In recent years, it has become feasible to use devices for assessing physical activity and sedentary behavior among large numbers of participants in epidemiologic studies, allowing for more precise assessments of these behaviors and quantification of their associations with health outcomes. Between 2011–2015, the Women’s Health Study (WHS) used the Actigraph GT3X+ device to measure physical activity and sedentary behavior over seven days, during waking hours, among 17,708 women (Mage, 72 years) living throughout the United States. Devices were sent to and returned by participants via mail. We describe here the methods used to collect and process the accelerometer data for epidemiologic data analyses. We also provide metrics that describe the quality of the accelerometer data collected, as well as expanded findings regarding previously published associations of physical activity or sedentary behavior with all-cause mortality during an average follow-up of 2.3 years (207 deaths). The WHS is one of the earliest “next generation” epidemiologic studies of physical activity, utilizing wearable devices, in which long-term follow-up of participants for various health outcomes is anticipated. It therefore serves as a useful case study in which to discuss unique challenges and issues faced.
I-Min Lee, Eric J. Shiroma, Kelly R. Evenson, Masamitsu Kamada, Andrea Z. LaCroix and Julie E. Buring
Michael J. LaMonte, I-Min Lee, Eileen Rillamas-Sun, John Bellettiere, Kelly R. Evenson, David M. Buchner, Chongzhi Di, Cora E. Lewis, Dori E. Rosenberg, Marcia L. Stefanick and Andrea Z. LaCroix
Background: Limited data are available regarding the correlation between questionnaire and device-measured physical activity (PA) and sedentary behavior (SB) in older women. Methods: We evaluated these correlations in 5,992 women, aged 63 and older, who completed the Women’s Health Initiative (WHI) and Community Healthy Activities Model Program for Seniors (CHAMPS) PA questionnaires and the CARDIA SB questionnaire prior to wearing a hip-worn accelerometer for 7 consecutive days. Accelerometer-measured total, light, and moderate-to-vigorous PA (MVPA), and total SB time were defined according to cutpoints established in a calibration study. Spearman coefficients were used to evaluate correlations between questionnaire and device measures. Results: Mean time spent in PA and SB was lower for questionnaire than accelerometer measures, with variation in means according to age, race/ethnicity, body mass index, and functional status. Overall, correlations between questionnaires and accelerometer measures were moderate for total PA, MVPA, and SB (r ≈ 0.20–0.40). Light intensity PA correlated weakly for WHI (r ≈ 0.01–0.06) and was variable for CHAMPS (r ≈ 0.07–0.22). Conclusion: Questionnaire and accelerometer estimates of total PA, MVPA, and SB have at best moderate correlations in older women and should not be assumed to be measuring the same behaviors or quantity of behavior. Light intensity PA is poorly measured by questionnaire. Because light intensity activities account for the largest proportion of daily activity time in older adults, and likely contribute to its health benefits, further research should investigate how to improve measurement of light intensity PA by questionnaires.