Image created by Angela A. Stanton, Ph.D. 

Fasting and Cardiovascular Disease? Huh?

I am sure you have heard by now that there is a new study out on TRE (time restricted eating) and that there is an association of increased cardiovascular disease risk (CVD). Right? If not yet: now you have!

Hold Your Horses!

Yep… just put your horses back in the stall and get a cup of coffee or tea or beer or whatever you drink when you are in pain… mental pain from stupidity! To offer a forerunner for you, here is what I wrote on Medscape as a message by healthcare providers, Since you may or may not have access to it, I copy-paste:

Yet one more absolute waste of data analysis and serious misinformation for the public. How do papers like this pass through the peer review process? Has everyone in research forgotten that association is not causation and that food questionnaires are inherently inaccurate and not useful at all? And, most importantly, hazard ratio of 1.91 (overall sample) does not meet the requirements of the Bradford Hill Criteria! In order to consider it to be a potential for even further investigation, the HR needs to be >2. And using the NHANES data generating very creative hypotheses about TRE where none existed. Might it be that people practicing TRE couldn’t eat because of medications or feeling sick?

AAS from Medscape (see here)

Let me elaborate just a bit on what I wrote above.

This study is not based on any clinical trial. It is based on an old database data analysis: NHANES, which ran from 2008 to 2013 and collected data from questionnaires. The food questionnaires contain questions like “How many cups of ribs you ate last year per week?”… quick! How many cups of ribs have you eaten last year per week? How about this year? How about just last month? Last week???? Can you EVER remember how many cups of ribs you ever ate? Who eats ribs in cups anyway? And who remembers how many servings of ribs one ate last year? Last month? Provided that you can even remember what “ribs” actually are?

OK.. assume you know precisely that you ate 1 cup of ribs per week last year. And you remember because you took notes about what you ate–and still do, all the time. And on some days you skipped breakfast because you had no time to eat or you felt sick. Whatever… You reported your ribs and all other foods you ate to NHANES and also the times you started your meals and how many meals a day very precisely to NHANES database.

Now along comes a researcher and looks at your data and says: Aha!! This person is not eating 3-5 times a day! They are keeping longer break than the usual 10 hours of not eating overnight and so, therefore, must be practicing TRE. So your record is now part of TRE. A year later you get a heart attack or stroke or you die. Why did that happen? Because you practiced TRE. And how did the researchers come to this conclusion? Well! Obviously! It was recorded in the NHANES that on some days you skipped a meal. There. Makes sense? I agree. It’s clueless.

Metadata and The Truth!

If I were a researcher police, I would make it illegal to analyze a database for things that were not initially incorporated into the database for testing. No one was thinking about TRE when this database was collected. For your information: “Time-restricted feeding (TRF) as an overt dietary intervention to improve health was first performed in mice 10 years ago in 2012.” TRE and TRF are the same things.

Next: the statistics. This paper was evaluated using the Bradford Hill Hazard Ratio (HR) analysis, which has certain requirements. One of the most important requirements is that whatever strength the data shows, the hazard ratio shows only association! Association means that 2 things move in the same direction, like the train and my car may both be heading to the same city but we are not connected in any way and neither the train caused my car to go to that city, nor did my car cause that train to go in that city. The two just happened to go to the same city. No causation, just association. Similarly, the Bradford Hill criteria requires that whatever HR is found, it is only purely association.

The second requirement of the Bradford Hill is that if the HR is < 2, it is not even associated. Meaning the train and my car are not even heading in the same direction, let alone to the same city. The overall HR in this study is 1.91, which by all account is <2, therefore… not even remotely associated. Yet, what did the authors say? Medscape said that the authors said (the article is not yet published) this: “In the overall sample, compared with an eating duration of 12-16 hours, 8-hour TRE was significantly associated with an increased risk for CV mortality (hazard ratio [HR], 1.91; 95% CI, 1.20-3.03).”

Once You Get This Far

Trash it! Yes! you read it correctly! Trash the paper! Even if some of the HR specifics they evaluated are >2, the request of the Bradford Hill Criteria is that this suggests “you can now test for causation” if there is any because association is shown. So some of their HR numbers go to between 2 and 3.6, and for those they need to use DIFFERENT statistical models to evaluate if there is any causation! The Bradford Hill statistics CANNOT establish causation! For reference, the studies showing the association of cancer and smoking had HR in the range of 20-30! And how many years of fight did it take to have the industry accept that there is even an association between smoking and cancer. And now, here we are, showing HR <2 and we take it as “causation”? Really? Why on earth?

Comments and opinions are always welcome and are censored for appropriateness,

Angela