From 3M Health Information Systems
Podcast Episode Transcript: Reducing inequities in health care delivery
Gordon Moore: Welcome to 3M’s Inside Angle podcast. This is your host, Gordon Moore. And with me today is Dr. Robert Phillips. He directs the Center for Professionalism and Value in Health Care, which aims to create space in which patients, health professionals, payers, and policymakers can work to renegotiate the social contract. The center will also seek to define value across the health care spectrum, how to measure it, how to improve it, and how to engage and develop leaders. Dr. Phillips is also responsible for overseeing the American Board of Family Medicine’s research and related collaborations, as well as the policy relationships related to family medicine and primary care. His bio was so long, it’s really quite daunting, but really interesting to read. Welcome, Dr. Phillips.
Dr. Robert Phillips: Thank you so much, Gordon. It’s great to be with you.
Gordon: Yeah. Thanks for joining me today. There’s a particular aspect in your broad purview of work that I think would be really interesting and captured by attention today, which is to talk about social deprivation and especially social deprivation indices and how to understand and quantify that, and why you would do that, and what do you do with that information in health care. So let’s get at that. Why don’t you tell me a little bit about what is a social deprivation index?
Dr. Phillips: So a social deprivation index is a collection of social determinants of health that have been constructed and weighted in a way that they are predictive of certain outcomes. So they are associated with certain things you would like to avoid. Poverty turns out to be a big piece of just about every social deprivation index, but level of education, employment, single-parent household, house ownership, a lot of these things tend to aggregate in neighborhoods so that some neighborhoods have a lot of those. And when that happens, those people are at significant risk of poor life outcomes, not just worse mortality, but more likely to go to prison, more likely to have job instability and housing and food instability. And so when you look at areas that have a high index, you’re looking pretty reliably in an area that’s going to suffer a lot of poor outcomes.
Gordon: Alright. So tell me, you mentioned there are these different factors, what does the weighting mean in this context?
Dr. Phillips: We take, or at least we took, a large basket of neighborhood-level measures that the Census Bureau collects in a couple of different surveys. And then we looked at outcomes, we looked at mortality, infant mortality, diabetes prevalence, a whole host of things, and then we come back to these predictors and we work with them. We add some in, we drop some out, we give them different weights until we come up with a short list of measures that are highly predictive of these outcomes. And then each of those elements tend to be weighted for their effect on those outcomes, which is why poverty in ours turns out to be about 40 percent of the index. It’s such a potent predictor of those outcomes that it’s more heavily weighted than the others.
So it’s both a collection, and then within that collection of measures, you weight them all with the goal of being able to more reliably predict the outcomes for that neighborhood.
Gordon: So when you mentioned the indices and their relationship to outcomes, I was curious about other social deprivation indices and weighting these factors. Is the weighting and the understanding of connection to outcomes universal across these different indices? Are they pretty similar in the way they work?
Dr. Phillips: They are. The other major index that we look at is the area deprivation index that was developed by Dr. Amy Kind at the University of Wisconsin. And while her collection of measures is slightly different than ours, the weightings of the ones that are overlapping are very consistent. And when you look at how these two indices perform across the United States, they are both pretty reliably and highly correlated. Her index was slightly different because she had access to different data than we did in some cases, and in others, she was using different outcomes, particularly looking at re-hospitalization rates or 30-day mortality rates. But the fact that they’re highly correlated is pretty important.
And we both use the same process that was used in New Zealand and the United Kingdom to develop their deprivation indices. So it is just no wonder that they’re very related.
Gordon: Let me pivot to a different aspect which is interesting to me. I look at these indices where you’re using survey data from the census and others and you are saying something about people who live in our neighborhood, but that may—that link, that imputation to the person may not always be true. What’s the risk of that and why don’t we just measure at the person level?
Dr. Phillips: So that’s often referred to as the ecologic fallacy, attaching the geographies, measures to the person, and that’s not always consistent, that’s a very important point. It turns out for these deprivation indices that the predictive risk for the people living there is extremely high. It may be a little different depending on what age they are or what outcomes they’re at risk for, but they’re incredibly reliable. It turns out, places that seem to deviate from that predictive model often have better social cohesion or social services than other places, it’s not the differences in the people that tend to predict those variances.
But where it also becomes important is when you have poor people living in wealthy neighborhoods, where the folks around them may make it hard to see that person and their risk, but, the work by Harvard in particular, Ichiro Kawachi and his team, have shown that the people who have those risks and locate into a wealthier neighborhood will have better life outcomes overtime. They’ve done longitudinal research demonstrating that not only for the adults, but for the children and sometimes the children of those children. So that is the greatest risk.
Our colleague, David Rakoff at Stanford also did work demonstrating that the risk for people in those neighborhoods that are of highest risk, highest index measures, that the risk of ecologic fallacy is extremely rare. So we think it’s a good mechanism for identifying people at risk and for channeling resources to those neighborhoods in particular. Individual-level measures become really important for understanding what someone needs. So you really need both of those. You need resources to come because the risk is there, and you need to know this person sitting in front of you in clinic, is their problem food instability, is it housing, instability, is it family chaos, is it job instability? What is affecting them that you need those resources to be assigned to?
And when I talk about this in a clinical context, I often say, we use adjustments based on people’s multiple health conditions to vary the payment that we give to clinics. So if someone has a kidney problem and they have diabetes and they have blindness, a lot of things that happen with severe diabetes, you will get a higher payment rate for that person and you will also not be judged as harshly in terms of their outcomes because they have underlying risk. That person, what you actually do for their health care is going to differ depending on which of their conditions is most severe, or whether they’re in a downward spiral or whether they’re doing relatively well. So it’s the same kind of risk assessment to bring resources, but personal assessment to decide what this person needs next.
Gordon: Yeah. So in a straight clinical model, I’m working with a person who is relatively healthy and has a single condition that’s time limited, it’s going to be very different working for another individual who maybe has three chronic conditions of high severity who’s in a downward spiral. I should be doing a lot more for that latter person because their clinical needs are so much greater. And you’re translating that to the nonclinical factors that influence health and outcomes.
Dr. Phillips: That’s right. And you may not just be doing more for that person. They may have a care coordinator. They may have a social worker if they have depression as part of that, or a mental health nurse. So it actually brings more team resources and maybe referral resources to that person. So you need more resources to be able to bring those additional capacities to that person’s care.
Gordon: I mean, that’s where I was going to go. I was thinking, I’m trained in family medicine, which means I’ve got a pretty broad view of what I would like to be able to do for the people coming to my practice, but typically I’m focusing on the medical stuff. And when I think about these other factors, I’m thinking, this is way out of my league, way out of my training. We’ve been working in the last couple of decades on multi-disciplinary care teams and figuring out how to distribute the work accordingly. But now I’m thinking I need to add a social work, I need to have a health coach, a community health worker.
And when I think about that in the context of overburdened primary care stressed beyond their capacity in many situations to do even the basic blocking and tackling work for chronic disease management, now we’re just piling on. How do you do this?
Dr. Phillips: My answer to that is it’s going to happen regardless, so you might as well get the resources you need to build out the support system you need. More than 30 states now require the collection of social determinants data if you see Medicaid patients. Only one of them so far, a second one is in the process, but only one of them so far gives you additional resources to deal with what you find. And we know that clinicians, when they’re given a directive to assess something, if they don’t have the resources to deal with it, they are less likely to assess it, they’re less likely to assess it correctly, they underestimate it. And if they do assess it correctly, it frustrates them to no end that they now have additional problems that they have to manage without the resources to do it.
So our perspective is, this is happening, we need to realign the resources so that they’re not hit with this frustration without the capacity to do something about it.
Gordon: That reminds me, I was in a diabetes and depression collaborative in New York City and working with a bunch practices across the city as they were starting to use a simple survey tools to unmask on recognized depression. And about six months into it, the practices were coming to us and saying, we need to stop this. I have no place to send them. I have no capacity to manage that. And it makes me think about U.S. Preventative Services Task Force recommendation on screening for depression in adolescents where it says, screening for depression adolescents is a good thing only if you have the capacity to act on it, otherwise, don’t do it. And that’s depressing to think about, it’s like, we need those resources.
Dr. Phillips: It is. And for a long time, they recommended not assessing, not screening for depression in adults unless you had the resources to manage it, because it recognized that that was just going to generate frustration. The outgoing chair of the U.S. Preventative Services Task Force is one of my clinic mates. And he ran a weight reduction intervention that had the ability to give people free access to this external support system, and it went over like crazy. I mean, patients loved it, the physicians loved it because they had a place to send people when they counsel them about weight loss.
When that intervention ended and the resource went away, so did the assessments, so did the recommendations to patients. It just had a voltage drop because there was no resource there. It’s very much a human nature thing.
Gordon: So I’m in practice and I’m thinking I want to do the best I can for my patients, which is normal, this is why I get into health care. And I’m working hard on diabetes, and I’m frustrated as anything because I feel like I’m working with some people whose attention and capacity to manage diabetes is completely swamped by their life circumstances. And this is a mechanism to measure and understand that and bring resources to bear. What are the mechanisms for resources? How does that—how could that work? What would be a framework that makes that real?
Dr. Phillips: So I’m going to answer that. I’m going to back up just one second. When we introduced this to the clinics that are part of our national registry, we don’t call it a social deprivation index, it’s too wonky, we actually call it a community vital sign. And like any vital signs like blood pressure or body mass index, what it does is it’s supposed to identify this patient has particular risks that you may want to address. And so if this community vital sign or social deprivation index is high for a patient, the goal is that the clinician says, you know what, I need to ask you a few more questions about your life circumstances.
It’s supposed to actually kick off that conversation with the patient about if they have social needs and what they might be. And then the idea is that if we can adjust the payments that that practice, if it’s large enough, could have a behavioral health specialist, they could have a social worker. If you look at one of the large Milwaukee Medicaid Managed Care organizations, they have actually what they call an ambulatory ICU for very unstable families and patients. And so they have substance use disorder counselors, they have community health workers who can go out and assess people’s housing stability or the chaos of their families.
These teams are known, and they’re known based on the risks of populations. We just don’t have the resources to put them in place. And for small practices that may not be able to either pay for or have the real need to have an in-office social worker or a behavioral health specialist, Vermont’s Blueprint for Health, which was a Medicare waiver demonstration, did that beautifully, where they embedded these collective, shared resources in a community that multiple practices developed relationships with and could draw on for patients’ needs. So there are even different models of how those teams can be assembled, but you can’t assemble them unless you’ve got additional resources based on your patient population.
Gordon: So we spend north of $3 trillion a year in health care in the United States, and I’m hearing we’re getting into all sorts of public health stuff and hiring other people, doing all this work. We’ve implied, but haven’t talked explicitly about how we create interfaces with community-based organizations and have to constantly survey who is or isn’t doing certain things and what programs are, aren’t funded, what are the eligibility criteria. It’s a lot of complex work. Well, with that kind of spending, are we just piling more spending on top of everything?
Dr. Phillips: Well, you probably are initially, or you’re adjusting payments based on the risk profile of a patient panel, which will be unpopular for those who have very advantaged patients. But the goal is to reduce the overall costs. And I say the real goal is to help those patients meet particular needs that translate into health outcomes later. So the goal of addressing health care costs, it may be downstream several years, so you have to have the patience and tolerance to know that it’s going to have that effect down the road. It is a shame that in this country, we actually draw down resources from social services because we can’t afford them because we pay so much into health care.
And now to talk about putting more resources into health care to deal with social issues is it’s an embarrassment, but it is the only vehicle we really have. And it actually means that Gordon, at some point, you wind up transferring those resources to community-based organizations, for example. That the funding stream comes through health care, but it goes back out to the community to deal with those issues. It is because it is a vehicle we have unfortunately. In the UK and in New Zealand, they adjust payments for both health care and social services so that a neighborhood gets resources from both sides to address their needs. We don’t have the luxury of that right now.
Gordon: I think that the vision I have of the U.S. health care system is that we are really phenomenal in some ways for cleaning up a mess. We have PET scanners all over the place, we have phenomenal specialists who are terrific at dealing with very narrow, scary spectrums of disease at its most extreme presentation, but we missed the boat upstream so that we have people who because of a late diagnosed or unrecognized or poorly managed condition, end up with a hospitalization, end up with complications, end up needing surgery when medical management might’ve saved the day if started years before.
But the horse is out of the barn, barn’s burned down and the ground’s been salted. So I have the sense that we do have the resources we may need, initial shift to spending, but if we can manage this well, I see bits and pieces of examples of even in complex patients with metastatic cancer that we’re not going to cure, but helping them manage just aspects of getting to and from a doctor’s office or being able to work with a nurse around constipation, avoids hospitalization for lifestyle issues. So even in that setting, to say nothing of working on smoking cessation, it’s early recognition of management of substance use disorders, which are probably ramping right now in COVID times. So I have a sense we can do this. We’re just spending 10x or 100x downstream, as opposed to the 1x ounce of prevention upstream.
Dr. Phillips: That that’s right. And one of the concerns I have, Gordon, is that if we focus too much on the costs as the outcome to be moderated, we may wind up adjusting our payments based on the evidence around costs rather than the evidence around solving people’s needs and improving health equity or outcomes. I am so proud of the State of Massachusetts for having constructed what they call a neighborhood stress score, it’s a deprivation index that they use to combined with mental health to reweight payments based on where patients come from, which neighborhoods they come from, but they designed it based on the goal of reducing their downstream costs.
And the risk of that is that you may not adjust your payments appropriately enough, and that you may, if you’re lucky, wind up solving the downstream cost issue, but you never improve the health equity. So we have a challenge, as we talk about what are we going to use to adjust payments of also saying how much and to what end. Even if downstream costs are our goal, that may not be the target that we need to be aiming for.
Gordon: Yeah. You touched on, or maybe it’s somewhat tangential, but it triggered a thought for me about something we touched on earlier, which is the difference between risk versus need. And one of the definitions of need that I read about often is we can see that this person is needy because they’ve had a lot of health care resource expenditure utilization, so that they’ve done a lot of stuff, or a lot of stuff has been done for them or with them by the health care delivery system. But that makes me a little bit squirmy because there’s an inherent bias.
If there are certain classes or groups of people who have constrained access, let’s say they’re working, but they’re in a job that will not allow them any time off because they could get fired and they have kids to take care of at home, they may not be able to just find the time to go to practices, to get stuff done until things blow up and then we’re in a disaster recovery mode again. So access to care creates this bias around utilization, and if we use utilization as a predictor of need, we’ve made a huge mistake. Is this when we use social deprivation indices and do these surveys? Are we going to be able to mitigate against that?
Dr. Phillips: Well, it won’t solve a couple of fundamental problems, Gordon. The first one is that we don’t have universal access even to primary care in this country, and adjusting payments won’t address that necessarily unless they’re funneled into community-based clinics or community health centers. The second problem I think with that is that we have to create space in those people’s lives to get care. And some companies have dealt with that by building health care within their company, within their sites and within their workplaces. Federally qualified centers have done that by putting them in the neighborhoods where those people live and having atypical hours to allow those people to come in.
So I don’t think that what we’re talking about will necessarily address that unless it is either accompanied by some policy changes about, and I hate to say universal access because it has such a political charge to it, but a common good called access to care, or if it is allowed to be funneled into particular kinds of clinics or hours or capacity.
Gordon: I think about the common good as in when we provide access to care and especially a foundation of high-performing primary care, we know that on the average, populations do better and do better as defined by life expectancy, slower progression of chronic diseases, as well as presenteeism at work, which is part of the virtuous cycle of the greater good. And my understanding is that’s pretty well tested and validated in population studies. Am I on target with that?
Dr. Phillips: No, you absolutely are. There have been many studies that show it either by association or even some early in the 70s just by giving insurance as a benefit. Sir Michael Marmot, who really developed this whole field of social deprivation in recognizing differences between classes of people in terms of health outcomes, in a famous Lancet piece now, said that the differences between the U.S. and the UK, the UK has a much more compressed range of outcome differences between populations than the United States, where we have a much higher hurdle for health equity issues to overcome. He said, one of the key features of the UK system versus the United States is access to primary care.
So it is one of our failings, our primary care workforce is eroding, particularly in rural areas. The primary care to population ratios that we aim for in the shortage areas of this country—in a paper we have coming out next month, pretty well demonstrates that we’re under shooting, we’re not adequately aiming for the right primary care density that we need to improve health equity. So it is a big issue, and I’m, co-chairing a committee right now for the National Academies of Science, Engineering, and Medicine on implementing high quality primary care. It’s the first focus by the National Academies on primary care in 25 years. But it won’t come out until April, and it addresses this very issue.
Gordon: Makes me think that there are some big leavers that may be interesting to think about pulling, for instance, I’m reading a lot about just recently in the press, some—the increase in med student slots not matched by the increase in residency slots so that you can finish medical school, but without residency training, you can’t become a board certified physician, which is a threshold you need to be on insurance panels and take care of people and bill accordingly. But what about in these health shortage areas, why don’t we think about how to forgive the debt, the hundreds of thousands of dollars of these now physicians, give them slots in residency, increase the capacity of primary care residencies as long as they are willing and able to serve? These mechanisms exist though, don’t they?
Dr. Phillips: They do. Where we fall down is that we don’t have pipeline programs that bring people from rural areas or underserved areas into medicine, and the cost of going to medical school has increasingly locked them out. And those people are more likely to return back to their home communities or communities like their home communities. And the second issue that we have is that we’re not only not building more training programs in those areas, which has been demonstrated to really influence the likelihood of someone practicing in one of those areas, because they get comfortable, they see that they can, they see that they can make a difference, they get to love those populations.
We’re not only not building those, we’re tearing them down. So the two real big drivers of whether a physician is likely to take up a loan repayment program to go work in an underserved area, those two really potent ways of doing it are disappearing.
Gordon: Okay. So we have some big challenges. I’m looking forward to National Academy’s report in April to see what comes out with that, hopefully, we’ll be addressing that. But in the interim, maybe the lesson for today is that we can measure social deprivation, we didn’t spend a lot of time talking about why that’s important, but we’ve touched on it somewhat. And I know that that it’s talked about enough that I don’t think we need to dig into that today. But we know how to measure it, we know which of the attributes have greater impact on a person’s health and wellbeing, and what we need to think about is how do we use that to move resources to practices and people with greater need so that we can help address those need, and that’s the future.
And you’ve mentioned Massachusetts is tying it, maybe too much close to dollars saved as the measure of success, but there are others who are doing that in other parts of the globe but maybe also in the States?
Dr. Phillips: There’s a few states considering how to adjust payments, several are looking at Massachusetts model, but not many are thinking about deprivation index or indices or geography as the way to allocate the resources. A lot of them are still very stuck on this idea of assessing the individual’s need and therefore their resources. What the folks in the UK found is that when you really rely on individuals’ needs for adjusting payments, you further disadvantage the disadvantage because they don’t get care on a regular basis, so they don’t show up to be assessed. Their situations are much more influx than other people’s. They lose their jobs, they lose their homes. People are more likely to go to prison and their families or come back from prison.
So being able to reliably and regularly assess those populations in order to assign resources to them, it falls apart. So these indices give us a more reliable way to channel resources. And I just want to tie this back to your previous comment about workforce. Because you’re putting more resources into those areas, it creates an incentive for clinicians to locate there. And it’s not just a profit-making idea, it’s also the idea that they will have the resources to care for those populations and be working with more robust teams. So there’s a behavioral economics behind this as well that could help shore up the workforce in those areas.
Gordon: Yeah. And you’ve touched on a piece, which I appreciate, which is the professionalism, which is beyond the usual human weakness of just looking at money. I think a lot of people get into health care who have an interest in actually doing good things to help people. And of course, we’re all human and we all are attracted to income, but that’s not the sole driver, I think for many. And I think for a good number of our colleagues, it can be depressing to be reduced to that only as the goal that measures our success. I guess the frustration I’ve sensed from colleagues in practice is that if we’re only measured by the money, we’re really missing what it is we’re doing for people.
And our measurement is really skewed in terms of not recognizing the value of what we do beyond simple counting of volume and codes that generate revenue and things like that. So we need a different measurement system, but I think that’s got to be a topic for another podcast.
Dr. Phillips: Well, you have come right back to the mission of our center, which is to align the values of the professions with how they are valued. And it does tie into this discussion, it’s am I getting the resources I need to take care of a population that I have a passion for, or is this a fool’s errand that I’m going to wind up burned out in five years and feeling like I was unable to make a difference just handing out bandages? They are related.
Gordon: Crisis is a horrible thing to waste, and I think these issues have been around for a long time, but the pandemic has just made these issues excruciating, and so maybe there’s an opportunity. So in closing, do you have any last thoughts or wrap-up you’d like the audience to hear?
Dr. Phillips: Well, we have an opportunity. Congress passed the Impact Act in 2014, which directed our Secretary of Health and Human Services to either say that we don’t need a plan or to come up with a plan for how to adjust payments based on social risk. And we are really trying to point out that there is a process, it is reliable, it has been tested in other countries and at least one of our states, that we have the data, that we can launch this as a demonstration project if needed. But the idea is that if we can get Medicare to develop this process, and if we can signal to the states that there is a basement or default process for doing this, we can save a lot of chaos.
Particularly, if 50 states have to apply for their own particular plan for how to do this, and we lose all capacity to understand the impact. We think some of them would gladly follow a default policy in lieu of trying to figure out how to do it on their own. And that would give us a real capacity to see how does this play out in terms of equity and if need be, for cost. So there’s a policy window that’s open and we hope to jump through it and make this a reality.
Gordon: Here’s hoping. Thank you, Dr. Phillips.
Dr. Phillips: Dr. Moore, thank you so much.