Identifying Homeless Medicaid Enrollees Using Enrollment Addresses

Area of Science:

• Public Health
• Health Services Research
• Data Science in Healthcare

Background:

• Identifying homelessness among Medicaid enrollees is crucial for targeted interventions.
• Traditional methods for identifying homeless populations can be resource-intensive and may miss individuals.
• Mailing address data offers a potentially scalable approach to flag individuals at risk.

Purpose of the Study:

• To design and validate an address-based indicator for identifying homelessness among Medicaid enrollees.
• To assess the accuracy of this method against self-reported housing status.

Main Methods:

• Utilized enrollment and claims data from Medicaid expansion enrollees in Hennepin and Ramsey counties.
• Constructed address-based indicators using mailing address information.
• Compared address-based indicators with self-reported housing status from psychosocial needs assessments.

Main Results:

• Of 1,677 enrollees, 427 (25%) self-reported homelessness.
• Among those self-reporting homelessness, 328 (77%) had at least one positive address indicator.
• Sensitivity ranged from 30% to 76%, and specificity from 79% to 97%, depending on the indicator's criteria.

Conclusions:

• An address-based indicator effectively identifies a significant proportion of Medicaid enrollees experiencing homelessness.
• This method presents a valuable tool for researchers, states, and health systems aiming to identify and support homeless populations.
• The findings suggest address data can be leveraged for population health management strategies.