Repeatability of lifetime exercise reporting.
The purpose of the study was to determine the reliability of lifetime exercise data obtained through a structured interview. Interviews were conducted in 1992-1993 and repeated in 1997 in 150 monozygotic male twins, aged 35-69 years, from the population-based Finnish Twin Cohort. Exercise mode, frequency, duration, intensity and period of participation were solicited for each regularly performed exercise from 12 years of age to the present. Questions related to the most common exercise mode reported in the initial interview were repeated in all subjects and the entire exercise interview was repeated in a subgroup of 38 subjects. The repeatability was highest for exercise years and mean hours/ week by mode for the most commonly performed exercise (Mean ICC=0.63-0.90), and for the sum of all lifetime exercises reported (Mean ICC = 0.69-0.73). The lowest repeatability was found for exercise intensity (Mean Kappa = 0.33-0.48). Similarly poor reliability was found for whether or not exercise was performed at a competitive level (Mean Kappa = 0.25-0.63). Overall, the structured interview of lifetime exercise was most repeatable for years of exercise and mean hours/week. Thus, these exposure variables should be considered in retrospective studies of exercise effects.