Comparison of self-reported and accelerometer-measured physical activity in Canadian adults

Background: Self-reported and accelerometer-measured physical activity levels generally exhibit low correlation and agreement. The objective of this study is to compare estimates of physical activity among adults from a newly developed Canadian questionnaire with those obtained objectively by accelerometry. Data and methods: Data for 18- to 79-year-olds (N = 2,372) were collected in 2014 and 2015 as part of the Canadian Health Measures Survey (CHMS). Moderate-to-vigorous physical activity (MVPA) was reported on the household questionnaire by domain (transportation, recreation, and occupational or household) as part of the new Physical Activity Adult Questionnaire (PAAQ) and measured objectively using the Actical accelerometer. Correlation and mean difference analyses were used to assess the relationships between measured and reported physical activity variables. Linear regression was used to test the association between measured and reported physical activity and measures of obesity. Results: On average, Canadian adults reported more physical activity than they accumulated on an accelerometer (49 minutes versus 23 minutes per day). The highest correlation observed was between accelerometer-measured MVPA and the sum of self-reported recreation and transportation activity (R = 0.36, p < 0.0001). The sum of activity from all domains (recreation + transportation + occupational or household) exhibited a lower correlation with measured variables because the occupational or household domain was negatively correlated with MVPA (R = -0.04). The occupational or household domain was positively correlated with light-intensity physical activity (R = 0.20, p < 0.0001). Respondents in the least active quintile were more likely than those in the most active quintile to report more activity than was measured by the accelerometer. On average, the most active quintile reported less activity than was measured by the accelerometer. Interpretation: The newly developed Canadian physical activity questionnaire exhibited modest correlation and agreement with accelerometer-measured physical activity among adults. Accelerometers and questionnaires provide complementary information, about different aspects of physical activity (actual movement versus perceived time). Consequently, one should exercise caution in using estimates derived from these methods interchangeably.