Self-reported exercise often inaccurate, says study

Self-reports of physical activity level tend to be inaccurate when compared against objective measurements, a recent study has found. Such accuracy is affected by sex and body mass index (BMI).

The study included 156 participants (aged 38–43 years, 58.3 percent women) who were made to wear the SenseWear Armband (SMA) device for objective and continuous measurement of physical activity over a 7-day period. Participants were also asked to complete the Physical Activity Recall (PAR) questionnaire and the Active Australia Survey (AAS).

The SWA recorded a high level of physical activity in the study population, with 91 percent meeting the government-recommended guidelines of 150 minutes of moderate-to-vigorous physical activity per week. In contrast, the PAR and AAS self-reports only found that 85 percent and 73 percent of participants, respectively, met this benchmark.

Multivariate general linear models showed significant and positive correlations between SMA and both self-reported measures at most intensities of exercise. However, significant absolute differences in measurements were also detected, with both surveys under-reporting physical activity.

For example, participants reported 69-percent less light physical activity per week using PAR and 38-percent less moderate physical activity using AAS as compared with SWA findings. On the other hand, PAR and AAS self-reports overestimated vigorous exercise by 91 and 43 minutes per week, respectively.

Sex-stratified analyses revealed that the interaction between PAR and SWA-measured physical activity was weaker in women. Moderator analysis revealed that adding BMI to the models altered the interaction between PAR and SWA for moderate-intensity exercise, and between AAS and SWA for vigorous-intensity exercise.

“The use of objective measurements should be prioritized within research, particularly in samples where measurement noise may be loud. Where objective measures are not feasible, validated questionnaires should be implemented, with the choice of measure considering individual and sociodemographic factors such as sex and BMI,” the researchers said.