 Good afternoon. It's my privilege to introduce Professor Anup Malani from the law school who is going to be talking for us today. Let me just give you a little bit of background for those of you who do not know Anup. He is the Aaron Director Research Scholar at the University of Chicago Law School. He's also a professor of medicine at the University of Chicago Pritzker School of Medicine. He teaches health law, food and drug law, insurance law, bankruptcy, contracts, and corporations. His research interests include law and economics, the welfare evaluation of legal rules and the economics of product liability, health economics and policy, including control of infectious disease, the conduct of clinical trials, medical malpractice and drug products liability, as well as conflicts of interest in medical research and the placebo effect. Professor Malani graduated from the University of Chicago Law School and the University of Chicago with a PhD in economics under Gary Becker. He clerked for the Honorable Stephen Williams at the U.S. Court of Appeals for the D.C. Circuit and the U.S. Supreme Court Justice Andrew Day O'Connor. He has held faculty positions here at the University and the Health Evaluation Science Department at the University of Virginia Medical School. He was a visiting professor at Harvard Law and the interim director of the Petrie-Flom Center for Health Law Policy, Biotechnology and Bioethics at Harvard Law School. Currently he's on the Committee on Clinical and Translational Sciences here in the BSD and the Edit of the Journal of Law and Economics. Today his talk will be entitled, addressing racial disparities in hospitals, actually a new title, Civil Rights Litigation and Racial Disparities in Health Care. So I didn't have it on my sheet. Please join me in giving a warm welcome to Professor Malani. And before he starts, just a reminder, this is not a microphone that projects, but it's important for the taping. So when you want to ask questions, you've got to grab the mic. Okay, I guess the first thing I've got to write down is a notice to get you a shorter bio. Because that was a mouthful. Okay, so thank you very much for inviting me. I'm going to talk about a topic that, you know, that I'm not an expert in, as it turns out, but it's an area that I'm trying to start doing a little bit of research in. And in part it's because of an introduction to the field from a very good friend of mine, Amitabh Chandra, who's a professor at the Kennedy School. And he's been working on disparities for a decade now. And one of the nice things when you're new to a field is you look at it from a different perspective than somebody that's been in it for a while. And we're finding that there are a lot of synergies between, you know, his extensive experience and me asking kind of silly, simple questions about, well, why could this be the cause? Why isn't that a solution? And so we're finding this to be a lot of fun to work on together. And our third, the third author here is Michael Frakes, who's a PhD student out of MIT and also has a JD. He's currently, he's actually a student of mine at the Petrie-Flam Center, and that's how I met him. And again, also has a budding interest in civil rights. So this is a fun topic to work in. But I do want to clarify, I am neither a civil rights law expert, it's not a class that I teach, I don't teach Equal Protection or Constitutional Law. I do teach health policy, and I think that racial disparities in health care and in health policy are very important. So it's very, it's a big omission on my part not to have studied this topic a little bit earlier and hoping to, with this short paper to kind of remedy that failing. Okay. So I'm going to try in this talk to cover three topics. The first is I'm going to try to kind of give you a glimpse of how I look at the question, what are the source of racial disparities in health care treatment and outcomes? Then I'm going to ask a question that is more appropriate for a lawyer. And so I think that I might be able to shed some light on is, and that is, can we reduce the degree of racial disparity in health care and health outcomes by either improving enforcement of Title VI or by expanding the scope of Title VI? And when I went into this, you know, it's kind of optimistic. You know, I've, everybody is educated with, or educated in the great success stories of the Civil Rights Act 64. And so I thought, you know, why should health care be different? But the more I looked at it, the more I became pessimistic that that's really the tool that we need now to address the disparities that persist. And so I'm going to propose some alternative remedies that I hope, or I think at least at this point are going to be more effective at reducing those disparities. So it's a little bit awkward for me as a lawyer to say, I'm not the solution. I like to tell people I'm the solution and have them pay me. So this is a little bit awkward. But be that as it may. So let me start out with the Institute of Medicine's report on racial disparities. This is from 2003. And they did an extensive analysis of the literature on racial disparities. And they acknowledge that the disparities have a lot of sources. They're historical sources, they're institutional sources and things like that. But after reviewing the literature, they said that they would focus on what they think the key element is, and that is summarized here. The study committee focused on the clinical encounter and found that there was evidence of stereotyping biases and uncertainty on the part of health care providers, and they can all contribute to unequal treatment. The conditions in which many clinical encounters take place, high time pressure, cognitive complexity, and pressures for cost containment, there shouldn't be an S there, may enhance the likelihood of care poorly matched to minority patients' needs. And while they did acknowledge a lot of other causes, the focus was on issues when a doctor sees a patient. So on the individual provider-patient relationship. But when I think about the source of disparity, I want to actually say, while it may be true that there are disparities caused by problems in the relationship between, and that's a euphemism, problems in the relationship between a doctor and a patient, when you think globally, you want to kind of divide up the disparities into three bins. Kind of just thinking as an empiricist rather than anything else, a historian or a lawyer. I like to think of the world as having within disparities, not as having, but I like to think that I can break down the world into within disparities, between disparities, and issues of access. And when I say within disparities, that means that if a black individual and a white individual goes to a given provider, is there a difference in the treatment that the black individual will get relative to the white individual, or the difference in outcomes between these two people holding the provider constant? So it's within provider disparity between blacks and whites. And just to be simple, I'm going to use the term black and white, but broadly speaking it's minority, non-minority, but those are a little bit longer words, so I'm going to use black-white knowing that that's somewhat inaccurate. Then there's between disparities, and that is I'm going to see if there's a, I'd like to see if the part of the problem is that blacks go to a set of providers, whites go to a different set of providers, and the problem is that the providers are different. So it's between provider disparities. So this is within provider disparities, between provider disparities. Perhaps blacks are going to worse providers, or treated by worse providers. I don't want to assert that the active voice or passive voice there. And the last thing is an issue of access. So maybe blacks and whites have different rates of getting access to providers to begin with. So it's that blacks are less likely to go to a provider or be given access to a provider, whereas whites are more likely to do so. So we want to look at that margin too. So think of this as the extensive margin, and these are two dimensions of what an economist might call the intensive margin. And what we'd like to do is an empirical study, a large-scale empirical study. Some of this has begun trying to take all the disparities and break it apart into these bins. And then once you do break it apart into bin, you want to figure out why it is that there is, for example, a within disparity, or why there is a between disparity, or why there is an access disparity. And without knowing the data, we can kind of come up with some theories for why you have within disparities. Now, I'm trained as an economist, so I tend to put things into an economic framework. And for me, an economic framework is something like the following. I want to know, holding a provider constant, why is that provider treating, and by the way, the provider could be a doctor, or it could be a hospital. It could be a full institution. Why is it treating a black different than a white? Or why is the black getting a different outcome than the white? One reason is there could be lower financial or medical returns. So lower financial returns, perhaps the black individual has different rates of insurance or coverage by insurance. Alternatively, there could be different rates of medical return to particular treatment. That may explain the difference. For example, when I say biology, I mean maybe a treatment is less appropriate for one racial group than another racial group, or an individual of an individual of a racial group or not. It doesn't have to be based on race. It could be something like, you know, antioplasty is inappropriate for this person, and this person happens to be a member of a minority community. It could be there's differential rates of compliance. And again, it could be that those are correlated with race, but it's the compliance that could be the issue. Alternatively, there could be a higher cost of treatment. For example, there might be language barriers, and the IOM focuses a little bit on this and a little bit on cultural barriers. And then finally, there's also just the direct bias. You cannot, if you ran a regression on what is the difference in how a black is treated and a white is treated for a given provider, you put in all the variables you want to, insurance, biology, compliance, language, cultural barriers, et cetera. There's still a residual left that's identified by just being black. That is what you might call racial bias. Now, I want to be a little, before I answer that, I want to be careful about something. Even if I call only the thing that's captured by, you know, say the variable black in a regression as direct bias, I want to be very clear. That doesn't mean that there isn't black-white disparity through all the other mechanisms. So, for example, if you've got differential rates of compliance, perhaps because African-Americans don't trust providers as much as white patients do, then you're going to see a loading on compliance. You're going to find that compliance may explain stuff, but you're not really getting at the root cause, which may be that there's a reason why African-Americans are less trustful of doctors than non-African-Americans, right, or minorities less trustful than non-minorities. Also, to some extent, there could be language barriers or cultural barriers that also overlap with black-white. So you want to be careful not to say, hey, if I run a regression, I only find a waiting on, you know, an X amount of waiting on black after I control for other things. That other stuff means that that's not bias, or that's not disparate treatment. It could be. It's just that there's overlap here between being black and having some of these other features. So now we can ask the same sort, do the same sort of analysis with between disparities and say, okay, why are there between disparities? Why do some providers provide higher quality care and other providers provide lower quality care and some are more likely to be treating blacks and some are more likely to be treating whites? Okay. And so you might say, oh, quality has costs and higher quality facilities locate an area with higher profit. That's the general economic framework. The quality tends towards profit because it's costly. And so you want to say, why is it that communities with higher concentration of African-Americans are not getting the quality facilities? Quality doctors, quality hospitals, quality nursing homes, you name it. And one possibility is that those communities are poor or have less insurance. And so they're not able to offer the sorts of profits that are required to attract quality providers. Alternatively, there could be higher cost. But here you want to be a little bit careful because it's not obvious that African-American communities have higher cost. On the one hand, if it's true that there's more poverty, then in fact, you're going to get lower rents there. And so land's going to be less expensive. But the flip side is that you might be harder to get providers, for example, if there's bias or something like that, or some other issue, harder to get providers to come and work at a hospital in a minority community. And so you have to pay higher wages or you have to deal with lower quality providers. And then the last thing is you might actually have institutional bias, right? And you don't have to go back to 1964 in segregation to get direct bias. You could find other reasons for it as well. But again, the important thing to remember is just because today you find, even if you found a small coefficient on this, that doesn't mean that there isn't a lot of black-white gap. It could be seeping in through the fact that the reason you have to pay higher wages is because providers don't want to work in minority communities. Correctly rent correctly, they have bad impressions. It could be that there are gaps in insurance and poverty that are actually themselves driven by black-white gaps. There could be problems that lead to lower human capital investment for African-Americans that leads to poverty. Or it could be that blacks are systematically denied access to insurance. You're seeing that in an insurance variable, but you don't explain why it is that African-Americans have lower insurance or why they might have higher poverty. And those are, again, just hard to get at sources of the gap. And by the way, this is the same sort of analysis you deal with to address issues of not just why is a high quality provider locating in a minority community, but why is there sometimes no provider locating in the minority community. What we want to do with this is not just figure out the causes. As I said, we want to figure out the source disparity. And I will admit that it's extremely hard to disentangle this. This is a lifetime project. And as a novice in this area, I can't just come in and criticize. I should be doing something positive, but I do want to point out and acknowledge that it's a difficult problem. I don't know on how much of the disparity is within or between. And I think that's going to be important for policy. That is to say, is it that a given provider treats blacks and whites differently, or is it, on average, a situation where blacks and whites go to different providers, the black individual going to a worse provider in terms of quality. Now, let's start with some really obvious things. First, my microphone came off. That's the first obvious thing. Second obvious thing is that blacks go to different hospitals than whites do. What we indicate this is that we rank hospitals. So this is primarily relying on work done by Amitabh Chandra and various co-authors. We rank hospital by the number of black discharges. So if you've got a lot of whites, sorry, if you've got a lot of blacks and if you've got a lot of whites. And then I'm going to give you the CDF for how many, what percentage of discharges are available in those hospitals. And what you find is, as you go across these hospitals, slowly accounting for all the white discharges gradually, but you'll see that these set of hospitals are the initial set of hospitals are accounting for a very high fraction of the black discharges right away. So these guys are doing a lot of the work when it comes to accounting for the black discharges. And these ones are doing very little. In fact, at the end, there's going to be a lot of providers that are providing very little, lots of cares to white individuals, but very little in terms of changing the percentage that are black. So in short, blacks are going to a different set of hospitals than whites are. And partly that's because of where they live. So this is a map that has the ratio of black population to the average percentage of the population that's black. So if you have a place that has twice as many blacks as the average community, you will have a ratio of two and you'll have a darker color than if you have less. What you see is African-Americans are still relative to national average, still located largely in the southern Atlantic states. So when we think about, when we put these two last figures together, what we're getting is blacks and whites are going to different hospitals. Hospitals are organized by communities and so we're basically saying blacks are going to hospitals here and whites are going to hospitals here, roughly speaking. It's a geographic distribution problem. And the hospitals that blacks go to are lower quality. So this table is giving you 90-day mortality after a heart attack by the style of heart attack patients who are African-American admitting hospital. So we're turning things around and we're saying these are the hospitals that are mostly black, that have a high rate of black discharges and these are hospitals that have low rate of discharges. A higher number is a higher mortality rate so a higher number is bad. And the blue versus purple bars have different levels of controls for covariates such as age, race, sex, sometimes other hospital ownership and treatment characteristics. So in some sense the purple bar, this comes across as blue and purple to me, I guess, or purple and red, but these right bars are ones that have more controls in there. But no matter what you do, you see that basically hospitals that deal largely with blacks have much higher mortality post-heart attack than ones that deal with whites. Now you might say, okay, well that may be true but what's the black-white gap within the hospital? So are the bad hospitals treating blacks and whites differently? Are the good hospitals treating blacks and whites differently? Let's answer that question. So it turns out black hospitals are worse for all races that go there. So now we've changed things around and we're going to have the light bar be non-black and the dark bar be black and the difference between them says for a hospital in the second-death style, that is to say generally treating more whites than blacks, very few blacks, the difference between the bars and the difference is that within disparity. And as you go across, you're getting the between disparity. Okay? So I want to know if the problem is blacks are going to worse hospitals or whether it's any given hospital is treating a black person worse. And what you find is you find sizable differences as you go from predominately white hospitals to predominately black hospitals. Again, the outcome is 90-day mortality after AMI. So for example here, you're looking at the mortality, you go over here, you're getting around a 24% mortality. That's a meaningful difference. Okay? And you get a similar story for ambulatory care, right? This is not just inpatient. So here what we did is, and again I want to credit Amitabh for doing the brunt of the work here, you take the percentage, so here we're, you take Medicare data, you are looking at outpatient diabetes care, you group ambulatory facilities into three tiers of quality. The lowest tier, the middle tier and the highest tier of quality. And by quality I mean the percentage of recommended diabetes care that's actually received. Okay? So in the highest tier you're in the high 70s, low 80s. In the lowest tier you're in the high 60s, low 70s. In terms of the percentage of recommended care the patient's actually received. Within each of these quality tiers, by the way it's harder to do geographic matching for ambulatory care. It's easier to do for hospitals. That's why I'm not doing it just on that basis. I'm doing it on quality. I'm saying in a low quality hospital what is the difference in how a black is treated in terms of percentage of care they get, percentage of recommended care they get versus a black. Whites are getting, or non-blacks are getting the lowest quality facilities they're getting 71% of the recommended care. Non-blacks are. Blacks are only getting 65%. And there's a gap also here even at the highest one where the non-blacks are getting 82% and blacks are getting 77%. So there is a within disparity but there's also a sizable disparity across the different tiers. And the important thing to remember is 50% of the blacks are in this tier but only a third actually of whites or non-blacks are in this tier. So the real problem is you're sampling the worst hospitals the most and that's explaining a large portion of the disparity. I misspoke, not hospitals. Ambulatory care facilities for diabetes. Again, a similar story for hospitals and for outpatient facilities. So far, what's the punchline? So far between differences are a large source of the disparity. And in fact, if we wanted me to put a number on it I can do that for hospitals. For hospitals, if you take the black-white gap in health outcomes on various different measures, say 90-day mortality after AMI about 60% of the gap can be explained as a between gap. That is to say, if we eliminated the difference between the hospitals that blacks and whites went to the gap would fall by 60%. Okay? So the rest is within. Again, I'm not telling you why there's a between and why there's a win-in, just the allocation between the two. That's what we have decent data for right now. Okay, so now the question is turn the page. Do I think that civil rights laws can help? Well, to address this question let's turn to Title 6. We're going to look at Section 601, the main part and it says in no uncertain terms no person in the United States shall on the grounds of race, color or national origin be excluded from participation in, or be subject to discrimination under any program or activity receiving federal financial assistance. Okay, this doesn't just apply to a small group of facilities. This applies to a lot of facilities. Anybody receiving Medicaid or Medicare money is subject to Title 6. But you would think that this is a huge stick, right? And the larger the Medicare or Medicaid budget grows the bigger the stick. The more facilities they get it, that's bigger. And it was a huge, huge success. Right, so the civil rights laws passed in 64 and 65 you get Medicare. And the first thing that happens is the government agency it's no longer, it wasn't HHS back then responsible for administering Medicare said that you only get Medicare certification if you have integrated hospital facilities. And this was about as big I mean we spend all our time in law school thinking about the great civil rights successes in education we're studying the wrong things folks. This was a huge success. In four months you got a thousand hospitals to integrate. Unbelievable. And the success in terms of improving health outcomes just, you know, it's amazing. Right, it's just shocking that it wasn't done earlier. Six thousand blacks were saved between 65 and 75. But if you want to get a sense of it through numbers, I like to look at figures here's some figures. This is trend, and by the way Medicare's targeted old people I'm going to show you infant mortality results associated with integration. So this is trends in infant mortality by race between 1950 and 1990. This line is the one for blacks. The middle line is the one for whites and this line is the gap between them. So this is kind of like the treatment effect. Look at the huge so you've got pretty stable gaps up until 65 and then after 65 huge drop off. I mean this is unbelievable. Right, we can only dream about these sorts of things. The sad thing is that we needed this. Right, one would like to achieve that sort of result again. Okay. So do I know that certification is responsible? This is a great study by Doug Allman, Ken Che and Mike Greenstone back in 2007 when I saw this my eyes just popped. So what they wanted to figure out was what was going on? After 65 was it something else or was it something specific about Medicare and certification and thereby specific to the civil rights laws? And so they said let's just figure out if certification was the key thing. Let's look at the number of I think they're looking at time relative to Medicare certification of a hospital in a county. These are years and these are years before and after. And what you find is before you're certified, remember in order to be certified you've got to integrate. So this is really just a proxy for the date that you integrate. You look at all cause, this is by the way these are black-white differences so it's like the treatment effect from the last graph. And you're looking at all cause post neonatal mortality and preventable cause post neonatal mortality. Stable gaps up until you get certified i.e. until you integrate and then you integrate and bam! Amazing reductions. This is why the civil rights laws inspire hope. Now that was a great accomplishment so then they went on to by the way gaps didn't die out, there's still gaps. I know this is just infamortality but so you want to start addressing this. So you want to look at other sources of disparity. And so we turned we said okay not by litigation by the way just by regulatory action we had a huge reduction in racial disparities. Now let's look at what's remaining. Let's start using the litigation tool and we're going to go after hospitals that are denying admitting privileges to black doctors. We're going to want hospitals that have prepayment requirements for black so white person comes in doesn't have to prepay but a black person comes in has to prepay. And maybe we'll also talk about hospitals where this happened with schools where people took their kids out of schools and sent them off to suburbs where there happens with hospitals as well. Hospitals may leaving inner cities where there's high density of minorities and going to suburbs where there's a relatively low density minorities to white suburbs so basically the hospital version of white flight and we started challenging all these and we were extremely successful here but we failed to stop the relocation using Title 6 and the question is why? And by the way I'm just giving you history right now. So in order to answer this I think you have to think about how Title 6 cases are prosecuted. Title 6 cases have three stages okay have three stages and here P is plaintiff and D is defendant. I lapse into these abbreviations because I'm a lawyer. So plaintiff the first thing that happens, plaintiff goes to court presents a satiate evidence of discrimination you discriminated against me or I say you've done some activity and I've noticed that blacks are doing worse than whites are at your facility. So I don't have to show that you discriminated I could show either that you discriminated or there's just disparate impact and so this allows in disparate impact or disparate treatment into court supposed to make it easier lower the hurdle but once you provide evidence of disparate impact then you go to the second stage and the second stage the defendant gets to come up and say actually there is a legitimate reason for my activity now if they can't come up with a legitimate reason then you're done but if they can't come up with a legitimate reason then you go to stage three the burden shifts back to the plaintiff the plaintiff has to show that the reason one of two things the reason that the defendant offered is pretext for discrimination it's bologna it's black same goal without having a disparate impact on minorities so let me give you see how this would play out in these two cases so in privilege and prepayment cases the defendant lost in stage two they couldn't give a reason for why they needed to deny privileges to black doctors there's no reason why it's got to be on black you can give quality reason but you can't do it on the basis of race if you think that there's different degrees of creditworthiness do it on creditworthiness but you can't do it on race those were easy to win because the defendant just couldn't come up with something in stage two the problem with the relocation cases is the defendant could come up with a stage two response what seems like a legitimate reason and the standard answer has been financial distress I can't survive it's not to the suburbs or I need to move into a more profitable section of the city or I need to just close my facilities down in the minority community it was very difficult for the plaintiff to come back in stage three and say there's a way that you can become financially viable while still remaining here and that was the challenge the civil rights law didn't give them the evidence they just said you've got to come up with the evidence that shows that the legitimate reason is wrong so that was a big hurdle and the relocation cases are not succeeding and so you can imagine a completely you can come up with whatever you think the source of disparities is and try to work through the logic of the three stage litigation process in title six cases and you might be able to see why it is that you have difficulty so imagine a situation where you've got a hospital that's doing and I'm not a doctor so I could be completely wrong with this assertion but I'm just going to come up with something so let's imagine that I say that an African American patient with angioplasty is at the University of Chicago then a white patient and the black patient wants to sue he or she he shows that in fact he didn't get angioplasty at the hospital in fact maybe even is able to show on average blacks are getting less angioplasty I don't mean to pick on the University of Chicago we're a great institution but I just want to come up with an example the question is what is the defendant going to do at the second stage what is the hospital going to defend at the second stage and they're going to say things like well in your case we made a medical judgment so we don't think that you are appropriate for X, Y and Z reasons or we thought that with angioplasty it's a little bit hard but if there's a treatment that requires some sort of compliance we looked at your history of compliance we don't really think it's appropriate and unless that patient has solid reasons for saying for sure everybody else would be giving me an angioplasty they're going to encounter problems in stage three and they might say stuff on aggregate but even staying stuff on aggregate you have to go through a lot of files right of people that are not involved in litigation and say actually you just did it because I was black even when you look at all these other people it's a very difficult litigation posture to be in so differential treatment cases would be extremely difficult to prosecute in title six relative to prepayment and privilege cases which are just much more stark much more stark so you might ask okay there's still a problem how do we respond to this do we want to improve enforcement of title six leave it as is in terms of the tax and the jurisprudence but expand enforcement spend more money on enforcement people are not spending enough either the government or individuals and I have three problems with this the first is as I told you I don't think title six can handle what I've called mixed motive cases that is to say there's no treatment differential behavior but that you can come up with a legitimate reason there might be a legitimate reason and an illegitimate reason for it as long as there's some legitimate reason you're stuck in stage three and that's just a hornet's nest explains the relocation cases not only that you've got a problem which is if you've got a hospital and you're looking at some disparities eliminated here let's suppose there isn't a decent facility in Kenwood I know this is right next to Hyde Park so you don't really need a separate facility in Kenwood but let's suppose you're trying to get one in and there isn't one there who are you going to sue so you can't increase access go from zero to some positive provider using civil rights you can only go after an existing provider you can't say there's a low quality hospital in Kenwood I want it to get better even though it's treating blacks and whites poorly and equally poorly nothing in civil rights law says a bad quality provider has to get better for everybody so that's a limitation in the way Title VI is done it addresses inequality within institutions within a provider a second problem is Title VI, actually the federal government basically substantially limited the scope of Title VI so it applies to institutional providers but it does not apply to doctors they define Medicare Part B patients payments, I should say as contracts of insurance and then say okay we don't deal with contracts of insurance that's not interstate commerce and so you carved out docs and the reason we carved out docs we thought was, well, docs are just not that important they're only 5.4% of Medicare payments in 1970 but now they're big, they're 38% and so what seemed like a trivial carve-out it's not such a big gain to include them so it's no big deal to carve them out and get some political benefits for it now that carve-out has a huge cost maybe you could revisit that carve-out but if we're just thinking about expanding enforcement without changing Title VI we've got a problem then the last thing is courts have hobbled Title VI so what are the remedies when you win a Title VI case so if the government wins you can get, for example, decertification of a hospital you can also get damages you can also get an injunction telling the facility you need to change or the provider if you could go after a hospital do you need to change your behavior but the government doesn't have a lot of money and sometimes it doesn't have the political will to spend a lot of money prosecuting these cases you might think, oh, that's not such a big deal let's have individuals go after institutions or institutions that discriminate on a between basis but the government but the Supreme Court has limited the ability of individuals to do that if they can prove intentional discrimination they can get damages but if they're only trying to argue disparate impact they can only get an injunction and this is a case Guardians Association vs. Civil Service Commission of New York the big limitation let's think about the incentives for an individual so if I got denied I'm African American I get denied angioplasty at the University of Chicago what I want is money what I want my case changed I don't just want an injunction saying future people should benefit they're getting the benefit I'm paying the cost of litigating so injunctions don't incentivize litigation in the way damages do so that's a big limitation so you might say, okay hey, just because Title VI is limited that's the remedy we shouldn't abandon civil rights laws so for example, maybe we should eliminate Stage 3 right and maybe even Stage 2 in fact you can just say you just have to prove some disparate impact or we might say let's extend it to doctors let's extend it to doctors doctors are subject to Title VI but I don't think that's going to give us as much as we think and there are two basic reasons the first is that litigation is slow, uncertain and costly that's actually three sub-reasons I've put under one bullet but you don't have to be a litigant to realize the date from filing a suit to actually getting some remedy whether it's damages or injunctions it could be months, if you're lucky it's typically years particularly if you count for appeals so I want an angioplasty today I don't want to sue tomorrow if I get figured out by tomorrow and then three years later if I get a remedy that doesn't help me I need the answer today and not only that, a lot of times when I don't get the angioplasty I might just think I didn't get the angioplasty because I was inappropriate I have to figure out that it was actually because of discrimination most people don't discover this so I think that there's a lot of people that may be suffering discrimination that technically might be Title VI but you just don't know enough to actually sue and the last thing is litigation is costly what I like is I don't get the angioplasty it's because I'm black I want you to spend the money to give me an angioplasty I don't want to spend three times that amount of money on a lawyer to then get you to give me an angioplasty it's a very costly process it is not that look, I love lawyers I am one but I don't think that's the best way to solve social problems paying them all the time I'd much rather just provide the better care if we could figure out a way to do that the other problem is and this is way more fundamental than litigation even if we can improve litigation you get a lot of settlements you could have just a much more effective litigation process you still have the problem that this is an inadequate penalty and here's the reason why and this is why I suspect that Title VI was successful early on and not successful now if we look at the penalty of certification or decertification I should say that's a good answer so let's suppose that University of Chicago is not providing angioplasty sorry, I got to pause for a second in general we think we overuse angioplasty so I'm not sure it's a bad thing for people to get less angioplasty but just roll with me on this example there's a different line of research that I do where I wonder about that problem but I've chosen it so I'm going to stick with it so if the problem is I African American patient is not getting angioplasties the question is does decertification help as a remedy and the answer is probably not because what decertification does is reduce the amount of money that the hospital is getting leaving both blacks and whites at the hospital worse off okay and even if I got damages by the way that may help me but what about the other African American patients that are going to this hospital now there's less resources available for them and another way to phrase this this statute redistributes money from let's say non-black patients to black patients it doesn't increase the amount of resources at the hospital overall it doesn't cause an outside source to pump in money now if you're in a world where hospitals have tons of extra funds right so you're in the 60s and early 70s when the federal government is funding huge capital investments by hospitals to deal with the fact that you've got this increased demand through Medicare and what not then hospitals are flushed with funds and you can say oh yeah take a dollar that you're giving to a white and give it to a black person and it's not the big deal because it's the return on the white person is lower than the return on the black person but now you're in a situation where that money is not coming in we're trying to cut back expenditures hospitals are in dire financial straits or relatively speaking now you're taking a poor hospital and you're saying let's take money from the white person give it to the black person in that hospital right you're hurting the white way more than you could have in the past you had a lot of funds in any case what that result might be is that you're giving inadequate treatment to both okay so it really works this is a good strategy for a cash flush environment it is not as good a strategy for an environment in which you have a dearth of resources okay in part I'm saying it's not just a problem of targeting it's a problem of targeting and resources so here are my alternate revenues some of them are not mine they're being done by other people so take the ACA the ACA may potentially reduce disparities in a big way simply because it is providing access to access to health care and health insurance for low income individuals expanding Medicaid and providing tax credits okay and if it's true that African Americans have a higher rate of poverty or a lower average rate of average income then although it's a race neutral statute it's going to disproportionately benefit minority communities okay now you have to be careful though remember how I said the disparities are in some sense driven by the rate of return that you're going to get it's not enough to just to give access you've got to be a little bit careful about the other things that the ACA does for example if the ACA is trying to come back on reimbursements by Medicare to doctors then doctors are not that excited about Medicare so it uses a lever to provide better care to blacks so you want to not just provide more access to insurance but also make sure reimbursement is there so there's a financial return to taking care of African Americans oh the Affordable Care Act sorry so it's the big health care legislation from this year Affordable Care Act it's the PP ACA and PAPACA is an odd sounding acronym so I use the term ACA other people prefer Obamacare I like the ACA but it's the big bill passed earlier this year okay now there are other things that are in part due to the ACA in part due to just reform and health care generally that might help so anything that lists the increases or lifts the average quality of hospitals like for example affordable care organizations vertical integration between hospitals and doctors if you've drunk the Kool-Aid and you think that's going to improve quality overall then it should improve quality for blacks and whites and if it has a bigger impact on lower quality providers then it's going to disproportionately help African Americans relative to non-African Americans so that's all great but I think or at least I strongly suspect that the best remedy may be to actually target the worst off the worst off not by raise the number of hospitals in the country and if you target the lowest quality hospitals in the country you're going to disproportionately help blacks because blacks end up at the lowest quality hospitals in the country and so let me give you some kind of preliminary data on this right so before I argued that between disparity has a larger impact I gave you a 60% number this is just another illustration of this same sort of data except I'm going to look at both AMI results and pneumonia results I'm going to look at percentage of hospital discharges there are some hospitals that just really have only treat whites because there are no blacks in the neighborhood to ones that 80% of the population to 100% of the population they treat discharges are black you're going from white to black you see the percentage of quality for either AMI or pneumonia falls as you go to the black hospitals and what you want to think about is what do I care about the between differences you know those are meaningful I guess not I guess they are meaningful what is the difference instead between the average of being at a good quality hospital bad quality hospital that's much bigger okay and that's the reason why I say between disparities has a larger impact what I'm saying is let's send funds to these hospitals we don't have to do it on the basis of whether they're black just have to do it on the basis of whether they're low quality because they both happen to be both low quality and black you just target the worst hospitals in America you're going to disproportionately help minorities and so here's a little simulation that gives you a sense of what you can do with targeted interventions so the previous chart by the way was for hospitals I'm going to turn to I've done this experiment for diabetes centers I should say this experiment is done for diabetes centers so this is the current disparity which is to say this is the percentage of recommended care received by diabetics so we're looking at ambulatory care facilities that are treating diabetics sorry to switch back and forth between hospitals these types of facilities but if you're a white person you're going to get 77% of the recommended care if you're black you're only going to get 70% of the recommended care 7% is the disparity what happens if you eliminate all within network or within facility disparities you'll leave the whites where they are at 77 you'll raise the blacks from 70 to 75 that's good it's a 5% percentage point gain but now let's suppose instead you took that money instead of focusing on all networks whether they have low quality or high quality reducing the gap between blacks and whites within them instead you just say I'm going to target my funds at the low quality facilities and try to raise them up and specifically you have to tell me how much I'm going to raise them up let's suppose I can raise them up to 85% of the recommended care no change to within network disparities what do you get? you basically are going to shove up blacks to 81% okay and whites are going to benefit a little bit as well just because there are some whites that go to facilities where they're predominantly black you get much greater bang for your buck this way not only that if you just target the worst hospitals aiming for the between disparities you'll actually eliminate a lot of the within disparities because it turns out the biggest within disparities occur at the hospital instead of the worst so it's the worst hospitals that discriminate between blacks and whites the most in some sense and the way to see that is again we're looking at diabetes care centers this axis is going to give the percentage of recommended diabetes care received by blacks percentage of diabetes care received by whites if you're on a 45 degree angle that means you are giving equal treatment to both groups but we're finding that you're giving this is an oblong the square graph anyway but you'll see that whites are generally getting more than blacks non-blacks are getting more than blacks but you might want to say okay well let's figure out where I think the within disparities are the highest and that's the difference so each of these dots is a facility and plotting what is the fraction of the blacks what are the fraction of the whites are getting at that facility and you might say okay the distance between a dot and 45 degree angle a different degree line tells you the excess care that whites are getting that's why you see so many of these above the line whites are getting more but the difference is going to tell you how much of a disparity there is within and the larger that disparity the more likely it's going to be statistically significant you'll see the biggest statistically significant disparities they're marked in red by the way each facility with a red dot is one that has statistically significant within disparity you'll see that those that tend to be on the left hand side of all the drops so if you get higher quality facilities then if you have lower quality facilities so targeting between will get you some within now the last thing and this is the sort of question I would ask which is really throwing money at the problem you think that's going to help don't you think it's a more serious problem than that and I'll admit it's problematic so you know if you take for example disproportionate share funding for hospitals these are hospitals that are providing disproportionate share of care for poor Medicare populations one of the difficulties when the government said you know if a state allocates some money to a county hospital for DSH funds the federal government will match it one of the problems was the states would then just backdoor out some of the funds from that hospital through intergovernmental transfers and what we're typically finding is for every 50 cents that a state spent on DSH payments the federal government matched it with another 50 cents but then through backdoor measures the states took back 40 cents so in effect they were providing 10 cents but not giving a whole dollar okay so that's a concern so we got to make sure that we don't just throw money at the problem in a way that states can take and use for other purposes now obviously it's harder to do that if you're giving money to a private hospital than if you're giving it to a state hospital but let's suppose that you could control this through proper accounting measures I think you can make a huge impact on mortality here I'm going to cite research by Kate Baker and Doug Steger and now at Harvard and Dartmouth respectively and what they said is okay let's look at 28-day infant mortality rates it's very hard to read so I'm going to say the results here so I'm going to take a look at 28-day infant mortality and 90-day post-AMI mortality and I'm going to say how does that change in a hospital after they get DSH funds and they try to separate out DSH funds either by looking at all DSH funds whether or not it's diverted right? and they sometimes try to break it down into okay let's suppose that I can figure out how much was not diverted and how much was diverted and so look at effective versus ineffective DSH funds and here's what they find if you don't control for the appropriation you still find that if you give hospitals $100 you're going to prevent 0.65 deaths per thousand births and you're going to reduce you're going to avoid 9.2 deaths just off of $100 now if you break that down into effective funds ones that are not diverted to ones that are diverted you're going to get better results you find basically you're going to have 1 death 1.06 deaths per thousand births or 2 21 deaths per thousand artifacts that's meaningful in fact if you go further and you say okay does it make a difference whether I go to private or not no as it turns out whether you send the money to a private hospital or a public hospital you're getting this fund just by throwing money at the problem you're letting the hospital decide what to do with the money but that's what you're getting the main thing is to avoid the ineffective DSH because that money itself is not doing a lot of work statistically insignificant work I don't like throwing money at problems I'd like to have an incentive compatible scheme that makes sure that people are doing the right things but if you can't get that you can go with the second best and the second best might be throwing money at the problem I just want the money to be thrown at the worst hospitals I think that's going to lift reduce the disparity between blacks and whites both blacks and whites will benefit there but it's not there so let me just quickly finish up by saying civil rights laws are not a great tool targeted funding I think is better when I say indirectly targeting I mean you're not targeting it at blacks you're targeting it at poor hospitals but the important thing to recognize is there are still persistent gaps in healthcare treatment and outcomes and however you want to deal with it we got to deal with it I just think that there's a different remedy that's going to be more effective than the old remedy we used to use so you started by emphasizing how much interaction there is between the terms that might go into the model related to socioeconomic status bias, compliance, et cetera I believe that's true and then using AMI mortality at 90 days at least one acute as a window into this is interesting because that's a disease process where access is an issue when the patient presents with a chest pain syndrome they're admitted it's not any choice on that and then they're out they're almost never in the hospital for 90 days and their mortality is determined by acute intervention but their ability to get follow-up their ability to get drug therapy and a bunch of things so I just wonder how much you've been able to sort those factors out I almost wonder whether it wouldn't be better to use in-hospital mortality and I don't know if you've analyzed that because making the hospital process different if the outcome is dictated by post-hospital access would be throwing money at the wrong problem I agree and we want to be careful so part of this problem is you know I'm the bum that looks for his or the drunk guy that looks for his keys under the lamppost that's the only place where there's light and so I got to look at outcomes that I can actually measure and I will admit that 90-day mortality is a bad outcome in part for the reasons that you suggested but the alternative is for example in-hospital mortality I have two problems with that one is the incidence that the rate at mortality is much lower so I've got other problems that also affecting that sub-population I would guess and I'm not a doctor you are, I would guess is that the question is how quickly you respond to the chest pain and so it could be that the rate at which an ambulance picks up a person makes a big difference on the differential in dying in-hospital because of an AMI and so again instead of measuring the post with the 90-day now I'm measuring the pre there are flaws in each one of these I'd like to look at a composite of these outcomes one of the biggest complaints I have as a researcher is I want to see Medicare data with more outcomes and not just death unfortunately I'm not as persuasive as I'd like to be and I can't get those measurements of course it's costly to measure them but I'd like to get more so I think more work needs to be done there I'm going to just work with what I have right now to give us a clue as to what's going on the thing that I'm really worried about is if I think that the outcomes I look at the channels to which black-white gaps persist there I'm really worried so if we've got ideas for why certain outcomes supposed to others are going to exacerbate black-white differences then I've got to be particularly cognizant of that in this sort of analysis I haven't got that just yet thanks that was great Anup, this is really interesting a couple of sort of suggestions one would be to sort of jump on Jesse's question I think some of the outcomes you choose seem odd to me and I think that outcomes that might be better would be outcomes of practice as opposed to outcomes of death so for instance if you look at that baby example that you talked about it's going to through the infant mortality and the effect was going to be 0.65 per thousand that turns out to be about 10% of all of the black-white difference in birth mortality and so in other words, black white death is about 6 per thousand and black death is about 12 per thousand and if you're going to say 0.65 per thousand it's not a big phenomenon and so at some level that was a depressing number for me to see and so the the point that I'd probably make is you know Jack Wenberg's but it's only a hundred dollars right but you have no way to scale up I mean it's not the case that a million dollars would die I've asserted a linear parameter it could be concave you know Jack Wenberg's stuff about practice variation yeah it strikes me that that would be a better outcome measure for you to look at if you can I don't like that because even that literature when people say there's practice variation one of the criticisms that it's getting is that the variation doesn't show up as big differences in outcomes because basically what you're getting is flat of the curve medicine regardless of what specific practice you're engaging in that's one problem the other thing is the latest kind of cutting edge research actually as it turns out by Amitabh Chandra and another co-author basically says what's happening is you know take somebody that deals with heart condition through intensive surgical treatments versus ones that use predominantly medicine right so intensive versus non intensive what you're finding is both are flat of the curve and they actually have remarkably different ignoring costs different similar returns and the reason is because there's specialization niche and in fact what you don't want to do is eliminate necessarily eliminate the variation by saying the person that's doing angioplasties ought to be doing more medicine if you say that then they're not trained in that or they're not specialized in that so you can actually worsen outcomes so given that stuff I don't I want to be very careful about using just the fact of variation as an outcome measure I really want to I don't want to use that as a biomarker for the real thing which is outcomes but I do think you're right that and I'll get back to the to the Kate Baker study in just a second Baker and Steger study I do think that we want to look at look at outcomes and I do understand that these might not be the appropriate outcomes now getting back to Baker and Steger and one of the one of the great things here and this is why I think that there's important important role for collaboration between doctors medical researchers and and social scientists particularly economics researchers is because the economists don't know if there are procedures or other indicators other than death that are good measures of outcomes doctors do and so they need to talk to each other we can offer some statistical methodology you can offer facts in reality and so that's great but back to Baker and Steger they're not interested so beyond the scaling up they're not interested in getting at the black-white gap what they're just interested in finding out is if there's an overall effect one of the criticism that you can have for that is that you know it could be that it lowered the averages but it only everybody focused the dollars just on lights right that would be the biggest criticism I have with the throwing money at the problem issue is if we actually did think that there were some institutional biases giving a hospital more money could be that they just give the money to white patients and not the black patients so I don't think that this is a panacea that's the area where I think the biggest weakness is here the 6.5% that's why I'm a little bit careful about putting that in the context of a black-white gap because I don't know if that's just compressed everything or increase the gap I just want to be a little bit careful about that so at the end your conclusion was to devote more resources to the poorest performing hospitals but there are a lot of different ways to do that and a lot of different ways to set up the incentives and potentially penalties so I want to talk a little bit more about more specifics about how you would design that transfer system in your ideal world yeah so you can actually couple it with a measurement of black-white outcomes so pick some outcomes that you like and it could be inputs like the rate of which procedures are given or it could be outputs in terms of outcomes and you could say I'm going to give payments to the poorest performing hospitals but the degree of payments that you'd get varies depending on how much you're reducing that gap the same thing we do just switch gears for a second and think about schools and what we want to do is we want to get schools to do better teach their students a little bit better and we can not just give them money but we could give them money that's geared towards the rate at which they progress and instead of just looking at the rate at which test scores progress in the hospital context you might want to think about how the black-white gap progresses or I think even better than that instead of fixing it a gap because I don't want the gap to be addressed by taking money just away from whites what I'd like to do is I'd like to say this is where black infant mortality rates are now conditional on risk right so you knew some risk adjustment I want to see this rise if you can make it rise to you know by 10% I'm going to give you or to a fixed absolute value I'm going to give you X amount of dollars if you can raise it a little more I'm going to give you 2X or 2.5X and things like that those sorts of incentive schemes I think are good if you want to make it you want to make sure that they're not just taking the money and spending it just on the whites but I don't think that that's the I mean here's the one way to answer how important that sort of gap is and why you might think that at the poor hospitals I want to work even harder on making sure that it's incentivized towards helping blacks and not just helping everybody look at the black-white gaps that's the QMI oh sorry I'll go back to this is that black-white? no one more so here's what you want to ask you want to say okay do I need to target make sure that the money is just sent to these hospitals or do I need to make sure it's sent to these hospitals and the black population within these hospitals but you want to address that question you want to say okay do I think the black-white gap is largest at the worst facilities and this chart tells you no there's another chart that's going to tell you yes so let me explain what this chart says this chart says basically if you look at the black hospitals predominantly black-serving hospitals this is the gap if you go to one that's not predominantly black-serving surviving some more whites the gap's very similar so the key issue is if I can just send money here to lower both things I'm going to get a lot of the work done I don't need to disproportionately say if I send it to them I need to focus a lot on the black-white differences and this is this chart which says that if you look at the worst hospital this is now by the way the last one was hospitals this is ambulatory care facilities treating diabetes here you're finding the gaps largest at the lowest quality facilities so if you're targeting money here and the gaps are the biggest within gaps are the biggest here then you might want to do kind of a more nuanced scheme where I'm not just throwing you money but I'm saying I'm going to give you money you should not be penalties you see it should be pretty obvious why it should not be penalties because you've already got a low performing facility by saying I'm going to penalize you more if you don't provide help to the blacks you're undoing the money that you gave them they're already financially constrained in some sense that's what the civil rights statutes do they penalize the worst hospitals but this is a hospital without the money in the first place so that's not helping the blacks at those hospitals what you want to do instead is reward gains in reducing the black-white disparity or just raising the level of quality for blacks but at the worst hospitals does that make sense? so that's the kind of nuanced answer I would give I have to think a lot more about mechanically how I'd implement that we're just now starting to seriously experiment and I'll put seriously in quotes with pay for performance within Medicare we've got to do a lot more of that I think that pay for performance is generally a good idea although I want to be a little bit careful about that I'd like to see some more experimental data pro side rather than the con side but I want to see some more stuff before I start saying let's do pay for performance on black-white it's very important that you get your outcomes right the metrics we learned this from No Child Left Behind so everybody believes exactly what I said this is fantastic thank you very much