Accuracy and Systematic Biases of Heart Rate Measurements by Consumer-Grade Fitness Trackers in Postoperative Patients: Prospective Clinical Trial

Abstract

Background Over the recent years, technological advances of wrist-worn fitness trackers heralded a new era in the continuous monitoring of vital signs. So far, these devices have primarily been used for sports. Objective However, for using these technologies in health care, further validations of the measurement accuracy in hospitalized patients are essential but lacking to date. Methods We conducted a prospective validation study with 201 patients after moderate to major surgery in a controlled setting to benchmark the accuracy of heart rate measurements in 4 consumer-grade fitness trackers (Apple Watch 7, Garmin Fenix 6 Pro, Withings ScanWatch, and Fitbit Sense) against the clinical gold standard (electrocardiography). Results All devices exhibited high correlation (r≥0.95; P<.001) and concordance (rc≥0.94) coefficients, with a relative error as low as mean absolute percentage error <5% based on 1630 valid measurements. We identified confounders significantly biasing the measurement accuracy, although not at clinically relevant levels (mean absolute error<5 beats per minute). Conclusions Consumer-grade fitness trackers appear promising in hospitalized patients for monitoring heart rate. Trial Registration ClinicalTrials.gov NCT05418881; https://www.clinicaltrials.gov/ct2/show/NCT05418881

Reliability of continuous vital sign monitoring in post-operative patients employing consumer-grade fitness trackers: A randomised pilot trial

“…>30 bpm) from the ground truth references. Reassuringly, our findings are consistent with our previous observations (r > 0.95 and bias <1.21 bpm) in a study on resting patients under controlled conditions, 8 and also in agreement with complementary literature. 24 Employing our predefined thresholds (i.e.…”

“…Standard attributes, including sex, age, BMI (including height and weight), heart rhythm, wrist circumference, skin tonality on the Fitzpatrick scale, hair density according to the scale of an in-house categorisation (0 = no forearm hair, 1 = minimal, 2 = moderate, and 3 = extensive hair density on the forearm) and type and length of surgery were recorded for each participant. 8 The duration of a surgery was defined as the time span from the beginning to the end of anaesthesia. Furthermore, the Charlson comorbidity score, the ASA score and Barthel score were documented in order to assess pre-existing diseases and the overall physical state.…”

“…However, state of the art clinical standard bedside monitors used in perioperative or intensive care units, seems unsuitable for tackling the challenge of continuous monitoring, as it generates disproportionate cost and as an even more important reason severely impairs the mobility of patients. With this in mind wrist-based consumer-grade fitness trackers based on the innovative technology of reflective pulse oximetry employing photoplethysmography (PPG) sensors 8 could offer an auspicious possibility of continuously monitoring vital signs at manageable costs and without impairing the mobility of patients. In short, PPG derives vital parameters based on the intensity of multi-spectral light reflected by the tissue, which is altered by shifts in the absorption spectra caused by physiological processes.…”

“… 11 Previous validation studies of fitness trackers have shown encouraging results for measuring different vital signs, but prevailingly were limited to on-demand measurements under controlled or even under laboratory conditions. 8 , 12 , 13 Longitudinal validation studies examining the performance of automated vital sign measurements by fitness trackers under real-world conditions in hospitalised patients are essential but have yet to be conducted.…”

Evaluating blood oxygen saturation measurements by popular fitness trackers in postoperative patients: A prospective clinical trial

“…So far, some consumer-grade devices have demonstrated acceptable accuracy for the heart rate monitoring in hospitalized patients, 9 as well as for measuring Sp O 2 over a broad range of oxygen saturation levels in resting healthy subjects (Bias +0.0% LoA [-4.9; 4.9] in hypobaric chambers 10 ; Bias +0.98% LoA [-4.66; 6.62] while breathing a hypoxic gas mixture). 11 Also employing the Apple Watch in outpatients with chronic lung disease suggests a promising measurement accuracy for Sp O 2 (Bias +0.8% LoA [-2.7; 4.1]).…”