In Reply.
—Drs Pahor and Carbonin query our use of different classification criteria in analysis of physical activity and longevity in a cohort of Harvard alumni. We used three schemes for categorizing physical activity: (1) increments of 2100 kJ/wk; (2) quintiles of total energy expenditures (<2524, 2524 to <4738, 4738 to <8001, 8001 to <13142, and >13142 kJ/wk), and (3) five categories each of vigorous and nonvigorous energy expenditure, using identical cutoffs (<630, 630 to <1680, 1680 to <3150, 3150 to <6300, and >6300 kJ/wk). Since vigorous and nonvigorous energy expenditure add up to total energy expenditure, the cutoffs for category 3 are, of course, lower than those for category 2.Ideally, we would have used category 1 for all analyses; however, using increments of 2100 kJ/wk led to small numbers of deaths (<50) within some categories (Table 2). Thus, for age-adjusted analysis, stable estimates for mortality rates could be