Measuring physiological parameters of stress in horses during groundwork, for example when involved in equine-assisted interventions, is important to gain insight into the stress levels of the horses. Heart rate and heart rate variability can be used as physiological indicators of stress in horses.

28 horses (15 mares and 13 geldings).

Average height was 154 ± 12 cm, average age was 10.6 ± 6.0 years. The horses were housed in individual boxes with straw bedding and ad libitum hay and were fed pellets once a day.

The horses were led by hand by 8 different handlers, the weight of the handlers was 72.8 ± 13.9 kilogram and age was 35.0 ± 18.7 years. All handlers were experienced horse handlers (> 1 year of experience) that were either students of veterinary medicine or were involved in training (police) horses.

The Hylofit electrodes (Equinics, Tølløse, Denmark), Polar H10 Heart Rate Sensor transmitter and Polar M430 receiver (Polar Electro Nederland, Utrecht, Netherlands) were used to obtain interbeat interval (IBI) recordings.

Hylofit, containing 2 electrodes, was attached with a Velcro elastic girth after applying Aquasonic 100 conducting gel, Kruuse ECG Electrodes were connected to the Televet 100 (Engel Engineering, Heusenstamm, Germany) which was also attached to the girth to prevent movement of the equipment while the horse is walking.

*NOTE*

*This riding school accommodates mentally and physically disabled people to ride and interact with horses and also EAI for military veterans with post-traumatic stress disorder.*

Baseline recordings in the stable lasted for 5 minutes and were performed in the horses’ own stables while standing still and being haltered. Food was removed 30 minutes prior to the start of the baseline recording.

Horses were taken to the groundwork exercise with novel objects within 10 minutes after the baseline recordings. The horses were led with a halter and rope, in walking pace keeping them in low intensity exercise throughout the measurements. Minimal pressure was applied to the horses.

All objects were placed in an inside arena at the riding school, about 4 meters of the outside walls. The objects consisted of three elements that were placed 8 meters apart: the horse had to walk under an arch with ribbons, in between a row of flags and umbrellas and over a plastic floor cover.

Recordings during groundwork were at least 2 minutes long. Most horses took more than 2 minutes to complete the groundwork exercise, but the horses that took <2 minutes were led through another time to gain at least 2 minutes of measurements.

Measurements with both ECG and HRM were simultaneously performed during baseline conditions (B) and groundwork conditions (GW) (before/after experimental design). Horses were given at least 5minutes to adapt to wearing the elastic girth and equipment before baseline measurements started.

Der SDNN-Wert in der Herzfrequenzvariabilität (HRV) steht für die Standardabweichung der RR-Intervalle im Messzeitbereich und wird in Millisekunden angegeben. Dieser Wert beschreibt, wie stark die einzelnen RR-Intervalle um ihren Mittelwert streuen.

Eine höhere Standardabweichung deutet auf eine größere Streuung hin, während eine Standardabweichung von null bedeutet, dass alle Messwerte identisch sind.

Der SDNN-Wert gilt als Maß für die Gesamtaktivierung des vegetativen Nervensystems und zeigt an, wie gut dieses System die Abläufe im Körper regulieren kann. Es wird empfohlen, den SDNN-Wert unter gleichen Messbedingungen zu vergleichen, um Verfälschungen zu vermeiden.

Die Abkürung RMSSD steht für „Root Mean Square of Successive Differences“. Es misst die kurzzeitigen Veränderungen aufeinanderfolgender RR-Intervalle und dient als fundiertes Maß für die parasympathische Aktivierung.

Ein hoher RMSSD-Wert zeigt an, dass der Körper gut mit dem Wechsel von Belastung und Entlastung umgehen kann, während ein niedriger Wert auf physische oder psychische Belastung hinweisen kann.

Der RMSSD wird als Indikator für die Erholungsfähigkeit, Fitness und Gesundheit interpretiert.

The Televet data were manually corrected using Televet software version 6.0.0 before RR-intervals being exported. All raw IBI were then imported into the software (Kubios HRV, University of Eastern Finland, Kuopio Finland) for HRV parameter analysis after applying the strong correction filter.

Statistics were performed using SPSS statistics (version 26, IBM Corp., Armonk, NY, USA). Two-sided probabilities were estimated throughout. The exact Kolmogorov-Smirnov onesample test was used to check Gaussianity of the data and showed that for the HR, SDNN, and RMSSD, the differences between the two compared methods were normally distributed.

A one-sample t-test was performed to see whether the measuring methods differed significantly from the expected null hypothesis (there is no difference between the two; difference=0, P-value of<0.05 was considered significant). Since the P-values of the one-sample t-test were not significant, Bland-Altmann plots were constructed in which the differences were plotted against the mean and were then visually inspected.

A linear regression was applied to detect proportional bias within the datapoints of the Bland-Altmann plot. Differences in HRM data between the baseline and groundwork were assessed with paired Student’s t-tests for normally distributed parameters (HR and HRV) and P = 0.05.

Besides P-values, estimated effect sizes were calculated and reported. To estimate the relative magnitude of the normally distributed differences, Cohen’s d effect size coefficients were calculated.

With the number of horses used in this study (n = 28), a two-tailed one-sample Student’s t-test, a threshold of significance = 0.05, and a power of ≈0.80, we were able to detect an effect size |d| of 0.55 or more.

Both systems were tolerated well by the horses and no interference between the systems was observed. Recordings during groundwork were at least 2 minutes long and lasted on average 6.3 ± 3.3 minutes.

During baseline measurements (B) and groundwork (GW), the results indicated that both recording methods did not differ significantly for mean HR and had zero or nearly zero effect (B: df = 27, t = −1.493, P = 0.147, |d| = 0.069, GW: df = 27, t = −0.816, P = 0.421, |d| = 0.027).

Similar to the HR data, the recording methods also did not differ significantly and had a zero or nearly zero effect respectively for the SDNN (B: df = 27, t = −1.062, P = 0.297, |d| = 0.068, GW: df = 27, t = 1.405, P = 0.171, |d| = 0.093) and RMSSD (B: df = 27, t = 0.531, P = 0.600, |d| = 0.021, GW: df = 27, t = 0.524, P = 0.605, |d|= 0.032).

- During B, the mean of the difference between HRM and ECG was −0.2 ± 0.7 bpm, with
**25 of the 27 recordings (93%) within the 95% confidence interval.** - During B, the SDNN mean of the difference between HRM and ECG was −1.2 ± 5.8ms, with
**25 of the 27 recordings (93%) within the 95% confidence interval**. - During B, the RMSSD mean of the difference between HRM and ECG was 0.3 ± 3.4ms, with
**26 of the 27 recordings (96%) within the 95% confidence interval.**

During GW, the mean of the difference between HRM and ECG was −0.3 ± 1.8 bpm, with

**25 of the 27 recordings (93%) within the 95% confidence interval.**During GW, the SDNN mean of the difference between HRM and ECG was 1.6 ± 6.0ms, with

**25 of the 27 recordings (93%) within the 95% confidence interval.**During GW, the RMSSD mean of the difference between HRM and ECG was −0.4 ± 4.2ms, with

**25 of the 27 recordings (93%) within the 95% confidence interval.**

A paired t-test showed that the mean HR across horses was significantly higher with a large effect size (df =27, t =−11.729, P < 0.001; |d| = 3.036) during GW (55.6 ± 10.2 bpm) compared to B (32.8 ± 2.7 bpm).

The SDNN also was significantly higher with a moderate effect size (df = 27, t = −3.493, P = 0.002; |d| = 0.748) during GW (72.7 ± 16.8ms) compared to B (58.8 ± 17.6ms).

The RMSSD was significantly lower with a large effect size (df = 27, t = 3.892, P = 0.001; |d| = 0.841) during GW(47.2 ± 12.7ms) compared to B (60.0 ± 16.5 ms).

The results from HRM and ECG systems do not differ significantly and had a zero or nearly zero effect size for HR and HRV parameters such as SDNN and RMSSD, indicating that the HRM provides accurate results.

Kapteijn, C. M., Frippiat, T., van Beckhoven, C., van Lith, H. A., Endenburg, N., Vermetten, E., Rodenburg, T. B. (2022). Measuring heart rate variability using a heart rate monitor in horses (Equus caballus) during groundwork.* Frontiers in Veterinary Science, 9*. doi: 10.3389/fvets.2022.939534

https://knowledge.time2tri.me/de/articles/die-wichtigsten-hrv-parameter-teil-1-zeitbezogene-parameter paij/vol1/iss1/5