Article information

STRESS. 2025;33(3):163-175

Publication date (electronic) : 2025 September 30

doi : https://doi.org/10.17547/kjsr.2025.33.3.163

Woongsur Hur¹, Megan L. Rogers², Eun-Jung Shim^,3

¹Master’s degree graduate, Department of Psychology, Pusan National University, Busan, Korea

²Assistant Professor, Department of Psychology, Texas State University, Texas, United States of America

³Professor, Department of Psychology, Pusan National University, Busan, Korea

허웅서¹, 매건로저스², 심은정^,3

¹부산대학교 심리학과 석사 졸업

²텍사스 주립대학교 심리학과 조교수

³부산대학교 심리학과 교수

Corresponding author Eun-Jung Shim Department of Psychology, Pusan National University, 2 Busandaehak-ro 63beon-gil, Geumjeong-gu, Busan 46241, Korea Tel: +82-51-510-2159 Fax: +82-51-581-1457 E-mail: angelasej@pusan.ac.kr

Received 2025 August 26; Revised 2025 September 15; Accepted 2025 September 23.

Abstract

Trans Abstract

Introduction

As of 2023, South Korea recorded an age-standardized suicide rate of 24.8 per 100,000 individuals, the highest among member countries of the Organization for Economic Cooperation and Development (OECD average: 10.7) [1]. Although suicidal ideation is relatively common, not all individuals with suicidal thoughts progress to attempt suicide. A systematic review estimated that between 2.6% to 37% of individuals with suicidal ideation attempt suicide [2]; however, the majority do not act on these thoughts [3]. Therefore, understanding the factors that influence the transition from suicidal ideation to suicide attempt is critical for effective suicide prevention.

A meta-analysis of 23 studies investigating the relationship between rumination and suicidal behavior observed that rumination was significantly associated with both suicidal ideation (Hedge’s g=0.74) and suicide attempts (Hedge’s g=0.26) [4]. Specifically, rumination about the causes, consequences, and symptoms of negative affect has been linked to increased severity of suicidal ideation [5] and elevated risk of suicide attempts [6]. However, prior studies have predominantly focused on general forms of affective rumination rather than thoughts specifically centered on suicide. Given that rumination is characterized by persistent, repetitive, and uncontrollable thoughts, suicide-specific rumination may play a crucial role in this transition.

Recent findings suggest that the specific characteristics of suicidal ideation, including its frequency, severity, persistence, and uncontrollability, play a crucial role in the transition from ideation to attempt [7-10]. For instance, a study involving 506 adolescents aged 12∼21 years found that individuals with more frequent, severe, and persistent suicidal ideation were significantly more likely to attempt suicide 4∼6 years later, even after adjusting for psychiatric diagnoses and prior suicide attempts [7]. Similarly, a large-scale study of 3,916 individuals with a history of suicidal ideation reported that mental disorders such as major depressive episodes increased the risk of suicide attempts by 1.3∼2.1 times. Furthermore, longer durations of suicidal ideation and difficulty controlling suicidal thoughts were associated with a 2.4∼3.6 times higher risk of suicide attempts [8]. More recently, the perceived controllability of suicidal thinking has been linked to elevated short-term suicide risk [10]. Therefore, rumination, which specifically focuses on suicidal ideation, may serve as a key predictor of suicide attempt. Despite its theoretical and empirical relevance, most studies have relied on general rumination measures, such as the Response Styles Questionnaire [11], which assess rumination on negative emotions rather than suicidal thoughts (e.g., “think ‘Why do I always react this way?’” and “analyze your personality to try to understand why you are depressed”) [12]. These measures fail to capture suicide-specific rumination, underscoring the need for a reliable and valid tool specifically designed to assess this construct.

To address this gap, Rogers et al. [13] developed the Suicide Rumination Scale [SRS], an eight-item measure designed to assess the presence, frequency, and intensity of rumination concerning suicidal ideation, intentions, and planning. The SRS has been validated in individuals with a history of suicidal ideation, and the Farsi version has also been validated in Iran [14]. However, despite South Korea’s high suicide rate and the central role of suicide-specific rumination in suicidal ideation and behavior, no validated Korean-language measure of this construct is available, limiting both the research and clinical assessment of Korean-speaking populations.

This study aimed to validate the Korean version of the Suicide Rumination Scale [K-SRS] in a Korean-speaking population. The factor structure, reliability, validity, and screening performance of the scale were examined. We hypothesized that K-SRS scores would be positively correlated maladaptive forms of rumination (e.g., brooding and depressive rumination) but not with deliberate rumination, supporting convergent and discriminant validity. Additionally, we expected positive associations with suicide risk and related cognitive-affective factors (e.g., perceived burdensomeness, thwarted belongingness, attentional control, and attentional bias), supporting concurrent validity. We also examined incremental validity beyond established risk factors, known- group validity based on levels of depression and anxiety, and the ability of the scale to distinguish between individuals with and without a history of suicide attempts.

Materials and Methods

1. Participants and procedures

The participants were recruited from July 2024 to March 2025 via online and offline bulletin board posts at several universities and various websites. The eligibility criteria included: adults aged 18 years or older, computer literacy, and a lifetime history of suicidal ideation. A total of 124 participants (27 males and 97 females) participated in the study, with a mean age of 25.2 years (SD=6.7). Most had completed high school (n=69; 55.6%), were either currently attending school or unemployed (n=91; 73.4%), and were single (n=113; 91.1%; Table 1).

Table 1.

Descriptive statistics of participant characteristics and study variables by suicide attempt history

All the study procedures were conducted online. Upon providing informed consent, the participants received a link to the survey and behavioral task. The survey was administered using Google Forms, whereas the behavioral task was conducted using Pavlovia. This study was approved by the Institutional Review Board [IRB] of Pusan National University (PNU IRB/2024_54_HR). All the participants provided written informed consent.

3. Statistical analyses

Descriptive statistics were employed to examine the participants’ demographic characteristics and the item properties of the K-SRS. Confirmatory factor analysis [CFA] evaluated the factor structure of the K-SRS, with model fit evaluated using χ² test (p＞.05), Comparative Fit Index (CFI ≥ .95), Root Mean Square Error of Approximation (RMSEA ≤.08), and Standardized Root Mean Square Residual (SRMR ≤.08) [28]. Owing to the poor fit of the single-factor model (Table 2), bifactor exploratory structural equation modeling [bifactor ESEM] was conducted based on Rogers et al. [13]. Unlike CFA, ESEM allows all items to load on multiple factors [29], providing a more realistic representation of factor structure by integrating the exploratory flexibility of EFA with the confirmatory rigor of CFA within a structural equation modeling framework. Unidimensionality was evaluated using bifactor ESEM indices, including explained common variance [ECV], percent of uncontaminated correlations [PUC], and average relative parameter bias [ARPB]. Thresholds of ≥.70 for ECV/PUC [30] and ＜10∼15% for ARPB [31] were employed to support unidimensionality. Given the evidence of multidimensionality and high modification indices between Items 1∼4 and Items 5∼8 (Table 3, 4), a two-factor ESEM was conducted, with model fit evaluated using the same fit indices as the CFA.

Table 2.

Fit indices of each model of the K-SRS

Table 3.

Standardized factor loadings for each model of the K-SRS

Table 4.

Residuals correlations (above the diagonal) and modification indices (below the diagonal) for the CFA model

The validity analyses also included convergent validity, assessed via correlations between the K-SRS and other rumination measures (e.g., the K-RRS and K-ERRI), and discriminant validity, tested via correlations with the deliberate rumination factor of the K-ERRI. Concurrent validity was evaluated via correlations with suicide risk, suicidal ideation intensity, perceived burdensomeness, thwarted belongingness, depression, anxiety, attentional control, and attentional bias. Incremental validity was tested employing hierarchical regression analysis to determine whether the K-SRS explained suicide risk beyond depression, anxiety, perceived burdensomeness, and thwarted belongingness. Known-group validity was examined using a one-way analysis of variance [ANOVA] to compare the K-SRS scores across depression and anxiety severity levels. Internal consistency was assessed via Cronbach’s α and McDonald’s ω. A receiver operating characteristic [ROC] analysis was employed to determine the optimal cutoff score for identifying individuals with a history of suicide attempts. The Youden index (sensitivity+specificity−1) was used to determine the cutoff that maximized classification accuracy, and the cutoff yielding the maximum value of this statistic was identified as the optimal threshold [32]. The area under the ROC curve [AUC] was interpreted as acceptable (0.7∼0.8), excellent (0.8∼0.9), or outstanding (≥ 0.9) [33].

All the statistical analyses were conducted using Jamovi 2.3.28. The factor analyses were conducted using Mplus 8.11, and ROC analyses were conducted using the pROC package in R.

Results

2. Descriptive statistics, factor structure, and reliability of the K-SRS

All the eight items of the K-SRS satisfied the criteria for normality, with skewness values ranging from 0.02 to 0.86 (below the cutoff of 3) and kurtosis values ranging from −1.24 to −0.26 (below the cutoff of 10). The items also demonstrated strong item-total correlations (Table 5).

Table 5.

Exploratory structural equation modeling results and descriptive statistics for individual items of the K-SRS (N=124)

5. Validity of the K-SRS

1) Convergent and discriminant validity

The K-SRS total and subscale scores were significantly correlated with brooding (rs=.27∼.38), reflective pondering (rs=.19∼.35), depressive rumination on the K-RRS (rs=.48∼.57), and intrusive rumination on the K-ERRI (rs=.37∼.50), supporting convergent validity (Table 6). By contrast, deliberate rumination on the K-ERRI was not significantly correlated with the K-SRS scores (rs=−.01 to .01), supporting discriminant validity.

Table 6.

Correlations between suicide-specific rumination and other study variables

2) Concurrent validity

The K-SRS total and subscale scores were significantly correlated with suicide risk (rs=.57∼.62), lifetime suicidal ideation intensity (rs=.53∼.60), perceived burdensomeness (rs=.35∼.55), thwarted belongingness (rs=.35∼.48), depression (rs=.45∼.65), and anxiety (rs=.34∼.58), supporting concurrent validity.

The total score and the intrusive rumination about suicide factor were negatively correlated with attentional control (rs=−.39 to −.30), whereas the contemplation of suicide methods factor was not (r=−.17).

No significant associations were found between the K-SRS scores and attentional bias indices (orientation: rs=−.16 to −.08, ps＞.05; disengagement: rs=.11 to .18, ps＞.05). However, the intrusive rumination about suicide factor indicated a trend-level positive correlation with disengagement (r=.18, p=.051). This finding suggests that individuals with higher levels of intrusive rumination about suicide may disengage more slowly from suicide-related stimuli. The item-level analysis showed that Item 1 (“Cannot ‘turn off’ these thoughts”) and Item 3 (“Have trouble getting the suicidal thoughts out of my mind”) were significantly correlated with disengagement scores (r=.19, p=.037).

3) Incremental validity and known-group validity

The hierarchical regression analyses demonstrated that the K-SRS explained an additional 15.7% of the variance in suicide risk (ΔR²=.151, F(1, 113)=31.43, p＜.001), beyond established risk factors, including depression, anxiety, perceived burdensomeness, thwarted belongingness, and other rumination measures, supporting its incremental validity (Table 7).

Table 7.

Hierarchical regression analysis results for suicide risk: the incremental validity of the K-SRS (N=124)

Known-group validity was supported by significantly higher K-SRS total and subscale scores among individuals with clinical levels of depression and anxiety symptoms compared with those with minimal symptoms (Depression: F(3, 120)=19.19, p＜.001, η²=.32; Anxiety: F(2, 56.32)=14.09, p＜.001, η²=.24; Table 8).

Table 8.

Known-group validity of the K-SRS: total and subscale scores by depression and anxiety severity levels (N=124)

Discussion

This study evaluated the psychometric properties of the K-SRS.

CFA indicated a poor fit for the single-factor structure, whereas the subsequent bifactor ESEM suggested multidimensionality. By contrast, the two-factor ESEM demonstrated a good model fit for the K-SRS. The first factor, intrusive rumination about suicide, included items capturing the difficulty in disengaging from suicidal thoughts, feelings of entrapment, and the inability to suppress such thoughts. The second factor, contemplation of suicide methods, comprised items reflecting thoughts about suicide planning, such as contemplating accessible or easy ways to die. This structure partially aligns with the bifactor model proposed during the initial development of the SRS, which included one general factor and two specific factors [13]. Similarly, in the validation study of the Farsi version of the SRS [14], the single-factor model demonstrated a poor fit in CFA, leading to the inclusion of correlated residuals. In that study, correlations between the residuals of items 5∼8 ranged from .37 to .47, suggesting that a single latent factor may not adequately explain these items and indicating the potential presence of a second factor. These findings underscore the need for further research to clarify the factor structure of the K-SRS across cultural contexts.

The K-SRS demonstrated strong convergent and discriminant validity. Significant positive correlations with brooding, reflective pondering, depressive rumination, and intrusive rumination (as measured by the K-RRS and K-ERRI) supported convergent validity, whereas a lack of significant association with deliberate rumination indicated discriminant validity. These findings suggest that suicide-specific rumination is linked to maladaptive, uncontrollable thought patterns, rather than intentional or neutral forms of rumination. This aligns with meta-analytic evidence indicating that brooding is strongly associated with suicidal ideation and suicide attempts [4]. Among the rumination subtypes, reflective pondering was more strongly correlated with the K-SRS total score and contemplation of suicide methods factor than with intrusive rumination about suicide. As reflective pondering is typically deliberate and problem-solving-oriented, with lower emotional involvement [15], contemplation of suicide methods factor likely reflects a more intentional form of rumination. The non-significant association between K-SRS and deliberate rumination further underscores the role of uncontrollability in suicide-specific rumination, suggesting that its involuntary nature can be particularly detrimental.

Furthermore, the K-SRS demonstrated strong concurrent validity, with higher levels of suicide-specific rumination associated with greater suicide risk, a higher intensity of suicidal ideation, and elevated levels of key suicide risk factors, including perceived burdensomeness, thwarted belongingness, depression, and anxiety. In addition, a negative association between intrusive rumination about suicide and attentional control was observed, consistent with previous findings linking rumination to poorer self-reported attentional control [34]. This suggests that individuals struggling with persistent suicidal thoughts may have difficulty in regulating their attention and reinforcing their perseverative thinking patterns. These findings highlight the potential benefits of interventions targeting attentional control, such as attention-bias modification training and mindfulness-based approaches, in reducing suicidal rumination.

Although the K-SRS scores were not significantly associated with orientation indices of attentional bias, a marginal association was observed between the intrusive rumination about suicide factor and disengagement scores, suggesting that individuals with more intrusive suicidal thoughts may struggle to shift their attention from suicide-related cues. Supporting this, Items 1 and 3, which reflect uncontrollable and persistent suicidal thoughts, were significantly associated with disengagement scores. These findings partially align with cognitive models of suicidal behavior, which emphasize the role of attentional biases in the development and maintenance of suicidal thinking [35]. Suicide-specific rumination has also been conceptualized as a proxy for suicide-specific attentional biases/fixations in intensive longitudinal investigations of the cognitive model of suicidal behavior [36]. However, the overall weak associations observed in this sample may reflect limitations inherent in probe-based tasks, which provide only static, momentary snapshots of attention [37], and may not adequately capture the dynamic and context-dependent nature of attentional processing in suicidal individuals [38]. Future research must explore alternative methodologies, such as mood-dependent administration or dynamic measures such as eye-tracking, to better assess suicide-related attentional bias [26].

The incremental validity of the K-SRS was supported by its ability to predict suicide risk beyond established risk factors including depression and anxiety, perceived burdensomeness, and thwarted belongingness [39]. Notably, difficulty in controlling suicidal thoughts has been shown to be a stronger predictor of future suicide attempts than depression or posttraumatic stress disorder in 3,916 individuals with suicidal ideation [8]. Collectively, these findings suggest that suicide-specific rumination is an independent and robust predictor of suicidal ideation and attempts.

The results also support the known-group validity of the K-SRS, with significantly higher total and subscale scores observed among individuals with greater severity of depression and anxiety. The strongest group differences emerged for the Intrusive Rumination subscale, indicating its sensitivity to clinical levels of depressive and anxiety symptoms, as intrusive rumination is a core feature of both conditions [40]. ROC analysis further assessed the discriminatory power of the K-SRS for identifying individuals with a history of suicide attempts. The optimal cutoff score was 13.5, with a sensitivity of 0.68 and specificity of 0.75 (AUC=.77), indicating an acceptable level of discrimination [33].

Despite these promising findings, several limitations must be noted. First, although the sample size (N=124) adhered to the minimum guidelines for a factor analysis of an 8-item scale [41], it was relatively small compared with typical validation studies. Cross-validation, which is typically recommended to evaluate the generalizability of factor structures, was not feasible owing to the challenges of recruiting individuals with suicidal ideation. To address this limitation, ESEM and bifactor-ESEM were employed to evaluate the robustness of the factor structure beyond CFA. However, these approaches cannot replace formal cross-validation. Future studies with larger sample sizes are needed to replicate the two-factor structure and further evaluate the reliability and validity of the K-SRS. Second, the cross-sectional design precluded the assessment of test-retest reliability or sensitivity to change, limiting the causal interpretation of the relationships between the study variables. Future longitudinal research is needed to evaluate whether suicide-specific rumination prospectively predicts suicide attempts among individuals with suicidal ideation. Third, the sample was predominantly female and comprised young adults, raising concerns about the generalizability of the scale to other demographic groups. Although measurement invariance across genders has been confirmed for the SRS [13,14], further validation with a larger and more diverse sample is required to establish the factor structure and overall validity of the measure across populations. Fourth, in the dot probe task, suicide-related words were rated as significantly more arousing than neutral words. This heightened arousal may have influenced the response patterns, potentially confounding the assessment of attentional bias. Future studies must carefully select neutral words with comparable arousal levels to ensure a more accurate assessment of attentional biases toward suicide-related stimuli.

In conclusion, the K-SRS is a reliable and valid tool for assessing rumination related to suicidal thoughts, intentions, and plans. Its robust psychometric properties support its utility in identifying individuals with a heightened suicide risk. However, further validation through longitudinal studies and more diverse samples is needed to strengthen its generalizability and applicability.

Article information Continued