Effect Of Hand Dominance On Accuracy Of Wrist-worn Physical Activity Trackers

Abstract

The increasing public awareness regarding the importance of physical activity (PA) for health has contributed to the rising popularity of personal PA trackers that track energy expenditure, activity intensity, and steps. Many popular PA trackers are designed to be worn on the wrist, but it is unclear if their accuracy is impacted by whether the trackers are worn on the dominant (D) or non-dominant (ND) wrist. PURPOSE: The purpose of this study was to investigate the impact of D vs. ND wrist placement on estimates of energy expenditure (EE, kcals), active minutes (ActMin), and steps between two popular personal PA trackers during a semi-structured protocol. METHODS: Thirty adults (15 male; age 49±20 years) wore two types of PA trackers (FB and JB, both worn on each wrist) and a portable metabolic analyzer while completing an 80-min, semi-structured PA protocol. Participants performed at least 12 of 21 sedentary, household, and exercise/ambulatory activities, with at least half of the visit time spent in sedentary activities. PA tracker estimates for EE, ActMin (METS >3.0), and steps were compared to criterion values measured by the metabolic analyzer (EE and ActMin) and manually counted steps (steps). Repeated measures analysis of variance and dependent t-tests were used to analyze differences between D and ND wrist placement and the accuracy of the trackers compared to criterion measures. RESULTS: Estimates of EE, ActMin, and steps between the D and ND FB trackers were not significantly different (276+19 vs. 258+11 kcals; 12+1 vs. 11+1 ActMin; 2483+147 vs. 2502+141 steps, respectively). Similarly, EE, ActMin, and steps from the D and ND JB trackers were not significantly different (217.5+13 vs. 213+12 kcals, 19+1 vs. 19 +1 min, 2394+181 vs. 2341+167 steps, respectively). However, both the FB and JB on both wrists underestimated EE (13-28% p<0.001), ActMin (30-55%, p<0.001), and steps (24-29%, p<0.001) relative to criterion measures. CONCLUSIONS: Choice of wrist for PA tracker placement did not affect PA tracker accuracy during an 80-min, semi-structured protocol. Estimates from wrist-worn, commercial PA trackers should be interpreted with caution due to their underestimation of PA variables. Support from the Ball State University ASPiRE Student Research Grant, and CAST Internal Grant.