Stage Models for Major Depression: Cognitive Behavior Therapy, Mechanistic Treatment Targets, and the Prevention of Stage Transition

Abstract

Stage models encourage a longitudinal perspective on the care of those with major depression: supporting vigilance to the risk for stage progression and the selection of interventions to address that risk. A central goal for this article is to evaluate the role of cognitive-behavior therapy (CBT) in addressing stage progression in the treatment of major depression. We summarize the evidence supporting depression-focused CBT for: (1) preventing depression onset, (2) treating syndromal depression, (3) treating residual symptoms, (4) preventing relapse, and (5) addressing pharmacologic treatment resistance. In addition, consistent with the goal of aiding prevention and intervention development by refining mechanistic treatment targets, we evaluate the role of two specific risk-factors for stage progression: insomnia and rumination. These risk factors have a feed-forward relationship with stress, both being amplified by stress and amplifying the negative consequences of stress. Moreover, each of these risk factors predict depression stage transmissions across multiple stages, and both are modifiable with treatment. Accordingly, insomnia and rumination appear to serve as excellent mechanistic targets for the prevention of depression stage progression. These findings are discussed in relation to current limitations and future research directions for targeting these risk factors and furthering the effective treatment of depression.

Stage models in medicine typically denote a temporal progression of severity for a disease (e.g., cancer) as indicated by the level of treatment resistance or degree of morbidity/mortality. When applied to psychiatric disorders, stage models reflect the progression from prodromal symptoms to full episodes, to recurrent episodes of greater severity, and finally to treatment resistance (e.g., Cosci & Fava, 2013; McGorry et al., 2014; Muneer, 2016; Scott et al., 2013). One clear benefit of stage models is that they promote a broader consideration of treatment targets and strategies than traditional diagnostic models of psychiatric disorders. Rather than focusing on symptom reduction or episode resolution alone, these models promote a proactive focus on preventing stage progression over time.

This theoretical framework is valuable for major depression in part because each transition from a euthymic state to relapse is not necessarily uniform as the disorder progresses. Rather, the transitions differ for earlier versus later stages in the progression of depression. Indeed, the risk of relapse may increase for each episode of depression and is also higher for those with inadequately resolved episodes of depression (i.e., residual symptoms after meeting remission criteria), a history of recurrence, and greater episode severity (Buckman et al., 2018; Hardeveld, Spijker, De Graaf, Nolen, & Beekman, 2010; Warden, Rush, Trivedi, Fava, & Wisniewski, 2007). Unfortunately, as noted by McGorry et al. (2014), prior intervention work has not tended to focus on preventing depression because the disorder is not diagnosed with respect to stage progression: “[O]ur current diagnostic systems, with their focus on well-established, largely chronic illness, do not support a pre-emptive, let alone a preventive, approach, since it is during the early stages of a disorder that interventions have the potential to offer the greatest benefit” (p. 211).

Consistent with the goal of aiding prevention/intervention development by refining mechanistic treatment targets (Nielsen et al., 2018), the purpose of the current article is to: (1) review and evaluate current stage models for unipolar depression, (2) consider the role of cognitive-behavioral therapy (CBT) for depression in addressing stage progression, (3) consider two specific risk-factors for stage progression—insomnia and rumination—that may serve as valuable treatment targets.

Stage Models for Unipolar Depression

Table 1 presents a stage model for unipolar depression that summarizes current perspectives (Cosci & Fava, 2013; de la Fuente-Tomas et al., 2019; Fava & Kellner, 1993; Fava & Tossani, 2007; Hetrick et al., 2008; McGorry et al., 2014; Scott et al., 2013; Verduijn et al., 2015), and lists the most common stage characteristics across these models. Staging models differ somewhat in the specificity of symptoms noted during the prodromal phase (Stage 1) and the ordering of the residual-symptom phase (Stage 3), but they share a common focus on syndromic status emerging at the phase of the first major depressive episode (Stage 2), and chronic, treatment-refractory depression emerging at the final phase (Stage 4).

Table 1.

A stage model for unipolar depression.

Stage Number	Phase Description	Disease Progression
Stage 0	Presymptomatic risk phase	Genetic vulnerability or risk for developing a mood disorder prior to the onset of symptoms
Stage 1a	Onset of prodromal phase	Some subtle changes in overall functioning with mild, non-specific symptoms associated with anxiety and depression (e.g., worry, irritability, pain, impaired concentration, sadness, anhedonia, fatigue, sleep disturbance, altered appetite, psychomotor symptoms)
Stage 1b	Progression of prodromal phase	Increased but sub-threshold depressive symptoms (e.g., depressed mood) and increased functional decline
Stage 2	Major depressive episode	First full episode of depression with clinically significant symptoms lasting weeks or longer
Stage 3a	Progression of residual/remitted phase: Relapse or recurrence	Return of depressed mood or a full major depressive episode (either relapse after period of remission or recurrence after recovery)
Stage 3b	End of residual/remitted phase	State of dysthymia or several recurring depressive episodes
Stage 4	Chronic and severe major depressive illness	Recurrent, severe depressive episodes and repeated failure to respond to treatment

Note. Residual symptoms refer to subsyndromal symptoms of depression that persist despite achievement of remission status. Depression remission is defined by a relatively brief asymptomatic period (typically less than eight weeks). Recovery is typically defined by the absence of depression beyond eight weeks. Relapse represents a return to syndromal depression occurring during remission; recurrence refers to the return of depression following recovery (for empirical commentary on these distinctions see de Zwart, Jeronimus, & De Jonge, 2019).

Attention to disorder stage transitions deviates from the DSM-driven focus on threshold-based disorder status and instead provides a longitudinal model with dimensional characteristics (Frank et al., 2015). As such, stage models are a natural fit for the longitudinal responsibilities of clinical practice and the task of caring for patients over time (Cosci & Fava, 2013).

These considerations are especially apt for the treatment of major depression. Research on the course of major depression reveals the short-lived nature of many episodes relative to the risk of ongoing subsyndromal symptoms and recurrence. For example, large-scale monitoring studies indicate that the median duration of major depressive episodes is three months. Indeed, there are striking similarities across the Ecological Catchment Area Survey, the National Comorbidity Survey, and the Netherlands Mental Health Survey (NMHS) in observing a median episode length at three months (mean duration range: 22–30 weeks), i.e., 50% of these samples recovered by month 3 (Eaton et al., 1997; Spijker et al., 2002; Ustün & Kessler, 2002). Furthermore, this 3-month duration is observed regardless of where and whether patients seek treatment (e.g., primary care, specialty mental health clinics, no treatment; Spijker et al., 2002).

Powerful cohort effects are presumably at work here, where patients with more concerning depression (e.g., greater comorbidity, chronicity, severity, history of recurrence) may be more likely to seek referrals for specialty treatment (Hardeveld et al., 2010). Indeed, greater episode length and history of previous treatment distinguishes treatment-seeking depressed individuals from non-seeking depressed individuals (Blumenthal & Endicott, 1997). Episode length alone appears to offer a robust prediction of recovery. The rate of recovery declines sharply after the initial three months of depression. For example, data from the NMHS indicate that just 13% of the full sample reached recovery between months 3 and 6 of observation, followed by 13% recovery in the subsequent 6 months (months 6–12), and followed by a dramatic drop to 4% recovery in the next 12 months (months 12–24). This declining recovery rate leaves 20% of the original sample demonstrating chronic depression over 2 years (Spijker et al., 2002). Appropriate to these considerations, meta-analysis of 13 randomized-controlled trials and 6 primary-care studies indicate a median episode length closer to a year (i.e., 23% remitted at 3 months, with an additional 9% remitted in the next 3 months, and an additional 21% remitting over the subsequent 6 months; i.e., 53% remitted by month 12) for treatment-seeking individuals in waitlist conditions (Whiteford et al., 2013). As such, the benchmark for “spontaneous” recovery for certain treatment-seeking adults appears to be half that for broader community samples. Knowledge of this time frame provides clinicians with an important benchmark to consider in relation to evaluating treatment success: how is a particular treatment doing relative to the expectation derived from treatment-seeking samples that 23% of adults with major depression will enter a phase of remission within three months?

It is also important to consider that some of the early depression remitters do not stay well. Among those who have achieved at least six months of remission, rates of relapse, again according to NMHS data, are on average fairly linear over the next 10 years at a rate of approximately 2.5% per year (Hardeveld et al., 2013; for similar rates, see Eaton et al., 2008). Of course, in contrast to this average rate, relapse rates are higher for a substantial subset of individuals with more previous episodes, greater severity and chronicity of an index episode, residual symptoms, lower social support, and comorbid illness (Hardeveld et al., 2013; Spijker et al., 2004; Warden et al., 2007). For example, relapse rates for those in control groups in clinical trials are approximately twice as high as those from community surveys: 4% for non-active control groups across 4 clinical trials (Vittengl et al., 2007) and 5.9% derived from a meta-analysis of 10 clinical trials that included a range of control conditions, including continuation therapy (Clarke et al., 2015). Accordingly, a substantial history of previous depressive episodes and these additional markers for relapse serve collectively as a robust signal to clinicians that relapse prevention efforts will likely be required to manage stage transitions.

Given these considerations, the central tasks of depression management can be broadly characterized as (1) preventing initial depression in those at risk, (2) fully and rapidly resolving major depression in those with the disorder, (3) maintaining treatment success in those at greatest risk of recurrence, and (4) preventing recurrent, severe depression. These values are fully consistent with a stage model approach (Scott et al., 2013), and can be translated to stage-specific questions for treatment selection:

• Stage 0 to 1 – Do Efficacious Primary (Universal) and Secondary (Targeted) Prevention Programs Exist?

• Stage 2 & 3a – How well does CBT provide a full resolution of depression?

• Stage 3b – How do treatments differ with respect to the maintenance of prior treatment gains?

• Stage 4 – Which treatments prevent the development of chronic depression after initial treatment nonresponse?

In the sections below, we center our review on the role and efficacy of CBT across stages as well as risk factors for stage progression. Following this general review, we then emphasize those risk factors that are relevant to multiple stage transitions and thus should be considered potential trans-stage treatment targets.

Stage 0 to 1 – Do Efficacious Primary (Universal) and Secondary (Targeted) Prevention Programs Exist?

A core assumption of primary prevention efforts for depression is that general resilience skills can be taught to children and adolescents—before the age of heightened depression onset risk—that will reduce the likelihood of developing symptoms and depressive disorders. Many of these skills center on enhancing resilience: the ability to maintain or recover positive mental health after experiencing adversity (Dray et al., 2017). Because these skills are directed not just to samples at particular risk but to a general age-selected population, interventions are typically delivered at school, by school personnel.

A systematic review identified 57 randomized-controlled trials that investigated the implementation of universal, school-based, resilience-focused interventions (Dray et al., 2017). These studies represented 41,521 children and adolescents across 16 countries. The studied interventions included cognitive-behavioral therapy (CBT), coping skills, life skills, social skills, psychological well-being therapy, mindfulness, positive psychology, social and emotional learning, interpersonal and self-management skills, the affective-behavioral-cognitive-dynamic (ABCD) model, and mental health promotion. Most of the trials utilized a CBT-based intervention. Overall, the active interventions showed significant benefit over the control conditions for four of the seven examined outcomes—depressive symptoms, internalizing problems, externalizing problems, and general psychological distress—though effect sizes were small. Additionally, the interventions were evidently not efficacious over long-term follow-up for assessments beyond 12 months post-treatment. Two mitigating factors may partially account for the failure of prevention interventions over the longer term. First, at the time of learning, children’s low symptom severity may make skill application less immediately relevant, leading to low skill application and low retention. Indeed, post-treatment effect sizes were larger for secondary relative to primary prevention interventions for children and adolescents (Hetrich et al., 2015), suggesting that intervention engagement or utilization may be improved by the presence of at least some symptoms of depression.

Although common prodromal symptoms of depression can include symptoms distinct from those that characterize major depression (e.g., worry, irritability, pain), many symptoms do include DSM-5 major depressive episode criteria (e.g., impaired concentration, sadness, anhedonia, fatigue, sleep disturbance, altered appetite, psychomotor symptoms; Benasi, Fava, & Guidi, 2021), underscoring the relevance of studies of subsyndromal symptoms for intervening for prodromal symptoms. An evaluation of 18 trials compared a psychological treatment, most often CBT, with a control condition for the treatment of subsyndromal depression (Cuijpers et al., 2014). Treatments were delivered in individual, group, telephone-based, and guided self-help formats (number of sessions: 6 to 16). These treatments showed an overall small-to-moderate post-treatment advantage over the control condition (g = 0.35, 95% CI 0.23–0.47). In addition to effects on subsyndromal symptoms, treatment significantly reduced the incidence of major depressive episodes at 6-month follow-up and at a trend level at 12-month follow-up. In sum, selecting participants based on presenting symptoms (secondary prevention) may provide longer-term benefits, perhaps due to improved participant engagement and matching interventions to symptoms (Young et al., 2021).

Stage 2 & 3a – How well does CBT provide a full resolution of depression?

As informed by a stage model approach, the goal for initial treatment of a major depressive episode is more than to reduce acute suffering; it is to treat the depressive episode to remission and to attend to the resolution of residual symptoms (subsyndromal symptoms of depression that persist despite achievement of remission status) that figure so prominently in predisposing individuals to relapse (Buckman et al., 2018; Hardeveld et al., 2010, 2013; Warden et al., 2007). How well does CBT do for the first part of this goal, treating to full remission? Meta-analyses indicate that half of the patients achieve a response to CBT (51% combined rate of response or remission). Further, most of those who respond do indeed achieve full remission status. Specifically, a meta-analysis of 34 CBT-controlled trials places the remission rate at 45% (Santoft et al., 2019; see also Gartlehner et al., 2016), with just 4% achieving a response without remission. These remission rates for CBT compare well to those of pharmacologic alternatives, with evidence for CBT vs. pharmacotherapy having approximately equal depression severity scores (Gartlehner et al., 2016).

And how well does CBT treat residual symptoms? Meta-analysis also provides strong support that CBT can be used to treat residual symptoms after pharmacologic treatment of major depression (Guidi & Fava, 2021). Also, well-being-focused CBT has received validation across several trials to extend gains offered by traditional CBT (see Fava, 2016; Guidi, Tomba, Cosci, Park, & Fava, 2017). Well-being-focused treatment represents the metaphorical “other side of the coin” from standard CBT: instead of targeting symptom reduction, these interventions focus on tracking and enhancing well-being. Novel strategies include asking patients to keep a “hedonic diary” to track and review daily episodes of well-being, with therapist efforts placed on helping patients further evoke and extend these periods of well-being. Evaluation of well-being therapy for residual symptoms of depression in a controlled clinical trial indicates that it enhances the level of recovery and reduces the likelihood of relapse, making it a fitting strategy for addressing the risk of stage transition for those achieving only partial remission from depression (Fava, 2016). Efficacy for well-being therapy has also been provided for patients with residual symptoms of generalized anxiety disorder, supporting the transdiagnostic relevance of this approach (Fava, 2016).

Switching treatment modalities and combining treatment strategies can also be applied when partial or nonresponse is encountered. These strategies receive more attention in the section on addressing treatment nonresponse.

Stage 3b – How do treatments differ with respect to the maintenance of prior treatment gains?

Whereas rates of acute response may not differentiate CBT and pharmacologic treatment, a substantial body of evidence supports a strong role of brief CBT in halting stage progression over a longer term by preventing relapse (Hollon, Stewart, & Strunk, 2006). Although sample sizes are limited for longer-term follow-up assessments, the protective effects of CBT appear to last at least two years following treatment initiation (Furukawa et al., 2021; Hollon et al., 2005). In addition to the staying power of initial treatment with CBT, there is evidence that continuation interventions with CBT further reduce relapse rates (Vittengl et al., 2007), and have been recommended for those with specific risk for recurrence (e.g., those with greater number of episodes; Bockting et al., 2015). Further, for those who initiated treatment with pharmacotherapy, there is meta-analytic evidence that the addition of CBT can reduce annual relapse rates by 50% relative to treatment-as-usual (Zhang et al., 2018). Indeed, there is support for CBT used specifically as a strategy for those who wish to discontinue their pharmacotherapy for depression, providing equal protection against relapse as continuing medication treatment (Bockting et al., 2018). In summary, CBT is a well-supported strategy for reducing the transition from Stage 3 to Stage 4.

Stage 4 - Which treatments prevent the development of chronic depression after initial treatment nonresponse?

Research makes clear that CBT for depression can succeed with patients who have previously failed to respond to antidepressant treatment. For example, in the largescale (N = 469) CoBalT trial, depressed patients with at least one antidepressant medication treatment failure showed a clear benefit for brief (12 sessions) treatment with CBT, with a doubling in treatment response (22% for usual pharmacotherapy care vs. 46% for usual care plus CBT (Wiles et al., 2013). This responsivity is noteworthy given that the duration of the present episode of depression was 2 years or longer for most participants (59%), and 70% had been prescribed their current antidepressant medication for more than 12 months. The success of CBT for treatment-resistant patients in the CoBalT trial is consistent with a meta-analysis of six randomized trials showing that CBT offers efficacy to depressed patients who have failed to respond to antidepressants as assessed by reduced depression symptoms and response or remission rate (Li et al., 2018).

The success of CBT after psychopharmacologic treatment failures (i.e., after Stage 3–4 for pharmacotherapy) also encourages an interesting question on the value of combined CBT and psychopharmacologic treatment. Does combined treatment offer greater efficacy because: (1) some patients require both modalities of treatment or (2) because combination treatment provides a rescue treatment for nonresponders to the other treatment? More specifically, Otto and Hearon (2015) have hypothesized that combination treatment response should be evaluated in relation to 5 potential subtypes of depressed patients: Type 1 gets better with either modality of treatment; Type 2 gets better with antidepressant medication but not CBT; Type 3 gets better with CBT but not antidepressant medication; Type 4 gets better only with medication plus CBT; and Type 5 does not get better with medication, CBT, or their combination. Accordingly, in a traditional combination treatment trial offering antidepressant medication, CBT, or the combination of these treatments, Types 1 and 2 would be hypothesized to respond to antidepressant medication alone, Types 1 and 3 would be hypothesized to respond to CBT alone, and Types 1, 2, 3 and 4 would be hypothesized to respond to combination treatment. In this scenario, combination treatment can outperform either modality alone even without the existence of Type 4.

What is the value of attending to the separate (rather than truly combined) efficacy of CBT and antidepressant medications? If the proportion of Type 4 patients is low or non-existent, the sequential application of treatment would be more cost-effective than any routine application of combination treatment. Indeed, the success of crossover treatment trials (antidepressant nonresponders referred for adjunctive CBT; Li et al., 2018) suggests a robust presence for Type 3. Perhaps more importantly, this analysis underscores the importance of considering a shift in treatment modality if nonresponse is encountered with either monotherapy. Further, if repeated trials of antidepressant treatment run the risk of increasing treatment resistance as hypothesized by some (e.g., Fava, 2020; Sharma, 2001), then shifting treatment modalities to CBT should occur earlier in treatment algorithms rather than following a progression of pharmacologic interventions (cf. Fekadu et al., 2009; Petersen et al., 2005). The sequential application of pharmacotherapy and CBT are further discussed in relation to a stage model by Guidi and associates (Guidi et al., 2017; Guidi & Fava, 2020).

Considering Treatment Targets for Preventing Stage Transition

As reviewed above, CBT is supported as an efficacious strategy for addressing each stage of depression by focusing attention directly on prodromal symptoms or depression symptom resolution. In the next sections, we consider whether other specific treatment targets are important for preventing stage transitions. We review two risk factors—insomnia and rumination—that have shown promise for predicting stage transmissions across multiple stages: risk for initial episodes (progression to Stage 2), risk for chronicity/severity (progression to Stage 3c), and risk for recurrence (progression to Stage 4). Moreover, each risk factor shares a core relationship with psychosocial stress—a known risk factor for depression. As discussed below, these risk factors—insomnia and rumination—have a feed-forward relationship with stress, both being amplified by stress and amplifying the negative consequences of stress. Critically, these risk factors are modifiable with treatment. Hence, they serve as excellent mechanistic targets for the prevention of depression stage progression.

Identification of mechanistic targets for stage transition fits well with developments in behavior-change research that emphasize the importance of the experimental medicine approach. This approach is designed to enhance both prevention and treatment efforts by ensuring that causal mechanistic factors underlying interventions are elucidated. Identification and testing of the putative mechanisms of interventions form the basis of the National Institute of Health Science Of Behavior Change program (SOBC; Nielsen et al., 2018). According to the SOBC approach, a construct should be considered a potential mechanism of change if a valid, reliable measure of the construct covaries with the outcome of interest and if changes in this measure are associated with changes in the outcome. Furthermore, in line with SOBC, many behavioral clinical trials to be funded by NIH are now focused more directly on the engagement of targeted mechanisms of action (e.g., Insel, 2015; https://commonfund.nih.gov/behaviorchange/related). We argue below that insomnia and rumination each meet the requirements to be considered likely mechanisms that may be harnessed to prevent the stage progression of depression.

The Role of Insomnia as a Trans-Stage Risk Factor

As detailed below, sleep disruption and insomnia have powerful effects on mood and are reliably linked to depression stage transitions. Further, the importance of these associations is amplified by the marked prevalence of sleep difficulties. Insomnia itself is a major public health concern, with one-third of adults reporting difficulties with sleep at least three nights a week (Ohayon & Reynolds, 2009). Furthermore, approximately one in four US adult workers meet criteria for insomnia, with associated losses in work performance estimated at 63.2 billion dollars annually (Kessler et al., 2011).

The mechanism by which low sleep confers risk for depressive stage transition is not certain, but sleep is intimately related to mood regulation abilities (Watling et al., 2017). For example, in experimental paradigms, sleep deprivation is associated with greater amygdaloid responsivity to negative emotional stimuli (Walker, 2009; Yoo et al., 2007), as well as greater electrophysiologic reactivity and declarative ratings of unpleasantness (Baglioni et al., 2010). Further, poorer sleep duration and quality predict lower emotion-regulation ability assessed by difficulties disengaging from negative material/cognitive reappraisal ability in the laboratory (Mauss et al., 2013; Nota & Coles, 2018). These experimental study findings are reflected in naturalistic studies. For example, a naturalistic monitoring study found that sleep loss among medical residents was linked to more intense negative emotions to negative events and less positive emotions to positive events (Zohar et al., 2005). Negative emotional effects of impaired sleep also extend to changes in the nature of memory retrieval. Inadequate sleep appears to bias memories toward greater deterioration of positive and neutral stimuli but not negative stimuli (Sterpenich et al., 2007) and greater reactivity to the emotional content of encoded material (Walker, 2009). Together, these findings support a three-pronged influence of insomnia on mood: (1) enhanced emotional responsivity to negative-affective stimuli, (2) lower emotion regulation abilities, and (3) more negative emotional memories. This triple hazard may predispose individuals toward the onset and maintenance of depressive episodes.

Insomnia and Early Stage (0–2) Transitions

In terms of risk for new-onset depression, a meta-analysis of general population studies indicated that non-depressed individuals with insomnia, as compared to those with no sleep difficulties, have a two-fold higher risk of developing depression (Baglioni, Battagliese, et al., 2011). A subsequent meta-analysis confirmed this finding in a broader set of studies that included sleep difficulties and insomnia as the risk factor (Li et al., 2016). Again, insomnia/sleep impairments were associated with a doubling of risk for depression based on 34 cohort studies with a mean follow-up of 60.4 months. Additionally, these findings extend across developmental phases, with insomnia-related risk for depression evident in both young samples (Breslau et al., 1996; Johnson et al., 2006) and aged adults (Perlis et al., 2006; Roberts et al., 2000).

Insomnia and Later-Stage (2–4) Transitions

Once depression develops, comorbid insomnia is common and prospectively predicts a worse course of depression. Data from the largescale Project IMPACT intervention study indicates a linear relationship between insomnia symptom severity and treatment non-response in people with later-life depression: 44% of those with persistent insomnia, 29% of those with intermediate insomnia, and 16% of those with no insomnia continued to be depressed at a 6-month follow-up assessment (Pigeon et al., 2008).

An association between insomnia and treatment nonresponse is also evident in the analysis of claims databases. In a largescale (N = 230,801) study of adult subjects with newly diagnosed and pharmacologically treated depression in the United States, a previous diagnosis of insomnia (occurring in 6% of the sample) predicted pharmacologic treatment resistance over one year follow-up (Cepeda et al., 2018). Furthermore, among residual symptoms, insomnia symptom severity can act as a robust predictor of depression recurrence (Dombrovski et al., 2008; Taylor et al., 2010). Finally, and particularly important for the management of risk in depressed patients, a meta-analysis of 39 studies showing that insomnia more than doubled the risk of suicidal ideation (RR = 2.79), suicide attempt (RR = 3.54), and completed suicide (RR = 2.43) (Pigeon et al., 2012). Importantly, the presence of depression did not moderate the significant association between sleep and suicide-related outcomes, thus supporting the independence of sleep as both a predictor of suicide and a potential treatment target. Finally, there appears to be no difference between CBT and pharmacotherapy concerning rates of residual symptoms of insomnia. High residual insomnia rates after these treatments suggest that adjunctive sleep treatment to address insomnia specifically may be necessary for some patients (Carney, Segal, Edinger, & Krystal, 2007; McClintock et al., 2011).

The Value of Intervening with Insomnia

Several large-scale studies have provided experimental support for the causal role of insomnia in depression across stages by showing that treatment of insomnia helps resolve depression. Concerning the transition between Stages 0 and 1, a large-scale (N=658) randomized trial of university students, selected based on insomnia, found that digital cognitive behavior therapy for insomnia (CBT-I) vs. an online sleep education condition resulted in a significantly lower risk of depression among those who had received CBT-I (Cheng et al., 2019; see also Christensen et al., 2016; Leerssen et al., 2021). The causal role of treating insomnia is further supported by a pooled-study analysis showing that mid-treatment improvements in insomnia associated with digital CBT-I mediated improvements in depression across treatment (Henry et al., 2021). Furthermore, intervening with insomnia in depressed individuals (Stage 2) leads to better treatment response and may halt stage progression; meta-analysis (23 studies) indicates that insomnia treatment improves depression outcomes on the order of moderate-to-large effect size for changes on the Beck Depression Inventory and very large effect size for changes on the Hamilton Depression Rating Scale (Gebara et al., 2018; see also Cunningham & Shapiro, 2018).

Sleep interventions also appear to be effective for residual depression. In a study of adults with residual symptoms of depression and insomnia, following an adequate trial of pharmacotherapy, both sleep and depression improved significantly more among the CBT-I group than among those randomized to continued treatment with pharmacotherapy alone (Watanabe et al., 2011).

In summary, treating insomnia is a valuable prevention strategy for people having minimal or prodromal depressive symptoms who are selected on the basis of sleep disruption (secondary prevention). Treatment of insomnia in people with residual symptoms can reduce stage transition to recurrence. For these interventions, clinicians have a choice between psychosocial or pharmacologic interventions for sleep. Short-term outcomes from CBT-I and pharmacotherapy are equivalent as assessed by both meta-analytic comparisons (Smith et al., 2002) and direct comparison in randomized trials (Jacobs et al., 2004). However, there are specific concerns about the use of sleep agents for secondary management of depression (Kripke, 2007). Moreover, CBT-I has been found to offer strongly enduring effects (Wu et al., 2006), a factor that may be particularly important given the role of insomnia in depression recurrence reviewed above. Moreover, adherence to CBT-I is evidently acceptable across a range of depression severity levels (Manber et al., 2011), appropriate to its application at any stage in the model. Also, as noted, CBT-I can be delivered through a number of modalities (e.g., via the internet or by app; Zachariae et al., 2016; Cheng et al., 2019), thereby reducing issues of therapist availability and burden.

The Role of Rumination as a Trans-Stage Risk Factor

Rumination is another potential mechanistic target to consider with regard to preventing progression along depression stages. Rumination has been defined as a pattern of responding to distress in which an individual perseverates on negative aspects of their life and associated negative causes and consequences while failing to initiate active problem-solving (Nolen-Hoeksema et al., 2008; Watkins, 2015). Rumination is distinct from traditional cognitive biases in that it does not center on the negative appraisal process but rather on passively dwelling on negative content. The Ruminative Response Scales is the most commonly used measure of rumination (Treynor, Gonzalez, & Nolen-Hoeksema, 2003). This scale includes two subtypes: brooding and pondering. Brooding is characterized as more self-critical, whereas pondering is more reflective. Although most research has focused on the full scale, the brooding subtype is generally considered a more potent risk factor for depression and other adverse outcomes (Treynor et al., 2003). Rumination has more recently been conceptualized not as a state or trait construct but rather as a “mental habit” that becomes a reflexive response through positive feedback loops that reinforce it as a learned prepotent behavior (Watkins & Nolen-Hoeksema, 2014).

A large body of literature has now examined the role of rumination in developing and maintaining depression. This research suggests rumination significantly contributes to the onset of the incident depressive episode and to the risk of relapse, and that it is associated with greater severity, as indicated by the presence of suicidal ideation (Miranda & Nolen-Hoeksema, 2007; Rogers & Joiner, 2017). Response Styles Theory proposes that rumination maintains and exacerbates depression via increases in mood-congruent thinking, decreases in problem-solving and instrumental behavior, and decreases in social support (Nolen-Hoeksema, 1991; Nolen-Hoeksema, Wisco, & Lyubomirsky, 2008).

Rumination and Stage 0–1 Transition

Rumination is a powerful, reliable risk factor for the onset of major depression, suggesting it may be a useful target for stopping the Stage-0-to-1 transition. Multiple large prospective studies have shown that rumination predicts the subsequent development of major depressive episodes (Abela & Hankin, 2011; Just & Alloy, 1997). This relationship persists after controlling for other negative cognitive styles associated with depression (e.g., negative attributional styles, hopelessness, pessimism; Nolen-Hoeksema et al., 1994; Spasojevic & Alloy, 2001) as well as anxiety and depressive symptom severity (Wilkinson et al., 2013).

The pathway through which rumination confers risk for onset of depression appears to fit a diathesis-stress model in which higher rumination increases the risk of depression onset when it interacts with the emergence of a life stressor (Hilt et al., 2010; Nolen-Hoeksema et al., 1994; Rood et al., 2009). Longitudinal studies support the role of rumination as a mediating variable between negative life events and the development of depression (Michl et al., 2013). As such, rumination presents as a useful preventative intervention target. Moreover, rumination-targeted screening of youth appears to be an excellent strategy for identifying those at high risk for depressive symptoms (Young & Dietrich, 2014). A randomized trial of a 6-week CBT intervention targeting rumination and delivered via group or internet successfully reduced the 12-month prevalence rate of depression in adolescents and young adults (Topper et al., 2017).

Rumination and Stage 1–2 Transitions

The Stage 1–2 transition, however, presents a more complex challenge to understanding rumination’s role in maintaining depressive episodes. Evidence is mixed as to whether higher rumination is always associated with increased duration of depressive episodes (Just & Alloy, 1997; Morrow & Nolen-Hoeksema, 1990). For example, one study showed that lower rumination predicted less time to remission, but only for patients with mild-to-moderate depression. In contrast, severely depressed patients remitted faster with higher rumination (Jones, Siegle, & Thase, 2008). One hypothesis to explain the seemingly weaker or less universal role of rumination in depression recovery as compared to depression onset prevention is that once an individual succumbs to a full-blown depressive episode, other features of depression (e.g., vegetative symptoms) may exert a relatively stronger influence on disease course (Nolen-Hoeksema et al., 2008). However, these results also may be explained by a lack of sufficient variability in rumination levels within clinically depressed samples to detect differential effects (Nolen-Hoeksema et al., 2008). Nevertheless, the negative impact rumination exerts on mood, problem-solving, and social support provides a strong rationale for how rumination would extend and exacerbate an existing episode of depression (Nolen-Hoeksema, 1991; Nolen-Hoeksema et al., 2008). Additionally, evidence that rumination prolongs subclinical symptoms of depression is more consistent (Morrow & Nolen-Hoeksema, 1990), and, as discussed, residual symptoms place individuals at risk for stage progression (Buckman et al., 2018). Thus, more research may be needed to fully understand the role of rumination during recovery from a current depressive episode.

Rumination and Stage 2–3 Transitions

During Stage 3, once individuals have proven susceptible to recurrent depressive episodes, the clear benefit of targeting rumination is apparent again. Multiple studies suggest that rumination is a common residual symptom after depressive episodes subside (Nolen-Hoeksema et al., 2008) and is a risk factor for depressive relapse. One uncontrolled study showed that rumination levels at post-treatment predicted relapse at 12 months after adjusting for current depressive symptoms and other risk factors (Michalak, Hölz, & Teismann, 2011). Additionally, research shows that rumination is higher in individuals with a previous history of depression (McMurrich & Johnson, 2008), suggesting it is a relatively stable risk factor that persists beyond the end of a major depressive episode.

Rumination and Stage 3–4 Transitions

Rumination’s role in the final transition to treatment-resistant depression remains relatively unclear. This is partly due to inconsistencies in the definition of what constitutes treatment-resistant depression within the literature. Some studies require the failure of one medication trial, whereas others require the failure of multiple medications across drug classes or require additional failure of an evidence-based psychotherapy (Berlim & Turecki, 2007). Accordingly, additional research is needed detailing rumination’s role after treatment failures, particularly in response to multiple treatment modalities.

The Value of Intervening with Rumination

In summary, rumination seems a compelling prevention and treatment target for depression across much of the stage model. Currently, the strongest support exists for addressing rumination for preventing Stage 0–1 progression for individuals exhibiting risk factors or for individuals in Stage 3 who have experienced recurrent depressive episodes. In Stage 2 (i.e., first major depressive episode), more research is needed to determine the relative benefit of a specific focus on rumination versus other treatment targets. That is, rumination is often addressed by standard CBT for depression as well as mindfulness-based CBT (Bieling et al., 2012; Manicavasagar et al., 2012). Additionally, rumination-focused intervention protocols have been developed in recent years, such as Rumination-Focused CBT (RFCBT; Watkins et al., 2007). RFCBT includes standard CBT techniques for depression but replaces cognitive restructuring with specific techniques to support the identification of rumination patterns, stimulus control for rumination cues, and development of new behaviors. Such targeted interventions may continue to improve upon the rumination-related outcomes of standard evidence-based psychosocial treatments (Watkins & Roberts, 2020).

Preventative applications for rumination, highlighted by the stage model, are particularly exciting. One recent study evaluated the preventative effects of RFCBT in a sample of 251 adolescents with elevated worry and rumination but no current depression or anxiety diagnoses (Topper et al., 2017). Adolescents who received RFCBT vs. a waitlist control, showed significantly greater reductions in rumination as well as significantly lower rates of depression at a 12-month follow-up. Further, mediation analyses suggested that decreases in rumination explained 38.9% of the reduction in the prevalence of depression (Topper et al., 2017). Although more research is needed, these findings are promising.

Future Directions

The current body of literature is more suggestive than conclusive with respect to understanding how targeting different mechanisms at different times in stage progression may affect the course of depression. The reason is that much previous research has not explicitly taken the stage-model approach. Prior studies have often aggregated patient samples across multiple stages, thereby preventing definitive conclusions about progression for particular transitions in the stage model. For example, although the meta-analysis of longitudinal studies by Baglioni et al. (2011) examined insomnia as a predictor of the incidence of depression, their search strategy was not fully limited to Stage 0-to-1 transitions. Therefore, their approach did not isolate how insomnia predicted the stage transition to the first-incident depressive episode. Similarly, as another example, the randomized clinical trial by Topper et al. (2017) suggested that CBT’s effects on depression were mediated by rumination, but they did not determine participants’ depression stage (i.e., some participants were in Stage 0 and others in Stage 3a). They note this is a limitation: “participants’ past history of psychopathological diagnoses were not assessed upon entry into this study. Therefore, it cannot be established whether the effects observed concern the prevention of first episodes of depression and GAD vs. relapse/recurrence” (p. 133). Future research—both observational studies and clinical trials—should apply the stage model to participant selection based on clinical assessment to move beyond the broader categories of remission, residual symptoms, relapse, and recurrence to instead focus on particular transitions in the stage model.

As is evident from the current review, CBT has efficacy across multiple stages, and indeed it appears that insomnia and rumination may be particularly important trans-stage processes underlying stage progression. The further development of stage-specific interventions requires greater attention to stage membership in research samples to allow characterization when “one size does not fit all” for clinical interventions across stages. To aid this research, there have been calls for more refined assessments of psychopathology and euthymia to better characterize individuals relative to stage progression and associated preventive efforts (Guidi & Fava, 2020). As part of refining assessment, it will be important to understand which risk factors for stage progression are unique predictors vs. correlates of depressed mood or some other factor. Additionally, future research should test whether factors that predict stage progression show associations that are developmentally moderated. This may include attention to “scarring” hypotheses about ways in which initial episodes of depression change the individual in a way that sets the stage for subsequent episodes (Burcusa & Iacono, 2007). For example, there is evidence that dysfunctional attitudes may arise from initial experiences of depression and influence risk for subsequent episodes, but these patterns may be more likely when the first episode of depression strikes in youth rather than later adulthood (Otto et al., 2007). In sum, a fuller understanding of the relationship between stages of depression and their risk factors will depend on future studies that consider depression stages in their design and characterize their sample accordingly. The products of this research will help clarify the true clinical utility of adopting a stage model approach for depression.

Further, the literature of randomized controlled trials examining CBT has rarely used the experimental-medicine approach explicitly. That is, the associations between the measured mechanisms and the depression outcome are often not reported in these studies. Future clinical trials should include measurement of putative mechanisms—including rumination and insomnia—as they investigate CBT’s efficacy in preventing stage transition. Further, studies showing important associations between insomnia symptoms and depression have variably assessed insomnia symptoms, including full criteria for insomnia disorder status (e.g., Johnson et al., 2006), elevated sleep impairment scores on symptom checklists (e.g., Pigeon et al., 2008) markers evident in chart review (e.g., Cepeda et al., 2018), or a combination of these methods (e.g., Li et al., 2016). The consistency of findings across these various methods underscores the robustness of insomnia’s associations with depression but leaves unclear what severity or type of insomnia symptoms (e.g., difficulties initiating or maintaining sleep, early vs. late sleep disruption) might be most relevant for prevention or intervention efforts. More research is needed to clarify this issue (cf., Leerssen et al., 2021; Taylor et al., 2010), but at present it appears that risk of depression is conferred readily by any of several prominent insomnia symptoms.

Additionally, the presence and impact of sleep impairment and rumination on depression stage progression may differ by race and ethnicity. Racial and ethnic disparities in sleep duration and quality are well documented (e.g., Ahn et al., 2021; Jones et al., 2020). Likewise, rumination is elevated in socially disempowered groups (Nolen-Hoeksema, Larson, & Grayson, 1999; Hatzenbuehler, McLaughlin, & Nolen-Hoeksema, 2008). Accordingly, research on the mechanistic influence of these risk factors on depression stage progression should account for the differential impact on minority populations to avoid the further propagation of harmful disparities in depression care (Shao et al., 2016).

It is also important to note that mechanistic investigations of sleep disruption and rumination need to be expanded beyond depression to the onset and maintenance of other psychiatric disorders and related behaviors. Research on health behaviors (e.g., substance use) supports the transdiagnostic importance of attending to these risk factors (see Clancy, Prestwich, Caperon, & O’Connor, 2016; Dolsen, & Harvey, 2017; Freeman et al., 2017; Grierson, Hickie, Naismith, & Scott, 2016; Riley, Park, & Laurenceau, 2019).

Finally, the degree of independence between the risk factors of rumination and insomnia needs further study. Non-independence among risk factors for depression is a common issue (e.g., see Burcusa & Iacono, 2007) and appears to be the case for rumination and insomnia. Rumination is clearly implicated in understanding how sleep reactivity contributes to insomnia. Sleep reactivity refers to the degree to which individuals exhibit sleep-disruptive responses to stress (Drake, Pillai, & Roth, 2013; Kalmbach, Anderson, & Drake, 2018). Rumination appears to be an important mediator in this relationship (Amaral et al., 2018; Otto et al., 2022). Sleep disruption, in turn, may then further enhance ruminations (e.g., Nota & Coles, 2018). Although rumination is sometimes implicated in insomnia treatment’s efficacy for depression (Cheng et al., 2020), meta-analysis suggests the effects of CBT-I on rumination are generally subtle (Ballesio et al., 2020). As such, more research is needed to understand the interplay of these risk factors—rumination and disrupted sleep—and whether existing intervention strategies can reliably target both.

Limitations in the Scope of Review

The present review was limited to only two risk factors. In emphasizing the role of insomnia and rumination as cross-stage risk factors for depression, we are not suggesting that other factors do not share their broad applicability to the course of depression. Ongoing investigations seek to expand the range of risk factors under consideration as well as the types of models used in prognostic studies of depression outcome (Buckman et al., 2021). These studies, as applied across the stages of depression, have the potential to elucidate additional modifiable risk factors for stage progression.

An additional limitation is that we centered our review on the trans-stage efficacy of CBT, providing some comparisons to pharmacologic alternatives but without detailing the role of other psychosocial interventions. For example, interpersonal psychotherapy has shown trans-stage efficacy in meta-analytic reviews (Clarke, Mayo-Wilson, Kenny, & Pilling, 2015; Cuijpers et al., 2011). Likewise, exercise interventions (used alone or in combination with other empirically supported treatments) show trans-stage efficacy for depression as evaluated by meta-analysis (Kvam, Kleppe, Nordhus, Hovland, 2016; Lee, Gierc, Vila-Rodriguez, Puterman, Faulkner, 2021; Schuch et al., 2018). Further, the efficacy of CBT includes the delivery of both cognitive restructuring and behavioral activation elements, and either element alone or their combination can offer strong efficacy for treating depression (for a network meta-analysis, see Ciharova et al., 2021). Finally, we did not specifically review the promise of mindfulness-based CBT for both prevention and intervention applications to the course of major depression (see Beshai, Dobson, Bockting, & Quigley, 2011; Bockting et al., 2015; Piet & Hougaard, 2011).

Conclusions

In many ways, the function of a stage model for major depression is to help clinicians “lift their eyes to the horizon” to consider the broader course of a patient’s depression. This shift in focus is defined by a vigilance to the risk for stage progression and the empirically informed selection of interventions to address that risk. Current evidence supports CBT as an efficacious treatment for depression that offers many benefits: (1) reducing prodromal symptoms and episodes of major depression, (2) addressing residual symptoms, (3) preventing relapse, and (4) addressing pharmacologic treatment resistance. The available evidence also supports a broadening of potential treatment targets to include assessment and intervention with two particular risk factors for stage progression: insomnia symptoms and rumination. Additional documentation is needed regarding the stage-specific efficacy of intervening with these risk factors. Nevertheless, both potential mechanisms are known to be modifiable risk factors and likely can be strategically harnessed to prevent depressive episodes or relapse at sensitive transition points. Greater attention to these factors is needed in the practice of CBT in order to address stage progression toward treatment resistance.