INTRODUCTION
Stress and coping with it can influence the onset and recurrence of mood disorders [1-5]. Stress is known to impact the onset and recurrence of depression [5]. Additionally, in patients with bipolar disorder, environmental stress often precedes a mood episode [4,6]. Patients with depression tend to rely excessively on emotional coping in stressful situations [7]. People who used avoidant coping experienced more depressive symptoms than people who did not [8]. Patients with depression used more avoidant coping than the control group [9]. In addition, patients with bipolar disorder, especially those with high anxiety, had more emotional coping than healthy adults [10]. Emotional coping may increase the risk of recurrence of mood episodes, and avoidant coping may prevent the negative effects of stressful situations from being alleviated.
Stress coping refers to all activities an individual does in a stressful situation, and includes cognitive and behavioral attempts to control internal or external demands [11]. Although each person has a preferred habitual coping style, it can also be understood as a coping process that continues to change depending on the results of coping [12]. Therefore, changing maladaptive coping into adaptive coping will have a positive effect on the course of mood disorders [13]. It would be helpful if maladaptive cognitive coping strategies could be addressed in an adaptive and healthy way. However, in patients with mood disorders, mood symptoms can lead to difficulty concentrating and cognitive distortions. It may be more advantageous to analyze the patient’s behavior when assessing coping and providing interventions. In other words, behavior is easier to observe than cognitive aspects, and improvement goals can be clearly set. For this reason, behavior-centered approach will be helpful in managing stress in patients with mood disorders [14,15].
Despite the usefulness of behavioral approach, research on stress coping behavior in patients with mood disorders is not sufficient [14,15]. The Behavioral Checklist for Coping with Stress (BCCS) was developed to evaluate the stress-coping behaviors [15]. Previous studies showed internal consistency and structural validity in patients with mood disorders [15,16]. A factor analysis showed four factors: personal activity, pleasure-seeking activity, social activity, and compulsive behavior. According to the results of previous study, the bipolar patients engaged in fewer social activities than the control group, but used pleasure-seeking activities more frequently [15]. In addition, personal or social activities reduce depression and anxiety, and pleasure-seeking or compulsive behaviors are associated with high levels of depression, anxiety, and anger suppression [16]. As such, maladaptive stress-relieving behaviors are commonly observed in patients with mood disorders.
Therefore, it may be necessary to intervene in maladaptive stress-coping strategies for patients with mood disorders. Given that individuals have unique preferences for stress coping strategies [17], identifying individual coping types may facilitate adaptive coping interventions. Recent studies have used cluster analysis to explore coping types in the general population [17,18]. Therefore, this study aimed to explore the clinical types of stress-coping behaviors in patients with mood disorders using cluster analysis. Furthermore, sociodemographic and clinical characteristics were compared among clusters.
METHODS
Participants
The subjects were patients with mood disorders who visited the outpatient clinic of the Mood Disorder Clinic at Pusan National University Hospital from November 2011 to February 2021. The inclusion criteria were 1) bipolar I disorder, bipolar II disorder, or other specified bipolar and related disorder, major depressive disorder, persistent depressive disorder, or other specified depressive disorder, when diagnosed by a psychiatrist according to the 5th edition Diagnostic and Statistical Manual for Mental Disorders (DSM-5) [19], and 2) patients aged between 18 and 65 years. Exclusion criteria were 1) patients with current manic or hypomanic episode due to possibility of uncooperativeness or unreliability on self-assessment, 2) patients who refused stress-related clinical assessments or did not complete self-report assessments, and 3) patients with serious medical conditions that could interfere with the completion of self-report assessments. A total of 252 patients, consisting of 184 people with bipolar disorders and 68 people with depressive disorders, were included. This study was approved by the Institutional Review Board at Pusan National University Hospital (2412-012-146).
Measures
Sociodemographic and clinical characteristics
Electronic medical records were assessed to examine the sociodemographic and clinical characteristics of the subjects at the first visit to the mood disorders clinic. Sociodemographic data were examined including age, sex, education, marital status, and occupation. Clinical characteristics including onsets of symptoms and mood episode, and illness were examined. In addition, self-report questionnaires on stress-coping behaviors, global severity, bipolarity, depression, and anxiety were performed.
BCCS
Stress coping behaviors were assessed using the BCCS [15]. The BCCS was designed to evaluate stress coping behaviors in patients with mood disorders. The BCCS with 54 items asks, “How often do you do the following to relieve stress?” Each behavior item answered on a 5-point Likert scale, “I don’t do it at all” (0), “I rarely do it” (1), “I do it sometimes” (2), “I do it often” (3), and “I do it very often” (4). In previous studies, the BCCS showed acceptable internal consistency and excellent structural validity, showing the potential to be a useful tool for measuring stress coping behavior [16]. BCCS consists of four subscales: personal, social, pleasure-seeking, compulsive/avoidant activities [15].
Clinical Global Impression-Severity
Clinical Global Impression-Severity (CGI-S) was used to measure the severity of the subject’s overall state of current mental symptoms, with 1 (normal), 2 (borderline), 3 (mild), 4 (moderate), 5 (markedly ill), 6 (severe), and 7 (extreme condition) [20].
Mood Disorder Questionnaire
The Mood Disorder Questionnaire (MDQ) is a self-report scale developed by Hirschfeld et al. [21] to screen for bipolar disorder. Originally, MDQ consisted of Part A, B, and C. Part A asks manic symptoms in the past. Part B checks whether the symptoms examined in Part A were experienced at the same time, and Part C checks whether problems with daily functioning occurred due to the symptoms of Part A. In this study, the total score of Part A of the MDQ was used to quantify the characteristics of lifetime mood states. Previous studies also reported that evaluating only Part A had higher sensitivity and specificity [22].
Bipolar Spectrum Diagnostic Scale
The Bipolar Spectrum Diagnostic Scale (BSDS) is a self-report scale developed by Ronald for the early detection of bipolar disorder patients with mild symptoms, and assesses characteristics of mood states throughout their life [23-25]. The first part contains 19 questions that describe the symptoms of bipolar disorder, and you are asked to mark if the contents apply to you, and each mark is scored as 1 point. In the second part, indicate how much the content in the first part applies to you by indicating one of four options. The BSDS total scores range from 0 to 25, with higher scores indicating a greater possibility of the disorder [25]. In this study, the total score of BSDS (BSDS-T), depressive subscale score of BSDS (BSDS-D), and mania subscale score of BSDS (BSDS-M) were used.
Beck Depression Inventory-II
The Beck Depression Inventory-II (BDI-II) was used to evaluate the severity of current depressive symptoms [26-28]. The depression scale developed by Beck and Steer [29] is a self-report scale consisting of 21 items. The items evaluate the emotional, cognitive, motivational, and physiological aspects of depressive symptoms; the total score is 0 to 63 points. The BDI-II excludes four items from original version: previous weight loss, changes in body image, work difficulties, and physical preoccupation, and includes nervousness, difficulty concentrating, feelings of worthlessness, and loss of motivation [26,27]. Four items were added, and the questions about sleep and appetite were changed to ones that can evaluate both increases and decreases. The Korean version of BDI-II has shown excellent reliability and validity, and the optimal dividing point was suggested to be 18 points [26].
State-Trait Anxiety Inventory
The State-Trait Anxiety Inventory (STAI) was used to assess anxiety-related characteristics in the subject. STAI is a self-report scale developed by Spielberger et al. [30] to assess anxiety symptoms and is useful in clinical as well as general populations [31]. STAI consists of 20 state anxiety measurement items and 20 trait anxiety measurement items. State anxiety refers to anxiety about a specific situation, and trait anxiety refers to innate anxiety characteristics.
Analysis
Hierarchical clustering and K-means cluster analysis were conducted to classify types of stress-coping behaviors. A Euclidean distance matrix was computed based on BCCS scores, and hierarchical clustering was performed using Ward’s linkage method. The optimal number of clusters was initially explored using a dendrogram derived from the hierarchical analysis. In addition, the elbow method—a heuristic approach commonly used in behavioral research—was also considered. The dendrogram suggested 3 or 4 clusters, while the elbow method indicated 4 or 5. Taking these results into account, along with considerations of clinical relevance and interpretability, four clusters were ultimately selected. K-means cluster analysis was then performed using the predetermined four-cluster solution.
Socio-demographic and clinical characteristics of each group divided according to the results of the cluster analysis were compared. Categorical variables were analyzed by chi-square tests, and continuous variables were compared by post-hoc tests using the Scheffé method after one-way analysis of variance. Statistical analysis was performed using the Statistical Package for Social Sciences version 29.0 (IBM Corp.). All tests were two-sided with a significant level of 0.05.
RESULTS
Comparison of stress-coping behaviors among clusters
As shown in Table 1, Cluster 1 exhibited significantly higher levels of personal and social activities compared to Clusters 2/3, but significantly lower levels than Cluster 4. In other words, personal and social activities in Cluster 1 were balanced between those observed in Clusters 2/3 and Cluster 4. Additionally, unlike Clusters 3/4, Cluster 1 showed lower levels of pleasure-seeking activities. Considering these overall patterns of stress-coping behaviors, this cluster was labeled the “Balanced Cluster.” In this cluster, CGI-S and BDI-II scores were low, and MDQ, BSDS-T, and BSDS-D scores were also lower. The proportion of depressive disorder patients was the second highest among the four clusters.
Cluster 2 exhibited generally lower levels of stress-coping behaviors compared to the other clusters. Based on these characteristics, it was labeled the “Withdrawn Cluster.” In this cluster, CGI-S and BDI-II scores were high, while MDQ, BSDS-T, BSDS-D, and BSDS-M scores were all lower. The proportion of depressive disorder patients, especially major depressive disorder, was the highest among the four clusters.
Cluster 3 exhibited the high levels of both pleasure-seeking and compulsive/avoidant activities compared to the other clusters. Based on these characteristics, it was labeled the “Impulsive Cluster.” In this cluster, CGI-S and BDI-II scores were high, and MDQ, BSDS-T, BSDS-D, and BSDS-M scores were all elevated. Mood symptom onset occurred earlier in Cluster 3 compared to Clusters 1 and 2. The average age of participants in this cluster was lower, and the proportion of bipolar disorder patients was the second highest among the four clusters, with a notable prevalence of patients with unspecified bipolar and related disorder.
Cluster 4 exhibited the highest levels of personal, social, and pleasure-seeking activities among all clusters. Given its overall high engagement in stress-coping behaviors, this cluster was labeled the “Active Cluster.” In this cluster, CGI-S and BDI-II scores were the lowest, while MDQ, BSDS-T, BSDS-D, and BSDS-M scores were higher. Additionally, the proportion of bipolar disorder patients was the highest, particularly with a significant prevalence of bipolar I disorder patients.
Comparison of sociodemographic and clinical characteristics among clusters
Sociodemographic and clinical characteristics of subjects among clusters were presented in Table 2. There was a significant difference in age among clusters (p<0.001), with Cluster 3 being significantly younger than Cluster 1 (p<0.05) and Cluster 2 (p<0.05). However, there were no significant differences in sex, education level, and occupation among clusters. Overall, a higher proportion of participants were unmarried (p=0.04), but marital status did not significantly differ between clusters.
Regarding clinical features, there was a significant difference in mood symptom onset age among clusters (p=0.003), with Cluster 3 showing earlier symptom onset than Cluster 1 and Cluster 2 (p<0.05). Initiation of mood episodes also differed significantly among clusters (p=0.013), with Cluster 3 experiencing earlier mood episode initiation than Cluster 1 (p<0.05). However, the duration of mood disorder illness did not differ significantly among clusters. The proportion of bipolar disorder and depressive disorder differed significantly among clusters (p<0.001).
In Table 3, clinical self-assessment scores were compared among clusters. CGI-S and BDI-II scores showed significant differences among clusters (p<0.001), with Cluster 2 and Cluster 3 scoring higher than Cluster 1 and Cluster 4 (p<0.05). MDQ total score also differed significantly among clusters (p<0.001), with Cluster 3 scoring significantly higher than Cluster 1 and Cluster 2, and Cluster 4 scoring significantly higher than Cluster 2 (p<0.05). There was a significant difference between groups in BSDS-T (p=0.005), BSDS-D (p=0.006), and BSDS-M (p=0.014). BSDS-T and BSDS-D were higher in cluster 3 than Clusters 1 and 2, and BSDS-M was significantly higher in Cluster 3 than Cluster 2 (p<0.05). The presence of depressive symptoms and the possibility of bipolar disorder were compared using the BDI-II score (cutoff point 18) and the BSDS score (cutoff point 13), and there was a significant difference only in Cluster 3 (p=0.014). In other words, the proportion of depressed bipolar disorder patients was high in Cluster 3.
DISCUSSION
This study classified patients with mood disorders into four types based on stress-coping behaviors using cluster analysis, and compared their clinical features according to these clusters. Patients were categorized into four clusters: Balanced, Withdrawn, Impulsive, and Active types. The Balanced and Active types showed proper stress-coping behaviors and stable clinical features. Meanwhile, the Withdrawn Cluster exhibited the fewest stress-coping behaviors, particularly in terms of individual and social activities. Previous research showed a positive correlation between avoidant coping strategy and depression [32]. Depressive symptoms tend to underestimate an individual’s ability to cope with stress and exacerbate feelings of overwhelm in perceived stressful situations [9]. Moreover, as depressive symptoms worsen, the differences in coping behaviors between bipolar and depressive disorders are likely to be less pronounced [33].
The Impulsive Cluster, predominantly comprising bipolar patients in a depressive state, exhibited the high level of pleasure-seeking activities. Bipolar patients engage more in pleasure-seeking activities compared to healthy individuals, possibly due to their inherent impulsivity and tendency to seek novel stimuli [15,34,35]. Interestingly, this impulsivity is not limited to acute manic state but is also observed during euthymic and depressive states [35,36]. It could be an inherent trait in bipolar patients, related to comorbid alcohol or substance use disorders [37,38], or a coping mechanism to overcome depressive symptoms [39]. Even in the general population, stress-related behaviors such as eating, shopping, gambling, smoking, and alcohol consumption correlate positively with depressive symptoms [8], suggesting that pleasure-seeking behaviors may be maladaptive coping strategy. Therefore, interventions for the Impulsive Cluster should focus on reducing pleasure-seeking behaviors and promoting alternative adaptive coping strategies.
The Active Cluster predominantly engaged in both individual and social activities. In this cluster, depressive symptoms were minimal, and disease severity was low. This may reflect the adaptability of using both personal satisfaction and social relationships effectively for stress coping. A large-scale study in the general population showed a negative correlation between leisure activities, sports, and depression [8]. Additionally, in patients with depression, greater social interaction leads to a stronger sense of belonging [40], while positive social experiences enhance personal coping and mood regulation in bipolar disorder [41]. However, the Impulsive Cluster, despite having both individual and social activities, exhibited more severe depressive symptoms than the Balanced or Active Cluster. Excessive social activity can trigger mania in bipolar disorder [42], emphasizing the need for balanced social engagement. Therefore, a harmonious combination of social and individual activities is likely to be effective for coping stress in patients with mood disorder.
The clinical significance of this study lies in the potential for tailored stress management programs based on types of stress-coping behaviors. The Balanced Cluster requires minimal intervention, maintaining their current coping behaviors. In contrast, the Withdrawn Cluster would benefit from behavioral activation techniques to increase adaptive individual and social coping behaviors [43,44]. For the Impulsive Cluster, it seems helpful to reduce pleasure-seeking behaviors [35], avoid excessive social activities, and promote individual activities that are relatively lacking. This aligns with the finding that interpersonal and social rhythm therapy is helpful for patients with bipolar disorder [45,46]. The Active Cluster should manage their activity level appropriately, considering social rhythm therapy [42]. By intervening in stress-coping behaviors, we may improve the prognosis of mood disorders.
There are some limitations in this study. First, patients experiencing manic or hypomanic episodes were excluded, as self-report questionnaires may be unreliable in the presence of such symptoms. However, considering that most clinical cases present in remission or with depressive symptoms, the study’s findings could reflect typical clinical presentations. Second, the clinical assessments were conducted by using self-report questionnaires. Future study will need to be investigated through prospective evaluation. Third, the study sample consisted of mood disorder patients from a single university hospital, limiting the generalizability of the findings. Multi-center studies are needed to address this.
Despite these limitations, this study is the first to classify stress-coping behaviors in patients with mood disorders using cluster analysis. The findings suggest that stress-coping behavior patterns are associated with clinical characteristics. In particular, the Withdrawn and Impulsive Clusters appear to reflect state-dependent aspects influenced by depressive symptoms, while the Impulsive and Active Clusters may be more closely related to bipolar traits. These results indicate that stress-coping behaviors may vary depending on the interplay between current mood state and underlying bipolarity. Moreover, the characteristics of stress-coping behaviors may also be associated with underlying temperament or treatment outcomes. Future prospective studies are warranted to explore the relationship between stress-coping behaviors and temperamental traits, beyond mood symptoms alone. Furthermore, assessing these behaviors can expand diagnostic approaches beyond purely symptomatic criteria, providing a more comprehensive understanding of patients’ conditions. Identifying patterns in individual stress-coping behaviors is pivotal for effective intervention strategies tailored to each patient’s needs.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
