The study, led by Hua Fan of Capital Medical University Affiliated Beijing Anding Hospital together with colleagues from Capital Medical University’s School of Public Health, is notable for its methodological approach. Rather than treating aggression as a single binary outcome that either occurs or does not, the researchers applied the Andersen-Gill model, a recurrence data framework originally developed for analyzing repeated events in medical survival analysis. This choice matters because aggressive behavior among psychiatric patients is rarely a one-time phenomenon; it tends to recur, and each episode carries its own clinical and social consequences. By modeling the rate of repeated events rather than just the probability of a first occurrence, the study captures a more realistic picture of how social support shapes behavior over time.

Participants were recruited consecutively between December 2019 and September 2022, all aged between 18 and 64 years, all carrying an ICD-10 diagnosis of a mood disorder, and all with no prior systematic treatment for their condition. Of the 388 patients enrolled, 384 completed the full 24-month follow-up period. Social support was measured at baseline using the Social Support Rating Scale, a validated instrument that quantifies the breadth and quality of a person’s support network, and scores were divided into quartiles ranging from the lowest (Q1) to the highest (Q4). Aggressive behavior was then assessed every three months by telephone using the Modified Overt Aggression Scale, an instrument that rates aggression across verbal, physical, and property-related dimensions.

To ensure that the observed associations were not confounded by other variables, the researchers used Directed Acyclic Graphs, a formal graphical tool for identifying which covariates must be adjusted for in the statistical model. They also checked for multicollinearity using variance inflation factors and explored the shape of the dose-response relationship with restricted cubic spline functions, which allow the data to reveal whether the effect of social support is linear or follows a more complex curve.

The headline finding is striking. Compared with patients in the lowest quartile of social support, those in the third quartile showed a 50 percent reduction in the risk of aggressive behavior, with a hazard ratio of 0.50 and a 95 percent confidence interval of 0.25 to 0.97. Those in the highest quartile fared even better, with a hazard ratio of 0.40 and a confidence interval of 0.17 to 0.94, meaning they had roughly 60 percent lower risk. Both results were statistically significant, and the overall pattern suggested a dose-response relationship: the more social support a patient reported, the lower their risk of recurrent aggression over the two-year follow-up window.

But the story becomes more nuanced when the researchers stratified their analysis by disease type, gender, and age. The protective effect was most pronounced among patients diagnosed with depressive disorders, among male patients, and among those aged 30 years or older. For these groups, stronger social networks appeared to act as a genuine buffer against the recurrence of aggressive episodes, consistent with the idea that supportive relationships can reduce stress, improve treatment adherence, and provide early warning signs when a patient’s condition begins to deteriorate.

The most unexpected result, however, emerged among female patients. Contrary to the overall trend, women in the third quartile of social support actually showed an increased risk of aggressive behavior compared with those in the lowest quartile, with a hazard ratio of 2.96 and a confidence interval of 1.05 to 8.37. This finding raises important questions about the nature of social relationships and how they may interact differently with psychiatric illness across genders. The authors suggest that for some female patients, social networks may involve interpersonal dynamics that generate conflict or stress rather than buffering it, though the study was not designed to disentangle these mechanisms definitively.

The study’s single-center design at Anding Psychiatric Hospital in Beijing means the findings should be generalized with caution. The patient population, the cultural context of social support in China, and the specific treatment pathways available at a major urban psychiatric hospital may all differ from those in other settings. Nevertheless, the prospective cohort design, the relatively large sample size, the use of a validated aggression scale at regular intervals, and the application of a recurrence model all strengthen the internal validity of the results. The research was funded by the National Natural Science Foundation of China under grant number 81872688, and the study protocol received ethical approval from the Institutional Review Board of Capital Medical University, with all patients providing informed consent.

From a clinical standpoint, the implications are significant. If social support genuinely reduces the risk of recurrent aggression, then interventions aimed at strengthening patients’ support networks, whether through family therapy, peer support programs, community outreach, or structured aftercare planning, could become a meaningful component of discharge planning for patients with mood disorders. The finding that the effect is strongest among patients with depressive disorders and among men over 30 suggests that these groups might benefit most from targeted social support interventions, while clinicians working with female patients may need to assess the quality and dynamics of their social relationships more carefully rather than simply encouraging broader networks.

The study also highlights the value of recurrence-based modeling in psychiatric research. Aggressive behavior, like relapse in substance use or rehospitalization in severe mental illness, is a repeating process, and statistical frameworks that account for this repetition can detect associations that simpler models might miss. As the field moves toward more sophisticated longitudinal designs, the Andersen-Gill approach may find wider application in studying not just aggression but a range of recurrent psychiatric outcomes. For now, the Beijing team’s work adds a compelling piece of evidence to a growing body of research suggesting that the social environment is not merely a backdrop to mental illness but an active ingredient in its trajectory, one that clinicians ignore at their patients’ peril.

Subject of Research: The association between social support and recurrent post-discharge aggressive behavior in patients with mood disorders

Article Title: Social support and post-discharge aggressive behavior among patients with mood disorders: a single-center analysis of recurrent aggressive events

Article References: Fan, H., Yu, X., Cheng, P., Yang, Z., Wang, Q., Xu, J., Zhu, H., & Gao, Q. (2026). Social support and post-discharge aggressive behavior among patients with mood disorders: a single-center analysis of recurrent aggressive events. BMC Psychiatry. https://doi.org/10.1186/s12888-026-07984-4

Image Credits: AI Generated

DOI: 10.1186/s12888-026-07984-4

Keywords: mood disorders, social support, aggressive behavior, Andersen-Gill model, recurrent events, psychiatry, depressive disorder, Modified Overt Aggression Scale, post-discharge outcomes, prospective cohort study, Beijing Anding Hospital, hazard ratio