Developing a Claims-Based Case-Finding Algorithm to Identify Patients with Schizophrenia at Elevated Risk of Relapse

This poster was presented at the 29th Annual U.S. Psychiatric & Mental Health Congress, held October 21-24, 2016, in San Antonio, Texas.

Objective: Management of patients with schizophrenia is complex, and payers have few tools to assess elevated risk of relapse. The objective of this study was to develop a tool to aid in early identification of patients at greatest risk of relapse, using administrative claims data.

Methods: Envolve PeopleCare - Centene public sector claims spanning 7/1/12 - 6/30/14 were used to assess the relationship between schizophrenia patient instability and relapse.  Comorbid psychiatric diagnoses, antipsychotic medication usage patterns, and six recently developed proxies of instability (reported elsewhere) were calculated for all patients and served as predictor variables.  Subsequent hospitalization was defined as a relapse event.  Logistic regression was used to predict relapse based on measured variables.

Results: The final algorithm contained predictor variables which maximized amount of variance accounted for and the positive predictive power (PPP), and included the following: inpatient admission for schizophrenia; psychiatric emergency department visit; fills of psychotropic medication (excluding antipsychotics); Charlson Comorbidity Index; antipsychotic medication switch; diagnoses of depression, bipolar disorder, or other non-organic psychosis. Model fit was acceptable with Nagelkerke R2=0.307, as were classification statistics with a PPP of 64.0% and negative predictive power of 94.7%.

Conclusions:  A predictive model using administrative claims to estimate instability among patients with schizophrenia may help identify individuals most in need of treatment intervention to prevent relapse and reduce the burden on the behavioral healthcare system. This algorithm had acceptable PPP, meaning that 64% of cases identified by the algorithm went on to relapse, indicating that intervention resources will be used efficiently.