States working to improve the health of people experiencing homelessness can match their Medicaid data with Homeless Management Information Systems (HMIS) data to track which populations are using housing services and which have the greatest unmet need. HMIS are databases that housing service providers and Continua of Care (CoCs) community and state agencies use to collect and aggregate demographic and service-use information for individuals and families experiencing and at risk of homelessness. Recently, the National Academy for State Health Policy’s Health and Housing Institute interviewed Connecticut Coalition to End Homelessness’ Director of HMIS and Strategic Analysis Brian Roccapriore and Connecticut Department of Housing’s Director of Individual and Family Support Programs Steve DiLella for their insights into the successes and challenges of HMIS-Medicaid data sharing in their state.
What is the history of Connecticut’s HMIS-Medicaid data sharing?
There is a long history in Connecticut of providing permanent supportive housing to homeless populations. Ten years ago, the Corporation for Supportive Housing helped with data matching for the most vulnerable people in the state. The first match between HMIS and state Medicaid data was for a re-entry program for the criminal justice population. They used that data-matching model to identify housing needs and ultimately to decrease costs for high-cost Medicaid users. This cost savings resulted from a substantial reduction in the utilization of health care and shelter systems. The Connecticut Coalition to End Homelessness received a federal Social Innovation Fund grant to house the 160,000 highest-cost Medicaid beneficiaries who use housing services. Once this grant was in place, the state experienced a decrease in high-cost medical services use, such as emergency departments, ambulances, behavioral health care, and hospitals, but also an increase in outpatient and medication usage. The state is currently trying to address some of this increase through a proposed Medicaid 1915(i) plan option.
Connecticut previously had 13 homeless CoCs, which used three different HMIS software systems and six to seven iterations of this software. About four years ago, the CoCs merged their systems into one platform that allows for a single release of information. This process proved challenging for the state, and officials faced a variety of issues. The most important part of addressing these setbacks was getting collective buy-in from providers to move to an open system. There was a great deal of conversation with the state attorney general’s office and the state’s hospital coalition over a year about what the release of information should look like — specifically, how restrictive it should be in terms of what to share and with whom. [The state’s current release of information authorization form can be viewed here.] The current release of information form now allows the state, with a client’s consent, to match data for housing and health opportunities.
Who approves data-matching requests, and who does the actual match?
In Connecticut, HMIS and Medicaid provided their data to a third party, New York University (NYU). NYU completed the actual matching process, and then the matched names went back to the state’s HMIS.
HMIS is guided by a steering committee with representatives from Coordinated Access Networks – referral systems that link people to housing services from all regions of the state. Whenever there is a request to obtain data from the HMIS, the committee must approve it.
Did you have to go through a process of defining appropriate data to be included in the data-use agreements?
There was a discovery period to see what was in each specific data warehouse and what was needed to match it. In Connecticut, each warehouse is unique in how it collects data and the data might not match well. We needed to be sure the data was similar enough to match and that it was collected in a uniform way. We had to look at data dictionaries to find common elements and had to define the requests to specify what specific data was needed, such as enrollment and length-of-stay data for emergency shelters, or transitional shelters, and Medicaid claims data.
How important are agency leadership and buy-in?
It’s important. Connecticut has a broad interagency council to increase supportive housing and we had buy-in from the Department of Social Services. This data sharing became a natural fit with the support of our leadership.
Did you seek input from the US Department of Housing and Urban Development (HUD)?
HUD officials recently said that some HMIS releases are too restrictive in the amount of information they can share.
We received HUD technical assistance on the initial release of information. The trend nationally is toward a more open-sharing model. People often default to “we cannot share information” in fear of violating the Health Insurance Portability and Accountability Act (HIPAA), when the true concern should be ensuring informed consent. In fact, the HIPAA Privacy Rule supports the secure sharing of information for a range of purposes, including improving patient and public health and health care quality.
What roadblocks did you encounter?
Most roadblocks we faced were from individual providers, who are protective of the people they serve. It was also hard to find time in the schedules of officials from the attorney general’s office and the Connecticut hospital association to discuss the work. It is important to start this process as early as possible.
Any additional advice you would like to share with states?
Start early, as this process moves slowly. Washington State has a warehouse of data that could be useful to look at. Allegheny County in Pennsylvania also has an impressive data warehouse. It’s important to ensure that there is someone to drive the process.
Strategic use of data can help states maximize and streamline their efforts to improve the health of state residents by addressing social and living conditions. Connecticut’s data-sharing work is an important example to other states similarly seeking to leverage data to improve health through housing.
This work is supported through NASHP’s Cooperative Agreement with the Health Resources and Services Administration (HRSA), grant #UD3OA22891.