Validity and reproducibility of self-reported total physical activity-differences by relative weight

OBJECTIVE: Physical activity is hypothesized to reduce the risk of obesity and several other chronic diseases and enhance longevity. However, most of the questionnaires used measure only part of total physical activity, occup ational and/or leisure-time activity, which might lead to misclassification of total physical activity level and to dilution of risk estimates. We eva luated the validity and reproducibility of a short self-administered physic al activity questionnaire, intended to measure long-term total daily 24 h p hysical activity. METHOD: The questionnaire included questions on level of physical activity at work, hours per day of walking/bicycling, home/household work, leisure-t ime activity/inactivity and sleeping and was sent twice during one year (wi nter/spring and late summer). Two 7-day activity records, performed 6 month s apart, were used as the reference method. One-hundred and eleven men, age d 44-78, completed the questionnaire and one or two activity records. The p hysical activity levels were measured as metabolic equivalents (MET)xh/day. RESULTS: Spearman correlation coefficient between total daily activity scor e estimated from the first questionnaire and the records (validity) was 0.5 6 (deattenuated) and between the first and the second questionnaire (reprod ucibility) 0.65. Significantly higher validity correlations were observed i n men with self-reported body mass index below 26 kg/m(2) than in heavier m en (r = 0.73 vs r = 0.39). CONCLUSIONS: This study indicates that the average total daily physical act ivity scores can be estimated satisfactorily in men using this simple self- administered questionnaire.