To the Editor: The objective of the cohort study by Dr Forman and colleagues1 was to estimate the hypothetical fraction of hypertension incidence associated with dietary and lifestyle factors in women. I believe that the wrong formula was used to calculate the population attributable fraction or risk (PAR).
The study by Forman et al used the crude formula for PAR, which is only appropriate when the association between the exposure and the outcome (as represented by the hazard ratio [HR]) is not adjusted for confounding variables.2 However, the authors adjusted their HRs for 8 variables they believed were confounders of the association between dietary and lifestyle factors and hypertension incidence. The use of adjusted HRs in the crude formula for the PAR is a common error in calculating attributable fractions.2 A review of methods to estimate the PAR reported that using adjusted HRs in the crude formula was “ . . . shown to yield inconsistent estimates . . . and, accordingly, severe bias was exhibited in simulations. . . . ”3
Williamson DF. Population Attributable Risk of Incident Hypertension in Women. JAMA. 2009;302(23):2550-2551. doi:10.1001/jama.2009.1826