Physical activity assessment in population surveys: can it really be simplified?

BACKGROUND: Several studies have used a simplified approach for the assessment of physical activity such as the frequency of exercise-induced sweating. In this study leisure-time physical activity has been assessed using this and another more detailed measure. SUBJECTS AND METHODS: A sample of 4171 adults answered the Health Interview Survey of Barcelona in 1992. The respondents were classified into categories depending on participation in moderate and/or intense physical activity (> or =20 min) and also according to the frequency of exercise-induced sweating: 0, 1-2 and > or =3 times/week. Agreement between the two measures was calculated using the weighted Kappa (Kw) statistic with 95% confidence intervals (95% CI). Stratified analyses were performed. RESULTS: Prevalence of physical activity > or =3 times/week was lower with the sweat question (12.5%) than with the questions about the frequency of performance of selected activities (19.6%). The physical activity patterns by age, gender and overweight were similar for the two measures, but differed by month of the year. Agreement was lower among the older age categories and was higher among males (Kw = 0.59, 95% CI: 0.57-0.62) than among females (Kw = 0.48, 95% CI: 0.46-0.50). Overall, the agreement was higher in the hotter months (Kw = 0.72 among males and 0.58 among females). CONCLUSIONS: In the assessment of physical activity in the population by means of the sweat question there can be interference from other variables, apart from the intensity of the activity, which influence sweating during the exercise. Further assessments of the validity of exercise-induced sweating in representative samples of the general population would be useful.