A Comparison of Self-Reported and Objective Physical Activity Measures in Young Australian Women

Abstract

Background: The evidence for beneficial effects of recommended levels of physical activity is overwhelming. However, 70% of Australians fail to meet these levels. In particular, physical activity participation by women falls sharply between ages 16 to 25 years. Further information about physical activity measures in young women is needed. Self-administered questionnaires are often used to measure physical activity given their ease of application, but known limitations, including recall bias, compromise the accuracy of data. Alternatives such as objective measures are commonly used to overcome this problem, but are more costly and time consuming.

Objective: To compare the output between the Modified Active Australia Survey (MAAS), the International Physical Activity Questionnaire (IPAQ), and an objective physical activity measure-the SenseWear Armband (SWA)-to evaluate the test-retest reliability of the MAAS and to determine the acceptability of the SWA among young women.

Methods: Young women from Victoria, Australia, aged 18 to 25 years who had participated in previous studies via Facebook advertising were recruited. Participants completed the two physical activity questionnaires online, immediately before and after wearing the armband for 7 consecutive days. Data from the SWA was blocked into 10-minute activity times. Follow-up IPAQ, MAAS, and SWA data were analyzed by comparing the total continuous and categorical activity scores, while concurrent validity of IPAQ and MAAS were analyzed by comparing follow-up scores. Test-retest reliability of MAAS was analyzed by comparing MAAS total physical activity scores at baseline and follow-up. Participants provided feedback in the follow-up questionnaire about their experience of wearing the armband to determine acceptability of the SWA. Data analyses included graphical (ie, Bland-Altman plot, scatterplot) and analytical (ie, canonical correlation, kappa statistic) methods to determine agreement between MAAS, IPAQ, and SWA data.

Results: A total of 58 participants returned complete data. Comparisons between the MAAS and IPAQ questionnaires (n=52) showed moderate agreement for both categorical (kappa=.48, P<.001) and continuous data (r=.69, P<.001). Overall, the IPAQ tended to give higher scores. No significant correlation was observed between SWA and IPAQ or MAAS continuous data, for both minute-by-minute and blocked SWA data. The SWA tended to record lower scores than the questionnaires, suggesting participants tended to overreport their amount of physical activity. The test-retest analysis of MAAS showed moderate agreement for continuous outcomes (r=.44, P=.001). However, poor agreement was seen for categorical outcomes. The acceptability of the SWA to participants was high.

Conclusions: Moderate agreement between the MAAS and IPAQ and moderate reliability of the MAAS indicates that the MAAS may be a suitable alternative to the IPAQ to assess total physical activity in young women, due to its shorter length and consequently lower participant burden. The SWA, and likely other monitoring devices, have the advantage over questionnaires of avoiding overreporting of self-reported physical activity, while being highly acceptable to participants.

Keywords: exercise; physical activity; questionnaires; women’s health.

Figure 1

Equation used to calculate average weekday scores from the SenseWear Armband activity monitor.

Figure 2

Bland-Altman plot of natural log transformed score showing the difference between IPAQ and SWA, plotted against the mean. Note: the lines represent the limits of agreement (95%) and average difference between the two variables.

Figure 3

Bland-Altman plot of natural log transformed score showing the difference between MAAS and SWA, plotted against the mean. Note: the lines represent the limits of agreement (95%) and average difference between the two variables.

Figure 4

Bland-Altman plot of natural log transformed score showing the difference between follow-up IPAQ and MAAS, plotted against the mean. Note: the lines represent the limits of agreement (95%) and average difference between the two variables.

Figure 5

Bland-Altman plot of natural log transformed score showing the difference between baseline and follow-up MAAS, plotted against the mean. Note: the lines represent the limits of agreement (95%) and average difference between the two variables.