Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Aug 1.
Published in final edited form as: J Psychopathol Clin Sci. 2023 Jun 5;132(6):694–703. doi: 10.1037/abn0000826

Pre-Onset Predictors of Chronic-Intermittent Depression from Early Adolescence to Early Adulthood

Daniel N Klein 1, Greg Perlman 2, Scott M Feltman 2, Roman Kotov 2
PMCID: PMC10524144  NIHMSID: NIHMS1901602  PMID: 37276087

Abstract

Individuals with prolonged or frequent episodes account for a disproportionate share of the burden of depression. However, there are surprisingly few data on whether individuals at risk for developing chronic-intermittent depression (CID) as opposed to briefer, infrequent depressive episodes (time-limited depression [TLD]) can be distinguished before their first depressive episode. We followed a community sample of 465 never-depressed females on five occasions from age 14 to 20 years and examined whether 18 pre-onset clinical and psychosocial variables prospectively predicted CID. The CID group accounted for 40% of depressed cases but 84% of the cumulative time depressed in the sample. Participants with CID (N = 60) exhibited significantly higher pre-onset levels of 16 of the 18 risk factors than the never-depressed group (N = 315). The TLD group (N = 90) had significantly higher pre-onset levels of 9 risk factors than never-depressed participants. Finally, the CID group had significantly higher levels of 9 risk factors than the TLD group, 5 of which were similar in TLD and never-depressed participants. These findings indicate that differences between CID and TLD are evident before onset and suggest that the liability to CID may be both greater than, and somewhat different from, the liability to TLD. Moreover, they suggest that individuals at risk for a malignant course of depression can be targeted for prevention and early intervention.

Keywords: depression, persistent, chronic, recurrent, prediction

General Scientific Summary

A minority of individuals with a chronic-intermittent course account for a disproportionate share of the burden of depression. However, there are few data on whether they can be distinguished from individuals who will develop more time-limited depressions (TLD) before the onset of the disorder. We identified a number of pre-onset variables that distinguished individuals who later developed chronic-intermittent depression (CID) from those who developed TLD. Moreover, individuals who developed TLD did not differ from those who remained depression-free on some of these predictors, suggesting that the liability to CID may be both stronger and somewhat different than the liability to TLD.

Depressive disorders (DD) are remarkably common and account for significant suffering and impairment (Kessler, Nierenberg, Pennix, Wang, Wittchn, & Ziobrowski, 2022). However, despite an enormous literature, progress in reducing prevalence, understanding etiopathogenesis, and developing more effective treatments has been disappointing (Kapur, Phillips, & Insel, 2012; Ormel, Hollon, Kessler, Cuijpers, & Monroe, 2022). These points may be related. The high prevalence of depression masks extraordinary clinical and etiological heterogeneity. Critically, only a minority of cases of DD, as yet poorly defined, may account for a disproportionate share of its burden (Monroe & Harkness, 2022). Combining this subgroup with all others who have experienced depression likely obscures the effects of key etiological and pathophysiological factors and limits development of more targeted and effective interventions.

According to the World Health Organization, DD rank first among mental disorders, and near the top of all mental and physical disorders, in their index of global burden of disease (James et al., 2018). Depression’s prominent position stems from its high prevalence, often early (i.e., adolescent or young adult) onset, and often chronic or recurrent course (Kessler et al., 2022). However, the course of DD is extremely heterogenous. The majority of individuals experience only 1–2 episodes over the course of their lifetimes, and only a small minority of episodes last as long as 1–2 years (Klein & Allmann, 2014; Monroe & Harkness, 2011). Hence, a disproportionate share of the personal and societal costs of DD are due to the subgroup of individuals with a relatively early onset and a chronic or highly recurrent course (Pettit, Lewinsohn, Roberts, Seeley, & Monteith, 2009).

One of the major aims of research on DD is to reduce heterogeneity by identifying more homogenous subgroups. However, the past century of research on clinical subtypes has, with rare exceptions (e.g., psychotic features), failed to produce evidence of meaningful differences (Baumeister & Parker, 2012; Klein, 2022). As a result, a continuum view has become increasingly dominant, in which variation in depression is viewed as reflecting quantitative differences in severity (Conway, Krueger, & HiTOP Consortium Executive Board, 2021; Ruscio, 2019). However, severity is generally studied cross-sectionally, neglecting longitudinal course (Klein, 2008). Symptom severity at one point in time may not be representative of severity over a longer period. Indeed, the orthogonality of cross-sectional symptom severity and course is reflected in our current classification system where major depressive episodes (MDEs) are generally understood to be time-limited and can vary from mild to severe in intensity, and persistent depressive disorder (PDD) is defined as a chronic condition that can also vary in severity from mild (i.e., dysthymic disorder) to severe (i.e., chronic major depression) or both at different times (“double depression” - MDE “superimposed” on dysthymic disorder).

The majority of research on the etiopathogenesis of DD has disregarded the prior course of depressed participants. However, if there are important differences between subgroups with different course trajectories, it may obscure etiological and pathophysiological influences that are specific to each subgroup. Indeed, the literature indicates that persistence, and in many but not all studies, recurrence, are associated with subthreshold/minor depressive symptoms, comorbid anxiety and externalizing disorders, a family history of psychopathology, especially DD, and a history of childhood maltreatment and maladaptive parenting. In addition, persistence and recurrence are often associated with lower levels of extraversion, particularly the facet of positive emotionality (PE); higher levels of neuroticism, rumination, and self-criticism; and poorer interpersonal and academic/work functioning (Brouwer, Williams, Kennis, Fu, Klein, Cuijpers, & Bockting, 2019; Buckman, Underwood, Clarke, Saunders, Hollon, Fearon, & Pilling, 2018; Burcusa & Iacono, 2007; Hardeveld, Spijker, De Graaf, Nolen, & Beekman, 2010; Hölzel, Härter, Reese, & Kriston, 2011; Klein & Allmann, 2014; Nanni, Uher, & Danese, 2012; Schramm, Klein, Elsaesser, Furukawa, & Domschke, 2020).

These differences raise the possibility of identifying individuals are at risk for developing a persistent or recurrent course prior to onset of DD. This has enormous clinical and public health implications, as it creates an opportunity to intervene early to pre-empt a malignant course. Unfortunately, with the exception of several studies focusing exclusively on recurrence (Eaton, Shao, Nestadt, Lee, Bienvenu, & Zandi, 2008; Pettit, Hartley, Lewinsohn, Seeley, & Klein, 2013; Wilson, Vaidyanathan, Miller, McGue, & Iacono, 2014), this literature is based on studies that began after participants developed DD. As a result, it is unclear if the characteristics that distinguish individuals with persistent or recurrent DD from those with briefer or fewer episodes reflect liabilities to more chronic/recurrent course trajectories, if they are pathoplastic factors that influence the course of DD after onset (e.g., coping or emotion regulation), are consequences of experiencing a depressive episode (e.g., via stress sensitization or stress generation), or reflect reporting biases (e.g., “effort after meaning”; Brown, Sklair, Harris, & Birley, 1973) (Burcusa & Iacono, 2007; Klein & Allmann, 2014; Monroe & Harkness, 2022). Thus, prospective studies beginning before a first lifetime DD episode and continuing long enough to evaluate persistence and recurrence are critical.

Additionally, few studies of correlates and predictors of persistence and recurrence have included a comparison group of never-depressed participants. As a result, they cannot address whether differences between persistent and non-persistent or recurrent and non-recurrent cases are specific to these course-based subtypes or apply, perhaps to varying degrees, to depression in general. If one group of depressed individuals differs both from the other group of depressed individuals and the never-depressed group, and the latter groups do not differ from each other, it would provide preliminary evidence for specific liabilities.

The current study capitalizes on a prospective longitudinal cohort study of a community sample of never-depressed early adolescent females (M = 14.4 years) followed into early adulthood (M =20.3 years) (Michelini, Perlman, Tian, Mackin, Nelson, Klein, & Kotov et al., 2021). The Adolescent Development of Emotions and Personality Traits (ADEPT) study collected a rich battery of risk factors at baseline and conducted five subsequent waves of semi-structured interview-based assessments of DD. We have previously reported on predictors of first-lifetime onset of DD in the first three years of follow-up in this sample (Michelini et al., 2021). Although we presented preliminary data on predictors of persistent/recurrent DD in that paper, with only three years of follow-up there was limited opportunity to observe the course of participants who developed DD. The current paper extends this work with a novel approach to defining course-based subtypes and the greater resolution afforded by six years of follow-up.

The parent study focused on female adolescents to maximize the rate of first-onset DD. The adolescent period is also relevant for the present aims, as youth who develop early-onset chronic or recurrent DD often experience problems with key developmental tasks that become increasingly difficult to overcome with age (Hammen, Brennan, Keenan-Miller, & Herr, 2008), and their longer cumulative experience of depression makes a disproportionate contribution to the burden of the disorder (Kessler et al., 2022).

For this paper, we selected a set of clinical and psychosocial variables spanning the domains of pre-existing psychopathology, parental psychopathology, parenting, peer and academic functioning, and personality that have been linked to the course of depression in previous studies (Brouwer et al., 2019; Buckman et al., 2018; Burcusa & Iacono, 2007; Hardeveld et al., 2010; Hölzel et al., 2011; Klein & Allmann, 2014; Schramm et al., 2020). A key question in designing our analyses was how to define course-based subtypes. One option was to examine recurrent (i.e., ≥ 2 episodes) versus single episode cases. However, many argue that recurrent depression should be defined more stringently (i.e., a greater number of episodes), and a recent review concluded that empirical studies do not support existing conventions for defining its constituent elements (i.e., episode onset, recovery, recurrence; de Zwart, Jeronimus, & de Jonge, 2019). As a result, Monroe and Harkness (2022) argue that rather than defining recurrent depression by number of episodes, it should be conceptualized as a latent liability that is manifested as frequent episodes with relatively short inter-episode periods.

Another option was to examine PDD vs non-persistent DD. However, this is also problematic. In the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5, American Psychiatric Association, 2013), the cutoff for PDD is two years in adults and one year in children and adolescents. These durations are not grounded in evidence, and definitions based on the proportion of time with a DD have greater validity (Mondimore, Zandi, MacKinnon, McInnis, Miller, Schweizer, … & Potash, 2007; Silver, Olino, Carlson, & Klein, 2020).

Finally, and most importantly, it is problematic to examine either recurrence or persistence without considering the other. For example, it is possible to experience a single episode of DD that lasts for many years, or several brief episodes separated by decades. In the first example, a relatively poor course is classified as the more benign subtype, and in the second example, a relatively benign course is classified as the poorer outcome (Klein & Allmann, 2014). Additionally, many cases of DD are both persistent and recurrent. For example, individuals with dysthymia almost always experience recurrent superimposed MDEs (Klein, Shankman, & Rose, 2006). Moreover, the greater the number of episodes an individual has, the more likely it is that at least one episode will be prolonged. Thus, an approach that encompasses both persistence and recurrence is necessary (Klein & Allmann, 2014; Pettit et al., 2009).

For the present study, we adopted a provisional definition emphasizing the cumulative duration of depression regardless of whether it takes the form of one extended episode or multiple briefer episodes. That allowed us to include both persistent DD and frequent but less persistent MDEs. Given that the study period spanned ages 14–20, we followed the DSM criteria for PDD in youth requiring a minimum duration of 12 months of depressive symptoms meeting criteria for PDD. Notably, this is consistent with the empirically-based definition derived in a community sample of adolescents by Pettit et al. (2009). However, unlike the DSM, we allowed well-intervals of > 2 months to include highly recurrent cases. We refer to individuals with a cumulative duration of depression of ≥ 1 year as having “chronic-intermittent depression” (CID), and those with a cumulative duration of < 1 year as having “time-limited” depression (TLD).

Method

Participants

ADEPT includes 550 females age 13.5–15.5 years from Suffolk County, New York, recruited through a commercial mailing list, word-of-mouth, school districts, on-line ads, and community postings (see Michelini et al., 2021). Inclusion criteria were English fluency, ability to complete questionnaires, and a biological parent willing to participate. Exclusion criteria were intellectual disability and lifetime history of major depressive disorder (MDD) or PDD. Youth with a history of DD not otherwise specified (D-NOS), defined as depression episodes that did not meet criteria for MDD or PDD but were associated with impairment or treatment-seeking, were allowed to participate (N =29).

Participants completed assessments of DD at baseline and 9, 18, 27, 36, and 72 months later. Participation rates at each follow-up wave were 95.8%, 94.5%, 92.4%, 92.2%, and 85.3%, respectively. After excluding individuals who developed bipolar disorder or did not participate in the 6-year follow-up, the analysis sample was 465 participants.

Of the 23 baseline variables in this study (Tables 1 and 2), only one differed significantly between participants in (N = 465) and not in (N = 85) the analysis sample. Those in the analysis sample had a significantly higher grade-point average (M = 91.00 [SD = 5.70] vs. M = 88.38 [SD = 7.46]), t(94.71) = 2.97, p = .004.

Table 1:

Descriptive Characteristics of Sample

Variable Never-Depressed (N = 315) Time-Limited Depression (N = 90) Chronic-Intermittent Depression (N = 60) Test statistic
Age at W1, M (SD) 14.40 (0.62) 14.26 (0.59) 14.48 (0.63) F (2, 462) = 2.57, p = .078
Age at W6, M (SD) 20.34 (0.94) 20.21 (0.89) 20.38 (0.85) F (2, 455) = 0.80, p = .450
Non-White 34 (10.8%) 12 (13.3%) 8 (13.3%) χ2(2) = 0.64, p = .730
Hispanic 32 (10.2%) 13 (14.4%) 6 (10.0%) χ2(2) = 1.38, p = .500
Mother Bachelor’s Degree 170 (54.7%) 48 (53.3%) 28 (46.7%) χ2(2) = 1.29, p = .520
Father Bachelor’s Degree 159 (52.1%)a 44 (51.2%)a 20 (33.9%)b χ2(2) = 6.68, p = 0.035
Open in a new tab

Note. Groups with different subscripts differ significantly.

Table 2:

Premorbid Predictors of Chronic-intermittent and Time-Limited Depression

Variable Never-Depressed (1) Time-Limited Depression (2) Chronic/Intermittent Depression (3) 1 vs 2 1 vs 3 2 vs 3
% (N) % (N) % (N) Phi Phi Phi
Lifetime Depression Not Otherwise Specified by W1 2.2 (7) 7.8 (7) 25.0 (15) .126, p = .011 .355, p < .001 .238, p = .003
Lifetime Anxiety Disorder by W1 19.7 (62) 37.8 (34) 36.7 (22) .177, p < .001 .149, p = .004 −.011, p = .890
Lifetime Behavior Disorder by W1 5.4 (17) 8.9 (8) 13.3 (8) .060, p = .225 .117, p = .024 .071, p = .388
Parental Depressive Disorder 19.7 (61) 37.1 (33) 28.8 (17) .171, p = .001 .082, p = .115 −.086, p = .298
Parental Anxiety Disorder 43.5 (137) 52.2 (47) 51.7 (31) .073, p = .142 060, p = .243 −.005, p = .947
Parental Substance Disorder 19.7 (62) 27.8 (25) 38.3 (23) .082, p = .099 .163, p = .002 .111, p = .175
M (SD) M (SD) M (SD) r r r
Parental care 3.44 (0.45) 3.37 (0.48) 3.17 (0.58) −.063, p = .224 −.198, p < .001 −.185, p = .031
Critical comments 0.18 (.51) 0.16 (0.63) 0.39 (0.80) .014, p = .788 .136, p = .011 .159, p = .059
Poorer relationship with parents 1.79 (0.57) 1.91 (0.64) 2.18 (0.79) .084, p = .102 .222, p < .001 .182, p = .033
Poorer relationship with best friend 1.49 (0.43) 1.56 (0.45) 1.73 (0.60) .058, p = .250 .185, p < .001 .168, p = .045
Peer victimization 1.35 (0.38) 1.46 (0.46) 1.57 (0.56) .115, p = .021 .195, p < .001 .112, p = .175
Grade point average 91.46 (5.38) 90.93 (6.26) 88.81 (6.02) −.039, p = .441 −.177, p = .001 −.168 , p = .041
SNAP positive temperament 0.81 (0.16) 0.76 (0.20) 0.68 (0.25) −.112, p = .027 −.258, p < .001 −.183, p = .029
PANAS positive affectivity 4.13 (0.76) 3.93 (0.84) 3.56 (1.00) −.106, p = .035 −.254, p < .001 −.200, p = .016
Neuroticism 2.60 (0.75) 2.96 (0.77) 3.15 (0.83) .198, p < .001 .258, p < .001 .117, p = .154
Rumination 1.51 (0.51) 1.86 (0.65) 2.07 (0.76) .261, p < .001 .353, p < .001 .151, p = .071
Self-criticism 1.90 (0.77) 2.27 (0.92) 2.60 (0.98) .186, p < .001 .301, p < .001 .168, p = .042
Self-harm .04 (.10) .06 (.11) .16 (.25) .096, p = .055 .313, p < .001 .258, p = .002
Open in a new tab

Note. W1 = Wave 1; SNAP = Schedule for Nonadaptive and Adaptive Personality; PANAS = Positive and Negative Affect Schedule.

Informed assent and consent were obtained from minor participants and their parents, respectively, and consent was obtained from participants after age 18. Participants were financially compensated. The study was approved by the Stony Brook University Institutional Review Board.

Measures

Depressive disorders.

At each wave, participants were administered the Kiddie Schedule for Affective Disorders and Schizophrenia, Present and Lifetime Version (K-SADS-PL; Kaufman, Birmaher, Brent, Rao, Flynn, Moreci, … & Ryan, 1997). The K-SADS-PL is a semi-structured interview for current and lifetime psychiatric diagnoses in youth based on the DSM-IV (American Psychiatric Association, 1994). Follow-up interviews covered the period since the prior assessment. Information about the onset and duration of each depressive episode was elicited. Interviews were conducted by trained staff supervised by licensed clinical psychologists. At waves 1, 3, 5, and 6 interviews were conducted in-person; interviews at waves 2 and 4 were by telephone. An independent rater derived diagnoses from recordings of 48 interviews selected from several waves to assess interrater reliability. Kappas for MDD and PDD were .73 and .85, respectively. Following Keller, Lavori, Friedman, Nielsen, Endicott, McDonald-Scott and Andreasen (1987) and Lyketsos, Nestadt, Cwi, Heithoff, and Eaton (1994), we constructed monthly ratings of DDs throughout the follow-up period.

We defined a CID course as ≥ 12 months of meeting criteria for MDD or PDD. A TLD course was defined as < 12 months meeting criteria for MDD or PDD. Months with D-NOS during the follow-up were not used to define either group. History of D-NOS at baseline was our measure of pre-existing subthreshold/minor depression.

Anxiety and Behavior Disorders.

The baseline K-SADS-L assessed prior history of anxiety and related disorders (specific phobia, agoraphobia, and generalized, social, and separation anxiety, panic, obsessive-compulsive, and post-traumatic stress disorder) and behavior disorders (attention-deficit/hyperactivity, conduct, and oppositional-defiant disorder, and disruptive behavior disorder not otherwise specified [DBD-NOS]). DBD-NOS was included due to the low rate of behavior disorders in this sample of adolescent girls. Interrater reliability based on 29 audio-recorded interviews at baseline that were independently scored by a second rater ranged from kappas of .51 (social anxiety) to .83 (specific phobia).

Parental Psychopathology.

Parental lifetime psychopathology was assessed with the Structured Clinical Interview for DSM-IV (SCID) (First, Spitzer, Gibbon, & Williams, 2002). The SCID was administered at baseline to the biological parent accompanying the participant (93.9% mothers) by trained interviewers who were closely supervised by clinical psychologists. Interrater reliability (kappa) for 25 audio-recorded interviews ranged from .62 (generalized anxiety disorder) to 1.00 (dysthymic disorder). For lifetime psychopathology in the non-participating biological parent, the participating parent was interviewed using the Family History Screen (Weissman, Wickramaratne, Adams, Wolk, Verdeli, & Olfson, 2000). We examined co-parent diagnoses of depression, anxiety, and substance disorders. Diagnoses of both parents were combined to reflect whether at least one parent ever had the disorder.

Personality Traits.

The facet of extraversion most closely linked to depression is PE (Watson, Stasik, Ellickson-Larew, & Stanton, 2015). Hence, we selected two measures of PE from the baseline assessment. The 26-item Positive Temperament scale from the Schedule of Non-Adaptive and Adaptive Personality (SNAP; Simms & Clark, 2006) is rated using a True/False format. Alpha in this sample was .87. Respondents also completed the 8 items of the PANAS-X Positive Affectivity scale (Watson & Clark, 1994) based on “how they feel in general, that is, on the average” using a 5-point scale. Alpha was .94.

We selected four measures from the domain of NE/neuroticism: a broad measure of neuroticism and three measures of clinical traits reflecting more specific aspects of NE: rumination, self-criticism, and self-harm. The Big Five Inventory (BFI; John & Srivastava, 1999) neuroticism scale includes 8-items rated on a 5-point scale; alpha in this sample was .83. The Ruminative Responses Scale (RRS) of the Response Styles Questionnaire (Nolen-Hoeksema & Morrow, 1987) consists of 22 items. Respondents rate how often they experience thoughts and behaviors when feeling sad on a 4-point scale; alpha was .95. Bagby, Parker, Joffe, and Buis’ (1994) revision of the Self-Criticism subscale of the Depressive Experiences Questionnaire consists of 9 items rated on a 5-point scale; alpha was .86. The SNAP Self-Harm scale (Simms & Clark, 2006) includes 16 True/False items reflecting self-destructive thoughts and behaviors and low self-esteem; alpha was .87.

Academic Functioning.

Participant’s grade point average (GPA) during the 9 months before the baseline assessment was obtained from school transcripts.

Parents and Peers.

At baseline, adolescents completed the Parental Bonding Instrument (PBI: Parker, Tupling, & Brown, 1979), which assesses each parent’s behavior towards them throughout childhood. We used the 12-item Care scale, rated on a 5-point scale, which is the subscale most consistently associated with depression (e.g., Plantes, Prusoff, Brennan, & Parker, 1988). To reduce the number of variables, we aggregated ratings for both parents. Alpha was .89.

Additionally, the participating parent was administered the Five-Minute Speech Sample (Magana, Goldstein, Karno, Miklowitz, Jenkins, & Falloon, 1986), a widely-used measure of Expressed Emotion (EE). Parents are asked to talk in an unstructured fashion about the participating adolescent for 5 minutes. Research assistants, trained by the instrument’s developer, scored the number of critical comments from audiotapes. This is the component of EE that consistently predicts the course of psychopathology (Ma, Chan, Chung, Ng, Hui, Suen, & Chen, 2021). Interrater reliability, assessed with the intraclass correlation from 100 independently coded audiotapes, was .92.

Adolescents’ perception of their current relationships with their parents and best friend were assessed with the Network of Relationships Inventory – Relationship Qualities Version (NRI); Furman & Buhrmester, 2009). We included four 3-item scales for each parent: satisfaction, approval, conflict, and criticism. The four scales for mothers and fathers were summed and aggregated. We included two additional 3-item scales for the best friend: pressure and exclusion. The six scales for the best friend were also aggregated. Higher scores indicated worse relationships. The composite scales for parents and best friend had alphas of .83 and .94, respectively.

Finally, participants completed the Revised Peer Experiences Questionnaire (RPEQ; De Los Reyes & Prinstein, 2004), which assesses peer victimization and bullying perpetration in adolescents. We used the three victimization scales: reputational and relational victimization and overt bullying. Items were answered using a 5-point frequency scale. We combined the three scales to form a 12-item measure of peer victimization; alpha was .86.

Data Analysis

Participants with CID, TLD, and no history of depression through wave 6 were compared on descriptive characteristics using chi-square, t-tests (equal variance not assumed when Levene’s test was significant), and one-way analysis of variance (ANOVA) tests. Significant omnibus effects were followed with pairwise Fishers Exact and Tukey HSD tests.

To examine differences between the three groups on the 18 premorbid risk factors, we ran three sets of analyses, one comparing the two depressed groups, and two comparing each depressed group to the never-depressed group. We used Pearson correlations for continuous risk factors (correlations have the same interpretation as t-tests but provide a more direct index of effect size) and phi coefficients for categorical risk factors.

Results

Sixty (12.9%) participants met criteria for CID, 90 (19.4%) met criteria for TLD, and 315 (67.6%) never experienced an episode of MDD or PDD during follow-up. The groups did not differ significantly on age at baseline and final assessments, race (white vs non-white), ethnicity (hispanic vs non-hispanic), and whether their mothers had a bachelor’s degree or higher at study entry. However, fathers of participants with CID were significantly less likely to have graduated from college than fathers of TLD and never-depressed participants (Table 1).

By definition, the CID group met criteria for a DD for a significantly greater number of months (M = 33.1; SD = 19.3) than the TLD group (M = 4.2; SD = 2.7), t (60.6) = 11.55, p < .001. Nonetheless, it is striking that although the CID group comprised 40.0% of participants with a DD, they accounted for 84.2% of the total months meeting criteria for a DD.

The CID (M = 1.55: SD = 0.89) and TLD (M = 1.31: SD = 0.63) groups did not differ on number of episodes of DD (defined, by convention, as meeting criteria for MDD or PDD followed by at least 2 months with no more than 1 or 2 mild symptoms; Keller et al., 1987), t (97.6) = 1.80, p =.075. However, the age of onset of DD was significantly earlier for the CID (M =15.83: SD = 1.79), than TLD (M = 18.24: SD = 1.74), group, t (147) = 8.21, p < .001.

Comparisons between the CID, TLD, and never-depressed groups on the 18 premorbid risk factors appear in Table 2 and are summarized in Figure 1. The CID group differed significantly from the never-depressed group on 16 risk factors. Participants with CID had higher rates of W1 (wave 1) D-NOS, anxiety, and behavioral disorders and parental history of substance disorders; lower parental care; more parent critical comments; poorer relationships with parents and best friends; greater peer victimization; lower GPA; lower SNAP positive temperament and PANAS-X positive affectivity; and higher neuroticism, rumination, self-criticism, and self-harm.

Figure 1.

Figure 1.

Open in a new tab

Summary of Comparisons Between Never-Depressed (ND), Time-Limited Depressed (TLD), And Chronic-Intermittent Depressed (CID) Groups on Baseline (Pre-Onset) Risk Factors

The TLD group differed significantly from the never-depressed group on 9 risk factors. Participants with TLD exhibited higher rates of W1 D-NOS, anxiety disorders, and parental history of depression; greater peer victimization; lower SNAP positive temperament and PANAS-X positive affectivity; and higher neuroticism, rumination, and self-criticism. Except for parental depression, all risk factors that distinguished the TLD and never-depressed group also differentiated CID from never-depressed participants.

Finally, in direct comparisons, the CID and TLD groups differed significantly on 9 of the 18 premorbid risk factors. Participants with CID had a higher rate of W1 D-NOS; lower parental care; poorer relationships with parents and best friends; lower GPA, SNAP positive temperament and PANAS-X positive affectivity; and higher self-criticism and self-harm than participants with TLD. The TLD group did not exhibit significantly greater risk on any of these variables than the CID group. Of the 9 risk factors that differentiated the CID and TLD groups, participants with TLD did not differ significantly from never-depressed participants on 5 (parental care, relationships with family and friends, GPA, and self-harm).

Discussion

Individuals with a persistent and/or recurrent course account for a disproportionate share of the burden of DD. Identifying those most likely to have a chronic-intermittent course should be a major research focus (Monroe & Harkness, 2022). Although numerous studies have examined correlates of persistence and recurrence, there is a surprising paucity of data on pre-onset predictors of CID. Rather, almost all studies have focused on individuals who have already experienced an onset of DD, typically many years before the study commenced (Monroe & Harkness, 2011). Predictors of the course of an established DD may differ from those that predict the development of CID in the first place. This is one of the only studies to prospectively examine a wide range of variables in a sample with no history of DD to identify predictor of who will develop DD with a more versus less chronic-intermittent course.

By definition, the CID group was depressed for more time than the TLD group. Still, the magnitude of the difference was striking. Over the 72-month follow-up, participants with CID were depressed for a mean 33 months, whereas those with TLD were depressed for an average of 4 months. Although the CID group comprised 40% of participants who developed first-onset DD, they accounted for 84% of the sample’s cumulative time depressed. Interestingly, participants with CID and TLD did not differ significantly on number of episodes during follow-up, indicating that it was the duration of episodes that distinguished the groups.

At baseline, the group that subsequently developed CID differed significantly from the never-depressed group on 16 of 18 risk factors examined. Participants who subsequently experienced TLD also differed from never-depressed participants on a number of premorbid risk factors (9 of 18), but there were fewer differences than observed for the CID group. Effect sizes were generally small, but significant differences were all in the expected direction. Only 1 risk factor, parental depression, significantly distinguished TLD, but not CID, from never-depressed participants (although the depressed groups did not differ significantly). Finally, the CID and TLD groups differed significantly on 9 premorbid risk factors. On each variable, participants who subsequently developed CID exhibited higher levels of risk than those who developed TLD. Moreover, for 5 of the 9 variables distinguishing the CID and TLD groups, participants with TLD did not differ significantly from never-depressed participants. Thus, participants with CID had significantly higher levels of 5 premorbid risk factors than both TLD and never-depressed participants, who did not differ from each other.

This pattern indicates that some risk factors are common to depression in general, although in some instances the effect is greater in CID than TLD. However, in addition to these shared risk factors, a number of risk factors are specific to CID, suggesting that this group has distinctive vulnerabilities (Figure 1). Importantly, findings indicating specificity for CID are not a function of statistical power, as the CID group was substantially smaller than the two other groups, so the TLD versus never-depressed comparisons had considerably greater power than comparisons between the depressed groups.

The 5 variables distinguishing the CID group from both the TLD and never-depressed groups but not differing between the two DD groups were parental care throughout childhood, current relationships with parents and best friend, GPA, and self-harm. Although most variables were derived from adolescents’ reports, GPA was from school records.

Our results for lower parental care and poorer current relationships with parents are consistent with retrospective (Nanni et al., 2012; Nelson, Klumparendt, Doebler, & Ehring, 2017) and prospective (Widom, DuMont, & Czaja, 2007; Wilson et al., 2014) studies reporting that childhood maltreatment and maladaptive parenting are among the strongest predictors of chronicity and recurrence in DD. Additionally, our finding that a poorer premorbid relationship with a best friend specifically predicted CID is consistent with theory and research indicating that problematic close relationships are associated with persistence of depression, although few studies have assessed this before the onset of DD (Hames, Hagan, & Joiner, 2013; Hölzel et al., 2011; Klein & Allmann, 2014). Poorer school performance was also specific to CID, consistent with evidence that DD with a more persistent and recurrent course is associated with greater functional impairment (e.g., Buckman et al., 2018; Schramm et al., 2020). However, this goes further in suggesting that poor academic performance precedes the onset of CID. Notably, the CID group’s mean GPA was in the high B range, indicating only subtle impairment that may reflect psychopathology and depressogenic traits already evident at W1. Finally, SNAP self-harm, which includes items tapping self-destructive thoughts and behaviors and low self-esteem, was the only personality variable predicting the development of CID but not TLD. Interestingly, in factor analyses of the SNAP, self-harm loads (in opposite directions) on both the negative and positive temperament factors (Simms & Clark, 2006), a combination that may predispose to a particularly malignant course.

Four premorbid risk factors, history of D-NOS, lower SNAP and PANAS PE, and higher self-criticism, were present to a significantly greater degree in CID than TLD, but also distinguished TLD from the never-depressed group. This first finding is consistent with evidence that subthreshold/minor depression not only predicts the first onset of DD (Klein, Glenn, Kosty, Seeley, Rohde, Lewinsohn, 2013; Lee, Stockings, Harris, Doi, Page, Davidson, & Barendregt, 2019), but also heralds a more recurrent course (Pettit et al., 2013). Thus, they are a particularly attractive target for early intervention.

Cross-sectional studies have consistently found that low PE is associated with DD, and follow-up studies report that it predicts a poorer course of depression in individuals with DD, but research examining the prediction of first-onset of DD has yielded inconsistent results (Klein et al., 2011). These latter studies have not considered the subsequent course of depression. In a rare exception, Wilson et al. (2014) found that premorbid PE was lower in a community sample of never-depressed early adolescents who subsequently developed recurrent MDD by young adulthood compared to adolescents who developed a single episode of MDD. Similarly, we found that premorbid PE, assessed using both the SNAP and the PANAS-X, was significantly lower in adolescents who developed CID than in adolescents who developed TLD. In a comparison that was not reported by Wilson et al. (2014), we also found that adolescents with TLD exhibited lower premorbid PE on both measures than adolescents who never developed DD. Finally, we found that higher premorbid self-criticism predicted the development of both CID and TLD, but that self-criticism was even greater in CID than TLD. Thus, along with subthreshold/minor depression, our results suggest that low PE and self-criticism are quantitative, or graded, risk factors that are associated with DD in general, but are evident to an even greater degree in individuals who develop a more malignant course.

CID and never-depressed participants differed on 3 premorbid risk factors that did not differentiate TLD from either of the other groups: W1 behavior disorders, parental SUD, and parental critical comments. The first 2 effects are concordant with the common but inconsistent finding of greater externalizing psychopathology in PDD and recurrent MDD (Burcusa & Iacono, 2007; Klein & Allmann, 2014; Schramm et al., 2020). Similar to our findings, in the 2 studies most like ours, neither Pettit et al. (2013) nor Wilson et al. (2014) found that adolescents with recurrent and single episode MDD differed on premorbid externalizing psychopathology.

The pattern of results for parental criticism is consistent with our other parenting measures except that the difference between the CID and TLD groups was only a trend (p = .059). Trends should not be overinterpreted but as critical comments were coded from parents’ free-speech samples, it lends confidence to our other findings on parenting based on youth self-report.

Four risk factors distinguished both depressed groups from never-depressed participants but were not associated with the subsequent course of DD: history of anxiety disorder before W1, and peer victimization, neuroticism, and rumination. Multiple studies have shown that each of these variables predicts subsequent DD and/or increases in depression symptoms (e.g., Cohen, Thakur, Young, & Hankin, 2020; Hill, Mellick, Temple, & Sharp, 2017; Klein et al., 2011, 2013). However, few studies have examined associations of premorbid levels of these variables with the course of DD. In rare exceptions, Wilson et al. (2014) reported that premorbid anxiety disorders distinguished recurrent from single episode MDD, but Pettit et al. (2013) found no difference. Our findings suggest that these 4 variables are general risk factors for DD.

We were surprised the CID group did not differ from the never-depressed and TLD groups on parental DD. The literature indicates that both PDD (Schramm et al., 2020) and recurrent DD (Burcusa & Iacono, 2007; Sullivan, Neale, & Kendler, 2000) exhibit greater familial aggregation of DD. Additionally, previous research indicates that both PDD and recurrent DD are specifically associated with higher rates of PDD and recurrent DD in relatives (Pettit et al., 2013; Schramm et al., 2020). Unfortunately, we did not assess the course of DD experienced by parents of our participants. Finer-grained DD phenotyping in parents may have revealed intergenerational transmission of CID. We were also surprised that none of the groups differed on parental anxiety disorders, as numerous studies have documented familial and genetic associations between anxiety and depression (e.g., Hettema, 2008; Lawrence, Murayama, & Creswell, 2019). Our reliance on the family history method to assess psychopathology in non-participating parents may have reduced the sensitivity of our analyses.

Although our definition of CID is broadly consistent with the DSM and with Pettit et al. (2009), it is somewhat arbitrary. Unfortunately, existing criteria (e.g., DSM PDD and recurrent MDD) are also arbitrary and may not be optimal (de Zwwart et al., 2019; Klein, 2008; Mondimore et al., 2007; Monroe & Harkness, 2022). Research designed to inform decisions about criteria for course-based classification (e.g., Pettit et al., 2009) is sorely needed. In addition, there is likely considerable heterogeneity remaining in the CID and TLD groups. In future studies, it will be important to parse this heterogeneity by examining severity and course patterning in a finer-grained fashion, as well as considering potential etiological and mechanistic variables. However, we believe distinguishing between broad CID and TLD groups is an important initial step in advancing research and treatment on depression.

This study had notable strengths, including a prospective longitudinal design with a wide range of clinical and psychosocial risk factors assessed before onset of DD. In addition, the cohort design allowed inclusion of a never-depressed group, affording insight into whether differences between the CID and TLD groups were specific to CID, common to depression but stronger in CID, or common to depression in general. Finally, we assessed depression at 5 points during the mean 6-year follow-up, which is important as single assessments covering long periods miss many depressive episodes Moffitt, Caspi, Taylor, Kokaua, Milne, Polanczyk, & Poulton, 2010).

However, the study also had limitations. First, we did not examine genetic and biological risk factors and life stress, all of which may further distinguish CID and TLD (Monroe & Harkness, 2022; Schramm et al., 2020). Second, we did not assess the severity of symptoms in each DD episode during follow-up. Hence, we cannot compare the CID and TLD groups on symptom severity. Third, we did not consider the course of other disorders that commonly co-occur with depression.

Fourth, we conducted direct interviews with only one parent, obtaining information about the co-parent using the family history method. This likely underestimated rates of psychopathology in co-parents (Andreasen, Rice, Endicott, Reich, & Coryell, 1986). Fifth, our sample was limited to females and was predominantly Caucasian and middle class, hence the findings may not generalize to other populations.

Sixth, the last follow-up was at age 20. With continued follow-up, a number of never-depressed participants will develop DD, and some of those with TLD will meet criteria for CID. Seventh, as we only followed participants to age 20, our results are limited to cases of DD with a relatively early onset (i.e., the early-onset subtype of PDD in the DSM-5 is defined as < age 21). As there are a number of cross-sectional and retrospective differences between early- and late-onset DDs on clinically- and etiologically-relevant features (Klein, Schatzberg, McCullough, Keller, Dowling, Goodman … Harrison, 1999; Korten, Comijs, Lamers, & Penninx, 2012), risk factors for the development of early- and late-onset CID may differ as well. Hence, both early vs late onset and course pattern should be considered in future studies (Klein, 2008; Pettit et al., 2009; Wilson et al., 2014).

Eighth, we conducted multiple statistical tests without correcting for number of analyses. However, risk factors were selected from reviews and meta-analyses based on evidence of associations with persistence or recurrence. Moreover, we have emphasized the pattern of findings across analyses rather than results of any particular tests. Finally, the CID group was relatively small, limiting the power of our comparisons.

Conclusions

The subgroup of individuals with prolonged or frequent DD episodes accounts for a disproportionate share of the time lived with depression. However, there is a surprising lack of data on pre-onset predictors of who is at risk for developing CID as opposed to briefer, infrequent depressive episodes. We identified a number of pre-onset clinical and psychosocial risk factors that distinguished early adolescent females who subsequently developed a chronic-intermittent course from those who developed a more transient course of DD. For some of these variables, youth who developed TLD also differed from never-depressed youth, but for other risk factors, the TLD and never-depressed groups were indistinguishable. This suggests that in addition to general sources of liability for depression, there are liabilities that are quantitatively stronger in CID than TLD, and liabilities that appear to be specific to CID. These findings highlight the value of a life-course subtyping approach to depression (Klein & Allmann, 2014; Monroe & Harkness, 2022) for advancing the understanding of etiopathogenesis, using prevention and treatment resources more efficiently, and reducing the vast personal and societal burden of DD.

Acknowledgments

This study was supported by National Institute of Mental Health Grants R01 MH093479 (Kotov) and R56 MH117116 (Kotov and Klein).

Footnotes

The authors have no disclosures to report.

This study was not preregistered. However, the data and study materials are available upon request from the corresponding author.

We are grateful to the anonymous reviewers of the initial version of this paper for their very thoughtful and helpful critiques.

1

Equal variances not assumed given a significant Levene’s test.

References

  1. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV). Washington, DC: American Psychiatric Association. [Google Scholar]
  2. Andreasen NC, Rice J, Endicott J, Reich T, & Coryell W (1986). The family history approach to diagnosis: how useful is it?. Archives of General Psychiatry, 43(5), 421–429. [DOI] [PubMed] [Google Scholar]
  3. Bagby RM, Parker JD, Joffe RT, & Buis T (1994). Reconstruction and validation of the Depressive Experiences Questionnaire. Assessment, 1(1), 59–68. [DOI] [PubMed] [Google Scholar]
  4. Baumeister H, & Parker G (2012). Meta-review of depressive subtyping models. Journal of Affective Disorders, 139, 126–140. [DOI] [PubMed] [Google Scholar]
  5. Brouwer ME, Williams AD, Kennis M, Fu Z, Klein NS, Cuijpers P, & Bockting CL (2019). Psychological theories of depressive relapse and recurrence: A systematic review and meta-analysis of prospective studies. Clinical Psychology Review, 74, 101773. [DOI] [PubMed] [Google Scholar]
  6. Brown GW, Sklair F, Harris TO, & Birley JLT (1973). Life-events and psychiatric disorders. Part 1: Some methodological issues. Psychological Medicine, 3(1), 74–87. [DOI] [PubMed] [Google Scholar]
  7. Buckman JE, Underwood A, Clarke K, Saunders R, Hollon SD, Fearon P, & Pilling S (2018). Risk factors for relapse and recurrence of depression in adults and how they operate: A four-phase systematic review and meta-synthesis. Clinical Psychology Review, 64, 13–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Burcusa SL, & Iacono WG (2007). Risk for recurrence in depression. Clinical Psychology Review, 27, 959–985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cohen JR, Thakur H, Young JF, & Hankin BL (2020). The development and validation of an algorithm to predict future depression onset in unselected youth. Psychological Medicine, 50(15), 2548–2556. [DOI] [PubMed] [Google Scholar]
  10. Conway CC, Krueger RF, & HiTOP Consortium Executive Board. (2021). Rethinking the diagnosis of mental disorders: data-driven psychological dimensions, not categories, as a framework for mental-health research, treatment, and training. Current Directions in Psychological Science, 30(2), 151–158. [Google Scholar]
  11. De Los Reyes A & Prinstein MJ (2004). Applying depression-distortion hypotheses to the assessment of peer victimization in adolescents. Journal of Clinical Child and Adolescent Psychology, 33, 325–35. 10.1207/s15374424jccp3302_14 [DOI] [PubMed] [Google Scholar]
  12. de Zwart PL, Jeronimus BF, & de Jonge P (2019). Empirical evidence for definitions of episode, remission, recovery, relapse and recurrence in depression: a systematic review. Epidemiology and Psychiatric Sciences, 28(5), 544–562. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Eaton WW, Shao H, Nestadt G, Lee BH, Bienvenu OJ, & Zandi P (2008). Population-based study of first onset and chronicity in major depressive disorder. Archives of General Psychiatry, 65(5), 513–520. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. First MB, Spitzer RL, Gibbon M, Williams J (2002). Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Non-Patient Edition (SCID-I/NP). New York State Psychiatric Institute, New York. [Google Scholar]
  15. Furman W, & Buhrmester D (2010). Network of Relationships Inventory Questionnaire Manual. Unpublished manual. Department of Psychology, University of Denver. [Google Scholar]
  16. Hames JL, Hagan CR, & Joiner TE (2013). Interpersonal processes in depression. Annual Review of Clinical Psychology, 9, 355–377. [DOI] [PubMed] [Google Scholar]
  17. Hammen C, Brennan PA, Keenan-Miller D, & Herr NR (2008). Early onset recurrent subtype of adolescent depression: Clinical and psychosocial correlates. Journal of Child Psychology and Psychiatry, 49(4), 433–440. [DOI] [PubMed] [Google Scholar]
  18. Hardeveld F, Spijker J, De Graaf R, Nolen WA, & Beekman ATF (2010). Prevalence and predictors of recurrence of major depressive disorder in the adult population. Acta Psychiatrica Scandinavica, 122(3), 184–191. [DOI] [PubMed] [Google Scholar]
  19. Hettema JM (2008). What is the genetic relationship between anxiety and depression? American Journal of Medical Genetics Part C (Seminars in Medical Genetics), 148C, 140–146. [DOI] [PubMed] [Google Scholar]
  20. Hill RM, Mellick W, Temple JR, & Sharp C (2017). The role of bullying in depressive symptoms from adolescence to emerging adulthood: A growth mixture model. Journal of Affective Disorders, 207, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hölzel L, Härter M, Reese C, & Kriston L (2011). Risk factors for chronic depression: A systematic review. Journal of Affective Disorders, 129, 1–13. [DOI] [PubMed] [Google Scholar]
  22. James SL, Abate D, Abate KH, Abay SM, Abbafati C, Abbasi N, … & Briggs AM (2018). Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. The Lancet, 392(10159), 1789–1858. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. John OP, & Srivastava S (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. In Pervin LA & John OP (Eds.), Handbook of personality (2nd. ed.) (pp. 102–138). New York: Guilford. [Google Scholar]
  24. Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, Williamson D, & Ryan N (1997). Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): Initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry 36, 980–988. 10.1097/00004583-199707000-00021 [DOI] [PubMed] [Google Scholar]
  25. Kapur S, Phillips AG, & Insel TR (2012). Why has it taken so long for biological psychiatry to develop clinical tests and what to do about it? Molecular Psychiatry, 17, 1174–1179. [DOI] [PubMed] [Google Scholar]
  26. Keller MB, Lavori PW, Friedman B, Nielsen E, Endicott J, McDonald-Scott P, & Andreasen NC (1987). The Longitudinal Interval Follow-up Evaluation: A comprehensive method for assessing outcome in prospective longitudinal studies. Archives of General Psychiatry, 44(6), 540–548. [DOI] [PubMed] [Google Scholar]
  27. Kessler RC, Nierenberg AA, Pennix BWJH, Wang PS, Wittchn H-U, & Ziobrowski HN (2022). Epidemiology and burden of mood disorders. In Nemeroff CB, Rasgon N, Schatzberg AF, & Strakowski SM (Eds.), Textbook of Mood Disorders (2nd ed). Washington, DC: American Psychiatric Association Publishing. [Google Scholar]
  28. Klein DN (2008). Classification of depressive disorders in DSM-V: Proposal for a two-dimension system. Journal of Abnormal Psychology, 117, 552–560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Klein DN (2022). Diagnosis and classification of depressive disorders. Invited chapter under review for Pettit JW, and Olino TM (Eds.), APA Handbook of Depression, Washington, DC: American Psychological Association Press. [Google Scholar]
  30. Klein DN, & Allmann AES (2014). Course of depression: Persistence and recurrence. In Gotlib IH & Hammen CL (Eds.), Handbook of depression (pp. 64–83). New York: Guilford Press. [Google Scholar]
  31. Klein DK, Glenn CR, Kosty DB, Seeley JR, Rohde P, Lewinsohn PM (2013). Predictors of first lifetime onset of major depressive disorder in young adulthood. Journal of Abnormal Psychology, 122, 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Klein DN, & Kotov R (2016). Course of depression in a 10-year prospective study: Evidence for qualitatively distinct subgroups. Journal of Abnormal Psychology, 125, 337–348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Klein DN, Kotov R, & Bufferd SJ (2011). Personality and depression: Explanatory models and review of the evidence. Annual Review of Clinical Psychology, 7, 269–295. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Klein DN, Shankman SA, & Rose S (2006). Ten-year prospective follow-up study of the naturalistic course of dysthymic disorder and double depression. American Journal of Psychiatry, 163, 872–880. [DOI] [PubMed] [Google Scholar]
  35. Klein DN, Schatzberg AF, McCullough JP, Keller MB, Dowling F, Goodman D …. Harrison WM (1999). Early- versus late-onset dysthymic disorder: Comparison in outpatients with superimposed major depressive episodes. Journal of Affective Disorders, 52, 187–196. [DOI] [PubMed] [Google Scholar]
  36. Korten NC, Comijs HC, Lamers F, & Penninx BW (2012). Early and late onset depression in young and middle aged adults: differential symptomatology, characteristics and risk factors?. Journal of Affective Disorders, 138(3), 259–267. [DOI] [PubMed] [Google Scholar]
  37. Lawrence PJ, Murayama K, & Creswell C (2019). Systematic review and meta-analysis: anxiety and disorders in offspring of parents with anxiety disorders. Journal of the American Academy of Child & Adolescent Psychiatry, 58(1), 46–60. [DOI] [PubMed] [Google Scholar]
  38. Lee YY, Stockings EA, Harris MG, Doi SAR, Page IS, Davidson SK, & Barendregt JJ (2019). The risk of developing major depression among individuals with subthreshold depression: a systematic review and meta-analysis of longitudinal cohort studies. Psychological Medicine, 49(1), 92–102. [DOI] [PubMed] [Google Scholar]
  39. Lyketsos CG, Nestadt G, Cwi J, Heithoff K, & Eaton WW (1994). The Life Chart Interview: A standardized method to describe the course of psychopathology. International Journal of Methods in Psychiatric Research, 4(3), 143–155. [Google Scholar]
  40. Ma CF, Chan SKW, Chung YL, Ng SM, Hui CLM, Suen YN, & Chen EYH (2021). The predictive power of expressed emotion and its components in relapse of schizophrenia: a meta-analysis and meta-regression. Psychological Medicine, 51(3), 365–375. [DOI] [PubMed] [Google Scholar]
  41. Magaña AB, Goldstein MJ, Karno M, Miklowitz DJ, Jenkins J, & Falloon IR (1986). A brief method for assessing expressed emotion in relatives of psychiatric patients. Psychiatry Research, 17(3), 203–212. [DOI] [PubMed] [Google Scholar]
  42. Michelini G, Perlman G, Tian Y, Mackin DM, Nelson BD, Klein DN, & Kotov R (2021). Multiple domains of risk factors for first onset of depression in adolescent girls. Journal of Affective Disorders, 283, 20–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Moffitt T, Caspi A, Taylor A, Kokaua J, Milne B, Polanczyk G, & Poulton R (2010). How common are common mental disorders? Evidence that lifetime prevalence rates are doubled by prospective versus retrospective ascertainment. Psychological Medicine, 40(6), 899–909. doi: 10.1017/S0033291709991036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Mondimore FM, Zandi PP, MacKinnon DF, McInnis MG, Miller EB, Schweizer B, … & Potash JB (2007). A comparison of the familiality of chronic depression in recurrent early-onset depression pedigrees using different definitions of chronicity. Journal of Affective Disorders, 100(1–3), 171–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Monroe SM, & Harkness KL (2011). Recurrence in major depression: a conceptual analysis. Psychological Review, 118(4), 655. [DOI] [PubMed] [Google Scholar]
  46. Monroe SM, & Harkness KL (2022). Major Depression and its recurrences: Life course matters. Annual Review of Clinical Psychology, 18, 329–357. [DOI] [PubMed] [Google Scholar]
  47. Nanni V, Uher R, & Danese A (2012). Childhood maltreatment predicts unfavorable course of illness and treatment outcome in depression: a meta-analysis. American Journal of Psychiatry, 169(2), 141–151. [DOI] [PubMed] [Google Scholar]
  48. Nelson J, Klumparendt A, Doebler P, & Ehring T (2017). Childhood maltreatment and characteristics of adult depression: meta-analysis. The British Journal of Psychiatry, 210(2), 96–104. [DOI] [PubMed] [Google Scholar]
  49. Nolen-Hoeksema S, & Morrow J (1991). Responses to Depression questionnaire. Unpublished manuscript, Department of Psychology, Stanford University. [Google Scholar]
  50. Ormel J, Hollon SD, Kessler RC, Cuijpers P, & Monroe SM (2022). More treatment but no less depression: The treatment-prevalence paradox. Clinical Psychology Review, 91, 102111. [DOI] [PubMed] [Google Scholar]
  51. Parker G, Tupling H, and Brown LB (1979) A Parental Bonding Instrument. British Journal of Medical Psychology, 52, 1–10. [Google Scholar]
  52. Pettit JW, Hartley C, Lewinsohn PM, Seeley JR, & Klein DN (2013). Is liability to recurrent major depressive disorder present before first episode onset in adolescence or acquired after the initial episode? Journal of Abnormal Psychology 122, 353–358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Pettit JW, Lewinsohn PM, Roberts RE, Seeley JR, & Monteith LL (2009). The long-term course of depression: Development of an empirical index and identification of early adult outcomes. Psychological Medicine, 39(3), 403–412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Plantes MM, Prusoff BA, Brennan J, & Parker G (1988). Parental representations of depressed outpatients from a USA sample. Journal of Affective Disorders, 15(2), 149–155. [DOI] [PubMed] [Google Scholar]
  55. Ruscio AM (2019). Normal versus pathological mood: Implications for diagnosis. Annual Review of Clinical Psychology, 15, 179–205. [DOI] [PubMed] [Google Scholar]
  56. Schramm E, Klein DN, Elsaesser M, Furukawa T, & Domschke K (2020). Review of Persistent Depressive Disorder: History, correlates, and clinical implications. Lancet Psychiatry, 7, 801–812. [DOI] [PubMed] [Google Scholar]
  57. Silver J, Olino TM, Carlson GA, & Klein DN (2020). Offspring of mothers with histories of chronic and non-chronic depression: Symptom trajectories from ages 6 to 15. Frontiers Psychiatry, 11, 601779. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Simms LJ, & Clark LA (2006). The Schedule for Nonadaptive and Adaptive Personality (SNAP): A Dimensional Measure of Traits Relevant to Personality and Personality Pathology. In Strack S (Ed.), Differentiating Normal and Abnormal Personality (pp. 431–450). Springer. [Google Scholar]
  59. Sullivan PF, Neale MC, & Kendler KS (2000). Genetic epidemiology of major depression: review and meta-analysis. American Journal of Psychiatry, 157(10), 1552–1562. [DOI] [PubMed] [Google Scholar]
  60. Watson D, & Clark LA (1994). The PANAS-X: Manual for the positive and negative affect schedule-expanded form. Department of Psychology, University of Iowa. [Google Scholar]
  61. Watson D, Stasik SM, Ellickson-Larew S, & Stanton K (2015). Extraversion and psychopathology: A facet-level analysis. Journal of Abnormal Psychology, 124, 432–446. [DOI] [PubMed] [Google Scholar]
  62. Weissman MM, Wickramaratne P, Adams P, Wolk S, Verdeli H, & Olfson M (2000). Brief screening for family psychiatric history: the family history screen. Archives of General Psychiatry, 57(7), 675–682. [DOI] [PubMed] [Google Scholar]
  63. Widom CS, DuMont K, & Czaja SJ (2007). A prospective investigation of major depressive disorder and comorbidity in abused and neglected children grown up. Archives of General Psychiatry, 64(1), 49–56. [DOI] [PubMed] [Google Scholar]
  64. Wilson S, Vaidyanathan U, Miller MB, McGue M, & Iacono WG (2014). Premorbid risk factors for major depressive disorder: are they associated with early onset and recurrent course? Development and Psychopathology, 26(4pt2), 1477–1493. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES