Measuring Physical Activity with Hip Accelerometry among U.S. Older Adults: How Many Days Are Enough?

• Published in: PloS one 2017-01, Vol.12 (1), p.e0170082-e0170082
• Main Authors: Kocherginsky, Masha, Huisingh-Scheetz, Megan, Dale, William, Lauderdale, Diane S, Waite, Linda
• Format: Article
• Language: English
• Subjects: Accelerometers; Accelerometry; Adults; Age; Aged; Agreements; Bias; Biology and Life Sciences; Biomarkers; Demographics; Elderly; Engineering and Technology; Exercise; Exercise - physiology; Female; Geriatrics; Health aspects; Health sciences; Hip; Hip - physiology; Humans; Linear Models; Logistic Models; Male; Measurement; Medicine; Medicine and Health Sciences; Nutrition; Nutrition Surveys; Older people; Palliative care; People and Places; Physical activity; Physical Sciences; Public health; Regression; Research and Analysis Methods; Sedentary behavior; Sedentary Lifestyle; Statistical methods; Studies; United States; Wear
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0170082

Accelerometers are increasingly used in research. Four to 7 days of monitoring is preferred to estimate average activity but may be burdensome for older adults. We aimed to investigate: 1) 7-day accelerometry protocol adherence, 2) demographic predictors of adherence, 3) day of the week effect, and 4) average activity calculated from 7 versus fewer days among older adults. We used the 2003-2006 older adult hip accelerometry data from the National Health and Nutrition Examination Survey (NHANES) sample. We determined proportions with 1-7 valid (10-20 hours) wear days and identified wear day correlates using ordinal logistic regression. We determined the day of week effect on 5 accelerometry measures (counts per minute, CPM; % sedentary behavior; % light-lifestyle activity; % moderate-vigorous activity, MVPA; total activity counts) using multivariate linear regression and compared averages estimated over 2 or 3 versus 7 days using correlations, linear regression, and Bland-Altman plots. Among 2,208 participants aged 65+, 85% of participants had ≥2 and 44% had 7 valid wear days. Increasing age (p = 0.01) and non-white race (p < 0.001) were associated with fewer days. Daily CPM, % MVPA, and total daily activity counts were similar Monday through Saturday, but significantly lower on Sundays (p < 0.001). Daily % sedentary behavior and % light-lifestyle activity were significantly different on Saturdays (p = 0.04-0.045) and Sundays (p < 0.001) compared to weekdays. Among participants with 7 valid days, 2 or 3 day averages were highly correlated with 7 day averages for all 5 accelerometry measures (2 versus 7 days: r = 0.90-0.93, 3 versus 7 days: r = 0.94-0.96). Protocols of 2-3 days, adjusting for Sundays (average CPM, % moderate-vigorous activity, and average total daily activity counts) or weekends (% sedentary behavior and % light-lifestyle activity), give reliable estimates of older adult activity.