The Washington Post told readers that “standing for at least a quarter of the day reduces odds of obesity.”
Inside the article, it was reported that the benefit of such standing is immense, with the smallest effect size at 32 percent. The analysis apparently found that men who said they stand a quarter of the time were 32 percent less likely to be obese than those who said they don’t stand at all.
If we were to take such a claim seriously, it would have been impossible to explain how obesity had become so widespread, as its curtailment should have been as easy as giving away some standing desks.
Of course, in the cited study, recently appearing in Mayo Clinic Proceedings, the authors appropriately drew a more modest conclusion that “standing more than a quarter of the time is related to reduced odds of elevated adiposity.” (Our italics.) The switch from “is related to reduced odds” in the research paper to “reduces odds” in the news headline changed the meaning entirely, inserting a cause-effect relationship that has not been proven in neither the present study nor other studies.
While the headline writer is the likely source of this word play, the study’s authors are not taken off the hook, either. Based on their analysis of the data of over 7,000 people, they suggested that “clinicians and public health practitioners should consider encouraging patients to … increase standing time for chronic disease prevention.”
So, despite a robust discussion about the limitation of their analysis, the research team proceeded to offer a recommendation that interprets correlational evidence as causal evidence. We have no intention of singling out this particular set of scientists, as such a narrative device is widely practiced in the medical literature, and it is also a staple of “big data” studies.
Using the same data, we offer our recommendation: disease prevention as a remedy for chronic sitting.
We say this only partly in jest. The scientific evidence consists of data from people enrolled in the Cooper Center Longitudinal Study, which was started in 1970 to investigate the relationship between physical activity and fitness and morbidity and mortality. Each patient was asked how much they stand during the day, and was physically measured on a number of common obesity measures, such as BMI, and waist circumference. An analysis revealed that selected obesity measures are inversely correlated with the amount of standing time.
What’s the problem? We think the authors of the cited study did a great job explaining it:
These data prohibit the ability to conclude whether more standing reduces the risk of obesity … or whether being obese … leads to less standing.
While there are many approaches to establishing causality, everyone agrees that the cause must precede the effect. There is nothing in the study to meet this most basic requirement. The data about standing habits and obesity were concurrent. The above quotation implies that the following pair of conclusions is equally valid:
“If you stand for longer periods of time, you are less likely to become obese, which implies you are less likely to suffer from obesity-related diseases.”
“If you are obese (or suffer from obesity-related illnesses), you are more likely to prefer sitting.”
Hence our recommendation. Seriously, we think it is plausible that the extra weight people carry around can lead to more sitting and less standing.
Waiters stand around a lot, as their profession demands. The average waiter is less obese than the typical person, as this analysis by The Atlantic suggests. Does this mean waiting tables reduces the odds of obesity? Or does this mean that restaurateurs preferentially hire slimmer waiters?
This problem of unmeasured confounders is what makes observational data so difficult to draw conclusions from. Because the researchers wanted to explore standing habits, the study participants were classified according to how much time they spent standing. If, instead, the researchers had asked about diets, the participants would have been classified according to what they eat. If dieting is correlated with standing, and both habits reduce the odds of obesity, then the present study exaggerates the effect of standing. (The authors disclosed this limitation.) In the worst-case scenario, standing yields no benefit or may even produce harm—but so long as dieting is beneficial, this type of analysis would conclude erroneously that standing reduces odds of obesity.
As mentioned before, the reported effect sizes were huge, ranging from 32 percent to over 50 percent. Those are averages, and in this case, readers should be advised about the uncertainty of these averages. We use the group of men who stand more than 75 percent of the time as an illustration. The analysis showed that this group is 39 percent less likely to have elevated body fat percentage compared to those men who almost never stand… with a confidence band that runs from 64 percent less likely to 3 percent more likely.
Said differently, if we had access to repeated samples, the effect of standing could plausibly range from 64 percent reduction to 3 percent elevation, depending on which sample was analyzed! This tells us the true effect might be much smaller than advertised. Thus, we would like to see additional more robust studies before drawing conclusions about standing.
Even if we accepted the newspaper’s conclusion, that “standing for at least a quarter of the day reduces odds of obesity,” we have trouble reducing this insight to a lifestyle alteration. How much more standing do we have to partake in? How long does it take for the additional standing to produce the desired effect? If we were already obese, should we start a regime of standing? For those of us at more advanced ages, have we built a deficit of standing?