Thank you for the opportunity to respond to the letter from Barreira and Schuna (1) regarding our publication on the association between various accelerometer metrics and physical activity energy expenditure (PAEE) measured by doubly labeled water (DLW) (2). We appreciate the interest in our study as well as the authors pointing out its strengths.
We agree that steps per day has many advantages as a metric for physical activity assessment, including that the “step” is a fundamental unit of human ambulatory activity (3). In addition, as many authors have previously noted, a step is a meaningful output that both researchers and the general public intuitively understand.
Studies have shown that the step counts of different commercial activity trackers have good reliability and validity, in general, for treadmill walking and under free-living conditions. Many of these devices produce mean values that are within 10%–20% of each other (4). This contrasts with another commonly reported metric in physical activity research, minutes of moderate to vigorous physical activity (MVPA), for which means range from 2.9 to 203.6 min·d−1 under free-living conditions, depending on which algorithm is used (5).
Research-grade accelerometers have also been evaluated for step counting accuracy. Studies have shown that, under free-living conditions, when the ActiGraph accelerometer is worn on the hip, steps per day is overestimated when the low-frequency extension is enabled, but underestimated when the low-frequency extension is not enabled (6,7). Other step-counting algorithms might provide a closer estimate of steps per day, such as ActiGraph’s “ActiGO” algorithm or “Moving Average Vector Magnitude” algorithm developed for wrist-worn ActiGraph. More research is needed to determine the most accurate step-counting algorithms for research-grade accelerometers.
The purpose of this study was to compare the correlation between accelerometer-measured physical activity and DLW-assessed PAEE when total volume of activity is considered versus MVPA alone. We investigated three metrics related to total volume and two thresholds for MVPA. We also wanted to present results for both vertical axis and triaxial counts given that some investigators still utilize vertical axis counts although current devices measure acceleration along three axes. Although we agree that it would have been interesting to look at other metrics such as “censored” steps and steps accumulated in different cadence bands, the manuscript was already quite long, and we unfortunately could not include all possible metrics.
It has not escaped our attention that the correlations between DLW-estimated PAEE and steps per day were among the highest correlations observed in our study, and they were just as strong as the correlations between DLW-estimated PAEE and total activity counts per day. Certainly, this lends support to “steps per day” being a useful metric for expressing overall physical activity volume. Medical researchers are now turning toward focusing on “steps” in lieu of “minutes of MVPA” (8). In part, this is due to the intuitive meaning of a “step” for researchers and the general public, but it is also due to the greater degree of standardization of steps per day and the fact that steps per day is closely associated with PAEE and health variables.