Accuracy of Heart Rate Obtained by Auscultation in Atrial Fibrillation
The accuracy of heart rate estimation by cardiac auscultation over a 15-second period, and the influence of clinical experience on accuracy were evaluated in a dog with chronic atrial fibrillation by test subjects of varying experience. Only 30% of all test subjects provided accurate heart rate estimates. Board-certified specialists, medicine residents, and experienced nurses were significantly more accurate in their estimates than surgery residents and students. Accurate estimates were provided by 12.5% of surgery residents and students, as opposed to 64% of the other test subjects. Auscultatory estimates of heart rate in atrial fibrillation may be significantly inaccurate, and under some circumstances they may not provide a sound basis for making clinical decisions.
Introduction
Heart rate is a useful, vital parameter to evaluate an animal’s psychological as well as physiological state. Tachycardia is induced by increased sympathetic tone and can be a sign of psychological or physiological stress (e.g., brought on by examination by a veterinarian or poor cardiac performance).1 Tachycardia not only results from but also potentially causes myocardial failure (tachycardia-induced dilated cardiomyopathy);2 hence, therapeutic interventions to reduce heart rate in chronic, pathological tachycardias are critical for cardiac health. Accurate assessment of heart rate is crucial to evaluate therapeutic success. The resting heart rate obtained in a veterinarian’s office may not reflect the true resting heart rate (so-called “white coat effect”); therefore, owners are sometimes instructed to count the heart rate at home. Atrial fibrillation is one of the most common pathological tachyarrhythmias in the dog, and it is treated most often by controlling the ventricular response rate (i.e., heart rate).1 Estimation of heart rate in atrial fibrillation with rapid ventricular response can be difficult, because the rhythm is irregular, marked pulse deficits are common, and pathological respiratory sounds may interfere with auscultation in dyspneic animals. This study was undertaken to answer the following two questions:
-
Can heart rate be obtained accurately by cardiac auscultation in a dog with atrial fibrillation with a moderately fast ventricular response rate?
-
Does the level of expertise and experience influence the accuracy of the obtained heart rate?
Materials and Methods
A male Irish wolfhound with dilated cardiomyopathy, congestive heart failure, and atrial fibrillation was used as a heart rate generator in this study. This dog was chosen because the authors considered his heart and respiratory sounds to be typical of those commonly found in dogs showing clinical signs of heart failure and atrial fibrillation, based on their clinical experience (i.e., easily heard, distinct, transient heart sounds [S1 and S2] with a moderately fast ventricular response rate; mildly labored respiration with a modestly elevated respiratory rate of 40 breaths per minute; and somewhat louder-than-normal bronchovesicular lung sounds). The dog’s clinical condition and demeanor were stable enough to allow all of the subjects to be tested over a 3-hour period in order to minimize the influence of clinical variability on the results. All test subjects were familiar with auscultatory heart rate estimation, and they felt comfortable and confident in their ability to perform the task requested. All test subjects were using a Littman stethoscope (either their own or one that was provided).a The subjects were placed in a quiet room, with the dog in a standing position; they were allowed as much time as they desired to acclimate to the auscultatory conditions. When they were confident that they were evaluating the heart sounds correctly, they signaled the start of a 15-second measurement interval that they timed on their personal watch. Each test subject performed only one 15-second interval. The order in which the test subjects examined the dog was random. The gold standard for obtaining the heart rate was a simultaneously recorded electrocardiogram (EKG;b paper speed, 10 mm per second). The signaled start of the counting interval was marked on the running EKG paper, and the end of the 15-second interval was marked later on the EKG paper by measuring a 15-cm interval.
Test persons were grouped into four categories, based on their level of training and current exposure to patients. Group 1 was composed of internal medicine board-certified veterinary specialists (ACVIM or ECVIM; n=4); group 2 was made up of medicine residents and experienced nurses (n=7); group 3 consisted of surgery residents (n=4); and group 4 was composed of final-year veterinary students (n=12) that had minimal exposure to patients at that stage of their training but had received instruction in heart rate measurement by auscultation. Absolute and relative errors (ignoring direction of error) of heart rate estimation were calculated for each individual count. Relative error was calculated for each individual by dividing the absolute error of the ausculted heart rate by the EKG heart rate multiplied by 100% (e.g., when an individual counted a heart rate of 156 at an EKG heart rate of 160, the absolute error was 4, and the relative error was 4/160 × 100% = 2.5%). Analysis of variance (ANOVA) was performed with Statview 5.0.b Post-hoc evaluation of differences was done using the Fisher’s Protected Least Significant Difference test, with the level of statistical significance set at P=0.05. The normal distribution of achieved results was tested by means of the Q-S-normality test in Statview 5.0.c There was sufficient normality to use parametric tests like ANOVA (P>0.99). The criterion for clinically relevant accuracy was set at 10%.3 The difference in accuracy between test subjects who routinely evaluate heart rate in dogs with arrhythmias (groups 1 and 2) versus those subjects with less experience in such evaluation (groups 3 and 4) was compared with a chi-square test.
Results
The dog’s true heart rate (based on the EKG) throughout the study ranged from 148 to 176 beats per minute (bpm). The mean heart rates obtained by EKG during auscultation by members of groups 1 to 4 were 162, 159, 159, and 163 bpm, respectively. The estimated heart rates ranged from 80 to 300 bpm. Under- and overestimation occurred, with 16 (59%) persons underestimating, 10 (37%) persons overestimating, and only one person correctly estimating the heart rate. The absolute and relative errors ranged from 0 to 140 bpm and 0% to 87.5%, respectively. Mean and ranges of absolute errors in the four groups were 16 bpm (range, 4 to 36 bpm) in group 1, 15 bpm (0 to 44 bpm) in group 2, 71 bpm (40 to 140 bpm) in group 3, and 54 bpm (12 to 100 bpm) in group 4. Mean and ranges of relative errors in the four groups were 9.9% (2.3% to 22.0%) in group 1, 9.6% (0% to 28.2%) in group 2, 44.6% (25.0% to 87.5%) in group 3, and 32.9% (7.1% to 64.1%) in group 4 [see Figure]. Accuracies of estimates did not differ between test persons in groups 1 and 2 or between test persons in groups 3 and 4. Estimates were significantly more accurate in groups 1 and 2 than in groups 3 and 4. When a 10% estimation error was considered to represent clinically relevant accuracy, 17 (63%) test persons estimated the heart rate inaccurately: 87.5% in groups 3 and 4 as opposed to 36.4% in groups 1 and 2 (P=0.01). When considering a 20% estimation error, 81% in groups 3 and 4 estimated the heart rate inaccurately as opposed to 18% in groups 1 and 2.
Discussion
The results of this study indicate that estimation of heart rate by auscultation is often inaccurate in atrial fibrillation with a moderately fast ventricular response rate, and clinical expertise significantly influences the accuracy of heart rate estimates. Accuracy appeared to be influenced by several factors. The medicine residents (group 2) performed significantly better than the surgery residents in group 3, although their duration of clinical training was similar. In addition, several medicine residents performed better than board-certified individuals. In a similar human study testing nurses with different levels of training, the most educated nurses were the least accurate in estimating heart rate.3 This finding may indicate that the frequency and currency of clinical practice affect the accuracy of heart rate counting more than the level of education.
Several factors may have negatively influenced the results of the authors’ study. Heart rate was counted only for 15 seconds as opposed to 30 or 60 seconds. With shorter counting intervals, error is compounded by multiplication, and more accurate results with a 60-second interval have been reported in a previous study.3 On the other hand, with longer counting intervals, inaccuracies can result because test persons are more likely to lose track of the heart sounds or counts.45 The dog in this study was chosen as a typical patient with clinically significant atrial fibrillation. He was relatively anxious and restless, and several test persons had to restart counting because they lost track of the counts due to the dog’s motion. For this reason, a 15-second interval was chosen, and the authors did not attempt to study the effect of different counting intervals on the accuracy of heart rate estimation. Results may also be more accurate in dogs with stabilized heart failure where pathological respiratory sounds do not disturb cardiac auscultation. Estimates in this study may have been poor because the test persons were not aware of the purpose of this study (they were instructed only to measure heart rate as accurately as possible, and all participants appeared to take the task seriously). However, this design may more accurately reflect how auscultation is practiced on a routine basis. During the investigation, the dog’s heart rate varied between 148 and 176 bpm, and faster ventricular response rates are more difficult to count.56 The mean heart rates obtained by EKG were essentially identical in the different groups (162, 159, 159, and 163 bpm), and hence variation in heart rate did not account for the differences between groups.
The criterion for clinically relevant accuracy was set at 10%. This level may be considered by some to be too rigid for clinical veterinary purposes. The results, however, remain essentially unchanged even if one considers 20% to be a more clinically appropriate level of accuracy.
The gold standard for obtaining heart rate in this dog was the EKG. The EKG may not correctly reflect true auscultable heart rate, because some electrical depolarizations may be associated with inaudible heart sounds. Based on the distribution of the estimated heart rates with a similar degree of over- and underestimation, and the accuracy in test persons with frequent and current experience, the authors believe that the EKG in this dog closely reflected the true heart rate.
For this investigation, the authors chose to examine one patient in as short a time as possible in order to minimize any differences in clinical or environmental conditions that might potentially bias the results among the different groups.
Accurate heart rate assessment in atrial fibrillation is crucial for therapeutic evaluation. Clients are commonly instructed to measure the heart rate at home as a reflection of therapeutic control. The results of this study suggest that auscultatory estimates of heart rate may be significantly inaccurate, and clinical practice influences the accuracy of heart rate estimates. The authors conclude that heart rate estimates obtained by untrained or unpracticed individuals may not provide a sound basis for clinical decisions. Because an EKG obtained in a veterinary office may also not reflect true resting heart rate due to the psychological stress associated with the examination, ambulatory EKG monitoring appears to provide the optimal means of heart rate assessment in such cases.
Limitations of the Study
Only one dog with atrial fibrillation and moderately fast heart rate was used to test accuracy of auscultation by different individuals. Examining other dogs of different breeds, with more or less respiratory distress, more or less ease of handling during auscultation, and faster or slower ventricular response may have yielded different results. Similarly, the number of individuals in each group was small and may not have adequately represented the respective population, and each individual had to count only once. Finally, recording of simultaneous phonocardiograms along with the EKG may represent a more appropriate gold standard. Nevertheless, the counts obtained by 27 different individuals gave a huge range of different results, indicating that accurate auscultation, indeed, may be difficult. Further studies to assess accuracy of cardiac auscultation under different conditions and to evaluate the minimal amount of training necessary to accurately estimate heart rates need to be done before home monitoring can be recommended.
Littman Classic II; Heiland, Lyssach, Switzerland
Cardiostat 31S; Siemens, Erlangen, Germany
SAS Institute, Inc., Wangen, Switzerland
Citation: Journal of the American Animal Hospital Association 39, 3; 10.5326/0390237
Mean±standard deviation of the relative error of the heart rate estimated by four groups under investigation in percentage of the heart rate obtained by electrocardiogram.
