File
Kobayashi M1, Fujii K1, Saito T1, Kasahara S1, Shimura T1, Tajima C1, Sakai Y1, Kurokawa K1, Shinohara T2, Usuda S3
1Geriatrics Research Institute and Hospital, Maebashi, Japan, 2Takasaki University of Health and Welfare, Takasaki, Japan, 3Gunma University Graduate School of Health Sciences, Maebashi, Japan
Background: A wrist-worn heart rate (WHR) monitor is a valuable and easily wearable device for estimating exercise intensity in patients with stroke. It is based on optically sensing blood flow, providing the HR during exercise. However, the accuracy of a WHR monitor is not clear.
Purpose: The purpose of this study was to evaluate the accuracy of a WHR monitor in patients with stroke during physical therapy.
Methods: This was a cross-sectional study. Nineteen inpatients (11 males, 7 females; mean±SD age 64.8±10.9 years) with stroke in the recovery phase were recruited. The types of stroke were infarctions (n=7) and hemorrhages (n=12). The numbers of patients in Brunnstrom Recovery stages I, II, III, IV, V, and VI in the legwere 0, 6, 1, 2, 3, and 7. The number of patients with assisted gait, supervised gait, and independent gait were 9, 3, and 7, respectively. Patients wore WHR monitors (Polar A370) on paretic and non-paretic wrists and a chest strap heart rate (CHR) monitor (Polar H10) connected to an Actigraph (wGT3X-BT) by Bluetooth. The CHR was the criterion measure. WHR and CHR were measured every minute during physical therapy sessions three times within a week. Paretic and non-paretic WHRs were compared by paired t-tests. Bland-Altman plots, Pearson's correlation coefficient, and mean absolute percentage error (MAPE) were used to examine relative reliability and absolute reliability. The 95% confidence intervals (95%CIs) were calculated to examine measurement agreement between CHR and WHR.
Results: A total of 3172 data points were measured by each of CHR and WHR. The criterion measure of CHR was 81.7±29.3 bpm, the paretic WHR was 82.4±14.5 bpm, and the non-paretic WHR was 82.4±14.1 bpm. There was no significant difference between paretic and non-paretic WHRs (p >0.05), and their difference was 0.03±0.14 bpm. The difference between CHR and paretic WHR was -0.7±10.2 bpm, and the lower and upper 95%CIs of the difference were -1.04 and -0.33 bpm, respectively. The difference between CHR and non-paretic WHR was -0.7±10.5 bpm, and the lower and upper 95%CIs of the difference were -1.02 and -0.29 bpm, respectively. The mean of CHR and paretic WHR was 82.1±13.3 bpm, and the mean of CHR and non-paretic WHR was 82.1±13.1 bpm. The MAPEs of the paretic side and the non-paretic side were 5.35 and 5.55, respectively. Pearson's correlation coefficients among CHR, paretic WHR, and non-paretic WHR were 0.723-0.844 (all p 0.01). Pearson's correlation coefficient of the mean and difference between CHR and paretic WHR was -0.05 (p 0.01), and that between CHR and non-paretic WHR was -0.012 (p>0.05).
Conclusion(s): Paretic and non-paretic WHR results showed fixed bias and a small proportional bias, but these values were slight in usual exercise. Therefore, the WHR monitor is valuable to estimate exercise intensity for patients with stroke.
Implications: We can use WHR monitors during exercise for patients with stroke to estimate fitness intensity. In stroke rehabilitation, aerobic exercise is encouraged to improve physical functions and prevent a second stroke. We can develop effective exercise using WHR monitors.
Keywords: Heart rate, Data Accuracy, Stroke
Funding acknowledgements: Not applicable
Purpose: The purpose of this study was to evaluate the accuracy of a WHR monitor in patients with stroke during physical therapy.
Methods: This was a cross-sectional study. Nineteen inpatients (11 males, 7 females; mean±SD age 64.8±10.9 years) with stroke in the recovery phase were recruited. The types of stroke were infarctions (n=7) and hemorrhages (n=12). The numbers of patients in Brunnstrom Recovery stages I, II, III, IV, V, and VI in the legwere 0, 6, 1, 2, 3, and 7. The number of patients with assisted gait, supervised gait, and independent gait were 9, 3, and 7, respectively. Patients wore WHR monitors (Polar A370) on paretic and non-paretic wrists and a chest strap heart rate (CHR) monitor (Polar H10) connected to an Actigraph (wGT3X-BT) by Bluetooth. The CHR was the criterion measure. WHR and CHR were measured every minute during physical therapy sessions three times within a week. Paretic and non-paretic WHRs were compared by paired t-tests. Bland-Altman plots, Pearson's correlation coefficient, and mean absolute percentage error (MAPE) were used to examine relative reliability and absolute reliability. The 95% confidence intervals (95%CIs) were calculated to examine measurement agreement between CHR and WHR.
Results: A total of 3172 data points were measured by each of CHR and WHR. The criterion measure of CHR was 81.7±29.3 bpm, the paretic WHR was 82.4±14.5 bpm, and the non-paretic WHR was 82.4±14.1 bpm. There was no significant difference between paretic and non-paretic WHRs (p >0.05), and their difference was 0.03±0.14 bpm. The difference between CHR and paretic WHR was -0.7±10.2 bpm, and the lower and upper 95%CIs of the difference were -1.04 and -0.33 bpm, respectively. The difference between CHR and non-paretic WHR was -0.7±10.5 bpm, and the lower and upper 95%CIs of the difference were -1.02 and -0.29 bpm, respectively. The mean of CHR and paretic WHR was 82.1±13.3 bpm, and the mean of CHR and non-paretic WHR was 82.1±13.1 bpm. The MAPEs of the paretic side and the non-paretic side were 5.35 and 5.55, respectively. Pearson's correlation coefficients among CHR, paretic WHR, and non-paretic WHR were 0.723-0.844 (all p 0.01). Pearson's correlation coefficient of the mean and difference between CHR and paretic WHR was -0.05 (p 0.01), and that between CHR and non-paretic WHR was -0.012 (p>0.05).
Conclusion(s): Paretic and non-paretic WHR results showed fixed bias and a small proportional bias, but these values were slight in usual exercise. Therefore, the WHR monitor is valuable to estimate exercise intensity for patients with stroke.
Implications: We can use WHR monitors during exercise for patients with stroke to estimate fitness intensity. In stroke rehabilitation, aerobic exercise is encouraged to improve physical functions and prevent a second stroke. We can develop effective exercise using WHR monitors.
Keywords: Heart rate, Data Accuracy, Stroke
Funding acknowledgements: Not applicable
Topic: Neurology: stroke; Robotics & technology; Outcome measurement
Ethics approval required: Yes
Institution: Geriatrics Research Institute and hospital
Ethics committee: Institutional Review Board
Ethics number: No 60
All authors, affiliations and abstracts have been published as submitted.
