• National Human Services Data Consortium
• National Alliance to End Homelessness
• National Health Care for the Homeless Council

• OPEN MINDS Technology & Analytics Institute
• National Coalition Against Domestic Violence

Traditional Data Integration

Communication Between Providers

Data Quality Analysis

HMISes and CoCs have coordinated entry, a best practice for responding to people experiencing homelessness. However, for coordinated entry to happen, there must be an intake.

Intake can’t be defined easily: there are about 14 types of housing and HUD projects in the United States, and the differences between them are often nuanced and plentiful. For example, there are two types of emergency shelter as defined by HUD, transitional housing, different types of permanent housing, grant per diem (GPD), HOPWA, and Ryan White.

Partly because of the many types of housing support available, HUD (and other regulatory entities) require a high level of data during intake. Many social workers will feel this level of data intrusive, perhaps even traumatizing to clients or vicariously to themselves.

To quote a recent article we wrote:

• After all, questions like, “Have you ever been raped?” or, “Have you ever committed sex acts in exchange for money or food?” can have an emotional cost to both the client and the worker performing the intake.

  Psychologist Stephen Fleming, a world-renowned trauma researcher, writes, “One need not directly experience or observe an event for it to be potentially traumatic; events experienced by others that one learns about also have this potential.”

  The operative word there is “potential.” In a recent interview, Fleming states he prefers to use the term “potentially traumatic event” (PTE) because it doesn’t assume trauma or PTSD symptoms, as many people can experience PTEs without any PTSD symptoms.

  On that same vein, potential harm can be caused when the intake worker assumes the new client must have been traumatized, which can prime that individual with a false narrative of how they “should” have reacted to the PTE.

  That said, reliving old PTEs can also cause temporary emotional distress. So that’s where the debate lives right now: do we cause more potential harm by not collecting enough data or by collecting too much?

  The fact that the United States has a threshold on spending toward vulnerable populations isn’t likely to change. Therefore, we still need a lot of that data.

Best Practice: Learn and Demonstrate Trauma Sensitivities

The client expectations might be different, depending on the nature of the service program. Shelter intake will be different from coordinated entry intake, so the intake questions will be varied. Case workers performing intake will necessarily ask a lot of intrusive questions, and some potential clients will find this process painful or triggering.

The Substance Abuse and Mental Health Services Administration (SAMHSA) calls the trauma-informed approach the Four R’s:

1. Realize the impact of trauma and the potential paths to recovery;
2. Recognize the signs and symptoms of trauma in clients;
3. Respond with policies that integrate knowledge about trauma; and,
4. Resist retraumatization when asking intake questions.

Service Provision

Fair distribution of services is a challenge for most CoCs and HMISes because what’s fair is a debate of ethics. Really advanced communities are pulling data like demographics, income, race, and the like to determine whether there’s a potential bias, but many service providers tend to struggle with this question of fairness.

That’s why system mapping is a useful exercise to follow where the services are being used most, but it’s extremely labor intensive and difficult to do manually. That’s one reason community-intelligence platforms like ClientInsight can provide greater efficiency and accuracy.

In short, we need an analytics revolution in social services.

Best Practice: Closed-Loop Referral

The initiator of the service request should be able to identify the provider and notify them before the client arrives or contacts the provider, passing along any documentation that was part of the referral. The agency then receives the referral, and the “loop” is “closed” when the client (ideally) shows up or there is a result indicated. Even “failed” or “ineligible” results can be a long-term silver lining, because at least that data helps improve the system.

ClientTrack facilitates this process with an eligibility engine and closed-loop referral workflow.

Exit and After-care

The line between traditional homeless management information systems (i.e., data points about support for homeless populations) and modern HMISes is becoming blurrier. The most forward-thinking HMISes now track after-care and have exit strategies in place for clients beyond providing permanent housing. That said, a successful exit to permanent housing is what is measured by HUD.

How do most clients leave an HMIS?

• Sometimes, clients suddenly disappear and lose contact with service providers. Others just leave for a period of time and come back.
• Sometimes, it’s a transitional housing project and the client fails out of it.
• Sometimes, the individual is sent to a substance treatment facility.
• Sometimes, there’s criminal activity that can’t be addressed through a harm-reduction model because of client behavior.
• Sometimes, clients pass away.

What does stability look like for HMISes and their clients?

Placing a client in permanent housing is not always successful. Some case managers will look for opportunities for clients to be reunified with their families, which could be the best outcome in many cases, especially where medical noncompliance for a severe psychiatric illness is an issue.

For HUD, the desired outcome is housing. However, case workers often need to maintain a connection with clients to mitigate recidivism. One helpful practice to supplement the case worker–client relationship is to place housed, former clients as peer mentors for new clients entering the CoC’s system.

After-care matters because HUD also wants to see a reduction in recidivism. The follow-up relationship helps ensure stability. In a data warehouse environment, you bring in other data sources from public services not in the CoC (i.e., unemployment offices, Social Security, housing assistance, SNAP) and share that data across the CoC for the benefit of case managers. However, it’ll probably be a more gradual process to get data-sharing agreements with those public services that aren’t specifically geared toward homeless populations.

Best Practice: Automate Processes Where Possible

Automation is not the same thing as “not caring” or “disinterest.” The reality is there is not enough time, capacity, or workforce to handle the amounts of cases in many communities. So, in many instances, it might come down to either choosing an automated, follow-up robo-call and, well, nothing. In that case, it’s better to automate the robo-call, albeit less intimate.

The principle is this: if left to short-term memory, many follow-ups can easily be forgotten. Human frailty in the CoC must be mitigated where possible.