Time to tackle IT and culture barriers to social data integration, says Jacob Reider
During a discussion at HIMSS19 this past month about the vast promise but big challenges of tackling patients' social determinants of health, a group of diverse professionals detailed the steps they're taking to ensure better health and living conditions for the populations they serve.
- An employee of one medical group described visiting people's homes to observe variability, with clinicians prompted to gather certain information to create standardized assessments of different ways of life.
- One expert pointed out that data feeds from homeless management information systems and criminal justice systems could be better put to use for population health management.
- The need for states, with help from federal funding, to potentially act as conveners helping ensure access to housing groups, food banks and other essential social services was discussed.
- The central importance of food security, birth plans, neonatal nutrition, transportation and many other social factors was a common theme. An attendee made the case that patients should be screened for loneliness the way they would be for any other medical condition.
- One healthcare professional likened the process of screening for social factors to "deep learning" – not in the artificial intelligence meaning of the word, but the sense of gleaning insights from patients' day-to-day lives when they're not presenting at the clinic or the ED.
The focus group was convened by Dr. Jacob Reider, former chief medical officer and deputy national coordinator at ONC – and now CEO of Troy, New York-based Alliance for Better Health, which partners with hospitals, physician practices, LTPAC and home care programs, social services and community groups to improve care for Medicaid members and uninsured patients by helping build a more tightly integrated community-wide care delivery system.
The time for better tracking and management of social determinants of health has arrived, said Reider at HIMSS19. (Even if "social determinants" may not be quite the right phrase, as he explains later in this Q&A.)
"Social care is an imperative part of improving the health of a population," said Reider.
There are basically three types of care – of increasing levels of acuity and, most would argue, expense, he explained: social care, healthcare and medical care.
Most might think it's "self-evident" that if a society invests more money on the lower end of that continuum, it will spend less money at the top, he said. And that is indeed the hope.
But in order for this historical moment, with social data finally getting the attention it deserves, to really come to fruition, CMS policymakers and other payers will need to be on board.
"And if we're going to get health plans to invest, we need to show ROI," said Reider. It will be important to "learn from the mistakes we made in medical care," he explained, and that includes doing a better job assessing KPIs and quality measures.
That, in turn, depends on first ironing out of some substantial technical challenges related to EHRs and interoperability – to say nothing of getting consensus answers to fundamental questions such as: "What's the data we need to capture and how do we aggregate it?"
The meeting at HIMSS19 was meant to gather like-minded people from widely diverse backgrounds with a "shared interest in doing the right things," said Reider, to help capitalize on the momentum building behind the value of this data. "It's really incredible what we can do," he said. "Now let's organize ourselves."
A few weeks after that meeting, Reider spoke with Healthcare IT News for a long discussion of how information on patients' social data can be better acquired and, ultimately, acted upon for more personalized care.
Q. What were some of your takeaways from the SDOH roundtable in Orlando?
A. There's a lot of interest in this in this domain. From different perspectives. Who did we have there? Folks from federally-qualified health centers, so on the frontlines of the medical side. We had folks from states, and/or or former state leaders who are looking at this from a public health perspective. We had folks from social care providers. We had health plans. So we had this really broad cross section of the care continuum, all saying, “Hey this is important and we would like to participate in some way, in figuring this out together.”
In the weeks that have followed, I've had a lot of communications with people who were there, or people who wanted to be. You may remember I talked about that cycle of the four As: acquire information, aggregate information, analyze information and then act on information. And a lot of the conversations that I've had since that meeting have been, to some degree, about the first three – and how important they are in order for the last one to occur.
So the first step is, how are we capturing information about identification of people's social care needs. Because in the medical universe we had ICD, and SNOMED-CT that we would use to capture information. I'm not saying we did it super well, but we had a reasonably good way of representing the fact that a patient had diabetes, using a certain coding system.
We don't have a very good consistent, predictable, repeatable mechanism for expressing that somebody has food insecurity needs. Or is a domestic violence victim. Or has a behavior health challenge that makes it difficult to leave the house. Or that they have transportation challenges. We don't have good ways of representing those things, and there's been a lot of interest since that meeting in identifying, OK, so what are the coding standards that we could use? What are the transmission protocols? There's been talk about FHIR. So how do we use FHIR to move this information. So that's been an interesting byproduct of that meeting.
People are starting to think about this in a way that I don't think they had been together before. There have been isolated instances of these conversations, but I don't think they've come together before.
Q. Or what about a challenge even as simple, or seemingly simple perhaps, as just a data field in an EHR for clinicians to note certain social determinant factors. This is something the vendors are going to have to get on board with, when it comes to developing ...
A. And that's my point. Right? If we make a field – and, you know, I think it's more than a field. But that's the problem actually. I could tell you horror stories, whether it's when I was wearing the Allscripts hat (ed note: Reider worked at Allscripts for four years as medical director and CMIO), or when I was wearing the federal hat, somebody would say, “Oh, it's just a field. Can you just add this to the EHR?”
It's so much more than that. Because what do we put in the field? If we make a field and it's free text, wow. Now we're creating all kinds of chaos because people are going to type in stuff that we wouldn't have expected. If we make it a field that's a checkbox and that maps to a semantic concept, then which semantic concept? Right? So, "food insecurity": How do I express that and is there a LOINC code for that? Well right now there isn't.
So we actually have to start at the nomenclature and vocabulary standards and say, “OK, here are the seven concepts that we want to make sure are captured.” Then the EHR potentially could capture it. And and then FHIR or some other transmission standard could transmit it. And then it would have a consistent meaning when the information was captured and then transmitted and then conveyed to other organizations so that it could be analyzed and acted upon. So yes, you get it: definitely, maybe, possibly, someday there's a field. But there's stuff that has to happen before that.
Q. It's kind of a similar fundamental problem that genomic precision medicine – the other half, along with social determinants, some say, of that broader concept of personalized medicine – is grappling with: How do you add that detailed data to the EHR in a way that's not just slapping on a PDF?
A. Sure. And I've had the same conversations with the EHR vendors and the geneticists. Because the geneticists, to them, well, maybe it's a PDF and maybe it's a 600-row spreadsheet of all the sequences they've identified that are relevant to a certain individual. So now we say, “OK, this 600-row spreadsheet, how do we how do we express that in the EHR?” Same question. We need a semantic concept that represents each one of those things. So we can say BRCA1 BRCA2, BRCA3 = increased risk of breast cancer. Now the EHR can take some action on it, or the doc can take some action on it. So yeah, it's exactly the same situation all over again.
Q. Well if there are some pretty big and fundamental IT challenges that we probably have to overcome early, with EHRs and interoperability, there are other technologies that seem to be very exciting and enabling for social determinants: remote monitoring, Bluetooth, stuff like that. Are you excited about some of those ancillary tools?
A. I think it's similar to what we were just talking about: What's the information that's going to be most useful for us to better understand and then cause us to make certain decisions. So OK, let's take the Bluetooth scale. Ninety-nine percent of the time, today's reading is the same as yesterday's reading. And so I certainly wouldn't want to drown all the physicians in the country with daily weight readings from all their patients.
However, some small percentage of the time there's going to be a delta: Grandma ate too much corned beef or ham last night, and she has congestive heart failure. It's a massive salt load and she gained 12 pounds. And tomorrow she's going to have to go to the hospital if we don't do something, but she doesn't feel bad today. Wow. That's huge.
So how do we identify the variance there and then and then understand it? Let's take it and rewind it a layer. And this might start to be sorta Big Brother, right? But what's the social determinant of that corned beef meal? Well, maybe it's her credit card bill and the fact that she bought this big thing. I don't know. Or is there a way for us to predict that, maybe from the home health aide who helped make the meal for her, and she was tracking her meals. Then we could anticipate her weight gain and prevent it.
You know, recently we've been talking about food as medicine. And if you know if we are tracking her food carefully we can actually understand what her health risks are – and not just predict something bad happening but prevent something bad from happening.
So that's exciting if we can figure out what's super important. My formula is priority equals value over work: What's the work of getting information, with the value of that information. Then we can prioritize which data elements we should add to the EHR. Maybe it's what you had for dinner. Maybe it's that you live alone. Because those are important. "Do you live alone?" is a really important question, it's a powerful predictor of depression and other things. But right now we don't have a way of capturing that. We do have a way of capturing weight, so we capture a weight. But maybe weight, 89 percent of the time, is completely irrelevant. So we're looking at only the data we know how to understand. The cool thing about social determinants is it starts to open up to other data sources.
Q. You've spoken about the need to show, for lack of a better term, ROI of social determinants data to the health plans and accountable care organizations that would be key to giving this stuff the momentum that it needs. Expound on that a little bit.
A. If the hypothesis that we're all testing is that making more investments in social care helps people be more healthy, and therefore we avoid preventable acute care utilization, then we actually have to demonstrate that hypothesis is valid.
I'll give you a tangible example that came from my staff meeting this morning. Some members of our team announced, to their own glee, that we have now facilitated more than a thousand rides for Medicaid members. We pay for transportation for Medicaid members for non-medical trips. Medical trips are already paid for by Medicaid, but we will pay for a trip to the pharmacy to pick up your medicine. We'll pay for a trip to the food pantry. We’ll pay for a trip to the supermarket. You name it, we'll pay for a trip. And the people who can order those trips are members of our social care or medical care community. So if you're a homeless shelter or a doctor's office, and somebody says, "Well, I can't make it to CVS and they don't deliver," we'll pay for that.
So we've done a thousand of those rides. And it's all Lyft and Uber that we're we're using. So I'm guessing, you know, $10 a trip, to just pick a number. So we've spent $10,000 on this. Now here's the question: Was there ROI in that $10,000? Did did I prevent $10,000 – or $20,000, or $50,000 – of ED utilization?
Now I have to look at the data. So I challenged the team. The team was all happy, right? They're like, "Yahoo, a thousand people! Aren't we great!" And I said, "Well, I don't know. Are you?" It rains on the parade a little bit when the CEO is questioning whether you should be so cheerful. And I'm mostly being provocative, sure. But I want them to think. It's not just about doing all these good things, and we're really nice. Because you know the the old saying: no margin, no mission. If there isn't return on investment here, we can't keep doing this.
So we look at the data a couple ways. We can look at this population as a control group: We know who we gave a ride to, so we can look at those thousand people and say, "OK, how many times did they go to the hospital, to the ER, this year?" And then we can look at the same people last year. And maybe it's not going to be exactly the same people, but we can look at that group as a control group and we can say, "Hey, they've used the ER 17 percent less this year than last year since we've done the transportation program."
Seventeen percent less, at, say, $1000 a visit? Boom. We did it, right? Seventeen percent of a thousand is 170. So I just saved $170,000 with my $10,000 investment. That's excellent! We don't have that data yet. But when I talk about ROI, that's what I mean. We have to figure this stuff out. Does screening for transportation issues, and then acting on those transportation issues, results in the outcomes that we seek to find?
Maybe we haven't been doing it long enough. Maybe it's only been three or four months that we've been providing rides. Maybe I need to do it for six months to have enough data to look at that. Or another way to look at it might be to just look at a different control group. Maybe it's a thousand people who are in our community who didn't get rides and starts to get complicated there. But there might be confounding variables, like the fact that they got a ride means they're connected to our system. And if they're not getting rides then they're not connected to our system. So maybe the the control group is not the right one.
Q. We've discussed some of the the technology challenges, but talk about some of the cultural changes that will have to take place. I was struck by one of the comments from a member of the focus group, who said she didn't think the onus should necesarily be on the providers to have to deal with some of this – that there should be other stakeholders who are also taking charge and leading the way. Talk about the challenges of building a culture where this stuff is valued across the board – which is increasingly now, even if there's still a way to go.
A. Yeah, very insightful of you to notice that that came up, and that she said that. So, we physicians – and I'm going to malign my species for a little bit – we are taught in medical school and in residency that we are in control. And many people go into medicine and have that gene, right? They want to be in control. And as long as everything's sort of between the lines we feel good. And we want to take responsibility for stuff. And often we are told that we are responsible. We're responsible for keeping the patients healthy.
And some to some degree that is a fading component of the culture of being a physician. And we've seen over the past decade that team-based care has been recognized as much more effective, and seen that some some practices have even flipped it, and said, "Hey, the doctor is just the technician. And the care manager is the driver, and they're in charge. The doctor does what the care manager says they should do, and the doctor is not the quarterback anymore."
But the key, especially with some of these social determinant things, is that now what we're seeing, which I think is fascinating, is that the team is is expanding beyond the practice. Beyond the clinic. So at one facility, we've seen the culture evolve to where the doc and the nurse and maybe there's a care manager there and that's new that they've all started working together. And then they recognize, "Gosh the food pantry is part of the team. The transportation of poor people, the housing coordinator, the social worker at the county Department of Public Health. These are all members of our team too. Wow."
And so we can shift the perspective from the doctor and the doctor's team being responsible for addressing the social needs of the patient to the doctor being responsible for recognizing these social needs, and then handing off to the community. And I think that's the shift: "I don't need to get this person to the food pantry. I don't even need to know where the food pantry is. I just need to identify the need, and this is it."