Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation.

The objective of this study was to evaluate the accuracy of short duration electrocardiographic (ECG) recordings extracted from ambulatory continuous ECG (Holter) to assess 24-hour mean heart rate in dogs with atrial fibrillation. In this retrospective study, Holter recordings obtained from 20 dogs with atrial fibrillation were selected for analysis. Ten out of 20 dogs were receiving drugs to control heart rate at the time of Holter evaluation. From the Holter recordings, heart rate averages were calculated for various sample durations (five-minutes, 30 minutes, one-hour, two-hours, and three-hours) for each dog. Percentage of these shorter duration ECG obtained HR averages that fell within ±10%, ±15% and ± 20% of 24-hour mean heart rate was determined for each sample duration and for each dog. Seventy five percent of heart rate averages obtained from three-hour ECG recordings fell within ±10% of 24-hour mean HR. All the heart rate averages obtained from two-hour ECG recordings fell within ±20% of 24-hour mean heart rate. Based on the results of this study it can be concluded that the duration of the ECG recording affects the prediction accuracy for 24-hour Holter mean HR. Only two and three hours of Holter recordings provided all heart rate averages within ±20% of 24-hour mean heart rate. No significant differences were noted in the prediction accuracy of shorter duration ECG recordings based on rate control therapy status. Further prospective studies are needed to assess the accuracy of HR obtained at home using various ECG recording devices to predict 24-hour mean heart rate in dogs with atrial fibrillation.

Introduction

Atrial fibrillation (AF) is one of the most common pathologic supraventricular tachyarrhythmias noted in dogs [1]. It frequently occurs with atrial enlargement secondary to an underlying cardiac disease but can also occur without any evidence of structural heart disease (lone AF or idiopathic AF) [2]. Atrial fibrillation is characterized by rapid and irregular atrial activity resulting in loss of atrial contribution to the ventricular filling and reduced cardiac output [3]. Additionally, long-term persistence of rapid and irregular ventricular response rate can deteriorate cardiac function and worsen hemodynamics [3]. Management of AF involves either cardioversion of AF rhythm into sinus rhythm or ventricular rate control using pharmacological agents [4-8]. Electrical cardioversion has been shown to be safe and effective in dogs, but recurrence of AF is common, particularly in dogs with underlying structural heart disease [5, 6]. Medical cardioversion using amiodarone has been reported in dogs but systemic side effects with long-term amiodarone use remain a concern [7]. Consequently, pharmacological HR control through reduction in atrio-ventricular nodal conduction is the most commonly used treatment modality for management of AF in dogs [4].

Various methods can be used to evaluate heart rate (HR) in dogs with AF including cardiac auscultation, pulse assessment, short-term in hospital ECG and 24-hour continuous ambulatory electrocardiography (Holter). A previous study in dogs with AF has shown that HR obtained by cardiac auscultation is inaccurate and cannot be relied upon for management of AF [9]. Additionally, pulse assessment for HR estimation is likely to be inaccurate due to frequent pulse deficits noted in dogs with AF. Short-term in-hospital ECG recordings are commonly used to evaluate the need for rate control therapy and to adjust medications for HR control in dogs with AF. However, HR obtained by short-term, in-hospital ECG may not represent the achieved HR control at home. It has been shown that HR obtained by the one-minute in-hospital ECG overestimated the 24-hour mean HR by 26 bpm compared to a 24-hour Holter recording in dogs with AF [10]. Hospital environments might cause an increase in autonomic tone and consequently result in a higher HR in dogs [10, 11]. This could result in inappropriate changes in therapy, potentially impacting control of clinical signs and morbidity.

A current recommendation is to perform 24-hour Holter monitoring to assess HR in dogs with AF [4]. Twenty-four-hour Holter recording provides a more complete representation of HR pattern in the home environment and provides information about the presence of any concurrent arrhythmias that may influence treatment decisions. Additionally, 24-Holter recording provides prognostic information [12]. A recent retrospective study in dogs with AF has shown that dogs with a mean 24-hour HR above 125 bpm had a shorter survival time compared to dogs with lower 24-hour mean HR averages [12]. Despite the benefits of 24-hour Holter recordings, the use of Holter monitoring may be limited by potential disadvantages such as cost, recording duration and owner compliance. Therefore, there is a need to identify a practical and cost-effective alternative to 24-hour Holter monitoring to estimate mean HR in dogs with AF. Short-term ECG recordings performed at home, unaffected by the stress of hospital visits, might provide an effective alternative to long term ambulatory ECG recording. Such short duration ECG recordings at home must be long enough to reflect the dog’s 24-hour mean HR and be unaffected by sporadic variations in HR. There are several at home HR monitoring tools (i.e. smartphone based single lead ECG devices, cardiac loop recorders (implantable and cutaneous), wearable HR monitoring technology that uses photoplethysmography or other pulse sensing technology) commercially available that can be used for HR assessment in dogs with AF. In general, the accuracy of non-ECG based technologies to estimate HR in AF has not been evaluated in dogs. Moreover, methods of HR estimation based on pulse detection or peripheral rate of blood flow are unlikely to be accurate given the frequent nature of pulse deficits in dogs with AF and requires further investigation. ECG based devices such as a smartphone-based ECG (AliveCor®, AliveCor® Inc., San Francisco, California) has been previously evaluated in dogs with AF and was noted to have good owner compliance in using these devices for short duration ECG recording [13]. While AliveCor smart phone ECG provided good rhythm evaluation in dogs, there were considerable differences between veterinary cardiologists for estimation of HR using AliveCor obtained ECG in dogs with both sinus rhythm and AF [13]. More importantly, the accuracy and adequacy of short-term ECG recordings that can be obtained using any device at home to estimate 24-hour mean HR in dogs with AF is not known. The goal of this retrospective study was to assess the HR averages obtained using shorter duration ECG recordings (five-minutes, 30 minutes, one hour, two hours and three hours of ECG recordings extracted from a Holter) for their deviation from 24-hour mean HR in dogs with AF.

Materials and methods

Animals

Medical records of dogs evaluated at the Veterinary Medical Center, College of Veterinary Medicine, Michigan State University between the years 2005–2012 and diagnosed with AF were retrospectively reviewed. Only dogs that subsequently underwent Holter monitoring as part of their clinical evaluation were included in further analysis. Holter recordings were only selected for further analysis if no changes were made to the current rate control therapy or no new antiarrhythmic drugs were initiated at the beginning Holter monitoring. Holter monitoring was performed using two or three channel recorders (Lifecard CF, Spacelab’s Healthcare, Snoqualmie, Wisconsin, United States) as previously described for Holter application in dogs [14]. The dogs were sent home for Holter monitoring and the owners were asked to maintain a record of normal daily activities and note any clinical signs.

Holter data analysis

The Holter recordings were analyzed using commercially available analysis software (Pathfinder SL, Spacelab’s Healthcare, Snoqualmie, Wisconsin, United States). The entire recording was visually inspected by a cardiology resident (TG) under the supervision of a veterinary cardiologist (RAS) and corrected for any automated errors. After manual verification, continuous RR interval data (defined as the distance between two consecutive R waves measured in milliseconds) for the entire Holter recording was extracted for each dog. This data was then edited such that analysis of R to R intervals began at midnight and ended twenty-four hours later to provide a uniform 24 hours of continuous RR interval data for all dogs. For example, if Holter monitor was placed on a dog at 3 pm during the day, from 3 pm to midnight the RR interval data was discarded. Discarding the initial portion of the Holter recording in this way provided a uniform, time adjusted, at home, continuous RR interval data for all dogs. Dogs were excluded if the edited recording duration was less than 24 hours, if there was more than five percent artifact in the edited recording or if more than two percent ventricular aberrancy was noted. From the 24 hours of continuous RR interval data, 24-hour mean HR was calculated for each dog. Heart rate averages were calculated for sample durations of five minutes, 30 minutes, one hour, two hours, and three hours from the continuous RR interval data for each dog using a commercial software (MATLAB and Simulink software, MathWorks, Torrance, California, United States). This resulted in 288, five-minute HR averages, 48, thirty-minute HR averages, 24, one-hour HR averages, 12, two-hour HR averages and eight, three-hour HR averages available for analysis for each dog.

Statistical analysis

Statistical analysis was performed using a commercially available statistics software (Stata, StataCorp LP, College Station, Texas, United States). Statistical significance for a type-1 error was set at <0.05. Normality was tested using Kolmogorov-Smirnov test. Categorical data was represented as number of observations and percentages. Normally distributed continuous variables were presented by mean and standard deviation. Non-normally distributed continuous variables were presented by median and range. The unpaired Student’s t test was used to compare differences in mean HR based on rate control therapy status. From the calculated HR averages for each shorter sample duration, percentage of HR averages that are within ±10%,±15% and ±20% of 24-hour mean HR for each dog was calculated. The Wilcoxon rank sum test was used to assess the differences in the percentage of short duration HR averages that fell within a specified range from the 24-hour mean HR based on rate control therapy status. The bootstrap method was used to construct 95% confidence intervals (percentile method) for the median percentage of HR averages failing within specific percentage of the 24-hour mean HR (5000 repetitions with replacement) as previously described [15]. Linear mixed model for repeated measures using 24-hour mean HR as an independent variable, percentage of HR averages that are within ±10% of 24-hour mean HR as the dependent variable and individual dog as random effect was modelled (random intercept model). To achieve an interpretable intercept for the model, 125 was deduced from each individual dogs 24-hour mean HR.

Results

Demographics

Entire study population consisted of 20 dogs with AF who underwent Holter monitoring. The median age was eight years (range = 5.0–12.0 years) for all the dogs. Twelve dogs were males (10 / 12 neutered) and eight dogs were females (8 / 10 spayed). The median body weight was 38.2 kg (range = 12.9–78.6 kg). The breed representation is as follows, Great Dane (3), Bulldog (1), Bullmastiff (2), Rottweiler (1), Newfoundland (1), Airedale (1), Scottish Deerhound (1), Keeshond (1), Golden Retriever (3), Labrador retriever (1), Irish Wolfhound (1), and mixed breed (4). Nine dogs were diagnosed with dilated cardiomyopathy, eight dogs with myxomatous mitral valve disease, one dog with severe subaortic stenosis, one dog with a left to right shunting patent ductus arteriosus, and one dog was suspected to have lone AF. Ten dogs were not receiving any rate control therapy at the time of Holter evaluation. Ten dogs were receiving rate control therapy for at least 2 weeks at the time of Holter evaluation. Of the ten dogs receiving HR control therapy, six dogs were receiving combination therapy with diltiazem (range 1.1–1.8 mg / kg PO TID) and digoxin (range = 0.003–0.0046 mg / kg PO BID), one dog was receiving combination therapy with diltiazem (1.4 mg / kg PO TID) and atenolol (1.2 mg / kg PO SID), one dog was receiving combination therapy with diltiazem (0.98 mg / kg PO TID) and sotalol (1.3 mg / kg PO BID) and two dogs were receiving monotherapy with diltiazem (range = 1.2–1.78 mg / kg PO TID). Concurrent cardiac medications administered included furosemide (9 / 20), pimobendan (9 / 20), enalapril (4 / 20) and spironolactone (4 / 20).

Holter results

The unedited Holter recording durations ranged from 32 hours to seven days. Twelve out of 20 (60%) dogs had ventricular premature complexes (range = 25 to 2378 ventricular premature complexes per 24 hour). None of the dogs had > 2% ventricular aberrancy on their 24-hour Holter recording. The 24-hour mean HR was 126 bpm ± 29 for all the dogs in this study. The 24-hour mean HR was not significantly different (p = 0.627) between the dogs that did not receive HR control therapy (134 ± 29 bpm) and dogs that did receive rate control therapy (mean HR = 121 ± 33 bpm). Table 1 depicts the percentage of HR averages falling within ±10%, ±15%, and ±20% of the 24-hour mean HR for a given sample duration in study dogs. Fig 1 depicts the percentage of dogs with all their HR averages falling within ±10%, ±15% and ±20% of 24-hour mean HR. Note that with three hours of continuous ECG recording, 70% of dogs had all their HR averages within ±20% of 24-hour mean HR. No significant differences in the accuracy of short duration ECG samples were noted based on rate control therapy status (p = 0.062).

Table 1

Accuracy of shorter duration ECG recordings in 20 dogs with atrial fibrillation.

Sample duration	% of averages within ±10% of 24-hour mean HR (Median (95% CIa)	% of averages within ±15% of 24-hour mean HR (Median (95% CIa)	% of averages within ±20% of 24-hour mean HR (Median (95%CIa)
5 minutes	51 (38–62)	74 (66–79)	86 (79–88)
30 minutes	61(39–71)	81 (70–88)	93 (81–95)
1 hour	67 (54–79)	81 (70–92)	92 (83–100)
2 hours	75 (58–83)	91(83–100)	100 (96–100)
3 hours	75 (50–87)	100 (88–100)	100 (94–100)

Table 1 depicts the percentage of mean heart rate averages that fall within ±10%,±15%, and ±20% of 24- hour mean heart rate for a given sample duration in dogs with atrial fibrillation. True mean heart rate is the mean heart rate obtained from a 24-hour Holter monitor.

aCI is boot strapped 95% confidence interval for the median.

b HR = heart rate

Fig 1

Line plot of percentage of dogs with all their heart rate averages within a specified range from the 24-hour mean heart rate in 20 dogs with atrial fibrillation.

Fig 1 represents the percentage of dog with all their heart rate averages obtained using shorter duration ECG recordings that fall within ±10, ±15 and ±20% of the true mean heart rate. True mean HR = 24-hour mean HR.

The 24-hour mean HR was a poor predictor (intercept = 60.6, standard error = 5.5, P = < 0.05; slope = 0.1127, standard error = 0.17, P = 0.5246) of percentage of shorter duration heart rate averages that fall within a specified range of 24-hour mean HR after adjusting for random effects of individual dogs. Seventy six percent of the unaccounted residual variances were attributed to random subject variances (random intercept variance = 372, standard deviation = 19.29). This suggests that at a 24-hour mean HR of 125bpm, on an average 60% of the HR averages obtained from short duration ECG recordings fall within ±10% of the 24-hour mean HR, with wide variation in accuracy due to random dog effect.

Discussion

The results of this study demonstrate that increasing recording durations improve the accuracy of shorter duration ECG recordings to predict 24-hour mean HR in dogs with AF. Also, when a wider accuracy (20%) from 24-hour mean HR is acceptable for a clinical situation, shorter duration ECG recordings may be adequate in providing approximate estimates of 24 hour mean HR. These findings should be interpreted in the context of an individual clinician’s judgement on what is an acceptable degree of accuracy for HR evaluation in a dog with AF. For example, if a 20% accuracy from the 24-hour mean HR is acceptable, then a five-minute ECG recording may be reasonable as 86% of the five-minute sample HRs (see Table 1) remain within ±20% of the 24-hour mean HR. If tighter accuracy is desired, such as 10%, then a longer duration ECG recording is necessary. In authors opinion, 20% accuracy from the 24-hour mean HR is likely not acceptable for a dog with a higher 24-hour mean HR, while it may provide a clinically acceptable estimation in a dog with a lower 24-hour mean HR below 100 bpm. However, at the time of initial evaluation of a dog with AF, 24-hour mean HR information is not available to the clinician and it is not possible to know a dog’s 24-hour mean HR without Holter evaluations. One exception is in situations of fast atrial fibrillation (as defined by one-minute in hospital obtained ECG HR of more than 155 bpm) the 24-hour mean HR can be predicted to be above 140 bpm which could potentially guide the clinician to make decisions on accuracy and duration of the ECG recording [10]. In addition, due to the presence of large random variation between dogs, it is difficult to predict an acceptable accuracy of a short duration ECG on an individual dog even with if prior 24-hour mean HR is available. Due to the above-mentioned reasons, selection of a specific shorter duration ECG recording for HR evaluation or an acceptable range of accuracy from 24-hour mean HR is difficult at the time of initial evaluation.

Limitations

The major limitation of this study is the small sample size and retrospective nature of the study. Additionally, there are several other limitations. Most of the study dogs were large breed dogs with only one dog weighing <20kg and only one dog diagnosed with lone AF in this study. These results may not be directly extrapolated to small breed dog population or to dogs with lone AF, but the results do represent the typical clinical demographic of canine AF population. We only used ±10%,±15% and ±20% within 24-hour mean HR as acceptable accuracy from 24-hour mean HR. It is possible that some clinicians accept wider variation from 24 hour mean HR for case management in dogs with AF. Moreover, we only used a maximum of hours of ECG recording to assess the accuracy of shorter duration ECG recordings. To the authors knowledge, currently available at-home tools for continuous, short duration ECG evaluation (including smartphone-based ECG platforms, event recorders or continuous loop recorders) do not have the capacity to record continuous ECG for more than one-hour duration [13, 16]. Therefore, at the time of publication, practically any continuous ECG recording duration longer than one hour may not be practically feasible without placement of a version of Holter monitoring. While some devices have the capacity to record repeated shorter duration ECG recordings (e.g. AliveCor smartphone ECG) the use of such repeated short duration ECG recordings to estimate 24-hour mean HR was not evaluated in this study. Also, this study only evaluated accuracy of shorter duration ECG recordings extracted from a 24-hour Holter. It must be pointed out that results of this study cannot be directly interpreted as the accuracy of shorter duration ECG recordings obtained using various ECG recording devices that require owner-dog interaction such as placement of ECG recording device on a dog. Such interactions can induce anxiety and may not represent an ECG recording that would be extracted from a Holter. Additionally, the findings of this study do not apply to any non-ECG based HR monitoring systems that have the capacity to record continuous HR information.

Interestingly, there was no significant differences in 24-hour mean HR between dogs that did and did not receive HR control therapy in this study. Particularly, the 24-hour mean HR for the untreated group can be considered low for dogs with AF secondary to structural heart disease. There are several reasons for the lower 24-hour mean HR noted in the untreated group of AF dogs in this study. First, none of the previous studies that reported 24-hour mean HR in dogs with AF used the initial editing process used in this study [10, 12, 17]. Removal of a larger initial portion of the Holter recording may have eliminated high heart rate periods prior to acclimatization of a dog after Holter equipment placement, resulting in lower overall mean HR. Secondly, to the authors knowledge, information about 24-hour mean HR for the untreated group has only been reported previously in two veterinary studies. In one study [10] only a range of 24-hour mean HR was provided for the untreated group, while in the second study [17] dogs with 24-hour mean HR above140 bpm were excluded from the analysis. Therefore, direct comparison to previous studies could not be performed. Finally, in this study, initial Holter evaluation of some untreated dogs provided a mean 24-hour HR that was not high enough to warrant rate control therapy. This resulted in dogs with low 24-hour mean HR to be included in untreated group resulting in lower overall mean HR for the untreated AF group.

Conclusions

Increasing the duration of a short-term ECG recording improves the prediction accuracy for 24-hour mean HR. The accuracy of each shorter duration samples should be thoroughly considered by the clinician as it has direct impact on the decision to initiate or modify rate control therapy in AF dogs. Inaccuracies of 20% from 24-hour mean HR are likely to over or underestimate the efficacy of rate control therapy for most dogs with AF, thereby impacting treatment decisions and potentially patient outcome. Further prospective studies evaluating the accuracy of short-term ECG recordings during specific times of the day and at times of rest are needed.

7 Aug 2020

PONE-D-20-23023

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

PLOS ONE

Dear Dr. Sanders,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The Authors should respond to all of the Reviewers' comments. One of the major comments of the Reviewers was the potential added benefit of shorter time periods for Holter monitoring over 24 hour monitoring that usually is carried out, due to the fact that animals would still have to undergo the procedures for placing the Holter. Therefore this, especially, would need to be addressed in a revised manuscript. Furthermore, the Reviewers also noted that the statistical analysis of the manuscript should be strengthened.

Please submit your revised manuscript by Sep 21 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Daniel M. Johnson, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

3. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors provide a study that examined whether periods substantially shorter than 24hours of Holter monitoring (ambulatory ECG monitoring) could reasonably accurately reflect the 24 hour average heart rate in dogs with atrial fibrillation.

I have several concerns about the analyses.

First, in previous studies of mean heart rate, investigators suggested trimming the Holter recordings to exclude the first 3 hours and final hour of the recording (https://pubmed.ncbi.nlm.nih.gov/19645836/, see figure 2B) as this removes the effect of stress and anxiety experienced by most dogs when applying the Holter and when returning to have the Holter removed. The residual 20 hours more accurately reflect the true “at home” average heart rate. The authors should consider this approach to see how this affects their interpretation of agreement. They would obviously remove the short periods from these peripheral time periods (because these, too, would be affected by anxiety etc).

The statistical analysis section needs expanding. This reviewer cannot determine what analyses were applied for what purposes. For example, what “study variables” did the authors examine by Spearman Rank Correlation?

The accuracy of short periods might be very dependent on specific dogs. Did the authors account for dog in their analysis? In other words agreement might be high for some dogs and low for others. Examining the % agreement by dog might reveal such a “between dog” effect.

A further approach the authors should consider is to randomly select a 5-minute, 30-minute, 1 hour, 2-hour and 3-hour segment from each dog and see how often these randomly selected segments agree within the predefined limits. This would represent the clinical reality of a client obtaining a heart rate at home.

Minor comments:

Line 88-90. The authors state the limitations of Holters (cost, duration, compliance). However, the study examines Holter recordings and does not provide or discuss what alternatives to Holter monitoring exist or that could substitute Holter monitoring (and this reviewer knows of no such alternatives). Therefore, this statement seems rather contrived and without a solution.

Lines 177-180. If the average rates are not different statistically, then one group cannot have a “slightly higher” rate than the other. Also, a P value should be provided for all statistical testing, whether significant or not. Please amend.

Line 177-180. The average HR for dogs with AF that were not being administered any rate control therapy seem considerably lower than previously documented in dogs with secondary AF. Do the authors have any explanation for this rather low rate? (most previous studies suggest rates exceeding 160-190 bpm). The rates for the group under rate control therapy appears to be consistent with previous observations.

Line 185-186. Could the authors provide scatter plots for each of the short duration recordings vs the Holter recordings plotted as average heart rates? Although the P value for their observation is “not significant”, with only 10 dogs in each group, the probability of this being a false negative is quite large.

Line 204-213. The authors suggest that a 5 minute ECG recording (in-clinic) would mostly agree to within 20% of the Holter-determined average HR. However, this is NOT what the authors showed. To demonstrate that specific idea, the authors would need to examine the 5 minute intervals within the 1st hour of Holter application (while the dog was still in the clinic and stressed) with the average at-home HR. How many of the 240 5-minute intervals obtained during the first hour came within 20% of the average at-home heart rate? If the authors suggest that clients could obtain a 5-minute ECG at home on their pets, the authors would have to suggest how such a feat would be possible, and also demonstrate that doing this would not alter the level of anxiety in the dog during at-home acquisition.

Line 215-220. Several of the “health collars” (e.g. PetPace) can record ECGs in dogs, at least in sinus rhythm. Whether they can accurately count heart rates in dogs with atrial fibrillation remains to be determined. If they can, then they provide continuous monitoring, or can record variable periods for as long as desired (limited by battery life). Therefore, such devices (similar to the iWatch etc, used in humans) are available, but in limited supply.

Reviewer #2: General comments – this manuscript describes a retrospective analysis of subsets of ECG data recorded during a 24 hour Holter recording. While providing interesting preliminary data, I wonder how much less practical difficulty there might be in practice in obtaining shorter Holter recordings compared to a 24 hour recording? The dog still has to have the monitor fitted and wear it for several hours (preferably from the point the dog is relaxed at home, not from the initial time of Holter fitting) – if the benefits are “cost, recording duration and owner compliance” how much difference would shorter recordings truly make? I don’t have a good feel for how much less companies would charge to analyse shorter recordings, unfortunately. The main advantage I can see would be in those cases that the leads became detached after less than 24 hours – if it could be shown that the data was valid if at least 2-3 hours of recording was available once the dog was at home and relaxed that could prevent the need to start all over again. My specific comments, questions and observations are detailed below.

Introduction

Line 62 – should read “…remain a concern.”

Lines 63-65 – this sentence is identical to lines 59-61. I recommend deleting the sentence in lines 59-61, as it makes most sense to keep this information at the end of the paragraph.

Lines 67-79 – you mention that pulse assessment can be used to assess heart rate, but do not discuss later in the paragraph whether this would provide an accurate reflection of heart rate (which seems unlikely, due to pulse deficits, etc.).

Lines 91-93 – how do you propose that the Holter is fitted without “the stress of hospital visits”?

Results

Line 151 – just an idle observation, but I’m surprised only 20 dogs met the inclusion criteria over a 7 year period. Were there a lot of recordings that were <24 hours when edited?

Lines 152-153 and 154-155 – personally, I would just report the range rather than the range and the IQR, as I don’t think the latter adds anything of real value.

Lines 177-180 – please don’t describe non-significant differences. This should simply read “There was no significant difference in heart rate between dogs receiving HR control therapy (121+/- 33bpm) and those not receiving HR control therapy (134 +/- 29 bpm).”

Lines 180-182 – Table 1 and Figure 1 essentially show the same information twice. Please remove either the Table or the Figures.

Lines 182-184 – I’m not sure why this correlation is clinically interesting or important – please explain why you performed this test.

Lines 184-186 – given that rate control status had no significant impact on heart rate, why would you imagine that it would have an impact on accuracy? Again, this seems to be a statistical test that isn’t testing a hypothesis.

Table 1

The information displayed represents “Median (95% CI)” NOT “(Median +/- CI)” – please edit this accordingly.

Discussion

Lines 206-207 – I don’t think that “agreement” changes – you are just being more lenient about what you are willing to accept as “accurate”.

Lines 209-215 – based on your stated mean heart rate of 126 bpm, a 20% inaccuracy gives a heart rate of 151 bpm, which is likely to be interpreted very differently than a mean of 126 by most clinicians. This is clearly potentially problematic – please expand the discussion to reflect this.

Lines 215-220 – this is the first time you have mentioned alternative methods of obtaining an ECG recording in the home environment – which potentially changes ho you might be positioning shorter recordings (although I don’t relish the idea of asking a client to hold an AliveCor against their dog for an hour!). Please add this information to the Introduction (with a brief summary of the work that has already been done looking at accuracy of such recordings in the home environment for monitoring of AF).

Lines 226-228 – is it not equally likely that these would consistently underestimate 24 hour averages (simply because they wouldn’t include the periods of exercise and excitement that also influence mean heart rate over 24 hours)? So you might be at risk of thinking that your rate control was better than it actually is?

Line 236 – please remove the grocer’s apostrophe (the “s” in ECGs denotes that this is plural rather than denoting possession).

Lines 237-248 – I am still not sure why you did this test in the first place (it feels like a fishing trip) so my preference would be to remove this from the discussion.

Lines 248-251 – this is possible, of course, but I think it’s very unlikely (see previous comment regarding lines 209-215).

Lines 255-257 – this would still rely on owners to record these times accurately – is there a way to mitigate this as a potential limitation to future studies?

Conclusions

Lines 260-262 – surely there is experimental evidence that would help determine guidelines for what is acceptable accuracy, rather than simply relying on the judgement of the clinician?

Reviewer #3: The purpose of this study was to determine if the heart rate measured for shorter time periods (5 minutes, one, two and three hours) would equate to the mean rate as determined by 24 hour electrocardiographic monitoring. The mean heart rate for each of these four time intervals was assessed for its ‘closeness’ (within 10%, 15% or 20%) to the 24 hour mean rate. The authors embarked on this investigation with the justification that performing a 24 hour Holter for shorter periods of time would be less problematic for owners and dogs. The conclusion of this paper is that none of the shorter time periods would equate to the mean 24-hour heart rate, but the longer the recording the closer the short surveillance heart rate was to the full recording.

This is a very focused study to determine if short recordings are adequate in the evaluation of rate during atrial fibrillation. The authors state that there are no other methods currently available to determine heart rates in a dog for longer than one hour. However, an Internet search will reveal that individuals are adapting noninvasive monitoring such as Fit Bits for dogs. Additionally, other devices have been developed. (See reference: Yasin Cotur, Michael Kasimatis, Matti Kaisti, Selin Olenik, Charis Georgiou, Firat Güder. Stretchable Composite Acoustic Transducer for Wearable Monitoring of Vital Signs. Adv. Funct. Mater. 2020, 30, 1910288. Doi.org/10.1002/adfm.201910288) These may serve as a way to get minimal information (and likely inadequate when monitoring arrhythmias) regarding heart rate. Such devices could provide data for specific questions such as heart rate. Comparison of these devices with established methods such as 24 hour Holter monitoring would be important to document. It is not clear how this study could be used in such a situation. Authors need to further justify this study.

From an implementation component there are no striking errors in this manuscript. It should be noted however that the average heart rates of the dogs examined are actually low for atrial fibrillation. Dogs that have high ventricular response rates are of course the more critical for monitoring and this paper does not address that kind of patient. Differences between low and high ventricular response rate heart rates over 24 hours is likely meaningful.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Mark Rishniw

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

26 Aug 2020

PONE-D-20-23023

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

PLOS ONE

Dear Dr. Sanders,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The Authors should respond to all of the Reviewers' comments. One of the major comments of the Reviewers was the potential added benefit of shorter time periods for Holter monitoring over 24 hour monitoring that usually is carried out, due to the fact that animals would still have to undergo the procedures for placing the Holter. Therefore this, especially, would need to be addressed in a revised manuscript. Furthermore, the Reviewers also noted that the statistical analysis of the manuscript should be strengthened.

Please submit your revised manuscript by Sep 21 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Daniel M. Johnson, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

Please find the DOI for the data.

Gunasekaran, Tamilselvam (2020), “Atrial fibrillation dogs”, Mendeley Data, v1http://dx.doi.org/10.17632/hbwy6bn48n.1

3. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: I Don't Know

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors provide a study that examined whether periods substantially shorter than 24hours of Holter monitoring (ambulatory ECG monitoring) could reasonably accurately reflect the 24 hour average heart rate in dogs with atrial fibrillation.

I have several concerns about the analyses.

First, in previous studies of mean heart rate, investigators suggested trimming the Holter recordings to exclude the first 3 hours and final hour of the recording (https://pubmed.ncbi.nlm.nih.gov/19645836/, see figure 2B) as this removes the effect of stress and anxiety experienced by most dogs when applying the Holter and when returning to have the Holter removed. The residual 20 hours more accurately reflect the true “at home” average heart rate. The authors should consider this approach to see how this affects their interpretation of agreement. They would obviously remove the short periods from these peripheral time periods (because these, too, would be affected by anxiety etc).

We did undertake such a measure to remove the initial portion of the Holter recording that is likely affected by the stress of a hospital visit. We removed the initial portion (> 3 hrs in every case) of the Holter recording from the time of Holter application to midnight on the day of Holter application. Therefore, for each dog 24 hours of Holter data starting at midnight to midnight next day was extracted for further analysis. We noted this in 136-137 lines. We have added additional lines (137-141) to the manuscript to explain this process in detail.

The statistical analysis section needs expanding. This reviewer cannot determine what analyses were applied for what purposes. For example, what “study variables” did the authors examine by Spearman Rank Correlation?

We have expanded on the statistics section. However, per reviewer 2 recommendations we have removed some portions of the statistical analysis performed.

The accuracy of short periods might be very dependent on specific dogs. Did the authors account for dog in their analysis? In other words agreement might be high for some dogs and low for others. Examining the % agreement by dog might reveal such a “between dog” effect.

We agree that % agreement to 24 hours mean HR is dog dependent. We have now added an additional summary figure (new fig 1) that highlights this information.

A further approach the authors should consider is to randomly select a 5-minute, 30-minute, 1 hour, 2-hour and 3-hour segment from each dog and see how often these randomly selected segments agree within the predefined limits. This would represent the clinical reality of a client obtaining a heart rate at home.

We did think about performing a random selection approach as recommended by the reviewer. One of the problems with such an approach is that a random selection of a short duration ECG recordings will result in inclusion of sampling periods that are not uniform among all dogs. For example, in one dog the sampling may occur during periods of rest while in others sampling may occur during extreme activity. We believe this would limit the usefulness of the direct clinical application of a shorter duration sampling. The authors believe that a better approach would be if prospective sampling was done under uniform conditions such as at rest or specific time of the day, as it would directly translate to clinical application. Unfortunately, given the retrospective nature of our study with incomplete Holter diary entries, we could not perform analysis limited to specific situations.

Minor comments:

Line 88-90. The authors state the limitations of Holters (cost, duration, compliance). However, the study examines Holter recordings and does not provide or discuss what alternatives to Holter monitoring exist or that could substitute Holter monitoring (and this reviewer knows of no such alternatives). Therefore, this statement seems rather contrived and without a solution.

Thank you for noting this. We understand how our statement is incomplete. We have added additional information clarifying our goal in the introduction and limitation sections.

Lines 177-180. If the average rates are not different statistically, then one group cannot have a “slightly higher” rate than the other. Also, a P value should be provided for all statistical testing, whether significant or not. Please amend.

Requested changes made.

Line 177-180. The average HR for dogs with AF that were not being administered any rate control therapy seem considerably lower than previously documented in dogs with secondary AF. Do the authors have any explanation for this rather low rate? (most previous studies suggest rates exceeding 160-190 bpm). The rates for the group under rate control therapy appears to be consistent with previous observations.

We did notice such a discrepancy from previous studies. There are several reasons for the lower HR in the untreated group in this study. The data was edited to remove the initial portion of the Holter recording which is different compared to previous studies (only first 3 hours in the previous studies). To our knowledge, there is only one study where the 24-hour Holter mean HR was reported for the untreated group in the past (Gelzer AM, etal. Combination Therapy with Digoxin and Diltiazem Controls Ventricular Rate in Chronic Atrial Fibrillation in Dogs Better than Digoxin or Diltiazem Monotherapy: A Randomized Crossover Study in 18 Dogs). In the Gelzer study only dogs above a 24-mean HR of 140 bpm were included. Therefore, the true mean 24-hour HR in a group of untreated dogs is unknown. Additionally, in our institution Holter monitoring is routinely performed in most dogs to assess the need for rate control therapy prior to initiation of any rate control therapy. Similar to previous observations, we have noted in some of these dogs with extremely high in-hospital ECG HR that the at home 24-hour mean HR is low and does not warrant any rate control therapy. Therefore, portion of our untreated group is formed by such patients with low 24hour mean HR at home. We have added a few lines in the limitation’s sections to clarify this.

Line 185-186. Could the authors provide scatter plots for each of the short duration recordings vs the Holter recordings plotted as average heart rates? Although the P value for their observation is “not significant”, with only 10 dogs in each group, the probability of this being a false negative is quite large.

We have removed the correlation analysis per reviewer 2 recommendations. From what we understand the reviewer is asking to provide scatter plot for pairs of each shorter duration ECG recording HR to 24 hours mean HR? It would be cumbersome to do. For example, for 5 min recordings there is 288 separate 5 min vs 24-hour plots needs to be done for each dog. Unless a specific 5 min sample is selected (the limitations of which we addressed in the previous comment) the utility of such a plot is limited. If the reviewer is asking for average of all 5 min HRs for each dog vs 24 hour mean HR scatter plot, then we believe that the average of all 5 min would be essentially same as the 24 hours mean HR.

Line 204-213. The authors suggest that a 5-minute ECG recording (in-clinic) would mostly agree to within 20% of the Holter-determined average HR. However, this is NOT what the authors showed. To demonstrate that specific idea, the authors would need to examine the 5 minute intervals within the 1st hour of Holter application (while the dog was still in the clinic and stressed) with the average at-home HR. How many of the 240 5-minute intervals obtained during the first hour came within 20% of the average at-home heart rate? If the authors suggest that clients could obtain a 5-minute ECG at home on their pets, the authors would have to suggest how such a feat would be possible, and also demonstrate that doing this would not alter the level of anxiety in the dog during at-home acquisition.

As the reviewers allude to there are several possible at home ECG monitoring devices that a client can use to obtain short term ECGs (AliveCor, cutaneous applied loop recorder (previously shown at our institution), patch recorders). While devices such as AliveCor requires owner-dog interaction the devices such as Cutaneous loop recorders can be activated to collect ECG with minimal interaction. With ever evolving technology there is likely further advances in ECG technology that would minimize the need for dog interaction thereby anxiety during data collection. We have added additional descriptions in introduction section and limitations section to clarify this.

Line 215-220. Several of the “health collars” (e.g. PetPace) can record ECGs in dogs, at least in sinus rhythm. Whether they can accurately count heart rates in dogs with atrial fibrillation remains to be determined. If they can, then they provide continuous monitoring, or can record variable periods for as long as desired (limited by battery life). Therefore, such devices (similar to the iWatch etc, used in humans) are available, but in limited supply.

Agree that they are available and agree that they are currently untested. We have that research planned.

Reviewer #2: General comments – this manuscript describes a retrospective analysis of subsets of ECG data recorded during a 24 hour Holter recording. While providing interesting preliminary data, I wonder how much less practical difficulty there might be in practice in obtaining shorter Holter recordings compared to a 24 hour recording? The dog still has to have the monitor fitted and wear it for several hours (preferably from the point the dog is relaxed at home, not from the initial time of Holter fitting) – if the benefits are “cost, recording duration and owner compliance” how much difference would shorter recordings truly make? I don’t have a good feel for how much less companies would charge to analyse shorter recordings, unfortunately. The main advantage I can see would be in those cases that the leads became detached after less than 24 hours – if it could be shown that the data was valid if at least 2-3 hours of recording was available once the dog was at home and relaxed that could prevent the need to start all over again. My specific comments, questions and observations are detailed below.

Dear reviewer, our goal with this study was not to replace a longer Holter recording with a shorter Holter recording. Rather we sought to assess the accuracy of a shorter ECG recording with available Holter data as a surrogate to shorter duration ECG devices for HR assessment in dogs with AF. While the data from this study could not be directly extrapolated to shorter duration device obtained HR information (due to factors such as human-dog interaction and such action on HR while collection). We have added additional information in introduction and limitations sections to further clarify this.

Introduction

Line 62 – should read “…remain a concern.”

Recommended changes made

Lines 63-65 – this sentence is identical to lines 59-61. I recommend deleting the sentence in lines 59-61, as it makes most sense to keep this information at the end of the paragraph.

Recommended changes made

Lines 67-79 – you mention that pulse assessment can be used to assess heart rate, but do not discuss later in the paragraph whether this would provide an accurate reflection of heart rate (which seems unlikely, due to pulse deficits, etc.).

Recommended addition made.

Lines 91-93 – how do you propose that the Holter is fitted without “the stress of hospital visits”?

See comments above

Results

Line 151 – just an idle observation, but I’m surprised only 20 dogs met the inclusion criteria over a 7 year period. Were there a lot of recordings that were <24 hours when edited?

Yes, the removal of the initial portion of the Holter recording resulted in less than 24 hours of Holter data in many AF dogs and therefore excluded from the study.

Lines 152-153 and 154-155 – personally, I would just report the range rather than the range and the IQR, as I don’t think the latter adds anything of real value.

Recommended changes made

Lines 177-180 – please don’t describe non-significant differences. This should simply read “There was no significant difference in heart rate between dogs receiving HR control therapy (121+/- 33bpm) and those not receiving HR control therapy (134 +/- 29 bpm).”

Recommended changes made.

Lines 180-182 – Table 1 and Figure 1 essentially show the same information twice. Please remove either the Table or the Figures.

Per recommendation the figure 1 is removed now.

Lines 182-184 – I’m not sure why this correlation is clinically interesting or important – please explain why you performed this test.

Lines 184-186 – given that rate control status had no significant impact on heart rate, why would you imagine that it would have an impact on accuracy? Again, this seems to be a statistical test that isn’t testing a hypothesis.

Excellent points; our hypothesis was that in dogs with higher mean 24-hour HR’s, the accuracy of shorter duration ECG will be better. We anticipated that dogs with higher 24-hour mean HR’s will have limited ability to reach lower HR ranges, and potentially less of a deviation from 24 hour mean HR in one direction resulting in narrower band width of shorter duration HR averages. But you are correct, since the HR was not different between two group the accuracy is unlikely to be different nor there likely be a relationship to the 24 hours mean HR. We have removed these analyses now.

Table 1

The information displayed represents “Median (95% CI)” NOT “(Median +/- CI)” – please edit this accordingly.

Recommended correction made.

Discussion

Lines 206-207 – I don’t think that “agreement” changes – you are just being more lenient about what you are willing to accept as “accurate”.

We have made modifications to this statement.

Lines 209-215 – based on your stated mean heart rate of 126 bpm, a 20% inaccuracy gives a heart rate of 151 bpm, which is likely to be interpreted very differently than a mean of 126 by most clinicians. This is clearly potentially problematic – please expand the discussion to reflect this.

We have added additional statements as recommended.

Lines 215-220 – this is the first time you have mentioned alternative methods of obtaining an ECG recording in the home environment – which potentially changes ho you might be positioning shorter recordings (although I don’t relish the idea of asking a client to hold an AliveCor against their dog for an hour!). Please add this information to the Introduction (with a brief summary of the work that has already been done looking at accuracy of such recordings in the home environment for monitoring of AF).

Recommended description on previous work is now added to the introduction section.

Line 236 – please remove the grocer’s apostrophe (the “s” in ECGs denotes that this is plural rather than denoting possession).

Recommended change made

Lines 226-228 – is it not equally likely that these would consistently underestimate 24 hour averages (simply because they wouldn’t include the periods of exercise and excitement that also influence mean heart rate over 24 hours)? So you might be at risk of thinking that your rate control was better than it actually is?

Lines 237-248 – I am still not sure why you did this test in the first place (it feels like a fishing trip) so my preference would be to remove this from the discussion.

Lines 248-251 – this is possible, of course, but I think it’s very unlikely (see previous comment regarding lines 209-215).

These are great points. Taking your comments into consideration we have removed comments about prospective evaluations performed at resting ECGs.

Lines 255-257 – this would still rely on owners to record these times accurately – is there a way to mitigate this as a potential limitation to future studies?

We have removed this comment now per recommendation above. But in general, the owners would only be asked to record ECGs related to specific activities such as resting, sleeping or activity. With the time stamps from the owner recorded ECGs we can certainly match to a concurrently applied Holter recorder.

Conclusions

Lines 260-262 – surely there is experimental evidence that would help determine guidelines for what is acceptable accuracy, rather than simply relying on the judgement of the clinician?

While there are some evidence on what an acceptable HR for an AF dog is, the authors are unaware of any studies that specifically looked at an acceptable accuracy for a ECG device for measuring HR in dogs with AF. In veterinary patients of AF we believe that the clinician judgement of an accuracy is likely vary and as such we would like to leave the conclusions statements in the current version.

Reviewer #3: The purpose of this study was to determine if the heart rate measured for shorter time periods (5 minutes, one, two and three hours) would equate to the mean rate as determined by 24 hour electrocardiographic monitoring. The mean heart rate for each of these four time intervals was assessed for its ‘closeness’ (within 10%, 15% or 20%) to the 24 hour mean rate. The authors embarked on this investigation with the justification that performing a 24 hour Holter for shorter periods of time would be less problematic for owners and dogs. The conclusion of this paper is that none of the shorter time periods would equate to the mean 24-hour heart rate, but the longer the recording the closer the short surveillance heart rate was to the full recording.

This is a very focused study to determine if short recordings are adequate in the evaluation of rate during atrial fibrillation. The authors state that there are no other methods currently available to determine heart rates in a dog for longer than one hour. However, an Internet search will reveal that individuals are adapting noninvasive monitoring such as Fit Bits for dogs. Additionally, other devices have been developed. (See reference: Yasin Cotur, Michael Kasimatis, Matti Kaisti, Selin Olenik, Charis Georgiou, Firat Güder. Stretchable Composite Acoustic Transducer for Wearable Monitoring of Vital Signs. Adv. Funct. Mater. 2020, 30, 1910288. Doi.org/10.1002/adfm.201910288) These may serve as a way to get minimal information (and likely inadequate when monitoring arrhythmias) regarding heart rate. Such devices could provide data for specific questions such as heart rate. Comparison of these devices with established methods such as 24 hour Holter monitoring would be important to document. It is not clear how this study could be used in such a situation. Authors need to further justify this study

This comment is similar to previous reviewers comments we have clarified our goals and added additional statements to introduction and limitations sections.

From an implementation component there are no striking errors in this manuscript. It should be noted however that the average heart rates of the dogs examined are actually low for atrial fibrillation. Dogs that have high ventricular response rates are of course the more critical for monitoring and this paper does not address that kind of patient. Differences between low and high ventricular response rate heart rates over 24 hours is likely meaningful.

We have added additional explanations for why our 24hour mean HR was considerably lower in this study to the limitations section.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

4 Sep 2020

PONE-D-20-23023R1

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

PLOS ONE

Dear Dr. Sanders,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Your manuscript was sent back to 2 of the previous Reviewer's who, together with myself, appreciate the modifications you have made to the the original manuscript. However, there are still a number of outstanding questions that should be resolved, particularly regarding statistical analysis.

Please submit your revised manuscript by Oct 19 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Daniel M. Johnson, PhD

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have addressed most of my concerns. The only concern that I have is that individual dogs might have had "better agreement" than other dogs. A more complex analysis, where "dog" is inserted as a random variable into the model, might help demonstrate this.

To highlight this, one dog might have had agreement within 10% for 70% of the 5 minute segments, while another dog (for whatever reason) might have had agreement within 10% for only 10% of the 5 minute segments. A dog with an average AF rate of 230 bpm would have a higher probability of agreeing than one that still had considerable vagal influence on the AV nodal conduction.

This type of approach will likely require consultation with a statistician/epidemiologist. The authors might consider plotting the % agreement for each shorter period against average 24hr HR to see if there is any relationship.

Reviewer #2: General comments: thank you for the opportunity to read this revised version of the manuscript – I would like to thank the authors for addressing the majority of my previous comments and concerns. I have some remaining questions, which are detailed below.

Abstract

Line 40-41 – you say in lines 37-38 that “nearly 100% of heart rate averages obtained from 2 hour ECG recordings fell within 20% range” – but here you says that all of them did – please clarify this inconsistency.

Materials and Methods

Lines 158-159 – I’m still not sure why you are reporting both the range and the IQR. It seems superfluous to me – I would just report the range.

Lines 161-163 – it is not clear from this whether you mean “within 10% of the mean” (i.e. mean +/- 5%) or “mean+/- 10%” (which is what it says in the abstract) – please reword this to improve clarity.

Lines 161-163 – I assume that you mean that you are reporting the percentage of HR averages falling within 10%, 15% and 20% of the 24 hours mean HR FOR THAT DOG (rather than the average mean HR for all dogs)? Please clarify.

Results

Line 174 – again, I don’t know why you are reporting both range and IQR.

Lines 198-200 – again, the “10% range” is not the same as “mean +/- 10%”; it is mean +/- 5%. Please make sure that the wording is clear throughout the manuscript.

Lines 198-201 – an additional way to look at this would be to report the percentage of measurements that would have resulted in a different clinical decision being taken. For example, if (for the purposes of the study) we take 125bpm as a cut-off for deciding whether or not to change medications, you could report what percentage of the shorter recording were above this cut off if the 24 hour mean was <125 (and vice versa). This would be easier for clinicians to interpret, I think.

Table 1 – the poor accuracy for 3 hour recordings in the 10% range column is intriguing – the range includes 50% (i.e. no better than flipping a coin) – can you comment on the possible reasons for this apparent anomaly?

Discussion

Lines 232-234 – I appreciate what you are saying here (clearly, 20% of a larger number is bigger than 20% of a smaller number). However, I still think that even at lower heart rates 20% is too inaccurate – even at 120bpm, if you are talking about +/20% (which remains unclear – see previous comments) then this would go as high as 144bpm (which would almost certainly be interpreted differently). Even if you mean “within 20% of the mean” that would take you to 132bpm, which may well be interpreted differently from 120bpm. Personally, on the basis of your results I would want at least a 3 hour recording.

Lines 247-250 – I sincerely hope that clinicians do not accept wider limits of accuracy! See previous comments.

Lines 251-257 – these two sentences appear to contradict one another – if the other at home tools have the capacity to record continuous ECG for more than one hour, why is it not feasible to use them to make recording >1 hour long? Should it not say “…do not have the capacity to record continuous ECG for more than one hour duration”?

Conclusions

Lines 289-290 – that is a fair interpretation of your results – but I still maintain that the level of inaccuracy is potentially dangerous in terms of its impact on how the clinician might choose to manage the case (e.g. make dose adjustments to rate control medications (or not)). I think that it is very important that you acknowledge this overtly in the text – this is not simply a mathematical problem, but something that potentially impacts on how well patients are managed clinically.

Lines 290-292 – I think that it is really important that you provide more information that helps the reader understand better what 20% inaccuracy actually means for their patients; otherwise how can they judge what an acceptable level of error is?

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

28 Sep 2020

Reviewer #1: The authors have addressed most of my concerns. The only concern that I have is that individual dogs might have had "better agreement" than other dogs. A more complex analysis, where "dog" is inserted as a random variable into the model, might help demonstrate this.

To highlight this, one dog might have had agreement within 10% for 70% of the 5 minute segments, while another dog (for whatever reason) might have had agreement within 10% for only 10% of the 5 minute segments. A dog with an average AF rate of 230 bpm would have a higher probability of agreeing than one that still had considerable vagal influence on the AV nodal conduction.

This type of approach will likely require consultation with a statistician/epidemiologist. The authors might consider plotting the % agreement for each shorter period against average 24hr HR to see if there is any relationship.

We have now added an additional analysis that accounts for the random effect of dogs and added necessary discussions.

Reviewer #2: General comments: thank you for the opportunity to read this revised version of the manuscript – I would like to thank the authors for addressing the majority of my previous comments and concerns. I have some remaining questions, which are detailed below.

Abstract

Line 40-41 – you say in lines 37-38 that “nearly 100% of heart rate averages obtained from 2 hour ECG recordings fell within 20% range” – but here you says that all of them did – please clarify this inconsistency.

“Nearly 100%” was used instead of “all” in the lines 37-38 since the CI for the median included 100%. We have changed the wording for consistency.

Materials and Methods

Lines 158-159 – I’m still not sure why you are reporting both the range and the IQR. It seems superfluous to me – I would just report the range.

Removed all applicable IQR throughout the manuscript

Lines 161-163 – it is not clear from this whether you mean “within 10% of the mean” (i.e. mean +/- 5%) or “mean+/- 10%” (which is what it says in the abstract) – please reword this to improve clarity.

Recommended changes made to improve clarity.

Lines 161-163 – I assume that you mean that you are reporting the percentage of HR averages falling within 10%, 15% and 20% of the 24 hours mean HR FOR THAT DOG (rather than the average mean HR for all dogs)? Please clarify.

Clarification provided

Results

Line 174 – again, I don’t know why you are reporting both range and IQR.

Recommended changes made

Lines 198-200 – again, the “10% range” is not the same as “mean +/- 10%”; it is mean +/- 5%. Please make sure that the wording is clear throughout the manuscript.

Recommended changes made throughout the manuscript

Lines 198-201 – an additional way to look at this would be to report the percentage of measurements that would have resulted in a different clinical decision being taken. For example, if (for the purposes of the study) we take 125bpm as a cut-off for deciding whether or not to change medications, you could report what percentage of the shorter recording were above this cut off if the 24 hour mean was <125 (and vice versa). This would be easier for clinicians to interpret, I think.

We do see practical utility in providing clinical decision cut-off and accuracy for such a cutoff above and below. The accuracy of a binary cutoff would be naturally better than accuracy for a specific percent range from 24-hour mean HR. For example, for a dog with a 24 hour mean HR of 200bpm, the 20% range would include heart rates from 140 to 240 bpm. If for instance this dog’s minimum and maximum heart rate ranges from 130 – 220 using 1-hour averages at home, 100% of the values will be above 125 bpm while a lot lesser percentage of values will be between 180-220. However, the practical utility may be limited given that a clinician may not know the predicted 24-hour mean HR at the time of any short-term ECG evaluation. We have highlighted this information as well in the discussion. As such given the limited practical utility we have not added this information.

Table 1 – the poor accuracy for 3-hour recordings in the 10% range column is intriguing – the range includes 50% (i.e. no better than flipping a coin) – can you comment on the possible reasons for this apparent anomaly?

There are at least two scenarios. One such scenario would be a dog with a high 24-hour mean HR and if such dog can reach really low HR values at rest. In contrast also can happen in dogs with lower mean HR but has the ability to reach extremely high max HR during activity. In either case, the analysis looks for HR values within an acceptable percentage from the 24 hour mean HR. The probability of falling out of that range may be lower than a coin toss in some dogs. If it a cutoff such as value above and below 125 bpm was used, then the probability (similar to a coin toss) would be expected to be above 50%.

Discussion

Lines 232-234 – I appreciate what you are saying here (clearly, 20% of a larger number is bigger than 20% of a smaller number). However, I still think that even at lower heart rates 20% is too inaccurate – even at 120bpm, if you are talking about +/20% (which remains unclear – see previous comments) then this would go as high as 144bpm (which would almost certainly be interpreted differently). Even if you mean “within 20% of the mean” that would take you to 132bpm, which may well be interpreted differently from 120bpm. Personally, on the basis of your results I would want at least a 3 hour recording.

We agree that it is bit of a judgement call. We have now modified the sentence to better reflect what we are attempting to say.

Lines 247-250 – I sincerely hope that clinicians do not accept wider limits of accuracy! See previous comments.

That is our hope too.

Lines 251-257 – these two sentences appear to contradict one another – if the other at home tools have the capacity to record continuous ECG for more than one hour, why is it not feasible to use them to make recording >1 hour long? Should it not say “…do not have the capacity to record continuous ECG for more than one hour duration”?

Yes it should say do not have the capacity. We apologize for the oversight.

Conclusions

Lines 289-290 – that is a fair interpretation of your results – but I still maintain that the level of inaccuracy is potentially dangerous in terms of its impact on how the clinician might choose to manage the case (e.g. make dose adjustments to rate control medications (or not)). I think that it is very important that you acknowledge this overtly in the text – this is not simply a mathematical problem, but something that potentially impacts on how well patients are managed clinically.

We have added an additional statement to clarify this.

Lines 290-292 – I think that it is really important that you provide more information that helps the reader understand better what 20% inaccuracy actually means for their patients; otherwise how can they judge what an acceptable level of error is?

We have added an additional statement to provide clarification.

Submitted filename: Response to Reveiwers.docx

Click here for additional data file.

7 Oct 2020

PONE-D-20-23023R2

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

PLOS ONE

Dear Dr. Sanders,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Although the Reviewers and myself feel the changes you made much improved the manuscript, Reviewer 2 has a number of outstanding issues that need to be addressed before the manuscript can be accepted for publication in PLOS ONE.

Please submit your revised manuscript by Nov 21 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Daniel M. Johnson, PhD

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: General comments – thank you for responding to the majority of my previous comments and questions. I have a few remaining comments, which are detailed below.

Results

Lines 224-227 – surely the purpose of measuring the shorter duration heart rate averages is to try to predict the 24 hour mean heart rate; this analysis is therefore back-to-front.

Lines 227-229 – what is the P value for this effect?

Lines 229-232 – apologies if I have missed something, but I don’t understand where this information comes from – please can you explain how you derived this (i.e. where the information that “on…average 60% of the HR averages obtained from short duration ECG recordings fall within +/-10% of the 24 hour mean HR” comes from – I can’t figure it out from the preceding sentences).

Discussion

Lines 244-257 – I appreciate what you are trying to say – but this argument feels somewhat circular (essentially you can’t know what level of accuracy you need until you know what the 24 hour mean is, by which point you have done a 24h Holter, so the question becomes moot). Given this, can you suggest any situations in which a shorter recording WOULD be valuable? After all, surely the point of doing a 24h Holter in dogs with AF is that you don’t know what the 24h mean heart rate is and you want to find out? After all, even if you have previously Holtered a dog, surely the point of re-Holtering them is to see if anything has changed, so you don’t know what the mean HR is going to be before you do the Holter.

Conclusions

Lines 307-309 – I would still prefer this sentence to be edited or removed, as I don’t think that clinical functionality is determined by the clinician’s judgement (if something is wildly inaccurate that doesn’t change just because the clinician has decided that they are happy to accept a wildly inaccurate result).

Lines 309-314 – thank for adding this information – I think it provides very valuable context.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Mark Rishniw

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

9 Oct 2020

Lines 224-227 – surely the purpose of measuring the shorter duration heart rate averages is to try to predict the 24 hour mean heart rate; this analysis is therefore back-to-front.

This analysis was requested by reviewer one to highlight the individual differences between dogs in accuracy of short duration ECG.

Lines 227-229 – what is the P value for this effect?

As such there is no p values associated with this effect. It was calculated as a % of individual variances (random effect) to total residual variance. We have modified the wording to reflect it. Prior to selecting the model, models with and without dog as a random effect was compared and addition of random effect was found to be statistically significant.

Lines 229-232 – apologies if I have missed something, but I don’t understand where this information comes from – please can you explain how you derived this (i.e. where the information that “on…average 60% of the HR averages obtained from short duration ECG recordings fall within +/-10% of the 24 hour mean HR” comes from – I can’t figure it out from the preceding sentences).

The 60% in this sentence applies to the intercept of the linear mixed model.

Discussion

Lines 244-257 – I appreciate what you are trying to say – but this argument feels somewhat circular (essentially you can’t know what level of accuracy you need until you know what the 24 hour mean is, by which point you have done a 24h Holter, so the question becomes moot). Given this, can you suggest any situations in which a shorter recording WOULD be valuable? After all, surely the point of doing a 24h Holter in dogs with AF is that you don’t know what the 24h mean heart rate is and you want to find out? After all, even if you have previously Holtered a dog, surely the point of re-Holtering them is to see if anything has changed, so you don’t know what the mean HR is going to be before you do the Holter.

We appreciate your input on this. Prior to this publication, the accuracy of shorter duration ECG to predict 24 hour mean was unknown in AF dogs. If the shorter duration ECG predicted 24 hour mean HR better than the what was noted in this study, then certainly it could have replaced Holter monitoring for all dogs. But in this study the accuracy widely varied between dogs and was not influenced by the 24 hours mean HR, complicating further on selection of a desirable accuracy. We wanted to state this clearly to avoid any misleading information to the readers on the applicability of the results. As we pointed out in the discussion, one situation may be when the in-hospital ECG HR is above 155 bpm, then the 24-hour mean HR can be predicted to above 140 bpm. Then we would think 20% accuracy would be too high and a tighter accuracy is desired.

Conclusions

Lines 307-309 – I would still prefer this sentence to be edited or removed, as I don’t think that clinical functionality is determined by the clinician’s judgement (if something is wildly inaccurate that doesn’t change just because the clinician has decided that they are happy to accept a wildly inaccurate result).

This sentence is removed

Lines 309-314 – thank for adding this information – I think it provides very valuable context.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

19 Oct 2020

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

PONE-D-20-23023R3

Dear Dr. Sanders,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Daniel M. Johnson, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: Thank you for your clear and comprehensive responses to my previous comments - I have no remaining questions.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

21 Oct 2020

PONE-D-20-23023R3

Comparison of heart rate obtained from shorter duration Holter recordings to 24-hour mean heart rate in dogs with atrial fibrillation

Dear Dr. Sanders:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Daniel M. Johnson

Academic Editor

PLOS ONE