Questions for Martin Love, CEO of North Coast Health Improvement and Information Network (NCHIIN), a not-for-profit health information exchange partnered w/the Humboldt County DHHS-SS and Humboldt Housing and Homeless Coalition working to reduce emergency department utilization by a utilizer population.
Background on North Coast Health Improvement and Information Network (NCHIIN): NCHIIN is a California non-profit providing health information exchange (HIE) in Humboldt, California and also a community health improvement organization. It was chartered by the Humboldt Independent Practice Association and it has a staff of dedicated and experienced Information Technology and management professionals drawn from the sponsoring organization.
While EHR adoption in the community is greater than 90% for local clinicians, significant barriers to connecting individual practices, hospitals, social and healthcare service providers exist. NCHIIN lowers those barriers by helping to build the capacity to provide multiple results interfaces with data from local hospitals, labs and ancillary providers. NCHIIN provides those connections between providers while safeguarding patients’ privacy and confidentiality.
NCHIIN began as many HIE’s did with support from the High Tech Act administered by the Office of the National Coordinator (ONC). Each state received funds to build HIEs acknowledging that individual EHR wouldn’t be interoperable (able to talk to each other), but rather would be little silos of data. HIE’s were developed to connect data & make it useful.
A grant from ONC through AcademyHealth got them to take this current project on. The ONC was funding projects to integrate health data with data from other sources (cross-sector data).
- AISP – Can you clarify “cross-sector” data.
- ML – Here are some links:
Q1. AISP – What data do you link?
ML Response – Right now we only link limited data from health facilities and a few elements from HMIS which contains the individual homelessness data (social sector data). Like many pilots, we are starting small hoping to include more and more data – mission creep is a good thing.
Q2. AISP – What was the motivation for you to start this project?
ML Response – Humboldt County is on the coast of CA near the OR border, and is a fairly isolated area. There are only 135k individuals living in the county. In 2007 we began working with RWJF on many care improvement projects for Humboldt. Our results were variable and towards the end of the initiative, we had a sense that we needed to work on larger scale, coordinating across the community. We decided that only by bringing together health and social care leaders would we be successful and so we started a group with those leaders. This has been very successful, and when this project came along, we were able to go to the county and say “this looks like a good idea,” and they readily agreed.
Q3. AISP – What led you to choose homelessness as a non-health topic?
ML Response – It was fairly strategic. We asked the county what area might benefit from cross-sector exchange. They were particularly interested in trying to link information systems on the two sides (homelessness and health care). We were experienced linking health care information systems so we thought we could do the same thing with county social service systems. It was a bit of a miss, but was a useful learning experience. Their systems are very different from a health care EHR. Second, we didn’t want to be health care data-intensive because we thought it would make it a lot harder. Third, we wanted a fairly constrained population for our intervention. Our experience taught us that you need to start small and try to grow. Finally, we had an intervention that we’d been working on that we thought we could re-use—ED (emergency department) notifications. The idea behind this was that as an HIE, we were getting ADT (admissions, discharge, and transfer) data from all hospitals. These are real time data that include patient demographics, and, generally have a message about something happening in a hospital that hospitals would like to track. As an example of one typical ADT message, “John Doe has been admitted” or, “he registered in ED/left ED” etc.. This information can be sent on to primary care once a day to let them know that someone in their practice had been in the ED the previous 24 hours. We thought we could repurpose this service for the county, and that we could let the county’s case managers know when their clients had been in the ED (with client consent).
Q4. AISP – How did you overcome the consent barrier?
ML Response –
- The local group, the Humboldt Housing and Homeless Coalition (HHHC), included a release of information question in their consent form—the county facilitated this.
- AISP – is that HMIS? Is it an active or passive consent?
- ML – It’s active consent—the organization gets a client signature after an orientation to the intervention.
- We put together patient facing information that describes the project so patients are knowledgeably consenting to the activity. We believe hospitals are allowed to release patient names & discharge/admittance information as directory service. For instance, anyone can call up a hospital and ask, “Is John Doe there?” and be given an answer unless the patient has specifically asked the hospital to not do this.
- HMIS then knows the clients, knows which program they’re in, knows the case managers (cm), and knows the cm contact info. This comes to us once/day, and we use patient name as our list to compare with facility ADT info. Anytime we get an ADT message from a facility, it’s automatically run against the list of individuals in the program.
- We are still working with some hospital facilities on the consent issue.
Q5. AISP – So, you get a list of anyone who’s consented to have their info shared who is in HMIS. Is this done on a daily basis? How many days of service users are you getting?
ML Response – The homelessness data set is specified by HUD, and is really an enrollment in a HUD-reimbursed program.
- It includes everyone in HMIS, which means permanent supportive housing (PSH), shelter housing, etc. as long as the providers receive HUD funding they enter data on clients in HMIS. Anyone who is still enrolled as a client is an open case.
- has that file, and then it’s matched up with ADT system. We get it daily.
ML – Initial project concept:
- Build interfaces between HMIS and HIE. We discovered people don’t use HMIS the way they use Electronic Health Records (EHR). Sending info back to HMIS was essentially useless. People don’t look into HMIS to find out day to day information. We aborted this attempt, and decided instead to support a new case management platform for the county and the county homelessness case managers.
- The are not well integrated, and they contain a lot of legacy and isolated systems. It is hard for the county to know (without looking case by case) who is in what programs. The County had aspirations to develop a system across all programs to support CM. We decided to use the grant money to purchase this CM platform and the intended users are homelessness case managers. Mission creep is happening though. The county looks at their set of clients, and sees that the individuals experiencing homelessness are a small portion of the whole population of county social service clients. This is frustrating county case managers since they can only use the platform for a small percentage of their clients—those experiencing homelessness. Right now they are looking into how to expand the system to include additional programs.
Q6. AISP – Will be tricky. Once you get into behavioral health/child welfare, you have two different silos that have their own requirements. What we’ve seen is that other case managers can see that someone is known to another department and can see the contact info of the other person working on the case, but can’t see details such as diagnosis, etc.
ML Response – This is basically our intention. We are working pretty closely with the vendor of the platform the county is using and, we recently viewed the version 2 platform which will add a lot of levels to permissions. We are becoming optimistic that we will be able to pull this off.
- AISP – When you say permissions—my familiarity is that a client can give consent that their information can be shared, but it’s usually at a fixed time and duration, is limited to specific data elements, and to whom they’re providing consent. The permissions process to track according to these things. Then, CM also has to have authorization to access this information. It tends to be expensive, but maybe not these days?
- ML – The cost has been reasonable so far. This is likely because we had a simple use case.
Q7. AISP – Who is the vendor?
ML Response – ACT.md They’re based in Boston. It has been an excellent relationship.
Q8. AISP – How did you develop the vendor requirements?
ML Response – We worked with the county and the program office provided introductions to other communities with CM platforms.
Q9. AISP – What specific legal council did you pursue for establishing the MOUs?
ML Response – The grant program office has provided legal support on a variety of issues and ideas of how this might work.
Q10. AISP – Are you getting individual client level consent?
ML Response – Yes, via the CMs and a modification of the existing consent.
We are currently working on a process/MOU by which all organizational participants would have the same agreement. Homelessness organizations’ are not necessarily under HIPAA so there is legal (cost) overhead to participate since the standard health care organization to health care organization agreements are a mismatch to this new kind of exchange.
- AISP – if there was some ways to have the CoC give their blessing to this agreement, it could work.
- ML – I’m not sure if this would work. I think the problem is that they don’t want to be part of an HIE (health care) agreement. Another problem—case managers do have access to PHI, but are not being supervised by clinicians. So, we have this mixed-team issue. The hospitals have a social work department in-house, and we are hopeful to work with them on this.
Q11. AISP – What lessons have you learned?
ML Response – Start at the beginning, spend a lot of time trying to understand the needs and methods of both sides of the exchange. Social care and health care have their own norms. Relationships are key; technology is important but only after relationships. The consent and other similar privacy and confidentiality issues are very difficult for small organizations. We need to sort this out nationally.