Abstract: FR-PO806

Assessment of Wear Compliance, Differences in Step Counts, and Activity Levels of Fitbit and Jawbone Devices to the Actigraph GT3X+ in Hemodialysis Patients

Session Information

Category: Dialysis

  • 601 Standard Hemodialysis for ESRD

Authors

  • Perez, Luis M., University of Illinois, Urbana, Illinois, United States
  • Fang, Hsin-Yu, University of Illinois, Urbana, Illinois, United States
  • Burrows, Brett, University of Illinois, Urbana, Illinois, United States
  • Mullen, Sean P., University of Illinois, Urbana, Illinois, United States
  • Wilund, Ken, University of Illinois, Urbana, Illinois, United States
Background

The Actigraph GT3X+ (ACT) is a research standard in measuring steps and physical activity, but does not have the convenience of newer commercial devices. The objective of this study was to determine wear compliance and compare differences in steps counts of the Fitbit Flex (FLX), Fitbit Charge HR (CHR), and Jawbone Up2 (JWB) to the ACT in hemodialysis (HD) patients.

Methods

We recruited 29 HD patients (age: 53+10 y, BMI: 32.4+11.4 kg/m2, sex: 67% M) to simultaneously wear the ACT (hip), FLX, CHR, & JWB (wrist) for 7 days. A total of four days (2 dialysis & 2 non-dialysis) were selected for validation. We used self-reported wear-time and a validated algorithm to identify and select valid days for steps and sedentary, light, and moderate-to-vigorous (MVPA) activity.

Results

Adherence rate was for patients wearing and completing 4 days of valid data was 50% (13 excluded & 3 drop-outs). The average daily steps by device were: ACT=2666+1202, FLX=3735+1869, CHR=3660+1728, JWB=3041+1801; with a significant mean difference in steps from the ACT to the FLX & CHR (p<0.01), but not for the JWB (p=0.35). Regression of the mean steps to the difference in steps was significant for the FLX, CHR, & JWB to the ACT (all p<0.05). Patients had a greater step count on average for non-dialysis days compared to dialysis days.

Each device had a significant correlation in steps to the Actigraph (FLX r=0.84, p<0.01; CHR r=0.81, p<0.01, JWB r=0.67, p=0.01). The correlation of light activity (min.) was significant for the FLX (r=0.72, p<0.01) & CHR (r=0.67, p<0.01) was significant, but not for sedentary & MVPA activity. Furthermore, the mean difference in light time was not significantly different for the FLX & CHR.

Conclusion

The FLX, CHR, & JWB provided similar step estimates, but each device overestimated steps compared to the Actigraph. The devices provided similar estimates of time spent in light activity, but not sedentary or MVPA activity. It is possible that the algorithms or different positioning of the devices contributed to this discrepancy. Additional studies are needed to confirm these data.