Identifying Homeless Medicaid Enrollees Using Enrollment Addresses

Abstract

Objective: To design and test the validity of a method to identify homelessness among Medicaid enrollees using mailing address data.

Data sources/study setting: Enrollment and claims data on Medicaid expansion enrollees in Hennepin and Ramsey counties who also provided self-reported information on their current housing situation in a psychosocial needs assessment.

Study design: Construction of address-based indicators and comparison with self-report data.

Principal findings: Among 1,677 enrollees, 427 (25 percent) self-reported homelessness, of whom 328 (77 percent) had at least one positive address indicator. Depending on the type of addresses included in the indicator, sensitivity to detect self-reported homelessness ranged from 30 to 76 percent and specificity from 79 to 97 percent.

Conclusions: An address-based indicator can identify a large proportion of Medicaid enrollees who are experiencing homelessness. This approach may be of interest to researchers, states, and health systems attempting to identify homeless populations.

Keywords: Medicaid expansion; determinants of health/population health/socioeconomic causes of health; homelessness.

Figure 1

Receiver Operating Characteristic (ROC) of Homeless Address Indicator* Compared to Self‐Reported Homelessness [Color figure can be viewed at http://wileyonlinelibrary.com]

1. Source: Authors analyzes of Minnesota Department of Human Services enrollment and claims data and self‐reported Life Style Overview data from Hennepin County Medical Center electronic health records.

1. Note: *Indicator of use of either general delivery address (GDA) or non‐GDA homeless address.