Another great article, David. Your curiosity and excitement comes through even in the written transcript of the interview. And thanks Bapu for encouraging me to be more curious about what natural experiments are hiding in plain sight.
Love the creativity. Another book to add to my ever growing reading list, but there are worse problems to have.
The random experiment framing makes me curious what it would be like to integrate this type of thinking into one of my courses. Effectively asking students to be attentive to what is around them and then leverage some mathematical skills and reasoning to try to pull an insight out that might be otherwise ignored.
Hey Josh, so glad you enjoyed! And I feel you on the ever-growing reading list. I realize I have far, far more books now than I could read in a lifetime. That said, I do tend to read a little here and there from a lot of those books, while only finishing a much smaller number, and I really value what I get. And I think Random Acts of Medicine is definitely the type of book where you can pick a chapter that appeals to you and dive in. ...I actually surf my own shelves quite a bit, and just pick something out and flip through a little, and it's kind of surprising how often it sparks something. I guess that book is on my shelf for some reason, but maybe I bought it a year ago and never opened it, but then suddenly something grabs me. ...Or partly this is me justifying the fact that I'm single-handedly keeping my local independent bookstore in business;)
Another 'natural experiment' that I feel has been under-analyzed is Medicare. At age 65, everyone has good health insurance. At age 64, some people have poor health insurance, others have none. Rates of elective procedures like knee replacements spike at age 65, as you get those who have been putting it off. But non-elective procedures go up, too, suggesting people find ways to put just about everything off. If you look at mortality, it's worse to be 65 than 64, because the mortality rate always goes up with age. However, it goes up by less than 64 vs 63 or 66 vs 65. In other words, 64 year-olds die a bit more than they should, all else being equal. (Caveat: I looked into these things about 10-12 years ago and I no longer believe I have access to the data that would allow me to revisit it. Hopefully the Affordable Care Act would have attenuated this effect somewhat, but I know people in my daily life who have, for example, $10,000 annual deductibles and thus try to 'bundle' two or three procedures in the same year to save some money, so it has not gone away).
Francis, that is so interesting, and troubling. I appreciate you making me aware of that. I may ask Dr. Jena about this observation to see if he has any thoughts to share.
Hey Srini, thanks for the kind words! I have Outlive, and have actually been reading it just this week. Finding it really interesting, especially given all the complexity he's grappling with. If you read it, I'd love to hear thoughts.
Awesome -- I just finished my first read but feel like I need another read through to take notes!
I enjoyed it and found it immensely informative. The whole book feels like a RANGE case study. Peter and his analyst team are birds that looking across what all the frogs are doing in their respective areas and pulling out the most interesting threads relevant for longevity. It’s so comprehensive and I’m surprised nobody has done this before.
There’s so much information that on Jocko Willink’s podcast, he accurately called it a textbook / reference guide.
Absolutely, right on with the "reference" comment. I love have a store of books like that. I'll have them on my shelf, and sometimes for a break just flip through a little, and get some idea spark. I also agree in a way with your surprise that nobody has done this before. Although, when I wrote my first book, I felt the same way; "How has nobody connected these clearly connected threads that orbit the same question??" I think one of the answers was that most of the relevant parties were frog-like, and they weren't — as Arturo Casadevall puts it— "standing up from their trench to look in the next trench over." The bright side was that it left some pretty obvious opportunities for birds to come along and add value. I think there are a lot of opportunities like that, waiting for birds!
Totally, I think that makes a lot of sense. I think that’s why RANGE spoke to me. I am so oriented towards the birds, bigger picture, deeper truths, the invariants between different situations, etc. I just can’t get excited about niche to the nth degree I guess but I’m glad there are other personalities who are!
Fantastic conversation that I hope, before much longer, does stretch on for 40 more questions and answers. My wife covers the orthopedic device industry so she talks fairly regularly with surgeons, and we recently moved from the West side of Cleveland to the East side so now I talk fairly regularly with surgeons (and ER doctors, and other docs) just because they're around, and these are fascinating topics I would have never considered. Will definitely be reading the book this year. Thank you, Bapu, for your research and writing, and thank you, David, for your endless curiosity.
Matt, thanks so much for the kind words! ...I keep generating these ridiculously long lists of questions, and then when I transcribe interviews, I realize I can get maybe one-fifth into them and already have 5,000 words, which would be a really long post. I'm not sure what the right length is (both for readers, and for me trying to put it together), but I'm very glad to hear that it didn't strike you as going on too long. More importantly: congrats on the move, and if you pick up the book, I hope it makes for some stimulating conversation with all those docs!
Kudos for your question about behavioral similarities or differences related to Covid and children’s birthdays! I have two more questions. First, regarding the increased incidence of cardiac events in the geographic area around marathons, did the authors of the original article exclude mortality or other serious events among the marathon runners themselves? For example, most very large running events, unfortunately have one or more cardiac events, some resulting in mortality. In addition, there may be incidences of severe heat, illness or other problems that could then cause a cardiac event. Granted, one can hope these numbers are small, but they do occur, and it would be helpful to know whether they were included in that study or not. Second, regarding the opioid prescription discussion, at least one more natural experiment did occur in the US. In the not-so-distant past, physicians were instructed that we were under-prescribing pain medication, and should be more liberal in the treatment of pain, using opioid medications. More recently (this varies to some extent by state) a great deal of education has been required for license renewal that includes use of non-addictive medications such as acetaminophen or ibuprofen for pain rather than opioids. Any thoughts?
Hi Angela, thanks for these questions and insights! One point of context that I happen to know off the top of my head: there are no deaths among participants in most marathons. Last time I recall reading a large study of the topic, I believe the death rate was approximately 1 per 100,000 marathon participants. So even the NYC Marathon should not expect an annual death. I believe running-related deaths were excluded, though. (That said, there are other, usually nonfatal but serious issues, like hyponatremia, that can be fairly common in marathons.) In terms of the opioid-prescribing, that's very interesting! I'm not sure I have anything insightful to add about that, but I'd curious to hear your thoughts on it. I'd be really interested to know whether there have been studies look at impact on individual prescriber tendencies from either the old instructions, or the new.
I don't have any information about prescriber tendencies as recommendations changed, but will be interested to see what other readers say, and will look it up when I have a chance later.
True about actual mortality, so I guess I didn't state the question very well. Those running a marathon typically get near immediate care for cardiac events, or even hyponatremia or heat illness that might lead to a cardiac event. Medical tents at almost all marathons or even some shorter distance events are well staffed and well prepared. This contrasts with those who are impacted by ambulance transport problems and may not get even early care for their potential cardiac event.
Hi Angela, Bapu here! Two great questions. On marathons, we did two things to ensure that the mortality increases were not from literal marathon participation itself. First, we looked at quite elderly people with multiple chronic medical conditions, ie people unlikely to be running a race. The average age in our sample was about 77 years old and the same findings held (a marathon day mortality effect) if we restricted to people with dementia or on dialysis, ie people unlikely to be running the marathon. So, this seems it's all about delays.
Re: the opioid question, we actually have some new research not yet published that looks at this exact question using a natural experiment. The challenge with looking at an overall national policy towards "pain as a fifth vital sign" is that it's hard to separate the potential impacts of that from anything else changing over time. But we think we have a good way to do it, so stay tuned!
I didn't want to go into too much detail in a long post, but the book does a really nice job of walking through the steps of each natural experiment to examine whether one can claim causality, or whether it remains an open question. In that sense, a lot of the book is sort of an exercise in scientific thinking.
Hi, Bapu here. Great question and something we think a lot about in our research and go into detail in in the book. It'll be fun and won't disappoint you. A basic test we always do is see whether those exposed to a "treatment" (say, a heart attack during a marathon day) vs a "control" (say, NOT having a heart attack on a marathon day) is truly random. We make sure characteristics of the heart attack patients are otherwise similar across a number of different dimensions like age, other demographics, chronic medical conditions, etc. They are in the natural experiments we report. We also do additional tests, called "falsification tests", to corroborate why we think the findings are causal. Check out the book and feel free to reach out with questions or ideas.
Marathons are nearly always on Sunday mornings, when traffic is lightest. Thus the comparison is not strictly marathon days vs. non-marathon days, but marathon days vs. light-traffic days. This should strengthen your findings more (albeit slightly). If you cover this in the book, my apologies!
Hi Francis, Bapu here. It's actually not just marathon vs non-marathon days for the reason you mention but the same day of the week on marathon and non-marathons weeks. Eg Boston marathon is on Monday so marathon monday is compared to surrounding mondays in boston. For other cities, if the marathon is on a saturday, that marathon is compared to outcomes on other saturdays in the same city. And so on for sunday marathons. Hope you have a chance to check out the book! All best. Bapu
Another great article, David. Your curiosity and excitement comes through even in the written transcript of the interview. And thanks Bapu for encouraging me to be more curious about what natural experiments are hiding in plain sight.
Love the creativity. Another book to add to my ever growing reading list, but there are worse problems to have.
The random experiment framing makes me curious what it would be like to integrate this type of thinking into one of my courses. Effectively asking students to be attentive to what is around them and then leverage some mathematical skills and reasoning to try to pull an insight out that might be otherwise ignored.
Hey Josh, so glad you enjoyed! And I feel you on the ever-growing reading list. I realize I have far, far more books now than I could read in a lifetime. That said, I do tend to read a little here and there from a lot of those books, while only finishing a much smaller number, and I really value what I get. And I think Random Acts of Medicine is definitely the type of book where you can pick a chapter that appeals to you and dive in. ...I actually surf my own shelves quite a bit, and just pick something out and flip through a little, and it's kind of surprising how often it sparks something. I guess that book is on my shelf for some reason, but maybe I bought it a year ago and never opened it, but then suddenly something grabs me. ...Or partly this is me justifying the fact that I'm single-handedly keeping my local independent bookstore in business;)
Another 'natural experiment' that I feel has been under-analyzed is Medicare. At age 65, everyone has good health insurance. At age 64, some people have poor health insurance, others have none. Rates of elective procedures like knee replacements spike at age 65, as you get those who have been putting it off. But non-elective procedures go up, too, suggesting people find ways to put just about everything off. If you look at mortality, it's worse to be 65 than 64, because the mortality rate always goes up with age. However, it goes up by less than 64 vs 63 or 66 vs 65. In other words, 64 year-olds die a bit more than they should, all else being equal. (Caveat: I looked into these things about 10-12 years ago and I no longer believe I have access to the data that would allow me to revisit it. Hopefully the Affordable Care Act would have attenuated this effect somewhat, but I know people in my daily life who have, for example, $10,000 annual deductibles and thus try to 'bundle' two or three procedures in the same year to save some money, so it has not gone away).
Francis, that is so interesting, and troubling. I appreciate you making me aware of that. I may ask Dr. Jena about this observation to see if he has any thoughts to share.
Hey David, love your work esp Range. I originally found you from Peter Attias podcast. Btw have you read his book, Outlive?
Hey Srini, thanks for the kind words! I have Outlive, and have actually been reading it just this week. Finding it really interesting, especially given all the complexity he's grappling with. If you read it, I'd love to hear thoughts.
Awesome -- I just finished my first read but feel like I need another read through to take notes!
I enjoyed it and found it immensely informative. The whole book feels like a RANGE case study. Peter and his analyst team are birds that looking across what all the frogs are doing in their respective areas and pulling out the most interesting threads relevant for longevity. It’s so comprehensive and I’m surprised nobody has done this before.
There’s so much information that on Jocko Willink’s podcast, he accurately called it a textbook / reference guide.
Absolutely, right on with the "reference" comment. I love have a store of books like that. I'll have them on my shelf, and sometimes for a break just flip through a little, and get some idea spark. I also agree in a way with your surprise that nobody has done this before. Although, when I wrote my first book, I felt the same way; "How has nobody connected these clearly connected threads that orbit the same question??" I think one of the answers was that most of the relevant parties were frog-like, and they weren't — as Arturo Casadevall puts it— "standing up from their trench to look in the next trench over." The bright side was that it left some pretty obvious opportunities for birds to come along and add value. I think there are a lot of opportunities like that, waiting for birds!
Totally, I think that makes a lot of sense. I think that’s why RANGE spoke to me. I am so oriented towards the birds, bigger picture, deeper truths, the invariants between different situations, etc. I just can’t get excited about niche to the nth degree I guess but I’m glad there are other personalities who are!
Couldn't have put it more eloquently myself!
Fantastic conversation that I hope, before much longer, does stretch on for 40 more questions and answers. My wife covers the orthopedic device industry so she talks fairly regularly with surgeons, and we recently moved from the West side of Cleveland to the East side so now I talk fairly regularly with surgeons (and ER doctors, and other docs) just because they're around, and these are fascinating topics I would have never considered. Will definitely be reading the book this year. Thank you, Bapu, for your research and writing, and thank you, David, for your endless curiosity.
Matt, thanks so much for the kind words! ...I keep generating these ridiculously long lists of questions, and then when I transcribe interviews, I realize I can get maybe one-fifth into them and already have 5,000 words, which would be a really long post. I'm not sure what the right length is (both for readers, and for me trying to put it together), but I'm very glad to hear that it didn't strike you as going on too long. More importantly: congrats on the move, and if you pick up the book, I hope it makes for some stimulating conversation with all those docs!
Kudos for your question about behavioral similarities or differences related to Covid and children’s birthdays! I have two more questions. First, regarding the increased incidence of cardiac events in the geographic area around marathons, did the authors of the original article exclude mortality or other serious events among the marathon runners themselves? For example, most very large running events, unfortunately have one or more cardiac events, some resulting in mortality. In addition, there may be incidences of severe heat, illness or other problems that could then cause a cardiac event. Granted, one can hope these numbers are small, but they do occur, and it would be helpful to know whether they were included in that study or not. Second, regarding the opioid prescription discussion, at least one more natural experiment did occur in the US. In the not-so-distant past, physicians were instructed that we were under-prescribing pain medication, and should be more liberal in the treatment of pain, using opioid medications. More recently (this varies to some extent by state) a great deal of education has been required for license renewal that includes use of non-addictive medications such as acetaminophen or ibuprofen for pain rather than opioids. Any thoughts?
Hi Angela, thanks for these questions and insights! One point of context that I happen to know off the top of my head: there are no deaths among participants in most marathons. Last time I recall reading a large study of the topic, I believe the death rate was approximately 1 per 100,000 marathon participants. So even the NYC Marathon should not expect an annual death. I believe running-related deaths were excluded, though. (That said, there are other, usually nonfatal but serious issues, like hyponatremia, that can be fairly common in marathons.) In terms of the opioid-prescribing, that's very interesting! I'm not sure I have anything insightful to add about that, but I'd curious to hear your thoughts on it. I'd be really interested to know whether there have been studies look at impact on individual prescriber tendencies from either the old instructions, or the new.
I don't have any information about prescriber tendencies as recommendations changed, but will be interested to see what other readers say, and will look it up when I have a chance later.
True about actual mortality, so I guess I didn't state the question very well. Those running a marathon typically get near immediate care for cardiac events, or even hyponatremia or heat illness that might lead to a cardiac event. Medical tents at almost all marathons or even some shorter distance events are well staffed and well prepared. This contrasts with those who are impacted by ambulance transport problems and may not get even early care for their potential cardiac event.
Ah I see! Great point...I'm going to run this by Dr. Jena and see if he has any thoughts to share.
Hi Angela, Bapu here! Two great questions. On marathons, we did two things to ensure that the mortality increases were not from literal marathon participation itself. First, we looked at quite elderly people with multiple chronic medical conditions, ie people unlikely to be running a race. The average age in our sample was about 77 years old and the same findings held (a marathon day mortality effect) if we restricted to people with dementia or on dialysis, ie people unlikely to be running the marathon. So, this seems it's all about delays.
Re: the opioid question, we actually have some new research not yet published that looks at this exact question using a natural experiment. The challenge with looking at an overall national policy towards "pain as a fifth vital sign" is that it's hard to separate the potential impacts of that from anything else changing over time. But we think we have a good way to do it, so stay tuned!
Thanks for answering questions, Bapu! And I'd love to share the results of that work here when it's ready.
I love the random correlations but still can decide how much causation to assign
I didn't want to go into too much detail in a long post, but the book does a really nice job of walking through the steps of each natural experiment to examine whether one can claim causality, or whether it remains an open question. In that sense, a lot of the book is sort of an exercise in scientific thinking.
Thanks for this comment! Makes me way more likely to pick up the book.
Glad you appreciated it. Now I'm thinking maybe I should've gone into this issue explicitly in the post... Well, useful feedback for the future!
Hi, Bapu here. Great question and something we think a lot about in our research and go into detail in in the book. It'll be fun and won't disappoint you. A basic test we always do is see whether those exposed to a "treatment" (say, a heart attack during a marathon day) vs a "control" (say, NOT having a heart attack on a marathon day) is truly random. We make sure characteristics of the heart attack patients are otherwise similar across a number of different dimensions like age, other demographics, chronic medical conditions, etc. They are in the natural experiments we report. We also do additional tests, called "falsification tests", to corroborate why we think the findings are causal. Check out the book and feel free to reach out with questions or ideas.
Marathons are nearly always on Sunday mornings, when traffic is lightest. Thus the comparison is not strictly marathon days vs. non-marathon days, but marathon days vs. light-traffic days. This should strengthen your findings more (albeit slightly). If you cover this in the book, my apologies!
Hi Francis, Bapu here. It's actually not just marathon vs non-marathon days for the reason you mention but the same day of the week on marathon and non-marathons weeks. Eg Boston marathon is on Monday so marathon monday is compared to surrounding mondays in boston. For other cities, if the marathon is on a saturday, that marathon is compared to outcomes on other saturdays in the same city. And so on for sunday marathons. Hope you have a chance to check out the book! All best. Bapu