Adjusting athletes' body mass index to better reflect adiposity in epidemiological research.

The aim of the present study was to identify when body mass index (BMI) is unlikely to be a valid measure of adiposity in athletic populations and to propose a simple adjustment that will allow the BMI of athletes to reflect the adiposity normally associated with non-athletic populations. Using data from three previously published studies containing 236 athletes from seven sports and 293 age-matched controls, the association between adiposity (sum of 4 skinfold thicknesses, in millimetres) and BMI was explored using correlation, linear regression, and analysis of covariance (ANCOVA). As anticipated, there were strong positive correlations (r = 0.83 for both men and women) and slope parameters between adiposity and BMI in age-matched controls from Study 1 (all P < 0.001). The standard of sport participation reduced these associations. Of the correlations and linear-regression slope parameters between adiposity and BMI in the sports from Studies 2 and 3, although still positive in most groups, less than half of the correlations and slope parameters were statistically significant. When data from the three studies were combined, the ANCOVA identified that the BMI slope parameter of controls (5.81 mm (kg m(-2))(-1)) was greater than the BMI slope parameter for sports (2.62 mm (kg m(-2))(-1)) and middle-distance runners (0.94 mm (kg m(-2))(-1)) (P < 0.001). Based on these contrasting associations, we calculated how the BMI of athletes can be adjusted to reflect the same adiposity associated with age-matched controls. This simple adjustment allows the BMI of athletes and non-athletes to be used with greater confidence when investigating the effect of BMI as a risk factor in epidemiological research.