Accuracy Of Personally Owned Wearable Activity Monitors For Estimating VO2max

BACKGROUND: Cardiorespiratory fitness is indicated by maximal oxygen consumption (VO2max). VO2max values are inversely related to all-cause mortality and the risk of developing cardiovascular disease, thus making it an important health variable. The gold standard method of assessing VO2max via indirect calorimetry requires laboratory equipment, trained test administrators, and vigorous exercise on behalf of the participants, making it largely inaccessible. Although submaximal exercise tests and questionnaires exist to circumnavigate the inaccessibility of indirect calorimetry, this variable remains under-utilized. To improve its accessibility, many consumer-grade (CG) wearable activity monitors offer proprietary methods of estimating VO2max based on tracked aerobic workouts. However, these predictive algorithms have not been rigorously externally validated. PURPOSE: The purpose of this study was to compare VO2max estimates from personally owned CG activity monitors to VO2max measured by indirect calorimetry. METHODS: Fifty-five (23.3 ± 4.3 yrs) healthy individuals who owned a CG activity monitor with a predicted VO2max completed a VO2max test using a treadmill-based ramp protocol. The ramp protocol included walking or running at a constant speed while the grade increased by 1% at the end of every minute. Participants were encouraged to continue exercising until maximal exertion. VO2max was assessed with a metabolic gas analyzer (CardioCoach); oxygen consumption (ml·kg − 1·min − 1), respiratory exchange ratio, rating of perceived exertion, and heart rate were recorded for each stage. Paired samples t-tests compared the monitor-predicted to measured VO2max for 1) all monitors grouped together, 2) Apple monitors, and 3) Garmin monitors. RESULTS: When all monitors were grouped together, VO2max estimates were 8.4 ± 8.2 ml·kg − 1·min − 1 lower than measured VO2max (P < .001). Apple monitors estimated 9.7 ± 9.1 ml·kg − 1·min − 1 lower than measured VO2max (P < .001), and Garmin estimated 5.9 ± 6.8 ml·kg − 1·min − 1 lower than measured VO2max (P = .003). CONCLUSIONS: VO2max predicted by CG activity monitors was significantly lower than lab-measured VO2max. This discrepancy between estimated and measured VO2max should be considered when assessing cardiorespiratory fitness using CG activity monitors.