Simultaneous validation of ten physical activity questionnaires in older men: A doubly labeled water study

OBJECTIVE: The purpose of this study was to simultaneously validate 10 physical activity (PA) questionnaires in a homogenous population of healthy elderly men against the reference method: doubly labeled water (DLW). DESIGN: Cross-sectional study. SETTING: Community-based sample from Lyon, France. PARTICIPANTS: Nineteen healthy old men (age 73.4 +/- 4.1 years), recruited from various associations for elderly people in Lyon, agreed to participate in the study. MEASUREMENTS: The questionnaire-derived measures (scores) were compared with two validation measures: DLW and maximal oxygen uptake (VO2max). With the DLW method three parameters were calculated: (1) total energy expenditure (TEE), (2) physical activity level (PAL), i.e., the ratio of TEE to resting metabolic rate, (3) energy expenditure of PA. RESULTS: Relative validity. Correlation between the questionnaires and TEE ranged from 0.11 for the Yale Physical Activity Survey (YPAS) total index to 0.63 for the Stanford usual activity questionnaire. This questionnaire also gave the best correlation coefficients with PAL (0.75), and with VO2max (0.62). Significant results with TEE measured by the DLW method were also obtained for college alumni sports score, Seven Day Recall moderate activity, and Questionnaire d'Activite Physique Saint-Etienne sports activity (r = 0.54, r = 0.52, and r = 0.54, respectively). Absolute validity. No difference was found between PA measured by the Seven Day Recall or by the YPAS and DLW, on a group basis. The limits of agreement were wide for all the questionnaires. CONCLUSIONS: Only a few questionnaires demonstrated a reasonable degree of reliability and could be used to rank healthy older men according to PA. Correlation coefficients were best when the Stanford Usual Activity Questionnaire was compared with all the validation measures. The two questionnaires reporting recent PA, the Seven Day Recall, and YPAS accurately assessed energy expenditure for the group. The individual variability was high for all the questionnaires, suggesting that their use as a proxy measure of individual energy expenditure may be limited.