Establishing a long-term predictive model for aggressive behavior in schizophrenia: a 21-year longitudinal study in rural China.

Background: Although identifying factors contributing to aggressive behavior in individuals with schizophrenia is crucial for developing targeted prevention strategies and intervention, most studies were cross-sectional or short-term, and did not take into account the factor of urbanicity. This study aimed to develop a predictive model of aggressive behavior in individuals with schizophrenia in rural China.

Method: A total of 205 individuals with schizophrenia who were identified in 1994 and followed up in 2015 were included in the study. Aggressive behavior was assessed using the Modified Overt Aggression Scale (MOAS). The final predictive model was developed by backward stepwise regression. The model's predictive performance was evaluated using the C statistic and calibration curve.

Result: The rate of aggressive behavior in individuals with schizophrenia in rural China was 36.1% during 1994-2015. The final model of aggressive behavior incorporated the following factors: male, lower educational level, unmarried, with delusion, worse social functioning, and with previous treatment. The model demonstrated acceptable discriminative ability, with an AUC of 0.73, sensitivity of 0.82, and specificity of 0.53. The calibration curve indicated a good fit of the model.

Conclusion: The predictive model developed in this study showed good discriminative ability. A clinically practical nomogram was built to assess the risk of aggressive behavior in individuals with schizophrenia in rural China, which may facilitate early detection and intervention of these individuals, particularly in rural areas with limited resources. This approach may be relevant to similar settings internationally.