Predictors of Relapse in Patients with Schizophrenia and Schizoaffective Disorders in Real-World Data from a Large Health System

Purpose Schizophrenia is one of the top 15 causes of disability worldwide and among the most expensive mental disorders to treat. Relapse, commonly associated with healthcare utilization including hospitalization, is prevalent and progressive in schizophrenia. Electronic health record (EHR) data could be used for timely identification of those at highest risk for relapse. This study utilized a large sample of Midwestern healthcare patients (pts) diagnosed with schizophrenia and schizoaffective disorders to identify demographic, clinical, and utilization characteristics that predicted relapse. Methods This retrospective study includes EHR data from all pts between Oct 15, 2016, and Dec 31, 2021, who had at least 1 y of encounters. Patients’ first encounter with a schizophrenia or schizoaffective disorder diagnosis (ICD-10 F20 or F25) in this timeframe was defined as their index date, and all encounters up to 3 y post-index date (PID) were explored. Patient-level variables within the first 6 mo of follow-up (FU) were assessed as potential relapse predictors, and first relapse at or after 6 mo of FU within the system was explored as the outcome. Relapse was defined as occurrence of any behavioral health-related emergency room or inpatient encounter after 6 mo of FU within the system. Potential variables assessed include pt characteristics at index date, comorbidities diagnosed, healthcare encounter settings utilized, medication classes prescribed, and disorder-related outcomes experienced. Results The study sample included 8119 pts with 325,745 total FU encounters, with an average of 28.0 mo of FU data PID. Among all pts, 30.5% experienced relapse by this study’s definition. Analyses revealed differences in insurance type, race/ethnicity, and age, and differences across a breadth of comorbid diagnoses, healthcare encounter settings utilized, disorder-related outcomes experienced, and medical classes prescribed. Adjusted analysis revealed pts who relapsed were more likely to be younger (RR=0.99[0.99,0.99]; p<0.0001); identify as Hispanic or Latino (RR=1.15[1.03,1.28]; p=0.0121) or Non-Hispanic (NH) Pacific Islander (RR=1.83[1.16,2.89]; p=0.0090) vs. NH White; have Medicare (RR=1.22[1.08,1.39]; p=0.0018) or Medicaid (RR=1.33[1.17,1.51]; p<0.0001) vs. Private insurance; have diagnoses of substance use (RR=1.33[1.24,1.43]; p<0.0001) and EPS (RR=1.78[1.65,1.92]; p<0.0001); utilize more ER (RR=1.02[1.01,1.02]; p<0.0001) and BH inpatient (RR=1.10[1.06,1.14]; p<0.0001) encounters; experience more prior relapse (RR=1.05;[1.03,1.06]; p<0.0001), and receive more LAI prescriptions (RR=1.01[1.00,1.02]; p=0.0349). Conclusions An algorithm of these variables could conceivably be used to proactively assess relapse risk among pts with schizophrenia to ultimately decide on and implement appropriate relapse prevention plans. Further exploration may be required to better understand underlying modifiable factors that put pts at increased risk of poor outcomes. Funding Sumitomo Pharma America, Inc. and Otsuka Pharmaceuticals Development & Commercialization, Inc.