A national survey of suicidality and non-suicidal self-injury in bipolar disorder: insights from network analysis

Background There is a high prevalence of suicidal ideation (SI), suicide attempts (SA), and non-suicidal self-injury (NSSI) in bipolar disorder (BD). Understanding the nature of suicidality and NSSI in BD is an important way to inform optimal intervention for reducing suicide risk. We aimed to investigate the prevalence and correlates of SI, SA, and NSSI in patients with BD using data from a national survey. We used network analysis to explore the associations among suicidality, NSSI, addictive features of NSSI, and symptoms of BD. Methods Participants with BD were recruited from 20 research centers in China. Suicidality, NSSI, addictive features, and symptoms of BD were measured via a standardized electronic case report form. We used logistic regression and network analysis for data analysis. Results Of the 1,055 participants recruited, over 50% with depressive or mixed episodes had SI, more than 20% engaged in SA, and over 40% reported NSSI. The polarity of mood episodes was independently associated with SI ([hypo]mania vs. mixed episodes: odds ratio [OR] = 0.22, 95% confidence interval [CI] 0.14-0.33, p < 0.001) and NSSI ([hypo]mania vs. mixed episodes: OR = 0.33, 95% CI 0.19-0.58, p < 0.001; depressive episodes vs. mixed episodes: OR = 0.59, 95% CI 0.35-0.98, p = 0.040). Participants with mixed episodes reported the most severe addictive features of NSSI. SI was the most influential symptom within the network model of the total sample; NSSI, SI, and depressive symptoms were identified as key "bridge symptoms". Conclusions Suicidality and NSSI in BD were closely associated with depressive and mixed episodes, with mixed episodes showing the highest prevalence of NSSI and the most pronounced addictive features. SI was identified as the central symptom within the self-harm network, emphasizing its importance as an intervention target. These findings could offer potential insights regarding suicidality and NSSI in BD that can inform intervention strategies.