As a few commenters pointed out last week, I was insufficiently skeptical of the recent Duke University study that found an association between body-mass index (BMI) and workplace injuries. My biggest mistake was citing the dramatic numbers that were highlighted in the news coverage (and in the JAMA press release), comparing the workers' compensation claims of the most obese employees with those of employees in the "recommended weight" range. These numbers came from a simple analysis that did not control for occupational category. Not surprisingly, certain kinds of jobs, the ones involving a lot of manual labor and heavy lifting, had higher injury rates than others. And as the researchers note, "Employees in several of the high-risk occupations were heavier than average..., emphasizing [the need for] caution in the interpretation of the bivariate relationships."
In the multivariate analysis, which took job category into account, the differences between weight groups were much smaller: The heaviest workers had a claim rate only 45 percent higher than that of the "recommended weight" workers, compared to twice as high in the bivariate analysis. The rate for the merely "overweight" employees was essentially the same as the rate for their thinner co-workers. The number of lost workdays per claim and the costs per claim were still substantially higher in the heaviest group, but the ratios after adjustment were not as dramatic as the ones I cited. Sandy Szwarc notes the importance of adjusting for job category and mentions some other study limitations here.
It's not clear why heavier employees in the same jobs had higher workers' compensation costs. As I noted in last week's post, there are plausible reasons to think some of the difference may be due to extra weight per se, but fitness might also be a factor. The researchers mention a few other possible explanations:
If obese nurses (because of discrimination or for some other reason) are less likely to get promoted, they may be doing more of the heavy lifting. It is also possible that obese workers, given the same illness or injury, may be more or less likely to submit a workers' compensation claim than a colleague of recommended weight; we cannot assess this possibility based on our data. Furthermore, there are potential underlying unobservable factors that might make obese individuals more likely to get injured independent of their weight. Education may be such a factor, although educational differences are, at least to some extent, adjusted for through occupational category.
In short, the association is not as strong as the press coverage implied, and it does not prove a cause-and-effect relationship—points to keep in mind when anti-fat activists cite this study as evidence that obesity imposes costs on others. In any case, assuming excess weight does contribute to workplace injuries, the appropriate response—whether fitness programs, BMI limits for job candidates, or nothing at all—should be left to employers, who are in the best position to judge the impact on their own bottom lines.