Body mass index and mortality: comment on article by Wolfe and Michaud

To the Editor

We read with great interest the study by Wolfe and Michaud published recently in Arthritis Care & Research (1). One conclusion the authors made was that “overweight and obesity reduce the RR [relative risk] of all-cause and cardiovascular mortality across different age groups and durations of RA [rheumatoid arthritis].” We believe that this conclusion may need to be reevaluated.

First, we calculated the distribution of age categories across the 4 body mass index (BMI) groups based on the results shown in Table 1. As is shown, participants with a BMI <18.5 kg/m² had the highest percentage of people age >70 years (37.9%), followed by those with a BMI of 18.5–24.9 kg/m² (28.0%), a BMI of 25.0–29.9 kg/m² (23.4%), and a BMI of ≥30.0 kg/m² (13.3%). Assuming that other risk factors for mortality are perfectly balanced among the 4 BMI groups, one would still expect that participants with a BMI <18.5 kg/m² would have the highest mortality, followed by those with a normal BMI, those who are overweight, and then those in the obese BMI category, since age is one of the strongest risk factors for all-cause as well as cardiovascular mortality. In the statistical analysis section of the article, age at baseline was not included in the multivariable model for the mortality analyses. Even with age-specific RRs, there remains a potential for residual confounding, given the wide age range (i.e., 50–70 years).

Table 1.

Distribution of age categories across the 4 body mass index (BMI) groups

Age, years	BMI <18.5 kg/m2 (n = 577), %	BMI 18.5–24.9 kg/m2 (n = 8,632), %	BMI 25.0–29.9 kg/m2 (n = 7,939), %	BMI ≥30.0 kg/m2 (n = 7,407), %
<50	27.3	26.8	20.9	28.3
50–70	34.8	45.2	55.7	58.4
>70	37.9	28.0	23.4	13.3

Second, the RRs for “all ages” were lower than the age-specific RRs within the obesity group when compared with those with a normal BMI. For example, adjusted RRs for all-cause mortality were 1.0 for participants age <50 years, 0.9 for participants ages between 50 and 70 years, and 0.8 for participants age >70 years; however, when all participants were combined, the adjusted RR was 0.6. Similar results were also observed for cardiovascular mortality. If age was to be taken into account, one would expect that the RR estimate for “all ages” would be a weighted mean of the age-specific RRs and would fall within the range of the minimum and maximum age-specific RRs.

Finally, the finding that RRs in older people were attenuated compared with those in younger people among obese RA patients is interesting. A similar phenomenon has been reported and discussed in other fields as well. One explanation is that the risk of mortality in the referent group (i.e., participants with a normal BMI) increases with increasing age, resulting in a decline in relative estimates over the increasing age strata (2). Another potential explanation is depletion of susceptibility, where the participants most vulnerable to obesity-related complications died prior to the age at which they would have been eligible for study enrollment; therefore, those that survive into old age would be less susceptible to the complications of obesity (3).

The research question of whether obesity increases the risk of mortality among RA patients is interesting, since findings from previous studies have been conflicting. Given that age is such a strong predictor of mortality, it would be helpful if the authors could also provide an age-adjusted effect estimate to see whether such an adjustment might have any impact on the reported findings and conclusions.