Melissa Thomasson is an American economist. She is the Julian Lange Professor of Economics at Miami University in Oxford, Ohio, where she has also been the chair of the Department of Economics. She studies economic history, focusing on the evolution of health insurance and health care in the United States.
Episode Description
According to the Centers for Disease Control and Prevention hospital mortality rates in the US were on the decline in the early 2000s, even as total hospitalizations rose. This came after a 1999 U.S. Institute of Medicine report that suggested tens of thousands of individuals died in hospitals unnecessarily each year. The report focused attention on patient safety in modern hospitals. About 70 years earlier, an organization in the American South was also concerned with patient outcomes. That's the focus of this episode of Stats+Stories with guest Melissa Thomasson.
+Full Transcript
Rosemary Pennington
According to the Centers for Disease Control and Prevention hospital mortality rates in the US were on the decline in the early 2000s. Even as total hospitalizations rose. This came after a 1999 US Institute of Medicine report that suggested 10s of 1000s of individuals died in hospitals unnecessarily each year. The report focused attention on patient safety in modern hospitals. About 70 years earlier, an organization in the American South was also concerned with patient outcomes. That's the focus of this episode of Stats+Stories, where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington. Stats and stories is a production of Miami University's departments of statistics and media, journalism and film, as well as the American Statistical Association. Joining me is regular panelist John Bailer emeritus professor of statistics at Miami University. Our guest today is Melissa Thomasson. Thomasson is Associate Dean for Faculty Excellence and professor of economics at Miami University. She studies the economic history of health insurance and health care. Thomas is a research associate for the National Bureau of Economic Research serves or has served on the editorial boards of four journals in her field, and is on the Board of Trustees of the Economic History Association. She's also a co author of an American Economic Review paper on the legacy of the Duke Endowment on the modernization of hospitals. Melissa, thank you so much for joining us today.
Melissa Thomasson
It's my pleasure, Rosemary. And John, good to see you.
Rosemary Pennington
Why did you decide to look at the Duke endowments impact on hospitals?
Melissa Thomasson
Well, the project started with actually a Miami undergrad, another one of my co authors, Alex Hollingsworth, who graduated in 2010. And we were really interested in investigating the economic development of the hospital industry. And so really, you kind of have to go back to the turn of the 20th century then. And if you were to look at a hospital in 1900, it bore little resemblance to hospitals today. You know, at the time, only about 5% of women gave birth in a hospital. And those were the 5%, who were typically unwed mothers or similarly destitute, whose families had cast them out. And so you have to think about that in the context of the period and hospitals though, with advances in medical technology, we're rapidly changing over the first two decades of the 20th century. So you have this really interesting question of when did hospitals begin to have an impact on the lives of Americans, and also at the same time, what factors, you know, really led to this rise in hospitals. And the the Duke paper started, because as it turns out, James Buchanan, Duke, who we all I think we know, in Ohio, because of Duke Energy, but he really found his passion, or his his profit, at least, on inventing a machine that would roll cigarettes automatically. And that was the basis of his fortune, had been always interested in philanthropy. His father, Washington, Duke was a philanthropist. And he had amassed this fortune. And in the early 20s, when hospitals had really started to become the centers of science and technology said, we have to put more hospitals in the Carolinas. And so he set aside about a third of his fortune upon his death in 1925, to build and improve hospitals in the Carolinas, interestingly, the remainder went to Methodist churches, orphanages, Duke University, Furman college and Davidson College as well. So he had this he was interested in both education and hospitals.
John Bailer
This is this is fascinating background and context. So So you have this setting where you have this person that's, that's ready to kind of invest in, in the communities where, where he lives. Exactly. And then, so at this point in time, you're kind of looking at 1900s hospitals, and then thinking okay, there is going to be this injection of funds. And it's based on your your description here. Infant mortality was a was a huge kind of signal of kind of public health success. That that was one of the things you measured you were also going to measure mortality. Was that another aspect of of responses that you're considering?
Melissa Thomasson
Yes, so so, so true. So infant mortality is a really great measure of looking at for infant mortality is a great measure of health outcomes. And the reason why is that infants are among the most vulnerable members of our population. And so you know, in the data are also pretty well collected. Now, again, the birth registration area in the US doesn't really isn't complete until 1933. We're fortunate that North Carolina did have a birth registration area, but you can usually get data on infant births and infant deaths, and it's, it's reliably recorded. And so we looked at infant mortality. But the other thing is, of course, hospital care might save you if you're an infant. But a really good question is, if you're able to have hospital care over the course of your life, to someone who really couldn't, maybe they were born at the turn of the century instead of 20 years later, right? How can we look at that impact as well. And so we went to this, it's now publicly available with all the appropriate trainings and permissions, of course, but it turns out that the Social Security has a database called the new Medanta database that you can get information on individuals and where they were born, and how long they lived. And so you could actually observe, when these people died, we knew what county they were born in. And so you can kind of say, well, this person had exposure to a Duke Hospital over the course of his or her life, did it lesson, the did it prolong longevity. And so we were able to look at, at people who died, who had excess mortality at older ages as well.
Rosemary Pennington
So you have this infusion of funds from Duke that is going into this community to fund these hospitals, what sorts of things were hospitals doing when these these funds to, I guess, modernize themselves?
Melissa Thomasson
A lot of it too. And we talk about this, because it's not only money in the in the Duke stuff, but he really insisted that for hospitals to get the funds, they had to basically totally reinvent their bookkeeping, like he and actually invented, or at least, the people who ran the foundation, after his death, invented sort of modern accounting systems and brought those to hospitals. And, they looked at the efficiency of all the operations and and so they looked at, first of all, the number of beds, the quality of the beds, they looked at the number of medical personnel, the training, whether they should be medical personnel or not. And that's a whole nother subject for stats and stories on the quality of medical education at this period as well. And then they started looking at the technologies, right, does this have a laundry, does it have a kitchen, and then they did have an x ray, many of them don't. And a lot of the hospitals at the time were started, particularly in rural areas, by a physician who would just keep a bed upstairs in his house. And so you'll see if you go and you look at the hospital data, which aren't there, pretty reliably collected, interestingly enough, by the American Medical Association began in about 1920. And before that, you have sort of random surveys that go maybe back to around 1910. But a lot of these hospitals were six-bed, private hospitals, you know, and, and literally, it was just nursing care, it was a place where sick people could go, because, you know, in cities, perhaps they had, they didn't have anyone at home to care for them. Because people are working, maybe they were in a boarding house in a city and in a rural area, you know, for whatever reason, they just needed more attention. But it's really, you know, they're not hospitals, like we think of them. So
John Bailer
I was I was fascinated with the idea of, of where do you even get data to look at this. I mean, so you, you're, you're going back, you're taking this time machine back to, you know, to the 1920s plus or minus and tracking, how did you get the data to even start this this investigation?
Melissa Thomasson
Well, it's not it, we didn't get it online.
John Bailer
You didn't just do a browser search.
Melissa Thomasson
Colleagues, yeah, who have these datasets that are available online. And I'm very jealous, because the hard part of you know, history and economic history is, is trying to find the data. And a little bit of this was serendipity. So at first stumbled on the Duke Endowment report said annual reports, probably in around 1999, when I was an assistant professor, and I was at a conference in Cambridge and had happened to stop by Countway library and and had a look in their stacks. And so I saw that, that they had collected data on how much they spent in these hospitals that they built and the number of beds and how it changed over time. I guess that's, you know, you tuck that away, maybe I'll use that later. And then I have a wonderful, another colleague of mine at the University of Pittsburgh, and who did a lot of work in sort of early economic history, public health, health care. And he has a fascinating book called the pox of liberties, which is all about how our public health laws made us actually really vulnerable to pandemic, so I'll have to, he was before his time, unfortunately, he passed away in 2017. But he has always been interested in this stuff. And one of the things that he and I were talking about, again, like 20 years ago, was the fact that North Carolina was really on the vanguard of having public health reporting data that was published and kind of consistent you wouldn't North Carolina today might not think about like one of the more progressive states Data Wise, but it certainly was then. And so I had all this photocopy data sitting around for years with on birth deaths, and at the county level in North Carolina. And I have to say, too, that the photocopies each one of them had a pet mixture of foreigners hand in there. So I'd have these undergrads sit there and take these photocopied sheets and type them into Excel. And to this day, I have students out there who still talk about the hand data. So I had this hand data, I had this Duke data. And then Alex Hollingsworth, my co author, who is an undergraduate here, had gone to the University of Arizona for graduate school and studied with my dissertation advisor. And my dissertation advisor had mentioned, this project and Alex and I put our heads together and realized that there were a lot of interesting questions we could ask about the effect of medical care on health, and longer term health, and maybe other ways we could get at it. And so it's really one of those things where the data just came together over time, but really, by listening and thinking and being able to bring together lots of different datasets.
Rosemary Pennington
I'm sure John has a question about methodology. But you've brought up twice now. You brought tries twice. Now, the fact that North Carolina had all of this rich data at a time when many other states weren't collecting this? Why was North Carolina collecting this?
Melissa Thomasson
So I know that once once the endowment started paying out funds in the late 1920s. Part of it is that the people you know that they ended up hiring really excellent and progressive people who then pushed the state more in that direction. But in terms of why they had great public health data in 2015, I'm not sure you know, to be honest, it's but but I'm glad they did
John Bailer
Certainly give you an opportunity to explore this question and great detail. Yeah. So you've got this data set. That's that's sitting there. And you want to compare kind of what's happening in terms of trajectories of how the hospitals are staffed? And what's happening, some of the health outcomes of these communities that are impacted. So So can you talk a little bit about some of what you saw? You know, I'm not going to dive into the methods yet. I mean, I know that's, that's all Rosemary is all about the methods. See, she wants to talk to you about some of the, you know, the right modeling that you did later. But no, no, we won't do that. So let's let's talk about about some of the things that that kind of you learn from this, what are some of the takeaways before we dive in more into the methods.
Melissa Thomasson
So the first thing we learned is that the program had a huge impact, actually, in terms of, of putting it in terms of improving health outcomes, it lowered infant mortality by about 10% Overall, which is pretty huge. And, and pretty significant for the time. And importantly, it actually had a much greater effect almost double for black infants than for white infants. And that's also indicative of the fact that it's very likely that black admins had virtually no care before this happened, and then got some now it's not equal at all. And I don't want to, you know, I don't want to pretend that that Duke was this. You know, he was a man of his time, let me put it that way. But he was very strident about saying that this care should be available to everyone, regardless of race. So that was one of the things and then we actually similarly found that the effects were about a 9% reduction in mortality between the ages of 54 and 65, although not really differential effects by race, then. So just having access to that, and you know, we're not able to trace, what of that is important in terms of well, what what was it? Did you have an appendectomy, and you didn't die of that. I mean, we don't know what that counterfactual is, but that it was important. And we were also able to get at mechanisms. So one of the things that, that Duke said in the indenture of trust of the Duke Endowment, as he said, you know, what we really need here are doctors, and there have been a lot of changes in medical education. But how do you get doctors here, you build hospitals, who's kind of like Dukes own Field of Dreams. So we found another data set, the American Medical directory that lists every single doctor in the country, their year of license, their year of birth, and where they went to medical school. And we found another dataset that had this grading for the quality of medical schools. And we were able to put it all together. And we were actually able to show that he was right. When you improved hospitals, they got a disproportionate share of better doctors coming in and less qualified doctors kind of just shutting their doors.
Rosemary Pennington
You're listening to stats and stories. And our guest today is Miami University's Melissa Thomas.
John Bailer
So what are the ways that you came to these insights, I saw a couple of ideas, you used differences and differences was part of your investigation, and also kind of intent to treat. So I found myself sort of channeling kind of clinical statistics kind of aspects of the intent to treat model. So could could you talk a little bit about what is differences, two differences mean when you're comparing hospitals that have been impacted by this support and funding and those that have not?
Melissa Thomasson
So a difference in difference is a very long winded way of saying that we know that there are differences between places that just exist because of what those places are. Right? So you might have one county gets a hospital and one county doesn't and one county has lower mortality than the other. And you can't just say, well, it's simply that one got a hospital and one didn't. And so what you really say is, here's the difference in mortality between the county that got it, and the county that didn't, it, you know, just kind of this raw thing. And then you have to say, then there's another difference. So we look at like, how long did they get it? And you're able to, to basically have one county that gets it and then one county that doesn't, and then compare a difference. And the outcomes on another level? So it's, it's a way of not doing this?
John Bailer
Oh, no worries, no worries, this is a hard question.
Melissa Thomasson
Yeah. Take this part out, let me think of this is where I'm like, now, what are we doing the difference, the difference, because I got all excited about the two way fixed effects. So the difference is really, that we're comparing at different times. So if you basically said if they had a hospital, or if they don't, that's one difference. And then the county is just different, because it's a different county. And that's another difference. And so we're able to kind of subtract out the difference that occurs just because one county is different at its core than another and then attribute the residual difference to the differences because one got a hospital and one didn't. Now, that's a little bit of an oversimplification here, because what we're really doing is looking at a, what we call a continuous treatment timing. So it's more like most counties have eventually got a hospital. But we were leveraging in the fact is that there's this difference in the timing when they did and so we're kind of taking that into account as well.
John Bailer
So just as a quick follow up, did you have information on the hospitals that received the funding sort of their infant mortality before they received funding?
Melissa Thomasson
So we definitely had? Yes, we had information on infant mortality in the counties, yes, before and after. So that's what I mean, when we're saying the first difference, right is looking at the counties looking at the differences in infant mortality with and without, and just the baseline sort of difference?
Rosemary Pennington
You've said that you found that this endowment spending was able to help attract as these hospitals were improved, better quality doctors to the area, how do you measure historically better quality doctors?
Melissa Thomasson
That was? That's a good question. So you can't really look at outcomes, right? Because you don't know, you know, there's not, you don't measure who had, who saw the doctor and who didn't. But in another paper work with another Miami colleague of mine, Greg nimish, Carolyn Malin and Jarrett Traber, we had done a lot of work looking at reforms and medical education that took place in the first two decades of the 20th century, some of your listeners might be familiar with something called the Flexner Report, which, you know, depending on the historians will say, but it wasn't really flexure that the doctors were doing it beforehand, they hired Flexner, to draw attention to it. And these reforms continued. But there were big changes in medical schools in those first two decades in the fact, for example, that they required at least two years of undergraduate work before going to medical school. So in 1900, maybe you didn't even have to go to high school. For some medical schools, there's a lot of variation there. And in this previous work, we had identified the fact that the real differentiator tended to be that those schools that required two years of undergraduate medical education or undergraduate education, prior to go into medical schools are this high quality doctor. So those ended up being the better schools and the better trained doctors, and that's yet another paper.
John Bailer
I like that you looked at a variety of ways of sort of trying to quantify the impact mean, one, one part of the report, you discussed the idea that that it was a life saved for about every $17,000 spent this in 20 $17. So scaling, scaling the currency appropriately, so that this was less than the Value of a Statistical Life. So there's, there's a, there's sort of lots to unpack there. So can you talk first about what what did you mean by a Value of a Statistical Life? And then if we can see this, this more this two part questions that Rosemary wants to just smack me about the studio? So I'll come back to it that so first, just what is what is the value of statistical life? What does that mean?
Melissa Thomasson
So, you know, an economist, they're always willing to ask the ugly questions like, what's a life worth to you, John? There's likely been a lot of good work on this about thinking about and it's there are different methodologies, but it's really about saying, if somebody doesn't have the opportunity to live their life, like what is the cost of that to society, and you know, conservative estimates would be a million dollars a day. It kind of just depends on the study, but, but that's the Value of a Statistical Life that's actually used by actuaries. I'm giving an economist kind of a hard time here because you have to, you know, it's the basis for things like law Students like that person lost their life because of your negligence, you know, what's the basis for damages, and they look at people's willingness to pay to avoid risks and other ways to sort of get at that. But you know, so if it's at least a million in a very conservative way, then we know that, that 17,000 is a really, really cheap and especially by modern medical standards, where I think 17,000 is like Tylenol in a hospital or something. Right? It's really a cheap intervention.
John Bailer
Yeah, I asked that apart, because I was interested in in disability adjusted life years and quality adjusted life years, DALYs and Qualys. And looking at global burden of disease, and so some previous work that I had been interested in. And so this, this idea of saying that they they did this work, and in looked at a life saved that sort of based on this kind of intervention, this hog was funding from Duke Foundation. And then the cost of that is sort of what, when they looked at that over the total dollars that were invested, I guess that's where the 17,000 Exactly right.
Melissa Thomasson
Yeah. So we basically looked at the cost of the intervention, we're able to back out how many lives saved, given the population who was impacted, and then say, divide, right, so.
John Bailer
But just as one of the things that struck me early, it's sort of the background of the paper was the comparison of healthcare costs in 1990 versus versus 2020. And in your paper, you reported that 2% of the US GDP in 1990, went to health care. 2% in 19 9019 100%, vote no, no, jeez, Charles, I'm gonna need some help. You reported that in 1900 2% of the US GDP went to health care. But then coming into 2020, it was 20% of US GDP and 6% was for hospital health care. And so my, my immediate question, when I looked at that is how do we get to the US compare to other countries, in terms of percentage of kind of GDP are equivalent?
Melissa Thomasson
Well, we're about twice as much, for example, as Great Britain, and we're far more than any other country as a percent of GDP. And about twice as much in per capita GDP as Great Britain and some of the, you know, Germany's a little bit more expensive, but typically much, much higher. Oh, that's amazing. Yeah, no, it's it's, it's because of the way that you know, we do health care at individual health plans and, and incentives to do more, and insurance companies are paying the bills, and it's all this tax preference compensation. So there's, there are a lot of reasons for it. But
Rosemary Pennington
if if you could boil this work down to just a couple of main takeaways, we should all sort of get from this, what do you think they are? So
Melissa Thomasson
I think that one of the important things is to really think about the impact of rural hospital closures today. Now, it's a really different time, okay. They didn't have interstate highways and telehealth and things like that, in the 1920s, and 30s. And 40s, you know, which is the basis for the study. And so to some extent, it's not a directly parallel situation. But it's really clear that that we were finding, and like I said, not exactly sure what the mechanism is, but it attracts doctors to these areas, right. And there are vulnerable parts of our population who will not have access to doctors, if hospitals closed because doctors need to have hospitals to do their jobs. Right? You'd and so I think it has a certain big impact there. I mean, we're really able to say infant mortality went down, mortality at at older ages went down. And it's because of access to medical care. And so if anything, I think it's a cautionary tale of how do you think about preserving access, meaningful access to medical care? So
John Bailer
I'm curious, what's, what's the next burning question for you? Are there spin offs to this are kind of complementary projects that are associated with this?
Melissa Thomasson
Yeah, we have? Well, there's there are always some complementary projects. And I have I have some underway. I mean, one of the things that we're really interested in, is looking at this medical education data we have and and trying to identify how it impacts. So how did these changes in medical education, affect health outcomes? And it's a little bit tricky, because you have to wonder, you know, there's a chicken versus egg problem there, were they so did they build these hospitals and get these doctors in areas that that were sicker? So they needed them? Right? This is where the difference in difference comes in. Or, you know, if you could, I mean, ideally, just like, you know, an economist would like hospitals to be randomly dropped on people, and then we could measure it. But if they're not randomly dropped, you have to do some things out. And of course, in general, if other colleagues are really interested in looking more at this too, and extending it and thinking about the broader implications, it was hard to generalize, but yes, we've amassed a lot of data and thinking about how to use it is top of the enough Oh no, this
John Bailer
is this is really great, really fascinating. Was there any really big surprise to you, as you were doing this analysis, something that really kind of you went, Wow, I didn't expect that when you were doing doing this work.
Melissa Thomasson
I didn't expect it to be as impactful as it was, oh, you know, to be honest, I guess. At the time, so it's I think your point about healthcare being 2% of GDP. And, you know, so by 19 20x rays were firmly ensconced, they were starting to use anesthesia. They had no antibiotics, right, they had general aseptic technique underway for surgeries with surgeries were still really risky. And so you think well, what can what can a doctor really do? You know, some 25. It's nice. And I guess I just was really surprised that it's 10%. I was not willing to give doctors and 1925 that much of the benefit of the doubt, so surprised at just the magnitude of it. And to have that persist and really persist, no matter how we tried to explain the result away with lots of robustness and placebo tests and have it persist into older ages was also pretty shocking.
Rosemary Pennington
Well, that's all the time we have for this episode of stats and stories. Melissa, thank you so much for joining us today.
Melissa Thomasson
My pleasure. Thank you both. Great, thanks.
Rosemary Pennington
Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.