Establishing statistical stability for heart rate variability in horses

Abstract

Heart rate variability (HRV) is used increasingly as a quantitative marker for health in humans and welfare in animals. By evaluating the variability expressed in HRV measures, the information can be used to discern physiological stresses acting on the sympathetic and parasympathetic systems in humans and animals. In equine studies, HRV is used as a noninvasive method to assess stress under various circumstances; however, statistical reliability of HRV measures has not been proven. The purpose of the present study was to establish the statistical stability of HRV measures obtained with a portable heart rate monitor using a test-retest approach. Time and frequency domain measures were obtained for 22 horses during routine activities (free in a box stall, confined on cross ties, and during walking) on two separate days. Within each data set (box stall, n = 19; cross ties, n = 21; walk, n = 7) descriptive statistics were evaluated to determine the group effect between conditions and between days. A comparison of group means on day 1 and day 2 was evaluated using a paired sample t-test. Test-retest reliability estimates were determined with intraclass correlation coefficients (ICCs), which included 95% CI for the ICC estimates. The results indicate that mean values between day 1 and day 2, based on the total group, did not differ across two conditions: box stall and cross ties. When horses were being led at a walk, HRV measurements for mean RR and mean HR were significantly different from day to day; however, no other HRV measurements in either the time or frequency domain were significantly different. ICC estimates of time domain variables were strongest for the box stall condition followed by the cross tie condition and then the walk condition. ICC estimates for the frequency domain variables were strongest for the cross tie condition followed by the walk condition and then the box stall condition. All of the ICCs for frequency domain variables were moderate or strong except for the low frequency:high frequency ratio, which was fair for walking and poor for the stall condition. The results support the hypothesis that HRV measurements from the Polar V800 portable heart rate monitors are statistically reliable and can be easily included in an assessment regimen for horses in different environments. Furthermore, assessment of horses for stress and pain during various circumstances could be reliably conducted using this methodology.