Cristal Benitez1, Andrew Gomez2, Lindsay P. Toth2. 1University of Alabama, Tuscaloosa, AL. 2University of North Florida, Jacksonville, FL.

BACKGROUND: The purpose of this research was to investigate step count accumulation in research-grade activity monitors during gym-based aerobic exercise activities at moderate and vigorous intensities. METHODS: This was part of a larger parent study including consumer-grade activity monitors and heart rate (HR) accuracy. In this sub-study 22 participants (29 ± 3.8 yr) wore a Polar T-31 strap across the chest, StepWatch 4 activity monitor (SW) on the right ankle, and an ActiGraph GT9X (AG) on the wrist, randomly assigned left or right. Participants completed aerobic exercise on six exercise machines for 3-min at moderate and vigorous intensities. Exercise included step-like activities (SL-A; i.e., elliptical, stair climber, arc trainer) and non-step-like activities (NSL-A; i.e., row machine, Airdyne bike, and stationary bike). During SL-A, video recordings of the lower body were captured, and steps were subsequently hand counted (HC). From the AG, steps were recorded from the screen before and after each 3-min exercise bout to determine steps per bout (AGMAVM). Additionally, AG data were downloaded with (AGLFE) and without (AGN) the low frequency extension (LFE), and then were summed across the 3-min bouts. Steps were downloaded from the SW (initialized with default settings) and summed across each 3-min bout. For SL-A, steps were converted to percent of HC steps and for NSL-A, steps accumulated during 3-min bouts were divided by 3 to produce a steps/min value. One-sample t-tests were used to compare SL-A to 100% of HC steps and NSL-A to 0 steps (because zero step-like motions were completed during these exercises). RESULTS: For SL-A, the SW captured 95.1 to 99.0% of HC steps, while the AGMAVM, AGLFE, and AGN captured 0.1 to 68.5%, 55.7 to 71.9%, and 14.3 to 53.6% of HC steps, respectively. T-test showed all step counting methods to be significantly different from 100% of HC except for SW during elliptical (P=.74) and step climber (P=.36) activities. For NSL-A, the SW captured 1.1 to 131.5 steps/min, while the AGMAVM, AGLFE, and AGN captured 1.2 to 54.7, 48.2 to 56.0, and 6.9 to 40.8 steps/min, respectively. T-test results showed all step counting methods for all activities significantly over-estimated steps (P<.05), when compared to 0 steps/min taken. CONCLUSIONS: Many aerobic exercise machines, even those that mimic stepping, show large step count error and thus should be interpreted with caution.

This document is currently not available here.