I’ve written a blog post on “Body Mass and Risk from COVID-19 and Influenza”, available at https://radfordneal.wordpress.com/2020/04/06/body-mass-and-risk-from-covid-19-and-influenza/
Here’s the intro:
Understanding the factors affecting whether someone infected with COVID-19 will become seriously ill is important for treatment of patients, for forecasting and planning, and — with factors that can be changed — for personal decisions aimed at reducing risk. Despite our current focus, influenza also remains a serious disease, so understanding its risk factors is also important.
Here, I’ll look at some of the evidence on how body mass — formalized as Body Mass Index (BMI, weight in kilograms divided by squared height in metres) — influences prognosis for respiratory diseases. Information specific to COVID-19 is still scant, but there is more data on influenza and on other respiratory infections (which includes coronaviruses other than COVID-19). Information on how BMI relates to general mortality should also be helpful.
Below, I’ll look at two relevant papers, plus a preliminary report on COVID-19. To preview my conclusions, it seems that being underweight and being seriously obese are both risk factors for serious respiratory illness. Furthermore, it seems that “underweight” should include the lower part of the “normal weight” category as defined by the WHO. Official advice in this respect seems dangerously misleading.
Seriously, BMI is a terrible metric. It is hardly any better than just plan weight. In Ancel Keys’ original paper he was looking for a way to estimate subcutaneous fat. Two problems here a) subcutaneous fat is pretty benign—visceral fat is the problem, b) BMI badly estimates even subcutaneous fat. So it measures the wrong thing, badly.
In many studies BMI in the slightly overweight range has the lowest mortality. BMI ignores body composition. Thus people who are skinny-fat and metabolically healthy show up as a ‘healthy’ BMI. Athletes, fit muscular people show up as overweight or obese. BMI does not work for smokers and ex smokers, older people, the skinny fat, athletes and muscular people—in general it works for perhaps slightly less than half the population.
The continued use of BMI is a good example of the problem in medicine of incumbency bias—beliefs and practice that are not based on solid evidence persist and can only be displaced by overwhelming evidence, and maybe not even then.
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0039504 - in particular have a look at table 2.
Note in the first study referred to in OP’s link the excess risk from being overweight is not statistically significant and even the excess risk from being obese is small. But even a bad metric like BMI is enough to see that being morbidly obese is bad. The risk from being underweight is often higher and may be attributed (as OP mentioned) to current or past smoking, but also to prior or current illness.
In both studies we see the typical U shaped curve for BMI and bad things. ABSI typically has a monotonic curve and is robust to issues like smoking. It is a real pity these studies did not use a decent metric.
In the CCP Corona Virus study we see that people who are “overweight” are *less* likely to be in critical care than the general public.
There are various hypotheses about why morbidly obese people are more vulnerable to CCP Corona virus. Perhaps it is related to metabolic syndrome as high or high normal serum glucose tends to depress the immune system.
I certainly agree that BMI is not ideal, but that’s what we’ve got as far as most presently-available studies go.
In addition to all the problems that you mention with BMI, there is just systematic measurement error, which is seldom mentioned. Is the height measured with or without shoes? Do unhealthy people tend to slump rather than stand up straight, reducing their measured height, or does the person measuring height make sure to get them to stand up straight? Is weight with or without clothing? What time of day is weight measured (the morning, right after a big meal,...)? I think different answers to these questions could shift measured BMI by about 0.5, perhaps systematically for different studies. Of course, this would affect ABSI too, along with whatever similar problems there are for measuring waist circumference.
Thanks for the link to the paper on ABSI. It looks very interesting. From a first glance, it seems to support the view that the WHO “normal” category for BMI extends too far in the low direction. (Of course, using ABSI as well rather than just BMI might give better information.)
If one has measurements on height, weight, and waist circumference, and a reasonable number of subjects, I think one would actually want to look at all three measurements, not just a one or two dimensional condensation of them, using some suitably flexible model for the relationship with mortality.