Transcripts For CSPAN3 Dr. 20240703

Tonights speakers babu jinnah, md, ph. D. Is an economist physician and the Joseph Newhouse professor at harvard. Joanna, the freakonomics m. D. Podcast, which explores the hidden of health care. Christopher worsham, m. D. M. P. H. , is a researcher, pulmonologist and Critical Care physician at harvard. His research and writing have been published by the new england journal of medicine, the New York Times, the wall street journal, and the washington post. They are joined in conversation tonight by emily oster oster, a New York Times bestselling author whose books include expecting better crib sheet and the family firm. Last year, she was named the 22 Time Magazine hundred most influential people list, joanna and worsham presenting their new book, random acts of medicine. Why do kids born in the summer get diagnosed more often with adhd . How are marathons for your health even when youre not running . Would have and salesmen have in common which annual event made people 30 more likely to get covid 19. As a University Chicago trained economist and Harvard School professor and dr. Babu jenna is uniquely equipped to answer these questions and is a Critical Care doctor at Massachusetts General Hospital who researches health care policy. Christopher worsham confronts their impact on the hospitals sickest patients. In the singular work of, science and medicine, jenna and worsham show us how medicine really works and its effect on all of us. Were so pleased to host this event here at Harvard Bookstore tonight. Please join welcoming babu Jenna Christopher worsham emily oster and. Okay. So is. Is it on . Is it. No my. On. Your. Am i doing it right . Oh, yeah. There you go. Yeah. All right. Oh. Oh. Wasnt okay. Hello and welcome i am extremely excited to be here with bob and chris to talk about their book. We talked about how to what to do in this conversation. Bob told me that his goal for it to be entertaining. So that is what we are going for. Im here to tell you. So the gist is, yeah, just be clear to everyone else, okay. So. So this is a book about these sort of sometimes random moments that determine outcomes in medicine and. More than that, its about how we can use those random moments to understand deeper questions and to generate information about. Causal relationships. What i think is so and unusual about you, too, is that youre sort of uniquely good about generating ideas out of your own experiences. And ive known barbara for a long time, and i know that not all of his ideas are generated in this way, are good, but some some of them are good. And thats why theres a whole book here, which is a really book. So i actually want to start before we talk about the book, i want to start with a challenge for the audience. You can sort of think about it in your head. I want to think i want you to think of it an example from life where something seemed sort of at chance away, something you encountered, where you thought if this had gone a different way, everything would have been different, that theres theres an element of randomness in the world, that you might be able to use to learn something. So ill give you an example. Show me more concrete. So i work a lot on. When i was pregnant with, my daughter, i was told you should arrive in at the hospital right midnight because then youll get an extra day, the hospital. But of course like if youve had a baby, you know that you dont just like pick. And so people get their. 1159 1201 that pretty similar but theres an extra day in the hospital. So think about sort of your version of that. And at the end were gonna come back in a lightning round and were going to generate ideas for book two with, with these guys. All right. So strand book, i want to start i want mainly want give people a flavor of, of the book. So im actually going out some of my favorite chapter. So i like to run my, my favorite chapters and one about marathons. So i want you to start with a random thing bob would tell us, like, where did this idea come from . Before i do that, can i just make one observation . Dan said, there was an event that led to a 30 increase in covid 19. That was not a Harvard Bookstore book event. So thats chris. Chris, the Critical Care doctor that his medical advice my familys in the room. My wifes in the room. She ran this race, called the to remember and its because of that title that i remember the race. It was our first time running this kind of race and it started the seaport went through general and beacon hill into cambridge and then back to the seaport. And she wanted me to watch her on the race route. And i said, okay. So i was down store and was going to park at mass where chris and i both work and i couldnt get off store, get to mass general. So i had to turn around and go back home. And hours later i see my wife. You missed it. You missed the race. You have to the negative points of the story. But yes, i missed i missed the race. I would prefer to frame it in a positive light. Its not the race. And i didnt combine together at the time that i was planning on it. So i never, ever, ever so i go back home and she says, i tell you what happened. She said, well, what happened to all the people that needed to get to mdh day . And that was sort of an offhand comment that she made and. I thought, well, what did happen to all the people . So fast forward a few months, can i tell the finding or no . Yeah. Yeah. Okay, but theyre not going to read the book. Ill tell you the finding anyway. So we look at this only one chapter. Thats true. Only one chapter. Thats true. So we look at this. We look at ten different cities over ten years. And what we found is that when cities host marathons, roads get closed, and as a result elderly people who have cardiac conditions, a heart attack or a cardiac arrest, they cannot get to the hospital and the mortality rate for those individuals who live along the race route goes up about 15 on the day of the marathon compared to the surrounding days. And then the clincher was we got some ambulance and we showed that the immense transport times go up. So it was all about delays in care because of marathons. Okay. So i think whats whats great about this example is that it this very specific thing and it tries to speak to a bigger question. So i want you to back out and say, like, presumably we are not very much of a per se interested in marathons and heart attack. Theres not that many marathons and that is not that important. This is not that common. But what is the underlying question youre trying to answer here . What is what is the question that you want or the answer to . Yeah, i think most the best part about this study is that it is its really cool and. Anybody can understand it like the roads are closed. People cant to the hospital outcomes are worse. I think you dont need an m. D. To figure that one out. The question is, like you said, when we zoom out, what can this teach and what can other really fun, clever. Not that fun because were talking about something serious but like interesting studies like this teach us. And so what are some other things that might cause ambulances to be delayed that we could so okay marathons maybe parades kind of similar to a parade but what if interstate five collapses in philadelphia which it has . Right. You can use this study to say all well, when the road collapses, we know theres going to be more traffic. We know ambulances are going to be delayed. What might we expect to happen with with these cardiac conditions right and you can prepare for that. You can prepare for the Boston Marathon. Every marathon its not just a road closure. Its a whole elaborate thing. And any time theres any disruption, a lot of these things become predictable. And the study that that bapu and his colleagues did tells us a lot about like the numbers like what we actually prepare for what steps could we take. So i actually i have stepped even further. So for me, i whats really interesting about the study is that like there is a theres a very big picture question about how important being able to get to the hospital fast. And thats about events and changes but its also about levels. So some we close a rural hospital and it means that if you have a heart attack now its going to be an hour or 35 minutes instead of 15 minutes, and thats that is a question that is relevant. If youve never had any marathons, youve never had to break closures, nothing changed. Its just about this overall picture. And thats a really hard and this is i think whats whats interesting about these approaches in the book is thats a very hard question to answer. So let me say i youre going to Say Something about taylor swift concerts in a disruption, but you cant say that kind of thing. Right. But you think about medicine or just in life the most fundamental question we have to ask in medicine, how quickly do we need to do something now . Theres probably a lot of young parents in the room. Youve got a kid is three years old whos got a fever and a headache in the middle of the night. Do you call the pediatrician then do you look at google and wait for a few hours . Do you go to the e. R. Or youre an intern in the hospital and someone have shortness of breath . What do you do . How quickly do you need to act . But you could never randomize thousand people with chest pain to say, all right, you go to the e. R. And another thousand to say you to the Harvard Bookstore for a book event and just wait to see if it feels better. That would not be ethical right. But nonetheless, we need to know answer to those kind question. So how would you answer that question without the marathon . What is your like . Whats the problem . What is the typical way . One way to answer that question in medicine. Okay, im not going say taylor swift concert, but the problem is, is that if you look at people who have delays in care and you compare them to people who do not have delays, care you cannot attribute worse outcomes for the people who had delays in care necessarily to the delays because theres a lot of reasons that they may have chosen to delay their care. So if you want to get that causal question, youve got to find a natural experiment, some form. And weve thought about a of ones we that you could possibly do that we havent done but are maybe on the list so so if what if the oncology is under construction and someone who has newly diagnosed cancer has to wait a little bit to get their appointment, whereas if the the Oncology Clinic wasnt under construction, they would be able to get treated maybe one or two or three days sooner, depending on the cancer that might matter. It might not matter. We dont know. And a lot of oncologists will start treating as soon as possible depending on the cancers, you can wait. Sometimes you cant. And we actually dont have a good sense of types of cancers and in what situations we should do that. You have a natural experiment that kind of happens by accident and. You can learn something about. It how much time can we with . So what is your chapter of the book . Babu i mean, its all about natural experiments. What is your favorite experiment in the book . Oh, okay. I think probably one that we shared. Yeah, we both have kids with august birthdays. Theyre in here. Theyre here . Yeah you wouldnt know that. But a few years ago, i dont know if you know this, but you have a young kid, you take them to the pediatrician around their birthdays. So we had a kid who was three years old a few years ago, and his birthday was august. So we took him to the pediatrician in august for his annual checkup. And as im walking out of the pediatricians office, this is right around the time chris and i met as walk of the pediatricians office, the nurse says come back in a few weeks we, have the flu shot ready and thats because a flu shot is not ready until Early September or late august at best. And so the first thing i thought walk into that office was i wish our son had been born two weeks later so i could have gotten the flu shot in the office that day. And i went to the and chris and i talked about it. And what do we find . You come in like, oh, i got so hes like, i got a good one. I got good one. And then and we talked right and we talked about this and hes telling the story and i, you know, said i had the same thing happen. Like its not this is not unique to you. And barbara and i both have relatively schedules in getting to the pediatrician for that extra flu shot and not to mention covid, which is for shots for these kids like it just so many trips to the pediatrician. Its a lot of work and we figured if its hard for us to do imagine people who cant just like take off half a day willy nilly to go down to the pediatrician. So we were like we got a one here. And so we started looking into it and i think we found. 1. 2 million children across the us who were age 2 to 5 and we just looked at flu vaccination rates depending on the month they were born in and most kids at that age, particularly they go for their routine checkups around their birthday fact. The American Academy of pediatrics suggests that you use their birthday as a reminder to go to doctor. And we found that looking the data, indeed that was the case most of these kids do go around their birthday. So we were like, check. All right, this is not just us. And then we just based on their birth month, what is the probability that these kids got vaccinated against the flu . And we found some stark differences. So children born in may, june, july were getting vaccinated around 40, 41 of the time. Children and theres no flu vaccine available in those months. They go to the doctor. Children born in october. The pediatrician is stocked with flu shots. Were getting vaccinated around 5 of the time. So thats you, know, 14, 15 Percentage Points difference just because of the month they were born. And so then were kind of left with the question of why is this and what should we be . There are more looking at this. What should you do . Oh i mean, i think thats the issue. The the sort of most thing i was like, okay, yeah, thats cool. And i can see why bobba was so excited, why you want to allow that seems so random. I mean this it does seem it does seem random. It seems like a random act of medicine. Like what can we put on the left hand . So what did you like . Whats the outcome . Well, so. So then. So then were okay. We got something. So then we take it the next step. These kids arent getting vaccinated and. Vaccines work. They do. Then we should see kids born in the summer get more flu than kids born in the fall. And indeed did see that. And we even took it another step further, knowing that little kids are sticking their hands in their mouth and then wiping them on grandmas face. We wanted to see whether they were spreading it to older family members in their household and indeed were. So we saw that these summer kids were getting more flu and their family members getting more flu. So then that brings us to the question of what do we do about it . And i think i imagine anyone with with young kids or has had young kids has a thousand ideas of how to make it easier to get to the pediatrician. But we saw some of these in covid during covid times where, they made it easier for people to get vaccines pharmacies started allowing, Young Children to get vaccinated against, all kinds of things when they previously wouldnt let them down at boston medical center, they were actually sending idle ambulances out to households to bring the vaccines to the kids. We dont do them a lot in the us, but around the world they do a lot of sort of school based flu vaccine programs where the vaccination rates are much higher. So those are some of the ideas we had thought about. So my short answer is just make it easier like it is because you make it, yeah, bring it to the get, bring it to them, bring it, make. Yeah, that seems that seems right. I mean is an example of something were actually the outcome is already se interest which is different from the from the marathons where like we dont really care about it other than what it tells us about something so much. If you are to know. I know well like, i mean i just im very upset about your bad of marathons and im trying to, like, get, you know, get in there against so i think what what is unique about your relative to a lot of what we read in the medical literature is this focus on the and almost in ways starting with that right rather than starting with like heres the question want more often starting with like heres this random random act of medicine lets see what we can use it to, what can use it to to figure out. But and i think you guys exploit that very well. But there are that cannot be answered with this. And my favorite of these and youve ive talked about this and you guys wrote about in the times recently is is diet. So i think we are sort of in agreement that every claim made about diet is garbage can be said except for the ones involving turmeric stuff. Actually, im just joking. Im just dont thats not an endorsement. Its like going to have like a baboon turmeric brand is like looking for a partnership out there. Yeah. So so the reason that theyre flawed is because the choices, dietary choices are not random. Its the same thing you outlined with differences and in sort of how long it takes you to get people who are eating certain kinds of day, different in other ways, and all of the components of diet are correlated on the other hand, the question of what is in it a good thing to eat . What is an appropriate is a very, very important one that many of us probably everyone in this room would really like to know answer to. So is there is the only answer there . You have to do a randomized or do you think there are ways to these kind of clever identification to answer question that question or questions like where we really start with the question and look for the randomness . Yeah i mean, i say a few things. One is just stepping back. Why dont we why we have this data in the first place, right. Most that are really important to us. We get answers to, but we dont get answers to these kinds question because no private party has an incentive to generate that evidence. So its sort of a a public problem, i dont think in this case i dont think i dont agree with that at all. Oh, man. As i said, nurses have said, okay, tell me why i dont agree. So i think but i think gets at why this is a hard question, which is that that people dont make their diet choices randomly and they dont want to make them random. Well, you think it necessarily had said, right, were going to pay you 100 to partake in this diet every month that they could get that information . I think they would get that information, but they dont have any ip over it. Really difficult to get people to be credible incentives to change their whatever. Now were just like that. Okay. I mean, i think we have to answer your initial question before before we spiral out. Like if you imagine what a. Randomized controlled trial of diet would look like, like we dont we talk about lifetimes of diet, right . We dont talk about what did you this month the six months, even this year. So there are randomized trials of diets but they last months maybe a year to get a randomized controlled trial of diet. You would need to tell somebody what to eat. A science test says you eat this for the decade and that like thats not in the cards for anybody. Im not signing for that. Right. So then the question, how do we make the best of the data that we have available to us . Bapu and i, of course, say well, we need some natural experiments there are some good ones we talked about. I dont know, but might not call them good, but we can agree on that, right. So, so there was that a really interesting one that we mentioned in that piece. Youre talking about in the uk. So during world war two, they had sugar. So every household in the uk until the mid 1950s could only get a certain amount sugar in their household. And then one day it kind of and so children born before the rationing lifted grew up with a lot less sugar. Children born after the rationing was lifted. And they look decades into the future and. They saw that in their forties, fifties and sixties. These were continuing to eat sugar. Based on the sugar they had when they were a child that those habits stuck around so that was a really unique opportunity to see decades worth of dietary choices. They had higher of diabetes, they had higher cardiovascular disease. But thats a very unique opportunity and those things dont come up often. So were kind of left with making lemonade of lemons with some of these observe studies that that so are not worth the paper theyre printed on. Yeah. And i think that the challenge there for me is what do we do then . So theres no clever i mean, i love the, i love the sugar thing, but as you said, its sort of its limited. Its like one example. So we have we want to have we dont have randomness we do. We just rely on non garbage or do we say nothing . I would say the following. I mean, im curious to get your thoughts. Do you think weve adequately ruled out that randomness is impossible to assess here . Like, do you think . I mean, i dont think that nutritional researchers spend a lot of time thinking about natural experiments. I mean, i dont know. No, i mean, i think so actually. Think that the for me this idea of of sort of childhood differences is the most interesting. So we see. So you have this example of sugar theres also some research around habit formation about what grain you like in india that sort of people who grow up in a wheat growing region, a rice growing region, develop a taste for that particular carbohydrate and actually are willing to effectively consume fewer if they move to a with the other food, they will end up consuming fewer calories they than they really should because. Theyre willing to pay more for the preferred that they like before system work. Bye bye. I can add my teeth as an economist so i think this idea of habit formation in childhood as a way to look at variations in dietary choices probably under probably underutilized. Yeah i mean weve talked about scarce hours for example you go the doctor and youre like prediabetes or close to having high cholesterol someone whos just above that threshold just below that threshold may in theory change their behavior practice. But in practice, dont. Yeah, but i switched to oatmeal just to say mean im just a data point of one. But but i think he hasnt complaining about. Thats right. Thats right. But im curious i mean we could think using observation date and try to bound how much we think selection bias is in driving the results but i dont think that bad bad findings are useful for personally because it leads us to do the wrong thing. Yeah, no, i agree. And i think theres theres like a sort of fine line, especially in a space this where you kind of have a sense that theres some like if i put the diet in front of you, you could be like, know im a doctor, im telling you like that of the table is better for you than that other side the table. But looking at any individual piece of it is very difficult. Yeah. And in clinical practice, i mean so much of our is not just about people dont look at a plate of food and say, you know what, this doing for my body, theres culture and, you know, Family Dynamics and, history wrapped up in the food on your plate. Right. And so even if we did have all this great data telling people to general signs of Human History that led to the food on your plate is actually really really hard to do. So even when we know you have diabetes, youre eating way too many carbohydrates. You need to cut back on them. Even though i know these are a central part of your diet, its really hard to do for, i would say, the majority of people. I think why theres so much interest in things like know have some chia seeds and everything will be fixed because like that feels very doable. Well, you could tell me this. Im in a tv she shared this news now my life is crumbling. Yeah. All right. So lets lets get some other ideas, not just for me, okay . Im going to take anyone, any randomness that they want to. Random ideas. Yes. I had a baby in april. And the randomness of when i produce the milk versus something is totally random. And i dont know why anymore. But now i can i can pump sugar. I dont know. It feels totally random to me. What can you do with that . I, i look, i spent a lot of time with the mandela pump part, so thats where my expertise is. But it stops expertise stops there. But it does. You remember that study that we talked about, a ago that there are some questions about whether you believe it or not, the the uk is it. Yeah. You say it. All right. So. Okay, so you said something about breastfeeding. So here theres a study and we can talk about whether or not it makes sense or not but the basic was this is based the uk and they looked at lactation consultants in the uk consultants tend not to work on the weekends. And so what they observed was that women who delivered arguably by chance on the weekends had less access to lactation consultants and then were less likely to breastfeed. And then the researchers looked at the outcomes of those children whose moms were either born on the weekend, the weekdays under the idea that theyre otherwise similar, but for that single difference of access to lactation and then the continuation of breastfeeding. And so they found some differences. I dont know what do you think about that . So some empirical problems in the study but but as a as an approach i think its good and it sort of gets into this space of like theres some sort of variation that then the thing thats driving is continuation, which i think. Yeah, yeah. I mean you could take Something Like Something Like the mothers level is a predictor of how much milk there will be. So lets say its a really hot week. Give a baby in the summer its really hot out moms a little bit dehydrated. That baby might get of randomized based on the weather to getting less milk and that baby might be more likely to get more formula. And so then you have a natural experiment that says that can look at breast milk versus formula based on how hot that summer was compared to a baby born two weeks earlier or. That same time, a year before or a year after this all of that people think thats really thats very excellent. Okay. All right. What else we that was a good one. You guys had like 30 minutes. There. Yeah. Yeah. Oh, wait, wait. My grandmother. Yeah. Okay. All right. Heres random idea. So i am an older millennial. I was one of the, you know, first or second years of millennials, which basically makes me an it person, regardless of what my background of study is. So i now find myself in it for pure, random fact that i was in early eighties and i was the oldest person, so i had more knowledge about the subject matter than the people younger than me who the same amount about computers as me. But i have more computer than the people with more business experience than me, which has made me into an it person with no i. T. Background. I got some thoughts. First of all, i thought indian. So i thought the same thing about the 18th connection. Id be honest. You like, can i say that that also when hes hot, hes also gen x and am an older millennial and i have fixed his computer. Yes, thats right. But so the first thing first thing i thought when you were saying what youre saying is how do we figure out the causal impact of a career path . Right. So if you want to say what is the effect of going into i. T. Versus english versus economics, the people who select into those disciplines do so for a reason. They choose those ones, so you cant compare the outcomes. Educational wise people who went into the different parts. So youre searching for some sort of experiment where someone by chance enters a certain area. Thats what happened in my case so yeah biology i like chemistry researcher a bunch researcher and i am now in i. T with a chemistry. So could happen. No, i did not try to get in. No, thats my point. Just they need to be there because im the only person who understands youre older than me and understand they like get paid time. But it could be as random as you go to college. And there is a particular professor, a great teacher and by chance he or she is teaching that quarter. You you know, you didnt know that when you selected to take this course at that time but you happen to get access to that professor that changes the career that you go in or theres some interesting in economics right like if youre above or below a specific cutoff in the gpa, lets say 3. 4 versus 3. 5, you are allowed to do certain things in of what you study as long as you believe that cut off is pretty random, then you have these people who are quasi randomized to entering economics for a different discipline, and then that allows you to uncover whats the causal effect of studying one thing or the other. Yes, of course i like that. You dont like you. You dont dislike. I know. I like. Oh, ill take that. Ill take the day that i like it. All right. Yeah. How about. Yeah, yeah. How about government . Local, state, federal. Depending on where live and what the policy of massachusetts versus mississippi be is, you may have a better now, Better Health care better access to health care, maybe even better payment for your health care. So theres randomness that be addressed politically, but not. Yeah, yes. I mean, i would say within economics, the sort of state level policy variation like thats why were employed like empirical economists are state level, you know, state level policy variation is like our is our bread and butter because massachusetts got thing. Rhode island has another thing. And then you want to argue that like just over the border, its kind of its kind of similar, but you guys dont work much, that stuff. I mean, one of the challenges of that is the assumption that you have to work with is that massachusetts and rhode island are essentially the same other than this policy. If want to look at that specific policy and you can im not an economist i dont know if thats a point of pride or something i should be jealous, but well figure that out later. But right. Theres a lot of assumptions that go into that and some of them you can verify, some of them you cant. We have we often rely on making assumptions that we cant verify but sound reasonable. Right. In health care, we can actually we oftentimes can look at the ninth floor versus the 10th floor of, you know, the ninth floor surgery, surgical has one policy. The floor has a different one. I mean, anywhere where they are basically the same. And you can at slight differences is where you can draw cause and effect type inferences them we actually i think we have oftentimes an easier time doing that in health care because on the broad scale the patients that enter the Emergency Department the first week in october are going to be pretty similar to the patients who the Emergency Department the second week in october. And if we think theres some policy change that happened between those two weeks we have sort of that that window there. So in many because of the amount of data because of the granularity of data have with health care, we can do a little bit better often times. Then its across lines or across counties or what you. But i think it does get you to sort of a little of what we were talking about with diet, which is in a lot of these cases, the policies are of substantial interest. And so we are sometimes willing to give up a little bit of this granularity or precision in our estimates because we actually really want to know the answer whether this state level policy matters, especially if the differences are big enough, then you might say, well, we know that this isnt a perfect comparison, but that being said, we see a real difference here. This is really important. Or amitav one more. And then were open for general questions. Yeah. Wait, you had to wait for the boom. Yeah. So and harvard dining halls. We have, like trays that allow us to put like three or four plates of food on at once. I visited my brother, and his college doesnt have trays. They you have to get each plate stand, which is annoying, but found myself eating less food there because didnt want to make multiple trips to the surgery. So i was wondering if theres less food waste like colleges or places that is like obesity. Yeah, yeah. Well, you know, thats good. Yeah, i thought so. I thought about the obesity question because where i went to college, we had to pay per item and then i had friends who are to places where there is like a buffet style. And i always thought that that would have an effect it just turns out its difficult to get for every school on their payment policy is for food but that might be you know we talk about the freshman was it freshman 15 or up at that. Yeah we do about that. Well people talk about that so that might be different in places where its sort of fee for item versus a but that feels like something you could expect like that feels like an interesting experiment on paper there if you can convince them you just the thing is that i couldnt collect the data but. I know someone who could collect that data. Yeah, im looking at, lets say, you know, one day they, you know, the trays are getting a little ratty. So buy a new set of trays. These ones only fit three plates. Right. You got the week before for plates. The week after three plates. Yeah, this is harvard. These trays dont look ratty, by the way. I just. I. All right, so i will take other. So keep looking, and you can come up afterwards and tell them youre great ideas, but can just take some general questions about. Yeah, can you speak more about the Data Collection . I dont always hear in the podcast, but for example, with the ambulances showing up. So do you have do you get permission, you know, which things are publicly available . You know if you have to have it, what are the connections necessary we can be curious person get access to some of these things if theyre you know state local whatever level. So theres a variety of different types of data, but probably workhorse for what chris and i and and others work with. Its called insurance data. So any time you go to the doctor or youre hospitalized the doctor bills an Insurance Company, the hospital bills an Insurance Company to. The Insurance Company has a record of that information and they know a lot of things. For example, we do the pharmacy, you fill a prescription for medication. We would know what prescription medications. Now, all this is anonymized. So i dont know that this particular person filled a prescription, but you know that someone in a certain age and demographic who lives in certain area got medical care for. This condition was hospitalized has seen physicians for other conditions, has filled medication prescriptions. So thats sort of information. A lot of services, research as a health policy, which is an economist with and and then i mean traditionally how do you guys oh you cant just call them up but its not its not that difficult this thousand and thousands of people probably in this state who have access to that information. Its not as easy as just logging on and accessing it but its totally doable. Then on top of that, its really common Electronic Health record data thats becoming more commonly used because a Certain Companies out there are very prevalent and they can combine all their data anonymize and then you can get access to it as you know, as long as you have a purpose to do it. You signed the data use agreement saying, youre going to use it a certain way. That ambulance study, for example, uses the whole used ambulance emergency medical records that get combined a national database. Again, theyre anonymized, but that can get very detailed, so itll say this is the timeline. When one was called, this is the time the ambulance was dispatched. This was the time the ambulance got on scene, etc. , etc. So all of these different data sets get to Different Things depending on what youre looking into. You take a different data set off the shelf. We were this morning about this is a study we want to do. It goes back to the 1980s. We cant use medicare because our medicare data isnt that good going all the way back, that model. So we have to use something else. So i wonder if you could talk. Okay. All right. I wonder if you could talk about the translate of your findings into some sort of a policy, because you may do a really brilliant study. It may be published in a top journal. All of your colleagues may say that was a study. And yet, does it actually lead to sort of a change that improves Peoples Health . And what are the barriers to that translation from the finding to the actual policy implementation . You sound like reviewers this is reviewer to your review at three. Let me take a stab so my honest ive heard this this kind of question a lot and ive given it a lot of thought to answers. So the first answer is a of the stuff that i do and chris and i work on, i think we do it because its interesting to us, it may not have policy impact. And i kind of go in that eyes wide open, but i kind of think of it as being the basic science of the problem. So we didnt talk about the study that is in the book about cardiology meeting. So if you look at what happens to people with cardiac conditions during the dates of major cardiology conferences like the american heart association, i know we have some cardiologists in the room with apologize and the American College of cardiology patients actually do better during the dates of those their mortality falls even though cardiologists are at the meetings and so you might ask or someone might ask me, well, and chris, what do you do about that . Youre the is not to not have is not to have cardiology at meetings all year round maybe its a prescription but my answer would be know what people talk about less more in medicine, about how we can kind of pull back on procedures in certain patients, not do any harm. It actually might be better for them. But heres a way to actually show that. I think, in a creative and rigorous way, because we have, by the way, found that rates of stenting by about 30 during the dates of that meeting. So the policy impact for me is not clear, but it sort of tries to answer this more fundamental basic science question how do you know when more is more in medicine and how do we know when less is more . So when you tell cardio, i mean, thats that is an extremely interesting study when you talk to cardiology just about this, how do they react . They werent. They werent. I mean, you know what . Like i mean, like when you when have it as a personal attack, i can see that. But but to say like look, seriously like it looks like youre must be stenting too much. Do you get any movement there or is it just. Well maybe other people are but im not. Thats mostly cardiology. Yeah, theres at least one cardiologist in this room and maybe more they travel in packs by the way. Yeah, i think, you know, we so we were to on paper with a with a couple cardiologist actually. So thats two points of reference. And there we actually focus on these specific meetings. Were interventional cardiologist. These are the ones who do the procedures and. We saw the same sort of patterns there and. We are also able to unpack a little bit more where procedures are that are different and where might they be overuse. And theres a particular type of heart attack where its sort of a little bit more gray. Whether you should do this procedure. And thats where we found the sort of overuse. So, you know, i they push back and say well its hard to know who who this applies to and who doesnt apply to too which i would say, yeah, that might be true. But the data tells me that youre figuring it out during the day to these meetings. The outcomes are Getting Better and they figuring out which patients not to intervene on there. Worst kind of subconscious if that makes sense. Yeah. Questions of oh, theres one in the back. Leaves us alone. I dont know. Yeah. All well go in the back and then well go to you. I feel like covid produced millions of natural experiments, both in the Health Care Industry as well. As, like, innumerable other places. Im curious like, just broadly, what you might be thinking about in terms of like disruption, health care, like how that affected things or really anything else about that. You guys tell about the this what . So you can why dont you tell them about the birthday . Because i think is so one of the things that we rely upon natural experiments is the event or circumstances that led to patients going one way or the other is unrelated to the condition have right. So with covid, that was disruptive in a lot of ways. And so it actually it might be as many natural produced as we think because it was so disruptive we probably wouldnt be able to use it as well to study things like stuff that i do, respiratory failure or you pneumonia in the hospital because covid 19 actually complicated all that but theres actually already literature coming out looking at well cancer screenings got delayed or pediatric schedules got all messed up because people werent going to the doctor some of those things. Were not going see the results of those accidents. Medicine for some years, decades down the road, there will some interesting studies looking at colon cancer, Breast Cancer rates because of this disruption, lot of them are going to have to wait a while on bart who had did really cool study about actual 19. If you want to talk about that one just take a question and tell them the story do we have time . I think we have okay this we two kids who are. One of them was turning five or six a few years ago and we were thinking about doing a zoom birth. Everythings yeah exactly. Who can do 515 next thing you know theyre in college. Yeah. Anyway and we were thinking about whether or not to host a inperson Birthday Party or zoom Birthday Party and because we had some friends who were in this room who a zoom Birthday Party with someone named tim. Thats name. And hes amazing. By the way, as a magician, we decided to go zoom, but it did make me think, what if had chosen to have an inperson party . And that sort of spurred this idea. Well, look, early in the pandemic, we were trying to figure out whether or not small with people that you know and trust could be a source for the pandemic to spread. But its hard to study, right . Because you have to find data on who is gathering when theyre gathering, whether they get covid 19. And then you have to sort do the problem that people who gather might be different in a lot of other ways that preclude you from that. It was the gathering that caused them to get covid 19. So speaking about the data prom, its an enormous data prom. So the insight we had is maybe we could look at birthdays. So insurance companies, they know when someones birthday is. And if you at households in which a member that household had a birthday in any given city in any given week that household was 30 more likely to have a covid 19 diagnosis two weeks later than similar household in that same in that same week where no member had a birthday so it spoke to something about why we call it the birthday effect but its about the spread in people that you know and trust. That was sort of the interesting part of that to me. Yeah i like them zero zero. All right. Hi. My question is about your definition. Now, my understanding is that medicine journals like in you can only say the word causal if you do t in economics you can use recalls to describe rc to use or now to experiment to like how much of a challenge has it been to convince folks in medicine that what youre doing with these natural experiments, kosher . Have you heard of control . H so it works very well. You write a paper, you write the word effect, and then you press control h and substitute with association and then the same paper can get it. I wont use names of journals because i dont want to poop where we eat. Yeah, but you get the idea about that. I mean the and so i mean you raise a very good point and not an economist, but especially in medicine, we publish a lot of studies that really dont causal interpretations and the vast majority of the studies that get published outside of randomized controlled trials, which are also minority, dont have causal interpretations and should not have causal interpretations. E economists, when they publish an economics journals, not only are they doing the natural experiments that were talking about, but they do analysis after after analysis to. Try to check every box about what if this what if this . What if were wrong . What if we missed this . And they tend to have many more analyzes in those economics papers establishing than than we are medical, doctor, we dont have Attention Spans. So doctors have to have these shorter articles and so there are some larger reasons why you might place a little more heft, some of these economics papers than, the medical papers. That being said, you know, we try really hard to. Do all of those checkboxes to say, what about this . Do we consider this could it be explained by this . And and we try to fit as many those as we can. A lot of times we stick them the appendix. So dont forget to read the appendix because again, the short attention they make us put them somewhere else. But what we try to do as many as we can. I mean, for me, i think the biggest distinction between economics and medicine is not the Attention Span or the short and escalating. Sometimes were much to do, but because for a very long time, until quite recently economics never did randomized trials, we spent a huge amount of time developing methods that would try to do better with that was not randomized so natural experiments better control is better approaches to panel data pre this that and theres an enormous field econometrics basically about how do you learn causality from Observational Data i think in medicine because there was that well we have this thing the rc t like where we get the causality and so everything sort of either an rc t or the outer darkness, theyre like theres kind of very effort to distinguish among nonrandomized data into some things being, you know, better in terms of causality and others being worse. And i think theres some movement there. It strikes me as quite slow in getting people to to move to understanding distinction in some of the medical journals moving towards that direction. There are some statements that some of the major ones have released that that they might be rethinking that language. I mean, have you guys published in economic studies or to be focused mostly on medical journals, mostly in the last decade in medical journals, but prior to that, in a prior life in econ journals, many other. Yeah, so it like in the wake of the pandemic, a lot of and medical professionals are dealing with burnout. And so im wondering is burnout introducing any noise or confounding factors into studies or data that you might think is otherwise. Hmm. Thats a good one. I dont well, ill say two things. One, i dont think its a confounder in the type of work that we do, were giving an so one strategy that we often employ is to look at patients who are treated by different doctors by virtue of a scheduling thing. So you go on wednesday, you see tom in the ed, you go on thursday, you see tim in the ed, you go on friday, you see lisa. And so its random in respect. Now it possible that people who work particular days of the week have more or less levels of burnout . I think its possible, but typically think that those characteristics of the physicians are pretty across the time span, across the times that were looking. But theres a separate question of sort of what is the whats the causal effect of burnout, i mean, burnout. We talk about a but theres many occupations outside of medicine where burnout is real, but we dont about it in the same way that we talk about in medicine, i dont know. Its a little of a challenging area. Theres obvious reasons we would think about that. But i havent thought of a good cause of questions there. What about at staffing . I mean, theres been a lot of push on, like, you know, nursing shortages, physician shortages in the wake of the pandemic. I if theres a way to use that to figure out sort of impacts of on outcomes. Yeah, we do have something were working on now where we look at school closures. So as youve worked a lot on certain areas, close schools for a longer period of time and were looking at health care effects there. So the provider supply of nurses is impacted by a closure of a school. Nurses are 85 women in this country. And if you look at nurses with young kids, schools provide a lot of child care. So you see when school is closed, nurses are more likely to exit the labor force in those areas. And there might be Health Impacts that and perhaps burnout effects as well. All right. So going to one more question then. These guys will sign some books. Yeah. I have a phone question about the marathon. So its been written that. The Boston Marathon response in 2013 was so successful because, they were like armed and ready and patients werent in the hospital. Ambulances could help patients. So like did you balance or how do you balance in general marathons like the fact that theyre at the ready to help runners who have like heat exhaustion other problems that theyre experiencing because like for me as a marathon, im thinking more like, well, that old person was like probably going to die anyways and then say and then within the next year, maybe like im just being honest, like maybe their issue was really bad anyways and like 5 minutes delay made difference. What about the marathon runner whos like 36 and was healthy and they got to the hospital . So how do you how do you balance that . I think its a this a really important question that comes up a lot. So when we look at mortality city, the question is, is mortality relative to what . So if someone is more likely to die in 30 days, well, maybe they would have died in the past with the 35th day. So thats always important. Question maybe i say the following about the marathon study. One thing that was i say two things actually. One thing that was striking to me is that more people die because of road closures and delays in getting the hospital than died or injured in the marathon bombings. And then that the marathon bombings are so salient in peoples minds here as as they should be. But the number of people who die sort of behind the scenes is actually quite a bit greater, most my readers, someone asked about policy. Most of my research does not sort of touch with peoples lives in the way that someone other does. But i gave a talk once a woman came up to me afterwards. And heres what she told me, she says she has a problem with a column that predisposes her to bleeding and she was at her Office One Day and she had a gastrointestinal bleed and there was a marathon outside of our office. So she called the ambulance. And the ambulance could not get to her because the roads are closed. She calls her grandma. And grandma was like 75 or 80 and grandma decides to drive in the minivan on, the side of the road, literally the side of the road right here. People are moving scoops up. Granddaughter takes her to the hospital. She has a cardiac arrest in the hospital, which is very very bad. And she lived to tell me the story. So its sort of striking that these things that are happening underneath the surface, we dont always see, but nonetheless impacts on our health. And by the way, have against marathon runners nor do you write we support marathon 100 and taylor swift concerts. All right. Well on that note, its always good to end with taylor. Thank you guys so much. The book is great. Everyone should read it. Its fantastic. And yeah, thanks