Background: This study aimed to empirically derive and validate subtypes of nonsuicidal selfinjury (NSSI) based on the forms of self-harm behavior. Methods: Sixty young adults who exhibited repetitive NSSI completed a clinical interview and baseline measures, followed by a two-week ecological momentary assessment (EMA).A latent class analysis was administered to identify NSSI subtypes based on the forms of self-harm behavior. The subtypes were then compared to baseline clinical and EMA variables, including emotions and urges to self-harm. Results: The best-fitting model supported two subtypes: 1) substance abuse and suicide attempt subtype and 2) cutting and scratching subtype. The substance abuse and suicide attempt subtype reported a higher rate of engaging in lifetime suicide plans and suicide attempts and a higher number of lifetime suicide attempts. They also showed significantly higher severity of self-harm behavior, borderline personality traits, anger, posttraumatic symptoms, and difficulties in emotion regulation when compared to the cutting and scratching subtype. However, there was no inter-subtype difference in self-esteem. In the EMA data, anger toward others, feeling of rejection, loneliness, and helplessness were significantly high in the substance abuse and suicide attempt subtype. The variability of anger toward others was also higher in this subtype compared to the cutting and scratching subtype. Finally, in the two-week research period, the substance abuse and suicide attempt group showed a higher rate of reported urges to self-harm with or without suicidal intent. Conclusion Findings support the validity of the subtypes based on the forms of self-harm behavior, suggesting the clinical utility of addressing heterogeneity within individuals exhibiting self-harm for assessment and treatment.

Purpose: The prevalence rate of non-suicidal self-injury (NSSI) in the clinical population is higher than that in the community sample, necessitating the need to investigate the predicting factors of NSSI in this group. The present study aimed to develop a prediction model of NSSI among psychiatric patients in Korea. Materials and Methods: Decision tree analysis was conducted on a sample of 224 psychiatric patients. Emotion regulation strategies (rumination, cognitive reappraisal, and expressive suppression), impulsivity, problematic alcohol use, working memory, depressive mood, and gender were included in the model as predictors of NSSI. Results: Results indicated that rumination, problematic alcohol use, and working memory predicted lifetime NSSI engagement among psychiatric patients. The best predictor of lifetime NSSI engagement was rumination. Specifically, when the level of rumination was high, the level of working memory was lower, and the risk of NSSI was higher. In the case of low levels of rumination, the higher the level of problematic alcohol use, the higher the risk of NSSI. The highest prevalence of lifetime NSSI engagement was found in a subgroup of patients with high levels of rumination and low levels of working memory. Conclusion The major contribution of this study is finding a combination of factors to predict the high-risk group of NSSI among psychiatric patients in Korea. This study provides evidence on the effect of rumination, working memory, and problematic alcohol use on NSSI. It is suggested that clinicians and researchers should pay more attention to emotion regulation and related vulnerabilities in preventing and treating NSSI.