Background. Maximal oxygen consumption (VO2max) during exercise and the percentage of predicted VO2max (%pVO2max) are standard measures of cardiorespiratory fitness with established clinical predictive val-ue. They are usually measured during a cardiopulmonary exercise test (CPET), which requires expensive equipment and experienced personnel. Novel smartwatches estimate VO2max using proprietary algorithms, but their accuracy remains undetermined. Aims. To determine smartwatch device accuracy in estimating VO2max and %pVO2max using data from CPET as a reference.
Methods. 215 adults (44 (21%) male; median [interquartile range; IQR] 56 [32, 62] years old) were recruited from two population-based cohorts, the Avon Longitudinal Study of Parents and Children (ALSPAC) and TwinsUK. Participants performed a maximal CPET on a semi-recumbent bike to meas-ure VO2max and wore a Garmin Vivoactive 4s (GV4) smartwatch for 60 days following CPET. The first and last VO2max estimates provided by GV4 were compared to CPET measures. %pVO2max was measured using predict-ed VO2max from Wasserman and Whipp's anthropometric-based equations. In a subgroup wearing the smartwatch during CPET, peak heart rate (HR) from GV4 and CPET was also compared. Agreement was assessed using Bland-Altman analysis (bias and limits of agreement (LoA)), absolute per-centage error (APE), reported as median [interquartile range], and Pearson's correlation coefficient (cc). Results. VO2max and %pVO2max measured during CPET was 22.4 [17.5, 27.4] ml/kg/min and 90.9% [78.1%, 101.3%], respectively. VO2max esti-mates from GV4 were moderately correlated with CPET measures (cc=0.62 and 0.66 for first and last estimates) and showed a large positive bias ~14 ml/kg/min with LoA from 0 – 27 ml/kg/min. Correlation between VO2max from GV4 and anthropometric-based prediction of VO2max was high (cc>0.90). Agreement between %pVO2max from GV4 and CPET was poor (cc~0.15, bias ~52%, LoA 7-98 %).
Conclusions. GV4 provides estimates of VO2max that overestimates but moderately correlates with CPET measured VO2max. The agreement for %p VO2max is poor.