This study evaluates implantable loop recorders for detection of atrial fibrillation in horses with induced persistent AF and horses in sinus rhythm (study I) and also in a group of horses in field conditions (study II).

Overall, the study is well designed and well conducted and the conclusions are valid.

However, I think the manuscript could be improved by addressing the following points.

Major:
In sub-study 1, please describe in the methods exactly how horses were restrained for the 24 hour recording. Were they stabled? Was atrial fibrillation maintained and verified by the holter monitor at all times in all horses in the AF group?
Why were horses in sinus rhythm evaluated less frequently than horses in AF? Could this have affected your results? More frequent assessment of horses in SR might have improved the precision in determining the specificity.

There is a substantial discrepancy in the number of false positives in study 1 compared to study 2. This is mentioned in the discussion at lines 240 -243 and these were attributed motion artefacts that occurred during exercise. As you state, previous studies have also found that the recorders are not accurate during exercise. Are you able to report with more clarity whether or not horses were exercising at the time of the false positives in study 2? Could this information be included in Table 2? Or would you consider it appropriate to use heart rate as a proxy for exercise, for example determining the accuracy of the loop recorders in detecting AF at heart rates below 100 beats per minute? I also think for clarity it would be worthwhile reporting the sensitivity and specificity of the device in study 2. Are you able to determine the rate of true positives, true negatives based on visualising the ECGs recorded by the ILR? This would allow for improved clarity in the comparison of the device in stabled (?) horses vs the field study. As it is currently reported I feel like the reported sensitivity (and specificity in particular) are somewhat misleading as there were so few false positives in study 1 compared to study 2. Perhaps splitting study 2 into horses that were resting in the field vs exercising might better support that the loop recorder does perform adequately in the field except for when horses are exercising.

The reason why I think this is important is that, as the study is currently reported, it could be concluded that the device is accurate only in horses that are stabled.

Is it possible to limit the parameters of the recorded events to for example events at heart rates less than 100 beats per minute? Is this likely to improve the accuracy of the device? This could be important for readers to understand and greatly enhance the utility of the device. I understand that the point of the device is to capture episodes of AF that are infrequent and therefore long recording periods are required. Also, increasing the sensitivity of the device improves the chances of capturing infrequent episodes but that the device has a limit on how many events can be captured between downloads. If for example, the device was able to ignore events at higher heart rates (as a proxy for exercise), this might lead to fewer erroneously captured events (false positives). As reported in study 2, due to the high number of false positives, the captured events are an indication of possible AF, but all ECGs need to be reviewed to determine the actual heart rhythm. Reducing the number of false positives would improve device performance both because the event capture limits would be reached more slowly and also because clinicians would save time in reviewing ECGs that are be false positives.

Minor:
Abstract:
Line 9: Consider changing "presence" to "to be sustained"
Line 17: Consider specifying "The ILR ECGs were compared to SURFACE ECGS"

Introduction:
Line 55 - 56: Have any previous studies supported your hypothesis that breed and BCS will affect performance of the device or was this based on your clinical experience?

Line 59: Rather than "different breeds" use "WB and STB horses" as this is more specific. You could also include disciplines in this sentence ie. "cohort of WB nad STB horses of different disciplines, performance levels and body condition scores."

Methods:
Was baseline clinical examination, echocardiography and 24 hr ECG performed in all 28 horses in study 1 as per figure 2?
Were any horses in study 1 excluded at any time point, for example following clinical examination?
In study 1 were horses maintained in AF (sometimes through tachypacing) throughout the entire 24 hour recording or were there occasional periods of SR?
Line 71: Consider breaking up sentence as follows " Finally, an echocardiographic examination was performed to evaluate cardiac structure and function. Mild valvular etc"
Line 72: How many horses had mild regurgitation?
Line 78: Over what time period were horses recruited to the study? How were they recruited?
Line 100: consider changing to "enrolled in other PUBLISHED studies were included"
Lines 102-113 appear to be a repetition of the previous paragraph.
Is a subheading missing prior to line 123?
Line 128: Consider verified rather then noted in this sentence
Line 144: "Due to variations in enrolment timing, the AF horses did not contribute with an equal quantity of ECGs" Could you expand on this point to make the meaning clearer?
Line 150: consider describing this more clearly as "captured event ECGs"
Line 161 consider changing "occurrence" to relationship
Lines 162 and 164: Why was both a Mann Whitney test and linear regression analysis used to compare BCS and false positives?

Results:
Throughout the manuscript, please report the statistical test used in the results as well as in the methods section so that it is clear which statistical method is used for each comparison as per the EVJ manuscript checklist. Please also report confidence intervals.

Line 183: change at to on
Line 184: change where to were
Line 186 change on to of

Discussion:
Line 213: Would it be more accurate to say stabled horses rather than resting horses? Horses in study 2 were presumably also at rest (ie not exercising) for most of the time but there were a high number of false positives in study 2.
Lines 288 - 293: Both sentences in this section very similar, please choose one and remove the other. In relation to clinical significance of AF in horses performance limitation is much more important than stroke risk. It is likely than even very short episodes of less than 30 seconds will impact performance in highly athletic disciplines. This performance limitation has been previously quantified by Buhl 2018
Line 301: Is stabled more accurate than confined? How were horses confined?
Line 304: Please consider modifying "The device offers additional functions and settings FOR DETECTION OF OTHER ARRHYTHMIAS but these aspects were beyond the scope of this investigation."
Line 308: specify in this sentence that the device effects detects in "STABLED horses"
Line 313: identify rather than identifying
Line 313: Future validation also in horses with naturally occurring pAF

Figures:
Where statistical test P values are reported please also include the statistical test performed.

Thank you for addressing my previous concerns