Why you can't just ignore that new fasting study
The wellness world is reacting to alarming headlines about the impact of time-restricted eating (TRE) on cardiovascular health. At issue is an abstract presented during a poster session at last week’s American Heart Association’s annual meeting showing that TRE causes a 91% increase in death by cardiovascular disease. Here’s my take on the controversy.
Experts have long been divided on the benefits and risks of intermittent fasting.
It’s no surprise that this news is sending shock waves through the scientific and medical communities. But what I find fascinating is the degree of confirmation bias on display right now, both in those defending the results and in those refuting them. The pro-TRE camp’s position is that, since the study is preliminary, and not peer-reviewed or published, it should be summarily dismissed. The anti-fasting crowd, on the other hand, is embracing the results, responding with a smug, “we told you so.”
The pro-TRE camp’s position is that, since the study is preliminary, and not peer-reviewed or published, it should be summarily dismissed. The anti-fasting crowd, on the other hand, is embracing the results, responding with a smug, “we told you so.”
Both factions have valid points to make; however, the way I see it, neither group is giving enough credit to the scientific process: the process by which actual scientists (like myself) develop, analyze, and present our research findings. Of course, it didn’t help that the AHA released an inflammatory press release - which could honestly be a lesson in scientific journalistic integrity in itself. (Some have even suggested this move was motivated by the desire to draw visitors to their website to solicit donations).
There are two major issues with these dueling groups, and I’m going to address both. First, I’m taking a look at study design, data presentation, and the peer review process. Second, I’m teasing out what the data did – and didn’t – say, and where we should go from here.
For what it’s worth, I have no skin in the game with fasting or TRE. I’m following the data, just like everyone else. There have been studies showing the short-term benefits of intermittent fasting. There are some that have shown nothing. There are even some that have shown harm, particularly if you’re in caloric restriction for too long. Yet, there are hundreds of different and often conflicting “protocols” out there for “fasting” or “fasting mimicking,” or “time restricted eating,” or even the time of day when it’s best to fast. And the hype (and money to be made) is real – with “experts” writing books, selling programs, apps and gadgets, and ultimately sparking diet cults with battles brewing all over social media.
Why study design matters.
First, it’s important to note that this abstract was not a peer-reviewed study. It was a poster presented at a scientific meeting. This does not invalidate the results; however, it does indicate that they are preliminary (more on this below).
Second, the outcomes of the analysis are what we call correlational (remember, correlation does not equal causation). The researchers – three of whom are absolute pillars in the field of nutrition – looked at a group of 20,000 people in a population data set we call NHANES (the National Health and Nutrition Examination Survey) after an 8-year period. NHANES is a massive catalog of lifestyle behaviors of the general population in the United States that has been ongoing for decades. When looking at a population data set like NHANES, scientists use sophisticated statistical modeling to draw conclusions about whether lifestyle behaviors and demographics are associated with long-term health outcomes.
Scientists use this kind of analysis often to generate further hypotheses that can be tested with intervention studies like randomized controlled trials. For example, they may note an association between a vegetarian diet and reduced LDL in a big population-level data set. Then, other scientists will design a trial to administer study subjects a vegetarian diet, tracking their LDL over time to see if there is “causation.” Sometimes there is agreement between the two study designs, and sometimes there isn’t.
Let’s take a look at how scientists relay findings to the world.
Now, back to the AHA “study”. The NHANES data presented at AHA was written into an abstract that was presented as a poster at the annual meeting. Remember, an abstract isn’t a peer reviewed study. This means that the data is considered preliminary, as it has not been through the peer-review process (more on that below). An image of the poster that was presented at the AHA is available here.
Fun fact: when you present a poster at a scientific session, you print out a huge version of it, hang it on a wall in a massive convention hall filled with thousands of other posters. These sessions function as a massive science networking event. All the posters are understood to be preliminary data and people mill around reading through what everyone else has been working on in their labs. You stand by your poster and get peppered with questions from other scientists trying to poke holes in your work for four hours. It can be an excruciating experience, but it’s the first step in getting your data out in the world for other scientists to see and critique.
From here, the scientists write a complete manuscript (typically about 30-50 pages) and send it to an academic journal for peer review. Once submitted, the review process takes about four to 10 months of back-and-forth critique and revisions, until all reviewers are satisfied that the paper contains minimal flaws (as all science has gaps and limitations) and is being presented appropriately and responsibly.
It’s important to note:
A poster does not go through the peer review process. This is where I believe the pro-fasting camp has it right. The fact that this poster – and not a peer-reviewed study – sparked such sensational headlines has drawn the ire of TRE enthusiasts, and celebrity MDs like Peter Attia, leading him to conclude that he “does not put stock in the ‘study.’” However, neither bias nor the limitations of this type of study design or presentation invalidate the outcomes.
The scientists on the team stated their position quite clearly, both in the poster itself as well as in several interviews. They said they were “surprised” by the results and that “the findings do not mean that time-restricted eating caused cardiovascular death”.
All that being said, I think it’s safe to assume these scientists would agree with me when I say that, while the results of their study don’t align with some of the very limited data on intermittent fasting, it could be a real problem if the current TRE literature isn’t generalizable to a very vulnerable group in the population. Moreover, TRE studies have all been short term – none longer than a year. Is it also possible that after eight years, this kind of restriction leads to increased risk of death rather than longevity? Maybe it’s just me, but I think that’s an important consideration when interpreting these outcomes.
What the abstract said – and what it didn’t.
There has been a deluge of responses and summaries to the abstract (such as here, here and here), so I’m not going to go into the nitty gritty of the results. Rather, I’ll give you some actionable takeaways to consider as you decide what to do with your own diet.
Here’s what I believe TRE proponents and opponents are getting right and wrong:
There are some real concerns about the self-reported nutrition survey employed by this study (2 days of recall only), and whether the data can be relied upon to actually mean people were using a time-restricted eating model purposefully (unlikely) or due to other factors such as available time to eat, stress, job type (i.e. night shift workers), etc. However, the researchers did try to control for these things in their model (adjusting for age sex, race, total energy intake, education, income, food security status, smoking, alcohol, leisure time PA, diet quality, BMI and self-reported health conditions), and they still saw significance. That’s interesting.
Most people are not purposefully “fasting.” Frankly, the majority of the population has no idea what “fasting” or “time-restricted eating” even means. So, for the study to call eating in windows of less than eight hours per day, “time restricted” in the way the longevity crowd means it is likely NOT the same. However, this doesn’t mean that longevity biohackers’ eight-hour fasting is physiologically different than those who either choose or are forced to eat in eight-hour windows. (This is where the hypothesis generation comes in and when some strongly designed clinical trials need to answer this question.)
People have been “fasting” for a long time. It’s called “dieting” in most circles. The data has become abundantly clear that yo-yo style dieting leads to poor health, particularly if you go on and off diets – losing and regaining weight – your whole life. There is no way to figure that out with the NHANES data set for these folks. However, many of the popular fasting protocols out there recommend cycling fasting on and off throughout the year. Could this lead to weight loss/gain? Could this be problematic from a physiologic standpoint long term? Does it have to do with the actual timing of the eating, or caloric restriction, or some combination of both? Is this particularly problematic for high-risk individuals such as those with early stage cardiovascular disease or insulin resistance?
We don’t know the answers to any of these questions. We also can’t rule out the possibility of reverse causation (i.e. maybe being unhealthy led to reduced eating). And yet, people are jumping to conclusions and onto bandwagons. Either this poster is “garbage” and “should be dismissed”, or it’s validating and everyone should be “sounding the alarm for the imminent risk of death due to skipping breakfast”.
Here’s where I stand on time-restricted eating.
I think time-restricted eating is interesting. Truly. From a physiologic perspective, there have been some compelling reasons to do it, at least in the very limited and short-term studies we have. Simon Hill recently hosted a great (and long) podcast with some of the leading researchers on this topic and it’s worth a listen. However, restriction can also lead to problems as well – low energy availability, disordered eating behaviors, and even hormonal and skeletal muscle impairments.
So, we need to be responsible and thoughtful about this data - particularly because of the implications for clinical applications. Population level data is supposed to be “hypothesis generating” which means we won’t understand these results until the trials are done, especially in vulnerable populations. However, based on these outcomes, this study is making me think that fasting (or time-restricted eating within less than eight hours anyway) may not be for everyone – especially if it’s just another diet strategy masked in the alluring cloak of “longevity.”