Super Utilization of Healthcare Resources Among Schizophrenia Patients in United States Medicaid System Neeta Tandon Background: It is well documented that a small part of US population (top 5 to 10%) consumes 50% to 70% of the healthcare resources. Multiple emergency and hospitalization visits among schizophrenia patients may point to a poorly managed disease with multiple relapses. There is sparse understanding of the extent and causes of super utilization of emergency and hospitalization among patients with schizophrenia. Innovative population health management strategies targeting the super users with schizophrenia can alleviate the cost and improve the quality of care. This study aims to identify demographics and clinical characteristics associated with super utilizer of healthcare resource utilization among Medicaid schizophrenia patients. Methods: A systematic literature review was conducted to understand the economic cost and resource utilization in managing schizophrenia across all payer types in the US. The narrative systematic literature review was utilized to identify factors associated with resource utilization among Medicaid schizophrenia patients. The literature was also analysed to define “super utilizer” cohort. Using IBM MDCD Medicaid health insurance claims data (January 2001-Decemeber 2019), adult patients with schizophrenia claims, and ≥12 months of continuous insurance coverage before and ≥12 months after the first schizophrenia diagnosis in the database were included in the analysis. Based on literature review and the emergency and hospital visit distribution in the database, top 10% of resource utilizer were defined as “Super Utilizer”. Bivariate analyses were conducted to identify explanatory variables for a predictive analysis. Logistic regressions were used to identify the demographic and clinical characteristics of being in the top 10% of patients with the highest emergency and hospitalization use. As the database had 11% missing race information, models were tested with data with and without missing race. Single and Two-stage models with varied factors were tested to find the model with strongest predictive power. Results: Top 10% of Medicaid schizophrenia patients had at least 9 visits to ER and hospitals in a year. In the study population with missing race (n=253,495) and a single model approach, Charlson Comorbidity Index-Romano (CCI), young age (18-34 years), suicide ideation or attempt were strong predictors of high emergency and hospital resource use, with odds ratios (ORs; 95% confidence interval CI) of 12.57 (10.6-14.9), 3.29 (3.05-3.55) and 3.04 (2.93-3.14), respectively. In the absence of CCI data, having suicide ideation or attempt (SIA), End Stage Renal Disease (ESRD), and being young age (18-34 years) were the top three factors. All included factors were statistically significant, but the strength of odds ratios were small with race and gender. Given significant zero utilizers, a two-stage model approach was also applied. Being young, CCI, and being middle age emerged as top three predictors with SIA as close fourth. The odds ratios were 1.95 (1.76-2.16), 1.87 (1.39-2.51), 1.80 (1.67-1.94) respectively. In this model, alcohol and substance abuse and missing race data were not statistically significant. Without CCI as a factor, ESRD emerged as third strongest in effect size after young and middle age with SIA as close fourth. Limitations: Study was limited to data elements found in a payer sourced claims database. Thus, variables such as homelessness, perception of poor health, social isolation and other social determinants of health may be relevant but were not available. Conclusion: This study suggests that schizophrenia patients with super resource utilization have a high level of comorbidity burden along with being younger in age. Given the model’s high predictability in identifying super utilizer based on a claims data analysis, a practical, cost-effective, schizophrenia disease management is possible to improve quality of life among a significantly burdened patient and their caregiver population.
Neeta Tandon (Thu,) studied this question.