Factors Associated with Treatment Dropout in Patients with Mood Disorders Authors: Hwagyu Suh, Eunsoo Moon, Hyunju Lim, Kyungwon Kim, Young-Min Lee, Byung-Dae Lee Published: 2026 ABSTRACT ObjectiveTreatment dropout is common among patients with mood disorders and adversely affects long-term outcomes. This study investigated clinical and psychological factors associated with dropout, with a particular focus on stress coping strategies.MethodsThis retrospective cohort study included 321 patients with mood disorders who completed diagnostic evaluations and psychological assessments at a specialized mood disorder clinic between April 2010 and March 2021. Treatment dropout was defined as failure to attend scheduled outpatient appointments for more than 30 days without consultation, combined with no outpatient visit for at least 90 days.ResultsKaplan–Meier analyses showed significant differences in treatment retention according to diagnostic category. In univariate Cox analyses, younger age, diagnostic category, emotion-oriented coping, and pleasure-seeking activities were associated with dropout. In the clinically informed multivariable Cox model, younger age remained significantly associated with a higher hazard of dropout (hazard ratio = 0.987, 95% confidence interval = 0.976−0.999, p = 0.034). Diagnostic category was also significantly associated with dropout (Wald χ2 = 25.433, df = 4, p < 0.001). Compared with bipolar I disorder, bipolar II disorder, other specified bipolar and related disorder, major depressive disorder, and other depressive disorders showed significantly higher hazards of dropout. Emotion-oriented coping and pleasure-seeking activities were no longer significant after adjustment.ConclusionDiagnostic category and younger age were the most robust factors associated with treatment dropout. Coping variables showed exploratory associations in univariate analyses, but these associations were attenuated after adjustment for demographic and clinical factors. INTRODUCTION Mood disorders are among the most prevalent psychiatric conditions worldwide and are characterized by a chronic and recurrent course, contributing substantially to the global burden of disease [1,2]. In particular, major depressive disorder (MDD) and bipolar disorder are associated with an increased risk of relapses, symptom exacerbation, functional impairment, and suicide when treatment is discontinued, highlighting the importance of long-term treatment maintenance [2-4]. Consequently, treatment adherence has been recognized as a key determinant of long-term prognosis in patients with mood disorders [3,5]. Despite the importance of sustained treatment, premature discontinuation remains a persistent clinical challenge. Previous studies have reported that approximately 30−60% of patients with depressive disorders discontinue treatment during the early stages, and long-term follow-up studies have similarly demonstrated substantial rates of treatment dropout [6,7]. Early treatment discontinuation extends beyond poor medication adherence; it may hinder remission and contribute to chronicity and recurrent episodes, thereby adversely affecting long-term outcomes [8]. Various factors associated with treatment discontinuation have been extensively investigated, including demographic characteristics such as age, sex, and socioeconomic status, as well as illness severity, accessibility to care, medication side effects, and attitudes toward treatment [9-11]. More recently, increasing attention has been directed toward psychological processes underlying treatment adherence, particularly patients’ cognitive appraisal of stress and coping strategies. Coping refers to the cognitive and behavioral efforts individuals use to manage stressful situations [12]. Furthermore, it is closely associated with personality traits and emotion regulation styles [13]. Recent findings on cognitive reappraisal further suggest that depressive symptoms may influence how individuals select and benefit from emotion-regulation strategies [14]. Emotion-oriented or avoidant coping strategies, in particular, aim primarily to reduce emotional discomfort rather than address the stressor itself. Over time, such coping patterns may contribute to prolonged stress exposure and worsening psychopathology [15,16]. These maladaptive coping styles may also reduce engagement in treatment by promoting avoidance of emotional distress and self-reflection, both of which are often required during psychiatric care. Indeed, previous studies involving patients with mood disorders have reported associations between maladaptive coping strategies and poorer treatment adherence or reduced treatment persistence [17,18]. However, prior research has largely focused on short-term outcomes or dichotomous measures of treatment adherence, while relatively few studies have examined patterns of treatment discontinuation over time. In addition, comprehensive investigations into how different coping strategies influence long-term treatment dropout remain limited. Studies specifically evaluating time-to-dropout risk and its relationship with stress coping styles are still scarce. Therefore, the present study aimed to investigate the cumulative incidence of treatment dropout in patients with mood disorders through long-term follow-up and to examine the impact of various stress coping strategies on treatment discontinuation. By doing so, we sought to improve understanding of the psychological factors associated with reduced treatment persistence and to identify potential targets for clinical intervention aimed at enhancing long-term treatment engagement. METHODS Subjects This study was designed as a retrospective cohort study using electronic medical records (EMRs) from a specialized mood disorder clinic. A total of 321 patients with mood disorders who completed diagnostic evaluations and psychometric assessments at their initial visit between April 2010 and March 2021 were included. Demographic and clinical characteristics, psychometric scale scores, and time to treatment discontinuation were retrospectively obtained from EMRs and the clinic database. Patients were excluded if they met any of the following criteria: (1) missing data on major clinical scales; (2) diagnostic uncertainty with schizophrenia, schizoaffective disorder, or other psychotic disorders; (3) significant neurological disorders; (4) intellectual disability; (5) severe medical illnesses likely to interfere with psychiatric outpatient treatment; or (6) planned termination of treatment based on the treating psychiatrist’s clinical judgment. The study protocol was approved by the Institutional Review Board (IRB) of Pusan National University Hospital (PNUH: 2106-032-104). Written informed consent for the use of clinical and psychometric data for research purposes had been obtained from participants at the time of initial assessment. Data Collection and Measures Clinical information was obtained from EMRs. Collected variables included demographic characteristics (age at first assessment, sex, years of education, employment status, and socioeconomic status), clinical characteristics (psychiatric diagnosis and number of suicidal attempts), and stress coping strategies. Psychiatric diagnoses were classified into bipolar I disorder (BD-I), bipolar II disorder (BD-II), other specified bipolar and related disorder (OSBD), MDD, and other depressive disorders. Because only two participants were diagnosed with persistent depressive disorder (PDD), PDD was combined with other specified depressive disorder (OSDD) into a single “other depressive disorders” category to improve the stability of model estimates. Definition of treatment discontinuation (dropout) Treatment discontinuation (dropout) was defined as the absence from scheduled outpatient visits for more than 30 days without consultation with the treating physician [11], followed by no clinic visit for at least 90 days. This operational definition was based on previous studies of treatment continuity [6,19]. Time to dropout was defined as the interval from the first visit to the mood disorder clinic to the last outpatient visit. Patients who remained in treatment without dropout by the end of the observation period were treated as censored observations. Coping strategies Stress coping styles were assessed using the Coping Inventory for Stressful Situations (CISS) and the Behavioral Checklist for Coping with Stress (BCCS). The CISS evaluates coping responses to stressful situations and categorizes them into emotion-oriented coping, task-oriented coping, and avoidance-oriented coping [12,20]. The BCCS was developed to assess behavioral coping responses rather than broader cognitive or emotional coping tendencies [21]. The scale includes four behavioral domains: personal activities (e.g., meditation or solitary activities), pleasure-seeking activities (e.g., drinking alcohol, gaming, or other hedonic behaviors), social activities (interpersonal interactions), and compulsive activities (impulsive emotional release behaviors such as binge eating) [22]. CISS and BCCS subscale scores were treated as continuous variables in the analyses. In the present sample, the internal consistency of the CISS subscales was excellent to good, with Cronbach’s α values of 0.92 for task-oriented coping, 0.88 for emotion-oriented coping, and 0.80 for avoidance-oriented coping. Cronbach’s α values for the BCCS subscales were 0.81 for personal activities, 0.64 for pleasure-seeking activities, 0.79 for social activities, and 0.69 for compulsive activities. Statistics Descriptive statistics were used to summarize demographic and clinical characteristics. Differences between the bipolar disorder group and depressive disorder group were analyzed using independent ttests for continuous variables and chi-square tests for categorical variables. Kaplan–Meier survival analyses were performed to examine treatment retention according to diagnostic category, and differences between survival curves were assessed using the log-rank test. Univariate Cox proportional hazards regression analyses were first conducted to examine factors associated with time to dropout. To address clinically relevant confounding, a clinically informed multivariable Cox proportional hazards regression model was constructed [23]. Variables included in the multivariable model were selected a priori based on clinical relevance and previous literature. Age at first assessment, sex, diagnostic category, number of suicide attempts, emotion-oriented coping, and pleasure-seeking activities were entered simultaneously into the model using the forced-entry method. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated. Statistical significance was set at p < 0.05. All statistical analyses were performed using SPSS version 28 (IBM Corp.). RESULTS Demographic and Clinical Characteristics A total of 321 patients with mood disorders were included in the final analysis, comprising 221 patients with bipolar disorders (68.8%) and 100 patients with depressive disorders (31.2%). No significant difference in sex distribution was observed between the two groups (χ2 < 1, p = 0.894). However, patients with depressive disorders were significantly older at baseline than those with bipolar disorders (t = −3.94, p < 0.001). Years of education did not differ significantly between groups (p = 0.197). Significant differences were found in marital status and employment status: patients with bipolar disorders were more likely to be married (p = 0.001), whereas those with depressive disorders were more likely to be employed full-time or homemakers (p = 0.024). No significant difference was observed in socioeconomic status between the groups (p = 0.910) (Table 1). Baseline demographic characteristics were also compared between the dropout and treatment-retention groups (Supplementary Table 1; available online). The dropout group was significantly younger than the treatment-retention group (33.00 ± 11.81 vs. 36.61 ± 13.66 years, p = 0.015), whereas sex, marital status, employment status, socioeconomic status, and years of education did not differ significantly between groups. Factors Associated with Dropout Univariate Cox proportional hazards regression analyses were conducted to identify factors associated with dropout (Table 2). Among demographic variables, younger age was associated with a higher risk of dropout (HR = 0.988, 95% CI = 0.977−1.000, p = 0.043). The number of suicide attempts showed marginal significance (HR = 1.041, 95% CI = 1.000−1.084, p = 0.052). Sex and years of education were not significantly associated with dropout. Diagnostic category was significantly associated with dropout risk (χ2 = 27.915, df = 4, p < 0.001). Using BD-I as the reference group, patients with BD-II, OSBD, MDD, and other depressive disorders showed higher hazards of dropout. Among coping strategies measured by the CISS, emotion-oriented coping was significantly associated with an increased risk of dropout (HR = 1.015, 95% CI = 1.003−1.027, p = 0.016), whereas task-oriented coping and avoidance-oriented coping were not significantly associated with dropout risk. Among BCCS subscales, pleasure-seeking activities were significantly associated with an increased risk of dropout (HR = 1.026, 95% CI = 1.002−1.051, p = 0.033), while compulsive activities demonstrated borderline significance (HR = 1.037, 95% CI = 0.997−1.078, p = 0.071). Neither personal nor social activities were significantly associated with dropout. Kaplan–Meier Survival Analyses by Diagnostic Groups Kaplan–Meier survival analyses revealed significant differences in treatment retention across diagnostic categories (Fig. 1). The log-rank test indicated that survival distributions differed significantly among diagnostic groups (χ2 = 27.946, df = 4, p < 0.001). Patients with BD-I showed the longest treatment retention, with a median time to dropout of 3,545 days (95% CI = 2,536−4,554). In contrast, patients with OSBD demonstrated the shortest median time to dropout (659 days, 95% CI = 188−1,130). Median times to dropout were 1,377 days for BD-II (95% CI = 591−2,163), 1,088 days for MDD (95% CI = 493−1,683), and 792 days for other depressive disorders (95% CI = 0−1,849). Overall, patients with BD-I consistently demonstrated the highest treatment retention throughout the follow-up period. Kaplan–Meier curves demonstrated that dropout occurred predominantly during the early phase of follow-up, followed by a more gradual decline in treatment retention over time (Fig. 1). Multivariable Cox Analysis of Factors Associated with Dropout A clinically informed multivariable Cox proportional hazards regression model was constructed, including age at first assessment, sex, diagnostic category, number of suicide attempts, emotion-oriented coping, and pleasure-seeking activities (Table 3). The overall model was statistically significant (χ2 = 39.056, df = 9, p < 0.001). Younger age remained significantly associated with a higher hazard of dropout (HR = 0.987, 95% CI = 0.976−0.999, p = 0.034). Sex was not significantly associated with dropout (HR = 0.947, 95% CI = 0.695−1.290, p = 0.730). Diagnostic category remained significantly associated with dropout after adjustment for other variables (Wald χ2 = 25.433, df = 4, p < 0.001). Compared with patients with BD-I, those with BD-II (HR = 2.225, 95% CI = 1.282−3.862, p = 0.004), OSBD (HR = 3.027, 95% CI = 1.894−4.839, p < 0.001), MDD (HR = 3.171, 95% CI = 1.914−5.253, p < 0.001), and other depressive disorders (HR = 2.433, 95% CI = 1.350−4.386, p = 0.003) showed significantly higher hazards of dropout. The number of suicide attempts was not significantly associated with dropout after adjustment (HR = 1.021, 95% CI = 0.976−1.069, p = 0.369). Emotion-oriented coping (HR = 1.001, 95% CI = 0.988−1.014, p = 0.874) and pleasure-seeking activities (HR = 1.018, 95% CI = 0.992−1.044, p = 0.175) were also not significantly associated with dropout in the multivariable model. DISCUSSION The present study investigated clinical and psychological factors associated with treatment dropout among patients with mood disorders using long-term follow-up data. Diagnostic category showed the strongest association with dropout. Patients diagnosed with BD-II, OSBD, MDD, and other depressive disorders (OSDD and PDD) showed significantly higher hazards of dropout than those with BD-I, even after adjustment for age, sex, number of suicide attempts, and coping-related variables. Younger age was also independently associated with a higher hazard of dropout. In addition, emotion-oriented coping and pleasure-seeking activities were associated with increased dropout risk in univariate analyses, although these associations diminished after adjustment for other variables in the multivariable model. These findings underscore the clinical importance of diagnostic differentiation in mood disorder management. Research examining depression subtypes has demonstrated that diagnostic specificity influences treatment intensity requirements and clinical management approaches, suggesting that different diagnostic presentations warrant tailored clinical strategies [24]. Our results extend this perspective to treatment retention, indicating that diagnostic differentiation is equally important for identifying patients at elevated risk for treatment discontinuation and for implementing targeted engagement interventions. In the present study, patients with BD-I showed the highest treatment retention. One possible explanation is that patients with BD-I may have greater exposure to severe mood episodes, functional impairment, and hospitalization experiences, potentially increasing patients’ and caregivers’ awareness of illness severity and the perceived need for continued treatment. Patients with OSBD and depressive disorder diagnoses showed higher hazards of dropout than those with BD-I. This may reflect differences in illness course, symptom fluctuation, perceived need for treatment, or treatment expectations. Previous studies have similarly reported that insight into illness and recognition of treatment necessity are closely associated with treatment adherence among patients with bipolar disorder [25, 26]. The finding that patients with OSBD showed the highest dropout risk has important clinical implications. According to DSM-5, OSBD includes patients who do not fully meet criteria for bipolar I or bipolar II disorder but exhibit clinically meaningful bipolar-spectrum symptoms. In such patients, symptom variability and diagnostic uncertainty may complicate treatment planning and the development of shared expectations regarding long-term care. Furthermore, hypomanic experiences are often perceived as ego-syntonic, leading some patients to minimize or deny illness severity. Such denial may reduce treatment motivation and contribute to premature dropout [17,27]. Emotion-oriented coping and pleasure-seeking activities were associated with dropout in univariate analyses, suggesting that coping style may be clinically relevant to treatment retention. However, these associations were no longer significant after adjustment for age, sex, diagnostic category, and number of suicide attempts, indicating that coping-related effects may be closely linked to broader diagnostic and clinical characteristics. According to previous studies, as depressive symptoms become more severe, patients with both bipolar and depressive disorders tend to rely overwhelmingly on emotion-oriented coping regardless of their specific diagnosis [28]. Emotion-oriented coping tends to focus on reducing emotional distress rather than addressing stressors themselves and has been associated with rumination, self-blame, and emotional overwhelm [13,29]. Previous studies have shown that patients with mood disorders often rely more heavily on emotion-focused coping than problem-solving strategies, and such coping patterns have been linked to poorer illness acceptance, lower quality of life, and worsening depressive and anxiety symptoms [16,18]. These coping mechanisms may lead patients to avoid the self-reflection and emotional confrontation required during the treatment process, ultimately undermining long-term treatment retention. Pleasure-seeking coping may be understood as a strategy aimed at obtaining immediate emotional relief or impulsive gratification under stress. Although such behaviors may temporarily alleviate distress, they can delay problem-solving and reinforce avoidant behavioral patterns over time. In bipolar-spectrum disorders, impulsivity and reward-seeking behaviors are considered central clinical characteristics [30], which may interfere with sustained participation in outpatient treatment. The attenuation of the effect of pleasure-seeking coping in the multivariable model may indicate that these behaviors substantially overlap with underlying diagnostic characteristics. Another important finding was that most dropout events occurred during the early phase of treatment. This observation is consistent with previous studies of psychiatric treatment dropout and highlights the importance of establishing a therapeutic alliance and aligning treatment expectations during the initial stages of care [31]. Early evaluation of diagnostic characteristics, coping style, and treatment expectations, particularly among younger patients, may help identify those who require additional engagement strategies. Indeed, previous studies have demonstrated that structured psychoeducation improves illness awareness, medication adherence, and long-term treatment retention among patients with mood disorders [25]. Several limitations should be considered when interpreting the findings of this study. First, this was a retrospective study conducted at a single specialized mood disorder clinic, which limits generalizability to broader psychiatric outpatient settings or community populations. Patients with bipolar disorders were overrepresented in the sample, reflecting the characteristics of the specialized clinic. Second, dropout was operationally defined using clinic attendance records. Although this definition was useful for identifying sustained discontinuation of follow-up, the retrospective data did not allow us to determine the underlying reasons for dropout or to assess treatment-related factors such as medication adherence, medication adverse effects, therapeutic alliance, or clinician changes. Thus, dropout should be interpreted as clinic-attendance discontinuation rather than as a direct measure of poor treatment engagement. Third, although PDD was combined with OSDD to improve estimate stability, some diagnostic subgroups remained relatively small, and diagnosis-specific HRs should be interpreted cautiously. Fourth, coping strategies were measured only at baseline, although coping may change over time. Future prospective, multi-center studies incorporating longitudinal coping assessments, treatment-related variables, and direct assessment of dropout reasons are needed. Despite these limitations, this study provides meaningful insights by examining long-term trajectories of dropout among patients with mood disorders and exploring the relationship between coping strategies and treatment retention. The findings suggest that diagnostic category and younger age may help identify patients at increased risk for treatment dropout, while coping strategies may provide additional exploratory information regarding psychological characteristics related to treatment retention. Future prospective studies with larger samples are needed to evaluate integrated models incorporating psychological variables such as therapeutic alliance, impulsivity, illness insight, and emotion regulation.