Assessment of heart rate measurements obtained from a smart collar compared to 24-h Holter monitoring in healthy dogs

Abstract

Introduction/Objectives

The primary objective was to compare the 24-h mean heart rate (HR) provided by a smart collar with 24-h ambulatory electrocardiography (Holter) in healthy dogs. The secondary objective was to compare the 2-min HR values between the two methods during periods of activity and rest.

Animals, materials, and methods

Twelve healthy dogs were fitted with both Holter monitors and smart collars. Passing-Bablok regression and Bland-Altman (BA) agreement analysis were used to compare the 24-h mean HR between two methods. The BA analysis for repeated measures and bias plots were used to compare 2-min HR between methods.

Results

The regression analysis showed no significant differential or proportional bias between the methods to estimate 24-h mean HR. The BA analysis showed a mean bias of 2.2 beats per minute (bpm) (95% confidence interval = −0.2, 4.8) with an upper limit of agreement (LOA) of 9.6 bpm (5.1, 14.1) and a lower LOA of −5.1 bpm (−9.5 to −0.6). However, BA analysis of 2-min HR showed poor agreement between methods with wide LOAs at rest and during activity. The smart collar did not provide any HR information for 43% of the total possible recording duration (range = 24%–79%).

Study Limitations and Conclusions

The smart collar can provide a potentially clinically useful estimate of 24-h HR in dogs with normal sinus rhythm. The collar did not provide reliable 2-min HR measurements due to inaccuracies in HR estimation during periods of activity and the inability to report any HR during large periods of the recording.