The accuracy of self-reported physical activity questionnaires varies with sex and body mass index

Abstract

Background Factors contributing to the accurate measurement of self-reported physical activity are not well understood in middle-aged adults. We investigated the associations between two self-reported surveys and objectively measured physical activity in middle-aged adults, and the influence of individual and sociodemographic factors on these associations, at different intensities utilizing an observational study design. Methods Participants (n = 156) wore a SenseWear Armband™ (SWA) for a continuous seven-day period over the triceps of the left arm, to measure energy expenditure in metabolic equivalents. Participants also completed the Physical Activity Recall questionnaire (PAR) and Active Australia Survey (AAS). Associations were analyzed separately in general linear models for each intensity. The influence of individual and sociodemographic factors was assessed through moderator analyses. Results The PAR and SWA were significantly positively associated at moderate (β = 0.68, 95% CI 0.16–1.20), vigorous (β = 0.36, 95% CI 0.20–0.53), moderate-to-vigorous physical activity (MVPA) (β = 0.52, 95% CI 0.20–0.83), and total METmins (β = 0.63, 95% CI 0.35–0.90), the AAS and SWA were associated at all intensities (moderate (β = 0.41, 95% CI 0.15–0.67), vigorous (β = 0.32, 95% CI 0.19–0.46), MVPA (β = 0.42, 95% CI 0.18–0.65) and total METmins (β = 0.62, 95% CI 0.29–0.96). A significant interaction between the PAR and sex for vigorous-intensity unveiled a weaker association in women. Both surveys tended to under-report physical activity. The largest margins of error were present at light and moderate intensities. For the PAR, participants reported over 20 hours, or 69% less light physical activity than recorded by the SWA per week. For the AAS, participants reported over 7 hours, or 38% less moderate physical activity. Compared to lighter intensities, time spent at a vigorous intensity was overreported by participants with the PAR and AAS by 91 and 43 minutes per week, respectively. The addition of Body Mass Index (BMI) resulted in non-significant interactions between the PAR and SWA for moderate-intensity, and the AAS and SWA for vigorous-intensity; a significant interaction between AAS and BMI indicated that the strength of the association differed by BMI for vigorous-intensity. Conclusions The PAR and AAS are not equivalent to the SWA, and sex and BMI may alter the associations between the measures.