Identifying Disability with Medicaid Claims Data

Medicaid provides health insurance coverage and services and supports for people with disabilities, as well as groups of children and adults based on financial eligibility. While researchers have an understanding about some of the people with disabilities enrolled in Medicaid, such as those who qualify for Medicaid on the basis of a disability, less is known about those persons who might be disabled but may not have indicators in administrative data that allow for easy identification, such as the eligibility pathway indicators. Such identification is increasingly important, as states and researchers would like to utilize more readily available administrative data sources for budgeting and planning. Researchers have developed algorithms to identify people with disabilities in the Medicare population, but much less work has been done to develop algorithms to identify people with disabilities in the Medicaid population. This project developed a model that incorporates multiple claims-based indicators to identify Medicaid beneficiaries who have a disability using data from the Medicaid Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF). The model had strong predictive performance, but had low model sensitivity, underscoring the difficulty in capturing functional limitations from claims-based indicators.