumination, neuroticism, and internalizing disorders share genetic influences (du Pont et al., 2019), it is possible that individual differences in neuroticism may contribute to the covariation between rumination and depression. Prior studies have shown that the association between rumination and depressive symptoms is incremental to their shared association with neuroticism (e.g., Muris et al., 2005; Nolan et al., 1998; Roberts et al., 1998). However, these studies have generally been conducted with undergraduates, and as such, research is needed to evaluate the degree to which the associations between rumination and depressive symptoms is incremental to shared variance with neuroticism in other age groups. Building on existing longitudinal research, the current study used a

two-wave panel design to evaluate the bidirectional longitudinal associations between rumination and depressive symptoms in a populationbased sample of adults in Australia. In addition, we examined gender as a potential moderator of the association between rumination and depressive symptoms. Relative to men, there is a higher level of depressive symptoms (Wang et al., 2016) and depressive disorders (Girgus and Yang, 2015) in women, and a major tenet of the RST is that gender differences in rumination may help explain this gender difference in depression (Nolen-Hoeksema, 1987); indeed, women consistently report higher levels of rumination relative to men (for a meta-analysis, see Johnson and Whisman, 2013). Finally, given prior research that neuroticism is associated with both rumination and depression, we tested whether the longitudinal associations between rumination and depressive symptoms would remain statistically significant when adjusting for the influence of neuroticism in this probability sample of adults, similar to what has been found in studies of undergraduates (e.g., Muris et al., 2005; Nolan et al., 1998; Roberts et al., 1998). We hypothesized that (a) rumination would be significantly and positively associated with residual change in depressive symptoms; (b) depressive symptoms would be significantly and positively associated with residual change in rumination; (c) the longitudinal associations between rumination and depressive symptoms would be stronger for women relative to men; and (d) these longitudinal associations would remain statistically significant after adjusting for demographic characteristics and neuroticism. 2. Method 2.1. Participants Participants were drawn from the ongoing PATH Through Life project. The PATH project is an Australian longitudinal study that examines health and well-being across the adult life span by following three age cohorts (birth years 1975–79, 1956–60, and 1937–41, who were aged 20–24, 40–44, and 60–64 years old at the initial assessment) every 4 years (for additional sample details, see Anstey et al., 2011). Participants were recruited through the electoral rolls of Canberra, the national capital, and the neighboring town of Queanbeyan in Southeastern Australia; enrollment to vote is compulsory for Australian citizens. Randomly selected individuals were sent a letter describing the study and those individuals who were interested in participating were contacted by an interviewer (Anstey et al., 2004). Study participants provided written informed consent prior to their involvement at each wave of the study. The current analyses are based on data from a subset (N = 5891) of the foundational sample who completed measures of rumination and depressive symptoms at both Wave 2 (which was conducted between 2003 and 2006) and Wave 3 (which was conducted between 2007 and 2010) and a measure of neuroticism at Wave 2; these two waves were examined because the PHQ-9, used to assess depressive symptoms, was not administered in Wave 1 (which was conducted between 2001 and 2002). Attrition at either Wave 2 or Wave 3 was associated with the following Wave 1 demographic characteristics: being male, being in the oldest age group, having never been married, not having English as a first language, not participating in the labor force, and having lower levels of educational attainment. Importantly, rumination, depressive symptoms, and neuroticism were not independently associated with attrition after adjusting for general demographic characteristics (age, gender) and physical health. The sample consisted of 3036 women (51.5% of the sample) and 2855 men. The sample included 1888 people from the younger cohort (birth years 1975–79), 2114 people from the middle-aged cohort (birth years 1956–60), and 1889 people from the older cohort (birth years 1937–41); collapsed across cohorts, the overall mean age was 46.63 years (SD = 16.03). The study was approved by the Human Research Ethics Committee at the Australian National University. 681

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al.

2.2. Measures

Table 1 Means, standard deviations, and correlations among study measures.

2.2.1. Rumination Rumination was assessed using a 10-item version of the original Ruminative Responses Scale (RRS; Nolen-Hoeksema and Morrow, 1991). It consists of items from the RRS that, in a community sample of 1122 adults, (a) correlated most strongly with total scores on the original scale, and (b) were endorsed by more than 15% of the sample (as described by Davis and Nolen-Hoeksema, 2000). Prior research has found a strong correlation (r = 0.93) between the 10-item version and the 22-item RRS (Davis and Nolen-Hoeksema, 2000). Cronbach's alpha in this sample was 0.89 at Wave 2 and 0.90 at Wave 3.

Variable Wave 2 1 PHQ-9 2 RRS 3 EPQ-N Wave 3 4 PHQ-9 5 RRS

Mean

SD

1

2

3.74 7.74 3.87

4.09 4.99 3.28

0.63*** 0.56***

0.63***

3.54 7.01

3.97 5.05

0.58*** 0.50***

0.49*** 0.66***

3

4

0.45*** 0.52***

0.64***

Note. RRS = Ruminative Responses Scale. EPQ-N = Eysenck Personality Questionnaire–Neuroticism Scale. *p < .05. **p < .01. *** p < .001.

2.2.2. Depression Depressive symptoms were assessed using the Brief Patient Health Questionnaire (PHQ-9; Kroenke and Spitzer, 2002; Spitzer et al., 1999). Participants rated how often they had been bothered by nine depressive symptoms in the past two weeks on a scale from 0 (not at all) to 3 (nearly every day). Cronbach's alpha in this sample was 0.86 at both Wave 2 and Wave 3.

p < .001, d = 0.26). Using standard cutoffs for interpreting the PHQ-9 (Kroenke and Spitzer, 2002), 69.4% of the sample reported minimal depression (PHQ-9 scores ≤ 4), 21.8% of the sample reported mild depression (5–9), 5.9% reported moderate depression (10–14), 2.0% reported moderately severe depression (15–19), and 1.0% reported severe depression (≥ 20); corresponding figures at Wave 3 were 71.6%, 21.0%, 4.7%, 1.8%, and 0.9%, respectively. Scores on the RRS at Wave 2 were significantly and positively correlated with scores at Wave 3 (Table 1), which suggests that participants’ rank order remained stable over time, thereby providing evidence of the relative stability of rumination (Bagby et al., 2004). The absolute stability of rumination, which refers to the degree to which the mean scores for a group remain the same over time (Bagby et al., 2004), was also examined. Results from a paired t-test suggested that there was a small but statistically significant change in mean levels of rumination from Wave 2 to Wave 3, t = 13.49, p < .001, d = 0.14; the formula provided by Dunlap et al. (1996) was used for computing the effect size for this analysis. Inspection of the means in Table 1 indicates that on average, participants’ RRS scores declined over time. Finally, individual differences in stability of rumination was evaluated by examining the percentage of participants whose change in RRS scores from Wave 2 to Wave 3 were greater than would be expected by chance (i.e., if no actual change had occurred). Using the Jacobson and Truax (1991) formula for defining reliable change, 1344 participants (22.8% of the sample) had RRS scores that changed more than would be expected by chance: scores decreased for 870 participants (14.8% of the sample) and increased for 474 participants (8.0% of the sample). Change in rumination greater than what would be expected by chance was observed in 32.7% of the younger cohort, 22.8% of the middle-aged cohort, and 12.9% of the older cohort, χ2(2) = 210.56, p < .001. Rumination and depressive symptoms were positively associated within and across waves (Table 1). Neuroticism was also positively associated with rumination and depressive symptoms at Wave 2 (Table 1), thereby supporting our decision to examine neuroticism as a potential rival explanation for any longitudinal associations between rumination and depressive symptoms. Turning next to the regression analyses examining the longitudinal associations between rumination and depressive symptoms, examination of the distribution of the study measures indicated that baseline and follow-up scores on the RRS and the PHQ-9 were positively skewed. Therefore, square root transformations were used to better approximate a normal distribution on these two measures; following transformation, skewness for each of these measures was < 1. Multiple regression analyses were then used to examine gender as a potential moderator of the longitudinal associations between rumination and depressive symptoms. In analyses including the component terms and age, the Gender × Wave 2 Rumination interaction term was not significantly associated with Wave 3 depressive symptoms (b = −0.005, SE = 0.024, p = .835) and the Gender × Wave 2 Depressive Symptom interaction term was not significantly associated with Wave 3 rumination (b = 0.016, SE = 0.019, p = .384). Consequently, data were

2.2.3. Neuroticism Neuroticism was measured using the neuroticism scale from the short form of the Eysenck Personality Questionnaire Revised (EPQR-S; Eysenck et al., 1985). The scale has a binary response (no = 0, yes = 1), and a total score was calculated by summing the number of yes responses. Cronbach's alpha in this sample was 0.84. 2.3. Statistical analysis Pearson correlations were used to examine the cross-sectional associations among the study measures, and linear regression analyses were used to evaluate the bidirectional longitudinal associations between rumination and depressive symptoms. For the regression analyses, two separate analyses were conducted: one in which the outcome variable was Wave 3 depressive symptoms and the other in which the outcome variable was Wave 3 rumination. For both analyses, the outcome variable was regressed on Wave 2 rumination and depressive symptoms.1 Because interactions qualify the interpretation of main effects, we first tested whether gender moderated these longitudinal associations. A Gender × Wave 2 Rumination interaction was entered into the equation predicting Wave 3 depressive symptoms, whereas a Gender × Wave 2 Depressive Symptom interaction was entered into the equation predicting Wave 3 rumination. The Wave 2 RRS and PHQ-9 scores were mean deviated (i.e., “centered”) by subtracting the mean of the scale from participants’ observed score on the corresponding measure and the interaction terms were created by multiplying the meandeviated scores with a dummy-coded variable for gender (Aiken and West, 1991). Finally, to test whether the longitudinal associations were incremental to their shared association with demographic characteristics and neuroticism, age and gender were entered in the equations in Model 2, and Wave 2 neuroticism was added to the equations in Model 3. 3. Results Means and standard deviations for study questionnaires are presented in Table 1. Women reported significantly higher levels of rumination than men at Wave 2 (women: M = 8.57, SD = 5.20; men: M = 6.85, SD = 4.60; t = 13.42, p < .001, d = 0.34) and Wave 3 (women: M = 7.64, SD = 5.29; men: M = 6.34, SD = 4.71; t = 9.95, 1 Conducting two separate regression analyses yield results that are identical to a cross-lagged panel model with two observed variables (Newsom, 2015, p. 123).

682

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al.

Table 2 Results from regression analyses predicting Wave 3 depressive symptoms from Wave 2 variables. Predictor

Age Gender (Female) Neuroticism Depressive Symptoms Rumination

Model 1

Model 2 2

Model 3 2

b

SE

β

f

b

SE

β

f

b

SE

β

f2

– – – 0.46*** 0.25***

– – – 0.01 0.01

– – – 0.47 0.22

– – – 0.23 0.05

−0.00*** −0.02 – 0.45*** 0.24***

0.00 0.02 – 0.01 0.02

−0.04 −0.01 – 0.46 0.21

0.00 0.00 – 0.21 0.05

−0.00*** −0.04 0.04*** 0.41*** 0.18***

0.00 0.02 0.00 0.01 0.02

−0.04 −0.02 0.13 0.42 0.16

0.00 0.00 0.02 0.17 0.02

* p < .05. ** p < .01. *** p < .001. Table 3 Results from regression analyses predicting Wave 3 rumination from Wave 2 variables. Predictor

Age Gender (Female) Neuroticism Depressive Symptoms Rumination

Model 1

Model 2

Model 3

b

SE

β

f2

b

SE

β

f2

b

SE

β

f2

– – –

– – – 0.01 0.01

– – – 0.13 0.57

– – – 0.02 0.37

−0.00** 0.03 – 0.12*** 0.62***

0.00 0.02 – 0.01 0.01

−0.03 0.01 – 0.13 0.56

0.00 0.00 – 0.02 0.35

−0.00*** 0.01 0.04*** 0.08*** 0.56***

0.00 0.02 0.00 0.01 0.02

−0.03 0.00 0.13 0.09 0.51

0.00 0.00 0.02 0.01 0.24

.013*** 0.62***

* p < .05. ** p < .01. *** p < .001.

collapsed across gender for all longitudinal analyses. Tables 2 and 3 include the b, SE, and β from the regression analyses examining the longitudinal associations between rumination and depressive symptoms; the tables also include an effect size (f2) for each predictor. After adjusting for Wave 2 depressive symptoms, rumination was positively associated with depressive symptoms at Wave 3, and after adjusting for Wave 2 rumination, depressive symptoms were positively associated with rumination at Wave 3 (Model 1). These associations remained statistically significant when adjusting for age and gender (Model 2). These results supported the hypotheses that rumination would predict residual change in depressive symptoms and that depressive symptoms would predict residual change in rumination; significant associations were also observed when dummy-coded variables for age group were used instead of age. Finally, to test whether the longitudinal associations between rumination and depressive symptoms were incremental to their shared association with neuroticism, we included neuroticism as a covariate in the aforementioned regression analyses. Results from these analyses are presented in Tables 2 and 3 (Model 3). After adjusting for neuroticism as well as demographic characteristics and Wave 2 scores on the outcome variable, Wave 2 rumination remained positively associated with Wave 3 depressive symptoms and Wave 2 depressive symptoms remained positively associated with Wave 3 rumination. Combined, these variables accounted for 40% of the variance in Wave 3 depressive symptoms and 44% of the variance in Wave 3 rumination.2 Because the analyses were based on a community sample, in which the majority of participants reported minimal levels of depressive symptoms at baseline, a sensitivity analysis was conducted for people who reported at least a moderate level of depressive symptoms at baseline to assess the robustness of these findings for people with elevated depressive symptoms. Results based on the 518 people who scored ≥ 10 on the PHQ-9 at baseline paralleled those for the whole sample in suggesting that (a) gender did not moderate the longitudinal associations between rumination and depressive symptoms or depressive symptoms and rumination; (b) rumination predicated residual

change in depressive symptoms; and (c) rumination predicted residual change in depressive symptoms adjusting for demographic characteristics and neuroticism. In comparison, depressive symptoms did not predict residual change in rumination. 4. Discussion We examined the bidirectional, longitudinal association between rumination and depressive symptoms in a large probability sample of Australian adults. Rumination and depressive symptoms were significantly and positively associated within each wave and across waves. These findings extend prior longitudinal studies on rumination and depression by demonstrating that rumination predicts depressive symptoms and that depressive symptoms predict rumination four years later in a large probability sample of Australian adults. The RST (Nolen-Hoeksema, 1991) suggests that rumination may partially account for well-established gender differences in depressive symptoms (Wang et al., 2016) and depressive disorders (Girgus and Yang, 2015). Rumination may account for gender differences in depression in two ways: relative to men, rumination may be higher in women and/or the association between rumination and depression may be stronger for women. Our finding of mean differences in rumination is consistent with prior work (Johnson and Whisman, 2013), but we did not find evidence that gender moderated the association between rumination and depressive symptoms. Although some studies have found that gender moderates the association between rumination and depressive symptoms (e.g., Burwell and Shirk, 2007), our results are consistent with the majority of studies that have not found evidence in support of gender moderation (e.g., Abela and Hankin, 2011). Prior research has found that neurotcism is associated with rumination (e.g., Muris et al., 2005; Nolan et al., 1998; Roberts et al., 1998) and depression (for a meta-analysis, see Kotov et al., 2010), and it is possible that individual differences in neuroticism may contribute to the covariation between rumination and depression. Prior studies suggest that rumination may be a process that mediates the association between neuroticism and depression (e.g., Roelofs et al., 2008; Van Loey et al., 2014), whereas others suggest that rumination and neuroticism may have independent predictive utility (Nolan et al., 1998). In this study, neuroticism was positively associated with both rumination and depressive symptoms in both the cross-sectional (Table 1) and longitudinal analyses (Footnote 1). However, the 2-wave panel design

2 Adjusting for age, gender, and scores on the respective measure at baseline (Wave 2), neuroticism was significantly and positively associated with followup (Wave 3) depressive symptoms, b = .06, SE = .00, β = .19, p < .001, and rumination, b = .05, SE = .00, β = .16, p < .001.

683

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al.

of the current study precluded us from testing mediation (Kraemer et al., 2001). Additional multi-wave longitudinal research, ideally using a genetically-informed sample to adjust for unmeasured confounds (Rutter, 2007), is needed to fully examine the temporal associations between neuroticism, rumination, and depressive symptoms. Although we were unable to test mediation, we found that the longitudinal associations between rumination and depressive symptoms remained statistically significant adjusting for neuroticism. Our results suggest that rumination and neuroticism are independent predictors of depressive symptoms, which increases confidence in the potential causal role of rumination on depressive symptoms (and depressive symptoms on rumination). Prior research suggests that neuroticism and rumination share genetic influences but are differentiated by nonshared environmental influences (du Pont et al., 2019), suggesting that environmental factors that contribute to rumination (e.g., self-generated stressful life events, such as interpersonal conflict; McLaughlin et al., 2012) likely explain the independent association between rumination and depression. Future work should explore other constructs that could explain the longitudinal associations between rumination and depressive symptoms, as rumination covaries with other risk factors for depression, including negative cognitive styles, self-criticism, and dependency (for a review, see Nolen-Hoeksema et al., 2008). Evidence that the longitudinal associations between rumination and depression in probability samples remain statistically significant when adjusting for these other variables would provide additional evidence for rumination as a potential causal factor for depression. Although the focus of the present study was rumination and depressive symptoms in a large population sample, future studies should also examine the potentially causal associations between rumination and depression in clinically depressed samples. Relations between rumination and depression may vary based on depression severity. For example, one study examined rumination and depressive symptoms in a clinically depressed sample and found that rumination did not predict depressive symptoms 6 months later, but that negative affectivitytemperament (a construct simlar to neuroticism) did (Kasch et al., 2001). Similarly, another study found that rumination was associated with current depression severity and length of current depressive episode, but not the duration of prior depressive episodes (Riihimäki et al., 2016). In the current study, we found that when the anlayses were limited to people who reported at least a moderate level of depressive symptoms (i.e., >10 on the PHQ-9), rumination predicted residual change in depressive symptoms whereas depressive symptoms did not predict residual change in rumination. It may be that rumination is more stable for people with elevated symptoms or that these results are due to a restricted range of depressive symptom severity. Another important finding from the study concerns the degree to which rumination remains consistent over time. We found evidence for both relative (i.e., rank-order) and absolute (i.e., mean-level) consistency of rumination over a 4-year interval in a probability sample of adults, which builds on prior research on consistency of rumination in community samples over shorter intervals (Nolen-Hoeksema, 2000) or in undergraduate (Just and Alloy, 1997) or clinical participants (Bagby et al., 2004). Furthermore, the rank-order consistency of rumination obtained in this study (r = 0.66) is comparable to characteristics such as personality that are generally believed to be stable over time. For example, a meta-analysis of the consistency of temperament and adult personality traits based on longitudinal studies with testretest intervals greater than 1 year reported rank-order coefficients ranging from 0.57 to 0.75 across age groups divided by decade for people 22–73 years old (Roberts and DelVecchio, 2000). We also found individual differences in consistency of rumination insofar as the level of rumination changed more than would be expected by chance for nearly 23% of the sample. The percentage of people who exhibited reliable change in rumination differed significantly by age group, with percentages decreasing from younger to middle-aged to older age groups. Similarly, personality traits become increasingly consistent

with age (Roberts and DelVecchio, 2000). Therefore, although rumination appears to be relatively consistent over time for the majority of participants, there is reliable change in rumination for a relatively large subset of participants, particularly younger participants, which provides support for continued research on predictors of change in rumination. Our findings suggest that both depressive symptoms and neuroticism are independent predictors of residual change in rumination in this community sample. Although trait levels of rumination may be largely stable across time, there is evidence of fluctuations in state rumination throughout the day (e.g., Moberly and Watkins, 2008). State rumination, or the act of ruminating in response to a stressor, has been largely studied in the context of experimental studies using a rumination induction. Studying state rumination has been helpful in identifying potential mechanisms that may explain the role of rumination in the onset and maintenance of depressive symptoms, such as difficulty disengaging attention from negative emotional expressions (LeMoult et al., 2013). An experience sampling study found that state rumination moderated the association between negative life events and later negative affect (Moberly and Watkins, 2008), which is consistent with the RST's hypothesis that rumination may contribute to depressive symptoms by augmenting negative affect. Other studies have examined interactions between state and trait rumination; for example, one study found that women with high trait rumination had prolonged cortisol activation during a rumination induction (Shull et al., 2016). These studies underscore the importance of measuring both state and trait rumination to understand variation in rumination and the mechanisms that explain the association between rumination and depressive symptoms. The results from this study should be interpreted in light of several considerations and limitations. First, it is important to consider the length of time between assessments. Causal effects take time to unfold, and careful consideration is needed so that the time interval (i.e., lag) between assessments is long enough for prospective effects to occur, but not too long so that they have already faded (e.g., Selig and Preacher, 2009). The interval between assessments in the current study was 4 years, and additional research involving varying lags between assessments within or across studies is needed to identify the time interval for which rumination may have its maximum impact on depression (and depression have its maximum impact on rumination). Support for the importance of lag interval comes from a longitudinal study involving a community sample in which it was found that higher rumination predicted higher levels of depressive symptoms at 5 months and 3.5 years (and that the longitudinal associations did not differ in magnitude for these two time intervals), whereas higher depressive symptoms predicted higher levels of rumination only at 5 months and not at 3.5 years (Huffziger et al., 2009). Based on the findings from that study, the longitudinal association between depression and subsequent rumination obtained in this study may be smaller in magnitude than what may be expected for a shorter lag between assessments. Second, researchers have criticized the RRS as including items that may overlap in content with measures of depressive symptoms (e.g., Roberts et al., 1998; Segerstrom et al., 2000). Treynor et al. (2003) developed an abreviated version of the RRS that consisted of subscales measuring brooding and reflection. This version of the RRS was developed after the PATH study was initiated, and additional longitudinal research testing the hypotheses of this study using these subscales in probability samples in other countries is needed.3 In summary, results from this longitudinal study of a large probability sample of Australian adults indicated that there were bidirectional associations between rumination and depressive symptoms four

3 We conducted a sensitivity analysis using two items from Treynor et al.’s (2003) brooding scale included in the 10-item version of the RRS used in this study. Results parallel those presented for the full 10-item scale.

684

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al.

years later that do not vary by gender and that were incremental to demographic characteristics and neuroticism. These findings are consistent with the RST in suggesting causal relations between rumination and depression (Nolen-Hoeksema, 1991) and additionally suggest that rumination may be a consequence of depression. Taken together, results support the need for continued research on the role of rumination in the onset, maintenance, severity, and remission of depression.

psychopathology? J. Abnorm. Psychol. 128, 385–396. Dunlap, W.P., Cortina, J.M., Vaslow, J.B., Burke, M.J., 1996. Meta-analysis of experiments with matched groups or repeated measures designs. Psychol. Methods 1, 170–177. Eysenck, S.B.G., Eysenck, H.J., Barrett, P., 1985. A revised version of the psychoticism scale. Pers. Individ. Differ. 6, 21–29. Fanous, A., Gardner, C.O., Prescott, C.A., Cancro, R., Kendler, K.S., 2002. Neuroticism, major depression and gender: a population-based twin study. Psychol. Med. 32, 719–728. Gan, P., Xie, Y., Duan, W., Deng, Q., Yu, X., 2015. Rumination and loneliness independently predict six-month later depression symptoms among Chinese elderly in nursing homes. PLoS One 10, e0137176. Girgus, J.S., Yang, K., 2015. Gender and depression. Curr. Opin. Psychol. 4, 53–60. Grassia, M., Gibb, B.E., 2008. Rumination and prospective changes in depressive symptoms. J. Soc. Clin. Psychol. 27, 931–948. Grossmann, I., Kross, E., 2010. The impact of culture on adaptive versus maladaptive selfreflection. Psychol. Sci. 21, 1150–1157. Huffziger, S., Reinhard, I., Kuehner, C., 2009. A longitudinal study of rumination and distraction in formerly depressed inpatients and community controls. J. Abnorm. Psychol. 118, 746–756. Jacobson, N.S., Truax, P., 1991. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 59, 12–19. Johnson, D.P., Whisman, M.A., 2013. Gender differences in rumination: a meta-analysis. Pers. Individ. Differ. 55, 367–374. Johnson, D.P., Whisman, M.A., Corley, R.P., Hewitt, J.K., Friedman, N.P., 2014. Genetic and environmental influences on rumination and its covariation with depression. Cognit. Emot. 28, 1270–1286. Just, N., Alloy, L.B., 1997. The response styles theory of depression: tests and an extension of the theory. J. Abnorm. Psychol. 106, 221–229. Kasch, K.L., Klein, D.N., Lara, M.E., 2001. A construct validation study of the Response Styles Questionnaire Rumination scale in participants with a recent-onset major depressive episode. Psychol. Assess. 13, 375–383. Kenny, D.A., 1979. Correlation and Causality. John Wiley & Sons, New York. Kotov, R., Gamez, W., Schmidt, F., Watson, D., 2010. Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis. Psychol. Bull. 136, 768–821. Kraemer, H.C., Stice, E., Kazdin, A., Offord, D., Kupfer, D., 2001. How do risk factors work together? Mediators, moderators, and independent, overlapping, and proxy risk factors. Am. J. Psychiatry 158, 848–856. Kroenke, K., Spitzer, R.L., 2002. The PHQ-9: a new depression diagnostic and severity measure. Psychiatr. Ann. 32, 509–515. Lahey, B.B., 2009. Public health significance of neuroticism. Am. Psychol. 64, 241–256. LeMoult, J., Arditte, K.A., D'Avanzato, C., Joormann, J., 2013. State rumination: associations with emotional stress reactivity and attention biases. J. Exp. Psychopathol. 4, 471–484. Lewinsohn, P.M., Steinmetz, J.L., Larson, D.W., Franklin, J., 1981. Depression-related cognitions: antecedent or consequence? J. Abnorm. Psychol. 90, 213–219. Martin, L.L., Tesser, A., 1989. Toward a motivational and structural theory of ruminative thought, In: Uleman, J.S., Bargh, J.A. (Eds.), Unintended Thought. Guilford Press, New York, pp. 306–326. McLaughlin, K.A., Green, J.G., Gruber, M.J., Sampson, N.A., Zaslavsky, A.M., Kessler, R.C., 2012. Childhood adversities and first onset of psychiatric disorders in a national sample of US adolescents. Arch. Gen. Psychiatry 69, 1151–1160. Moberly, N.J., Watkins, E.R., 2008. Ruminative self-focus, negative life events, and negative affect. Behav. Res. Ther. 46, 1034–1039. Muris, P., Roelofs, J., Rassin, E., Franken, I., Mayer, B., 2005. Mediating effects of rumination and worry on the links between neuroticism, anxiety and depression. Pers. Individ. Differ. 39, 1105–1111. Newsom, J.T., 2015. Longitudinal Structural Equation Modeling: A Comprehensive Introduction. Routledge, New York. Nolan, S.A., Roberts, J.E., Gotlib, I.H., 1998. Neuroticism and ruminative response style as predictors of change in depressive symptomatology. Cognit. Ther. Res. 22, 445–455. Nolen-Hoeksema, S., 1987. Sex differences in unipolar depression: evidence and theory. Psychol. Bull. 101, 259–282. Nolen-Hoeksema, S., 1991. Responses to depression and their effects on the duration of depressive episodes. J. Abnorm. Psychol. 100, 569–582. Nolen-Hoeksema, S., 2000. The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. J. Abnorm. Psychol. 109, 504–511. Nolen-Hoeksema, S., Morrow, J., 1991. A prospective study of depression and posttraumatic stress symptoms after a natural disaster: the 1989 Loma Prieta earthquake. J. Pers. Soc. Psychol. 61, 115–121. Nolen-Hoeksema, S., Stice, E., Wade, E., Bohon, C., 2007. Reciprocal relations between rumination and bulimic, substance abuse, and depressive symptoms in female adolescents. J. Abnorm. Psychol. 116, 198–207. Nolen-Hoeksema, S., Wisco, B.E., Lyubomirsky, S., 2008. Rethinking rumination. Perspect. Psychol. Sci. 3, 400–424. Riihimäki, K., Vuorilehto, M., Jylhä, P., Isometsä, E., 2016. Rumination and primary care depression. Eur. Psychiatry 33, S521. Roberts, B.W., DelVecchio, W.F., 2000. The rank-order consistency of personality traits from childhood to old age: a quantitative review of longitudinal studies. Psychol. Bull. 126, 3–25. Roberts, J.E., Gilboa, E., Gotlib, I.H., 1998. Ruminative response style and vulnerability to episodes of dysphoria: gender, neuroticism, and episode duration. Cognit. Ther. Res. 22, 401–423. Roelofs, J., Huibers, M., Peeters, F., Arntz, A., van Os, J., 2008. Rumination and worrying

Role of the funding source Data collection was supported by grants from the National Health and Medical Research Council (grant numbers 179805, 418039). The funding source had no role in the study design; in the collection, analysis and interpretation of the data; or in the decision to submit the article for publication. Limitations Within-person analyses involving repeated assessments, shorter time intervals, and assessment of brooding rumination would provide a stronger test of the potential causal association between rumination and depression. CRediT authorship contribution statement Mark A. Whisman: Conceptualization, Formal analysis, Writing original draft. Alta du Pont: Conceptualization, Writing - original draft. Peter Butterworth: Writing - review & editing. Declaration of Competing Interest None. Acknowledgments The authors wish to thank other PATH investigators (Kaarin Anstey, Helen Christensen, Simon Easteal, Tony Jorm, Bryan Rodgers, and Andrew MacKinnon), Patricia Jacomb, Karen Maxwell, the PATH project interviewers and PATH participants. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jad.2019.09.035. References Abela, J.R.Z., Hankin, B.L., 2011. Rumination as a vulnerability factor to depression during the transition from early to middle adolescence: a multiwave longitudinal study. J. Abnorm. Psychol. 120, 259–271. Aiken, L.S., West, S.G., 1991. Multiple Regression: Testing and Interpreting Interactions. Sage, Thousand Oaks, CA. Anstey, K.J., Butterworth, P., Jorm, A.F., Christensen, H., Rodgers, B., Windsor, T.D., 2004. A population survey found an association between self-reports of traumatic brain injury and increased psychiatric symptoms. J. Clin. Epidemiol. 57, 1202–1209. Anstey, K.J., Christensen, H., Butterworth, P., Easteal, S., Mackinnon, A., Jacomb, T., Maxwell, K., Rodgers, B., Windsor, T., Cherbuin, N., Jorm, A.F., 2011. Cohort profile: the PATH through life project. Int. J. Epidemiol. 41, 951–960. Bagby, R.M., Rector, N.A., Bacchiochi, J.R., McBride, C., 2004. The stability of the Response Styles Questionnaire rumination scale in a sample of patients with major depression. Cognit. Ther. Res. 28, 527–538. Burwell, R.A., Shirk, S.R., 2007. Subtypes of rumination in adolescence: associations between brooding, reflection, depressive symptoms, and coping. J. Clin. Child Adolesc. Psychol. 36, 56–65. Calvete, E., Orue, I., Hankin, B.L., 2015. Cross-lagged associations among ruminative response style, stressors, and depressive symptoms in adolescents. J. Soc. Clin. Psychol. 34, 203–220. Davis, R.N., Nolen-Hoeksema, S., 2000. Cognitive inflexibility among ruminators and nonruminators. Cognit. Ther. Res. 24, 699–711. du Pont, A., Rhee, S.H., Corley, R.P., Hewitt, J.K., Friedman, N.P., 2019. Are rumination and neuroticism genetically or environmentally distinct risk factors for

685

Journal of Affective Disorders 260 (2020) 680–686

M.A. Whisman, et al. as possible mediators in the relation between neuroticism and symptoms of depression and anxiety in clinically depressed individuals. Behav. Res. Ther. 46, 1283–1289. Rutter, M., 2007. Proceeding from observed correlation to causal inference: the use of natural experiments. Perspect. Psychol. Sci. 2, 377–395. Sakamoto, S., Kambara, M., Tanno, Y., 2001. Response styles and cognitive and affective symptoms of depression. Pers. Individ. Differ. 31, 1053–1065. Segerstrom, S.C., Tsao, J.C.I., Alden, L.E., Craske, M.G., 2000. Worry and rumination: repetitive thought as a concomitant and predictor of negative mood. Cognit. Ther. Res. 24, 671–688. Selig, J.P., Preacher, K.J., 2009. Mediation models for longitudinal data in developmental research. Res. Hum. Dev. 6, 144–164. Shull, A., Mayer, S.E., McGinnis, E., Geiss, E., Vargas, I., Lopez-Duran, N.L., 2016. Trait and state rumination interact to prolong cortisol activation to psychosocial stress in females. Psychoneuroendocrinology 74, 324–332. Spasojević, J., Alloy, L.B., 2001. Rumination as a common mechanism relating depressive risk factors to depressio. Emotion 1, 25–37. Spitzer, R.L., Kroenke, K., Williams, J.B.W., Patient Health Questionnaire Primary Care

Study Group, 1999. Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. JAMA 282, 1737–1744. Thompson, R.J., Kircanski, K., Gotlib, I.H., 2016. The grass is not as green as you think: affect evaluation in people with internalizing disorders. J. Affect. Disord. 203, 233–240. Treynor, W., Gonzalez, R., Nolen-Hoeksema, S., 2003. Rumination reconsidered: a psychometric analysis. Cognit. Ther. Res. 27, 247–259. Van Loey, N.E., Oggel, A., Goemanne, A.S., Braem, L., Vanbrabant, L., Geenen, R., 2014. Cognitive emotion regulation strategies and neuroticism in relation to depressive symptoms following burn injury: a longitudinal study with a 2-year follow-up. J. Behav. Med. 37, 839–848. Wang, K., Lu, H., Cheung, E.F.C., Neumann, D.L., Shum, D.H.K., Chan, R.C.K., 2016. “Female preponderance” of depression in non-clinical populations: a meta-analytic study. Front. Psychol. 7, 1398. Wilkinson, P.O., Croudace, T.J., Goodyer, I.M., 2013. Rumination, anxiety, depressive symptoms and subsequent depression in adolescents at risk for psychopathology: a longitudinal cohort study. BMC Psychiatry 13, 250.

686