Services Effectiveness Research Program (SERP)

Duke University School of Medicine
Department of Psychiatry and Behavioral Sciences

Current and Recent Projects

Title: Antipsychotic treatment effects on reducing violence in schizophrenia

Sponsor: Eli Lilly and Company

Principal Investigator(s): Jeffrey Swanson, Ph.D

SERP Investigators and Collaborators: Marvin Swartz, MD; Richard Van Dorn, Ph.D

Project Description: Violent behavior is an uncommon, but costly and sometimes tragic problem among schizophrenia patients in the community. Violence may result from many factors in addition to psychotic symptoms. Information is lacking about the long-term effectiveness of antipsychotic medication in reducing violence among schizophrenia patients with different profiles of clinical and social-environmental risk factors. Also, better and more efficient methods of assessing violence and managing violence risk in schizophrenia treatment are needed.

The proposed research develops new scientific evidence on violence, schizophrenia, and treatment effectiveness. The project uses secondary analysis of existing de-identified data, the Schizophrenia Care and Assessment Program (SCAP) study, provided by Eli Lilly & Co. The original SCAP data were collected as part of a multi-site study in which all sites had approved IRB protocols. The SCAP is a large multi-site observational study of schizophrenia treatment and outcomes, which included a 4-week point-prevalence assessment of violent behavior and criminal victimization. A total of N=2,400 patients were enrolled from outpatient and inpatient treatment settings and followed for 3 years with clinical assessments every 6 months. The North Carolina site (n=400) supplemented the core SCAP protocol with a comprehensive assessment of violent behavior (MacArthur Community Violence Interview, clinical chart review, and arrest records) and also gathered detailed information on violent victimization for each 6 month-period

This secondary analysis project analyzes longitudinal information on violent behavior in the multi-site SCAP. Specifically, the aims are to:

  1. Assess the 4-week point-prevalence of violent behavior at all available time points for the individual and pooled SCAP sites.
  2. Evaluate the reliability and validity of the SCAP HQ self-report of violence by comparing its measurement performance to a comprehensive multi-source index of violence used in the North Carolina SCAP site.
  3. Develop a staged multivariable logistic regression model of the predictors of violent behavior in several domains: demographic characteristics, clinical characteristics, social-environmental factors, and mental-health services utilization and treatment.
  4. Estimate and compare the effectiveness of first- and second-generation antipsychotic medications on violent behavior in the multi-site SCAP sample, using lagged time-series regression analysis with propensity-score weighting to account for treatment selection effects.
  5. Analyze available SCAP information on criminal victimization experience and its relationship to violence perpetration, other risk factors, and treatment outcomes in schizophrenia.