How many days of pedometer monitoring predict monthly ambulatory activity in adults?

Purpose: To determine how many (lays of pedometer monitoring are necessary to estimate monthly ambulatory activity in adults. Methods: Two hundred and twelve adults (64% female; age = 38.3 +/- 13.3 yr: body mass index 27.9 +/- 5.3 kg.m(-2)) wore a pedometer (SW-200) for 28 consecutive days. About 76.4% were randomly allocated to a reliability group, whereas the remainder (n = 50) comprised a confirmation group. Mean step counts calculated over the 28-d period served as the criterion. Using the reliability group, intraclass correlations (ICC) were computed for the entire 4-wk period, for 3, 2, and 1 wk, and for different combinations of any 6, 5 4 3, and 2 d. The reliability of the recommended time frame was tested in the confirmation group using regression analysis, Results: In the reliability group, the ICC for any single given day was 0.41. All combinations including 6 d or more had ICC above 0.80. The inclusion of participant characteristics into a regression, alongside mean steps reported during 1 wk of monitoring, failed to strengthen the prediction. When tested in the confirmation group, there was a significant relationship between mean step counts calculated from the first week of monitoring and the criterion (adjusted R-2 = 0.91, P < 0.001). Conclusion: It is recommended that researchers collect pedometer data over a 7-d period for a reliable estimate of monthly activity in adults, A 7-d period is recommended, as opposed to 6 d (where ICC > 0.80) because: 1) step counts are characteristically lower on a Sunday; thus, for a reliable estimate of habitual activity, Sunday activity should always be included: and 2) in the event of missing data (1 d), data collected on 6 d will remain sufficiently reliable to estimate mean monthly activity.