The studies I have seen on intermittent fasting are all very poorly designed. The one in the NEJM cited here required calorie-restricted diets for both groups, which rather defeats the purpose of the time restriction by making the participants feel so hungry that they will eat whenever they can. Also, an intent-to-treat statistical analysis is probably premature. Isn't the goal is to learn what works when it is actually implemented? Both studies cited in this article found that the TRE group lost more weight than the group with no time restrictions, but the results were not statistically significant when compared with a group given different instructions. This would suggest that the sample sizes were too small, rather than that the TRE theory is unsound. It would be better to test the TRE system that seems to be more effective though, which is not eating at all for at least two days a week, rather than limiting eating to a specific time each day. The most successful TRE dieters eat only when they are actually hungry, AND do not eat even when hungry during a prohibited time.
I don’t feel as doctors we do well treating obesity. We never did well treating alcoholism. I think both deserve more thoughtfulness on our part. I think we need more training in addiction medicine. I recommend mindfulness app called eat right now by addiction Psychiatrist , Dr. Brewer and Overeaters Anonymous. Both similar and complementary. It is so hard for people who aren’t problem eaters to understand.
So many comments. I don't know if these two issues have been addressed.
1. In the TREAT study - the IF group was allowed ad libitum eating - this doesn't work. I have been doing IF for a few years now and you can't just go around eating 3000-3500 calories (trust me, it can be done in 6-8 hours). So that wasn't a particularly fair comparison.
2. In the NEJM study, both groups had caloric restriction, which is the driving factor. IF just allows easier calorie restriction, once you get used to the eating pattern.
The study we need is a large one comparing people practicing IF - ad libitum with a normal population (ad libitum) for metabolic parameters. The NEJM was a good one in that it shows that calorie restriction is the more important parameter, whether you achieve it without IF or with IF and with IF it is just easier, once the routine is established. But what about those who adopt IF as a lifestyle versus those who don't. How does that change your health parameters?
I don't practice IF to lose weight - I do it to maintain my weight along with significant levels of lifting and aerobic exercise. IF allows me better control over what I ingest, which I am assuming should help me live a little longer, healthier.
Leg 2 of the stool was not given justice (will need longer form text to explore).
Clinical experience is so much more specific than ‘what I’ve seen work with my eyes’ (in general) and more to do with ‘what’s different about this patient in front of me’ (situativity).
For the patient who has researched the heck out of IF, has created a plan with their spouse to lose weight together, and is waiting for a green light, I’ll say ‘yep it’s safe... go for it!’
(Vs)
For the patient who just heard it on the radio and wants to explore it further (not ignited, less momentum), I’ll pause and discuss EBM including efficacy vs risks.
This individualized approach is also employed with glucometers even though the evidence isn’t there to ‘routinely’ recommend it for those who aren’t on insulin. Yet some really do benefit... and they’ll usually self select :)
Something else to consider: is the placebo *really* a placebo? Consider the discovery of valproate by Pierre Eymard, who was using it as an "inert" vehicle to deliver other anticonvulsant drug candidates for testing.
In the NEJM study, both intervention & control groups were placed on a limited calorie diet, and both experienced a smiliar result. If the intervention had been: fasting vs no fasting, but otherwiseno change to their diets, might there have been a bigger delta?
And in the JAMA study, it was really two different interventions being tested (3 structured meals vs eat whenever then fast) but no true control.
You're not trying to figure out what worked for a gazillion different patients from a bunch of different studies - you're trying to figure out what works for the patient sitting in front of you.
Unlike most medicine which is not very scientific, fasting of all kinds has been around for thousands of years. The people who are harmed or die from fasting is extremely minuscule compared to those who are harmed or injured in the realm of modern medicine.
Thanks for this article. Diet studies, even the best ones, are fraught with methodological issues generally, and are hard to do well.
But ultimately, weight loss is really not rocket science. Expend more calories than you consume. No doubt that is easier said than done.
How you consume it, over how many hours, is probably less important. At least from a weight loss perspective.
And it’s far more important to adopt a habit that you can stick to, rather than worrying about the latest fad. The ones who slavishly swear by any diet in particular are merely poster children for placebo effect and self selection bias.
Sep 28, 2023·edited Sep 28, 2023Liked by Adam Cifu, MD
The study you cite on intermittent fasting showed a 3.74lb difference over a year been those who did and those who didn't fast. That seems like a nice difference, and the desired direction. Based on that I'd go with reasonable intermittent fasting without going crazy over it. My physical trainer does things like fast for 24 or even 36 hours. That can't be good: Americans tend to go berserk with every health tip.
Fasting from 9 pm to 9 or 10 am feels good to me. But what's really knocked some weight off me is cutting way down on rice and potatoes.
Sep 28, 2023·edited Sep 28, 2023Liked by Adam Cifu, MD
Great article. Always important to review probability since it so easy to neglect the basics, and we also forget how unintuitive it can be.
I'll always remember one day in second year of med school I was outside at one of the food trucks between classes, eating a delicious grilled cheese stuffed with mac n' cheese and bacon, all on buttered texas toast. The instructor for my clinical pathophysiology course walked up to me and looked at my lunch. It was especially embarrassing because we were currently on the cardiology section of the course. He said something like "you just walked out of my lecture on coronary artery disease and ordered this?" He laughed and half-jokingly threatened to fail me if he saw me eating this again.
At some point later that year I decided to try intermittent fasting while studying for Step 1. I did one meal a day at dinner and did regular weight training. Within seven or so weeks my LDL went from 150's to mid 90's. Lost something like 20 lbs, was in the best shape of my life. Every time I've tried it since then it works for me. But for a lot of people it really doesn't work. For what it's worth I usually advise people to do whatever diet/exercise regimen that is actually sustainable for them, whatever that is. Often I'll say something like "try to identify the 5% of the worst garbage that you eat, and just cut that out. Don't worry about eating more salads, just focus on the worst 5%".
Done IF for years off and on. Don't worry I have spoken to so many people that have said "Oh I could never do that" without even trying. It's won't catch on too well!
The first thing I wonder about the NEJM study (without reading it) is why did the control group loose any significant weight? Question: Does intermittent fasting work (does a single intervention work)? So 2 groups - fasting group and control group (ie. the propensity matched do-nothing group) - this would be the EBM one is looking for. Then one would move on to "Does intermittent fast work as compared to other 'diets' - but hey, what do I know...
The interesting thing about the NEJM study is that the control group was told to restrict calories just not in the Intermittent Fasting approach. Both groups lost weight, which is surely a positive from a clinical perspective; the study didn't show that IF was superior to calorie restriction. It did not compare either method to a better control, which would be randomly selected patients who weren't told to restrict their intake in any way.
Similarly, in the JAMA study, the Time Restricted Eating group lost weight and improved many of the metabolic parameters significantly (compared to baseline), just not significantly in terms of the gap between the two groups. Again, TRE worked and was associated with significant weight loss; the control lost a small amount of weight, so the difference between the groups was not significant.
The JAMA study raised an important point: the loss of lean body mass with fasting.
I don't find either study particularly convincing in that the key signal (fasting --> weight loss) was shown just not so much that it beat in statistical terms, the various control groups (even though the control groups did not lose statistically significant weight loss)
Sounds like we are nowhere near calling this "Evidence Based Medicine" -- it sounds more like we need to do larger studies for longer times before we can say for sure that IF doesn't work...
likely some benefit to gut microbiome as well. great point on the glucose/insulin effects. more than weight loss, I want to know if IF increases insulin sensitivity in those with metabolic syndrome or prediabetes/DMT2
Yes! I have heard that DMT2 and pre DMT2 is where there are the most benefits for IMF. Possibly through insulin sensitivity changes and maybe the microbiome?
The studies I have seen on intermittent fasting are all very poorly designed. The one in the NEJM cited here required calorie-restricted diets for both groups, which rather defeats the purpose of the time restriction by making the participants feel so hungry that they will eat whenever they can. Also, an intent-to-treat statistical analysis is probably premature. Isn't the goal is to learn what works when it is actually implemented? Both studies cited in this article found that the TRE group lost more weight than the group with no time restrictions, but the results were not statistically significant when compared with a group given different instructions. This would suggest that the sample sizes were too small, rather than that the TRE theory is unsound. It would be better to test the TRE system that seems to be more effective though, which is not eating at all for at least two days a week, rather than limiting eating to a specific time each day. The most successful TRE dieters eat only when they are actually hungry, AND do not eat even when hungry during a prohibited time.
I don’t feel as doctors we do well treating obesity. We never did well treating alcoholism. I think both deserve more thoughtfulness on our part. I think we need more training in addiction medicine. I recommend mindfulness app called eat right now by addiction Psychiatrist , Dr. Brewer and Overeaters Anonymous. Both similar and complementary. It is so hard for people who aren’t problem eaters to understand.
No papers. However, Cummins references leading researchers in this lecture:
https://m.youtube.com/watch?v=UZoQiDaWnuE&feature=youtu.be
So many comments. I don't know if these two issues have been addressed.
1. In the TREAT study - the IF group was allowed ad libitum eating - this doesn't work. I have been doing IF for a few years now and you can't just go around eating 3000-3500 calories (trust me, it can be done in 6-8 hours). So that wasn't a particularly fair comparison.
2. In the NEJM study, both groups had caloric restriction, which is the driving factor. IF just allows easier calorie restriction, once you get used to the eating pattern.
The study we need is a large one comparing people practicing IF - ad libitum with a normal population (ad libitum) for metabolic parameters. The NEJM was a good one in that it shows that calorie restriction is the more important parameter, whether you achieve it without IF or with IF and with IF it is just easier, once the routine is established. But what about those who adopt IF as a lifestyle versus those who don't. How does that change your health parameters?
I don't practice IF to lose weight - I do it to maintain my weight along with significant levels of lifting and aerobic exercise. IF allows me better control over what I ingest, which I am assuming should help me live a little longer, healthier.
Great example overall! One criticism though:
Leg 2 of the stool was not given justice (will need longer form text to explore).
Clinical experience is so much more specific than ‘what I’ve seen work with my eyes’ (in general) and more to do with ‘what’s different about this patient in front of me’ (situativity).
For the patient who has researched the heck out of IF, has created a plan with their spouse to lose weight together, and is waiting for a green light, I’ll say ‘yep it’s safe... go for it!’
(Vs)
For the patient who just heard it on the radio and wants to explore it further (not ignited, less momentum), I’ll pause and discuss EBM including efficacy vs risks.
This individualized approach is also employed with glucometers even though the evidence isn’t there to ‘routinely’ recommend it for those who aren’t on insulin. Yet some really do benefit... and they’ll usually self select :)
Article about your second point coming next week!
Something else to consider: is the placebo *really* a placebo? Consider the discovery of valproate by Pierre Eymard, who was using it as an "inert" vehicle to deliver other anticonvulsant drug candidates for testing.
In the NEJM study, both intervention & control groups were placed on a limited calorie diet, and both experienced a smiliar result. If the intervention had been: fasting vs no fasting, but otherwiseno change to their diets, might there have been a bigger delta?
And in the JAMA study, it was really two different interventions being tested (3 structured meals vs eat whenever then fast) but no true control.
You're not trying to figure out what worked for a gazillion different patients from a bunch of different studies - you're trying to figure out what works for the patient sitting in front of you.
So where do I come down on not intermittent fasting? It is probably harmless ...
Unlike most medicine which is not very scientific, fasting of all kinds has been around for thousands of years. The people who are harmed or die from fasting is extremely minuscule compared to those who are harmed or injured in the realm of modern medicine.
Thanks for this article. Diet studies, even the best ones, are fraught with methodological issues generally, and are hard to do well.
But ultimately, weight loss is really not rocket science. Expend more calories than you consume. No doubt that is easier said than done.
How you consume it, over how many hours, is probably less important. At least from a weight loss perspective.
And it’s far more important to adopt a habit that you can stick to, rather than worrying about the latest fad. The ones who slavishly swear by any diet in particular are merely poster children for placebo effect and self selection bias.
The study you cite on intermittent fasting showed a 3.74lb difference over a year been those who did and those who didn't fast. That seems like a nice difference, and the desired direction. Based on that I'd go with reasonable intermittent fasting without going crazy over it. My physical trainer does things like fast for 24 or even 36 hours. That can't be good: Americans tend to go berserk with every health tip.
Fasting from 9 pm to 9 or 10 am feels good to me. But what's really knocked some weight off me is cutting way down on rice and potatoes.
Great article. Always important to review probability since it so easy to neglect the basics, and we also forget how unintuitive it can be.
I'll always remember one day in second year of med school I was outside at one of the food trucks between classes, eating a delicious grilled cheese stuffed with mac n' cheese and bacon, all on buttered texas toast. The instructor for my clinical pathophysiology course walked up to me and looked at my lunch. It was especially embarrassing because we were currently on the cardiology section of the course. He said something like "you just walked out of my lecture on coronary artery disease and ordered this?" He laughed and half-jokingly threatened to fail me if he saw me eating this again.
At some point later that year I decided to try intermittent fasting while studying for Step 1. I did one meal a day at dinner and did regular weight training. Within seven or so weeks my LDL went from 150's to mid 90's. Lost something like 20 lbs, was in the best shape of my life. Every time I've tried it since then it works for me. But for a lot of people it really doesn't work. For what it's worth I usually advise people to do whatever diet/exercise regimen that is actually sustainable for them, whatever that is. Often I'll say something like "try to identify the 5% of the worst garbage that you eat, and just cut that out. Don't worry about eating more salads, just focus on the worst 5%".
Done IF for years off and on. Don't worry I have spoken to so many people that have said "Oh I could never do that" without even trying. It's won't catch on too well!
Why is the post-test probability 25% (after NEJM) and not 17%? ... didn't quite get the math there.
Prestudy probably here is your gestalt. I just guessed, 50/50 without data, cut in half by the results of that first study...
The first thing I wonder about the NEJM study (without reading it) is why did the control group loose any significant weight? Question: Does intermittent fasting work (does a single intervention work)? So 2 groups - fasting group and control group (ie. the propensity matched do-nothing group) - this would be the EBM one is looking for. Then one would move on to "Does intermittent fast work as compared to other 'diets' - but hey, what do I know...
The interesting thing about the NEJM study is that the control group was told to restrict calories just not in the Intermittent Fasting approach. Both groups lost weight, which is surely a positive from a clinical perspective; the study didn't show that IF was superior to calorie restriction. It did not compare either method to a better control, which would be randomly selected patients who weren't told to restrict their intake in any way.
Similarly, in the JAMA study, the Time Restricted Eating group lost weight and improved many of the metabolic parameters significantly (compared to baseline), just not significantly in terms of the gap between the two groups. Again, TRE worked and was associated with significant weight loss; the control lost a small amount of weight, so the difference between the groups was not significant.
The JAMA study raised an important point: the loss of lean body mass with fasting.
I don't find either study particularly convincing in that the key signal (fasting --> weight loss) was shown just not so much that it beat in statistical terms, the various control groups (even though the control groups did not lose statistically significant weight loss)
Sounds like we are nowhere near calling this "Evidence Based Medicine" -- it sounds more like we need to do larger studies for longer times before we can say for sure that IF doesn't work...
what about intermitant fasting for blood sugar control and not weight loss?
likely some benefit to gut microbiome as well. great point on the glucose/insulin effects. more than weight loss, I want to know if IF increases insulin sensitivity in those with metabolic syndrome or prediabetes/DMT2
Yes! I have heard that DMT2 and pre DMT2 is where there are the most benefits for IMF. Possibly through insulin sensitivity changes and maybe the microbiome?
It is probably beneficial.
that is what I thought. It was good for type II diabetes but not necessarily weight loss. Are you aware of a paper on this?