Nonlinear Analysis of Heart Rate Variability for Measuring Pain in Dairy Calves and Piglets, Heat Stress in Growing Pigs, and the Growing Pig Sickness Response to a Lipopolysaccharide Challenge

Heart rate variability (HRV), or the variation in time between adjacent heart beats over time, is a non-invasive proxy measure of autonomic nervous system (ANS) function that has been used regularly in studies focused on evaluating livestock stress and welfare. The autonomic nervous system regulates involuntary physiological processes (e.g. respiration and heart rate) and consists of two main components, the parasympathetic (PNS), and sympathetic (SNS) branches, which act to maintain bodily homeostasis (PNS) or stimulate the “fight-or-flight” response after exposure to a stressor (SNS). Traditional linear HRV measures provide an estimation of overall autonomic activity or changes to the balance between the PNS and SNS branches by evaluating changes to the mean, variance, or frequency spectra of the R-R intervals.

To interpret HRV data obtained via linear HRV measures, particularly spectral HRV analysis, a linear assumption has to be assumed where SNS and PNS activity act in a purely antagonistic manner. However, this assumption is not always met. In many cases, ANS activity is altered in a nonlinear manner, which is reflected to some degree in the variability of heart rate output. Therefore, HRV measures that evaluate nonlinear changes to organizational or structural aspects of the R-R interval variability may be a useful compliment to traditional linear HRV measures for distinguishing between stressed and non-stressed states. The purpose of this dissertation was to evaluate the use of nonlinear HRV measures for evaluating dairy calf disbudding pain, piglet castration pain, growing pig heat stress, and as potential indicators of the subsequent immune response to a lipopolysaccharide (LPS) challenge in growing pigs.

Chapter 1 provides a knowledge base for understanding HRV and its use as a measure of autonomic stress in studies with livestock species. A brief explanation of animal welfare science, measures used to evaluate an animal’s welfare, and a demonstration of need for non-invasive physiological measures is provided before discussing the physiological basis of HRV. Relevant linear and nonlinear HRV measures are explained and examples of their use in livestock stress research are provided. Finally, a rationale for the studies conducted in this dissertation is presented.

Chapter 2 evaluates the use of HRV as an indicator of castration pain in 9-d-old piglets over a 3-d experimental period. Compared to sham castrated piglets, surgically castrated piglets exhibited greater low frequency to high frequency ratios (LF/HF), reduced sample entropy (SampEn), and greater percent determinism (ÞT) during the post-castration period. However, postural behavior was not different between treatments and serum cortisol concentrations only tended to differ between treatments at 1 and 24 h post-castration treatment, with surgically castrated pigs having numerically greater serum cortisol concentrations at both timepoints. These results demonstrate the ability of nonlinear HRV measures (SampEn and ÞT) to complement the physiological interpretation of linear HRV measures (LF/HF) in response to castration. Specifically, pigs who were surgically castrated exhibited more regularity (SampEn) and periodicity (ÞT) in their HRV data, and potentially more sympathetic activity (LF/HF) compared to sham castrated piglets, indicating greater pain-related stress. Additionally, HRV was a more sensitive measure of the stress response to castration than readily identifiable behaviors such as posture and the serum cortisol response.

Chapter 3 evaluates the use of HRV as an indicator of disbudding pain in dairy heifer calves (4 to 7-wk of age) over a 5-d experimental period. Calves who were given lidocaine and meloxicam prior to disbudding exhibited lower mean R-R interval (RR) values and a greater short-term detrended fluctuation analysis scaling exponent (DFAα₁) than sham disbudded calves. Together, these results indicate that calves who received pain mitigation exhibited greater pain-related stress (RR) and reduced physiological complexity in their heart rate signal (DFAα₁). Calves who were disbudded without pain mitigation had an intermediate response compared to sham disbudded calves and calves provided lidocaine and meloxicam. However, their numerical values closely followed those of calves provided lidocaine and meloxicam. These results demonstrate the usefulness of nonlinear HRV measures (DFAα₁) for evaluating nonlinear and correlational aspects of physiological complexity in response to disbudding. Additionally, the HRV results suggest that the provision of meloxicam does not reduce the amount of pain-related stress experienced by calves following disbudding.

Chapter 4 evaluates the use of HRV as an indicator of heat stress in growing pigs exposed to an acute heat episode. Heat stressed pigs exhibited greater body temperatures and spent less time in an active position compared to thermoneutral control pigs. Additionally, heat stressed pigs displayed an altered nonlinear HRV response to the acute heat phase compared to non-heat stressed control pigs. Specifically, heat stressed pigs exhibited lower SampEn and tended to exhibit greater ÞT, but no alterations to linear measures were observed in response to the acute heat episode. The low frequency to high frequency ratio was higher in heat stressed pigs during the period following the acute heat phase. Therefore, nonlinear HRV measures (particularly SampEn) may be more sensitive to the immediate physiological stress response to increased environmental temperature than traditional linear HRV measures.

Chapter 5 evaluates the use of baseline HRV as a potential indicator of the subsequent cortisol and pro-inflammatory cytokine response to an LPS challenge in growing pigs. The time for a pig to approach a human (approach time) prior to LPS administration was inversely related to baseline standard deviation of the R-R intervals (SDNN), and directly related to RR and the mean length of diagonal lines in a recurrence plot (Lmean). This result may have implications for the use of HRV as a measure of temperament in livestock species, since pigs with lower baseline SDNN (i.e. greater stress) and greater baseline Lmean (i.e. increased periodicity length in HRV data; greater stress) values took longer to approach a human observer before LPS administration (which occurred 1 d after HRV measurement). Area under the curve values for approach time following LPS administration were inversely related to high frequency spectral power (HF) and directly related to body weight, where pigs with low baseline HF values (i.e. lower parasympathetic activity) and higher body weights were slower to approach a human observer following LPS administration. Additionally, pigs with greater Lmean values had a greater change in body temperature following LPS administration. In conclusion, while baseline HRV measures were not directly representative of the cortisol or cytokine response following an LPS challenge, HF and Lmean may be useful indicators for evaluating certain aspects (sickness behavior and fever) of the innate immune response to an LPS challenge.

In conclusion, these studies demonstrate the usefulness of nonlinear HRV measures for evaluating livestock stress. Measures such as sample entropy and those derived from recurrence quantification analysis (ÞT, Lmean) seem to be particularly useful for complementing traditional linear HRV measures and, in some cases, are more sensitive measures of the physiological stress response (see chapter 4). Therefore, their inclusion in future studies on livestock HRV is warranted. However, further work is needed to fully elucidate the physiological significance of nonlinear HRV measures and their response to stress.