The impact of skin preparation method on electrocardiogram quality in horses

Abstract

Objective

Several skin preparation techniques are used in electrocardiogram (ECG) monitoring of horses. The objective of this study was to determine which methods produce the greatest signal quality using textile electrodes and standard silver/silver chloride (Ag/AgCl) electrodes.

Animals and samples

Electrocardiogram data were collected using textile and Ag/AgCl electrodes simultaneously for 4 skin preparation techniques in 6 horses.

Procedure

The effects of skin preparation (cleansing with isopropyl alcohol, with or without shaving the hair) and the effects of the application of a conductive gel were assessed using metrics of signal quality.

Results

Shaving and cleansing with alcohol had no effect on signal quality for either electrode type. The Ag/AgCl electrodes contain a solid gel, and the application of additional gel did not affect signal quality. Data quality was significantly improved when gel was applied to textile electrodes. Furthermore, there was no difference in signal quality between electrode types when gel was used.

Conclusion and clinical relevance

This study suggests that skin preparation by cleansing and/or shaving does not have a significant effect on equine ECG signal quality. When gel is used, textile electrodes are a practical alternative for Ag/AgCl electrodes, as they produce ECG recordings of the same quality.

Equine cardiac health is typically assessed using electrocardiography, auscultation, and echocardiography, although definitive diagnoses of arrhythmias require high-quality electrocardiogram (ECG) recordings (1). However, ECGs can be difficult to interpret due to the presence of motion artifacts (MAs), which are caused by changes in the electrode/skin potential, where the electrode moves against the skin or the skin stretches, affecting the distribution of charges between the electrolyte and electrode (2). The primary approach for minimizing the occurrence of MAs in ECGs is to improve the application of electrodes to the skin to reduce electrode movement and improve electrical conduction (2).

There are a variety of electrode types used to collect ECGs in horses, including crocodile clips, adhesive electrodes, and smart textile electrodes. Crocodile clips are attached directly to the skin, which can cause discomfort and are only appropriate for short-term use (1). Adhesive silver/silver chloride (Ag/AgCl) electrodes are more commonly used. These electrodes contain a solid gel and may require shaving of the hair depending on coat length (1). During prolonged recordings, Ag/AgCl electrodes may become detached from the skin and the solid gel may dry. To address these issues, textile electrodes, in which conductive yarns are knit directly into textiles, have been developed. Work previously conducted on horses at rest (3,4) and during exercise (5,6) has shown that textile electrodes can produce data of the same quality as Ag/AgCl electrodes, indicating that they are a practical alternative if designed and used properly. However, these studies used various methods of skin preparation and, perhaps as a result, reported a wide range of data quality.

Standard preparation in human cardiovascular monitoring includes shaving the hair, cleansing the skin, and drying (7). Preparation may also include abrasion using a fine sandpaper to remove the stratum corneum (7). Finally, Ag/AgCl electrodes are applied with a conductive gel for optimal signal quality (7). Similarly in dogs and cats, the hair is shaved, the skin is cleansed using alcohol, and an adhesive electrode (containing gel) is adhered to the skin (8). A combination of cast padding and adhesive bandages are applied to minimize electrode movement and detachment (8). In horses, preparation methods include using conductive gel and/or alcohol, both with and without shaving of the hair (1,9). Elastic and/or adhesive bandages or foam patches are often applied for extended recordings (1). However, there are no standard skin preparation methods for equine ECG data collection, and the most appropriate methods may vary with electrode type. Therefore, the objective of this study was to assess different skin preparation techniques for short-duration, resting ECGs obtained using standard Ag/AgCl electrodes and novel textile electrodes. Based on sample size calculations done using a confidence level of 0.05 and a power of 0.80 from pilot data, 6 healthy horses [5 standardbred, 1 Morgan; mares; median age: 15.5 y (range: 9 to 19 y)] without a history of cardiac abnormalities were recruited. The study was done in accordance with the University of Guelph Animal Utilization Protocol (reference No. 4705). Horses were housed separately in standard box stalls (3.66 m × 3.66 m) and were permitted to freely move around the stalls, as is standard for non-exercising ECGs. Horses were provided ad libitum access to hay and water throughout the study.

The ECGs were simultaneously collected using adhesive Ag/AgCl electrodes containing a solid gel (SKINTACT W-601; Leonhard Lang GmbH, Innsbruck, Austria) and dry textile electrodes composed of carbon- and silver-coated yarns (Skiin Equine; Myant, Toronto, Ontario). The 5 textile electrodes were integrated into a girth band in a modified base-apex configuration to construct a 3-lead trace (3,6). Five Ag/AgCl electrodes were placed immediately above the textile electrodes, under the band. Silver-coated yarns were knit into the textile band to connect the textile electrodes to the recording device, and cable wires were snapped directly to the button of the Ag/AgCl electrodes. Identical devices (Skiin Equine; Myant) were used to record ECGs from both types of electrodes at a sampling frequency of 320 Hz. Both electrode types were tested on all horses, with 4 different skin preparation techniques in the following order: i) unshaved hair, isopropyl alcohol (control); ii) unshaved hair, alcohol, and conductive gel (Spectra 360; Parker Laboratories, Fairfield, New Jersey, USA); iii) shaved hair, alcohol (control); iv) shaved hair, alcohol, and gel. The techniques were completed one immediately after the other, separated only by the time needed to perform each technique. Alcohol was applied to gauze for cleansing of each electrode site. Following data collection with gel, the gel was wiped from the skin before shaving. Shaving was completed using electronic equine clippers without a guard comb, to achieve a close shave. In the case of Ag/AgCl electrodes, additional gel was applied to the silver connection of the electrode for the gel condition. Textile electrodes did not contain any conductive medium and were therefore considered dry electrodes. For the gel condition, the conductive gel was applied to the surface of the textile electrode once the band was positioned on the horse. All electrodes (textile and Ag/AgCl) were replaced between testing conditions to prevent contamination. Data were collected for 10 min per condition and were transmitted via a Bluetooth connection to 2 mobile phones for later analysis. Electrocardiogram data were analyzed using a custom Python script; data were filtered to remove baseline, high-frequency, and power-line noise, as has previously been described (3,5). Signal quality was assessed using kurtosis (k), kurtosis signal quality index (kSQI), and the percentage of MAs (%MA). Kurtosis is an excellent indicator of ECG signal quality, as the distribution of peaks is compared to a Gaussian distribution (10). Kurtosis values > 5 indicate good signal quality, whereas k values < 5 indicate the presence of MAs (5,10). The kSQI was calculated by dividing the number of windows where the k value was > 5 by the total number of windows, for a maximum value of 1 (5,11,12). The %MA was calculated by manually counting the duration of segments where P-waves and/or QRS complexes were not identified. Data from each electrode type were averaged across the 3 channels for a single value. The assumptions of normality and equal variance were assessed using Kolmogorov-Smirnov tests. Kurtosis and kSQI data were determined to be normally distributed and are presented as mean ± standard deviation. Repeated measures 2-way ANOVAs with Tukey multiple comparison tests were run. Data considered not to be of diagnostic value (> 10% MAs) were excluded from statistical analysis. All tests were run using GraphPad Prism (9.1.0) software (Dotmatics, Boston, Massachusetts, USA) with statistical significance set at P ≤ 0.05.

Overall, skin preparations of shaving and/or cleansing with alcohol did not significantly affect ECG signal quality for either electrode type when tested on unrestrained, resting horses (Figure 1 A, B). Shaving of the hair was not shown to have an effect for either electrode type for any of the metrics assessed. There was an effect of conductive gel for textile electrodes, where signal noise that obscured P-waves and/or QRS complexes was observed with dry electrodes (Figure 1 A). Good signal quality was observed, with median k values above the optimal value of 5 for all conditions tested, except when textile electrodes were used without gel (Figure 2 A, B). When values obtained for unshaved and shaved conditions were pooled, the average k values were 20.01 ± 4.46 and 6.31 ± 2.51 for textile electrodes used with and without conductive gel, respectively; and 17.10 ± 3.53 and 15.83 ± 5.79 for Ag/AgCl electrodes used with and without additional gel, respectively. Similarly, average pooled kSQI values were close to the maximum value of 1 when conductive gel was used for both electrode types (textile: 0.96 ± 0.05, Ag/AgCl: 0.96 ± 0.06) but were significantly lower when gel was not used with textile electrodes (0.44 ± 0.15) (Figure 2 C). However, no differences in kSQI were observed with Ag/AgCl electrodes when no additional gel was applied to the electrode (0.95 ± 0.07) (Figure 2 D). The %MA was very low for both electrode types when conductive gel was used: 0.25% (range: 0 to 1.05%) and 0% (range: 0 to 0.73%) for textile and Ag/AgCl electrodes, respectively. However, when textile electrodes were used without gel, the %MA was greater than 10% and was deemed not to be of diagnostic value. In comparison, when Ag/AgCl electrodes were used without additional gel, the %MA was much lower, at 0% (range: 0 to 0.49%).

Figure 1.

Three-lead electrocardiogram traces obtained from the same horse using textile electrodes (A) and Ag/AgCl electrodes (B), with preparation methods as follows: i) unshaved hair, isopropyl alcohol (control); ii) unshaved hair, isopropyl alcohol, and conductive gel; iii) shaved hair, isopropyl alcohol (control); iv) shaved hair, isopropyl alcohol, and conductive gel.

Figure 2.

Kurtosis (A, B) and kSQI (C, D) values for electrocardiograms obtained from horses using smart textile and Ag/AgCl electrodes, with no conductive medium (control; white bars) and electrically conductive gel (hatched bars), with and without shaving of the hair.

ns — No statistical significance (P > 0.05).

*** P < 0.001, **** P < 0.0001.

No significant differences were appreciated between the 2 electrode types for any of the 3 metrics when gel was used. However, when gel was not used, the k values for textile electrodes were 56% (P = 0.03) and 64% (P = 0.02) lower than those for Ag/AgCl in the unshaved and shaved conditions, respectively. Similarly, textile kSQI values without gel were 49% (P = 0.001) and 59% (P = 0.01) lower for unshaved and shaved conditions, respectively. For all horses, ECGs obtained with dry textile electrodes were considered not to be of diagnostic value, with %MA values exceeding 10%, whereas all ECGs obtained with Ag/AgCl electrodes had low degrees of MAs when additional gel was not used.

The results of this study indicate that the preparation methods that produce high-quality ECGs are dependent on the type of electrode used. With textile electrodes, the presence of a conductive medium had a strong effect on signal quality. Application of additional gel to Ag/AgCl electrodes did not affect any of the metrics assessed, indicating that gel already present in the electrode was sufficient, as has previously been suggested by Petrie for canine and feline ECGs (8). In comparison, textile electrodes are manufactured without any conductive medium, and the medium added must be able to sufficiently saturate both the skin and the electrode during recording. Previous work with textile electrodes has been conducted on horses, where alcohol was applied between the skin and electrode without shaving (9). However, variable data quality was observed, with kSQI values ranging from 0.3 to 0.9 during exercise on a treadmill (walk, trot, gallop) (5) and an average of 35% of the data corrupted by MAs at rest (4). In comparison, we previously observed < 0.5% of data being corrupted by MAs when gel was used with textile electrodes in resting horses (3). These differences in data quality are likely attributed to multiple factors. The present results indicate that, although skin preparation techniques of shaving and cleansing may not be required, dry textile electrodes are not suitable for equine ECG recordings. Therefore, it is recommended that alcohol alone not be used with textile electrodes, as evaporation leads to a dry electrode, resulting in reduced signal quality over time. It is important to note that this study did not evaluate ECG quality over an extended duration and was conducted only in resting horses. Further work is necessary to understand how signal quality may change over time or in the context of exercise. Additionally, no effect of shaving was observed for either electrode type when data were collected from horses with medium-length coats (autumn in Canada). Therefore, although we did not observe an effect of shaving, a difference may be appreciated when a very thick coat is present, particularly for certain breeds, such as Icelandic horses (1). In these cases, shaving may be warranted to ensure sufficient contact between the skin and electrode. However, when a thin-to-moderate haircoat is present, shaving is not necessary when using textile or Ag/AgCl electrodes.

As we previously observed in horses at rest and during submaximal exercise (3,6), there were no significant differences in signal quality between textile and Ag/AgCl electrodes when conductive gel was used. This indicates that textile electrodes are a practical and reliable alternative to adhesive Ag/AgCl electrodes for short-duration resting ECGs. Further work is required to determine if this is also the case for prolonged recordings and during exercise.