Download Full Text (449 KB)


Low physical activity is associated with heart failure with preserved ejection fraction (HFpEF). Step-counts, a measure of physical activity, can be measured via accelerometry. To date, few studies have examined validity of accelerometer-derived step-counts in the adults with HFpEF.

PURPOSE To assess criterion validity of the Actigraph GT9X accelerometer step-count function in adults with HFpEF via ankle, waist, and wrist placement, compared with manually counted, directly observed steps.

METHODS Six adults with HFpEF (age: 57.2 ± 9.4 y; African American: 50%; females: 100%) completed a cardiopulmonary exercise test (CPET) on a treadmill while wearing synchronized GT9X accelerometers on the ankle, waist, and wrist. Steps during CPET were measured by using the step-count function on the GT9X at 60 Hz sampling and data were downloaded into 1-second and 10-second epochs. Hand-tallied, directly observed steps (OS) was the criterion measure. Criterion validity was assessed via paired t tests to determine whether mean total steps (TS) from the three devices were significantly different from the mean TS from OS, and Pearson correlations were used to determine associations between device-measured TS and the total OS. Simple linear regression models were used to assess the effect of walking speed on absolute percentage error of the devices compared to OS. Agreement of the devices throughout the duration of the CPET was examined using Pearson correlations. Alpha was set at 0.05 for all statistical analyses.

RESULTS Mean TS from waist-worn (t = -5.29, p = .001) and wrist-worn (t = -12.50, p < .001) devices were significantly lower than mean TS from OS. Only TS from the ankle GT9X was significantly associated with TS from OS (r = 0.974, p = .001). GT9X-estimate steps from the ankle (r = 0.869, p < .001), waist (r = 0.550, p <.001), and wrist (r = 0.429, p <.001) were all significantly associated with OS-measured steps. Absolute percentage error was significantly and negatively associated with treadmill speed for devices on the ankle (b = - 10.70, p < .001), waist (b = -32.49, p < .001) and wrist (b = -10.08, p < .001).

CONCLUSION Our results suggest that accelerometer-derived TS may be a more accurate measure of TS when the device is worn on the ankle rather than waist or wrist, and that measurement error is higher at lower walking speed.

Publication Date



heart failure, physical activity, step count, accelerometry, HFpEF, validity


Cardiovascular Diseases | Exercise Science

Faculty Advisor/Mentor

Youngdeok Kim, PhD

Is Part Of

VCU Graduate Research Posters

Validity of the Actigraph GT9X Accelerometer Step-Count Function in Adults with Heart Failure with Preserved Ejection Fraction