Article Text

Original article
Is 7-day event triggered ECG recording equivalent to 7-day Holter ECG recording for atrial fibrillation screening?
  1. Laurent Roten,
  2. Manuel Schilling,
  3. Andreas Häberlin,
  4. Jens Seiler,
  5. Nicola G Schwick,
  6. Jürg Fuhrer,
  7. Etienne Delacrétaz,
  8. Hildegard Tanner
  1. Department of Cardiology, Inselspital, Bern University Hospital and University of Bern, Switzerland
  1. Correspondence to Laurent Roten, Kardiologie, Inselspital, Freiburgstrasse, 3010 Bern, Switzerland; laurent.roten{at}insel.ch

Abstract

Objective Prolonged ECG monitoring is standard for atrial fibrillation (AF) screening. This study investigated whether 7-day event triggered (tECG) ECG recording is equivalent to 7-day continuous Holter (cECG) ECG recording for AF screening.

Design Both a cECG (Lifecard CF) and a tECG (R.Test Evolution 3) were simultaneously worn for 7 days by patients with known or suspected paroxysmal AF.

Results In 100 corresponding recordings, median effective duration of monitoring was 165 h (range 10–170 h) for cECG and 137 h (0–169 h) for tECG (p<0.001). Median number and total duration of arrhythmias (AF, atrial flutter or atrial tachycardia) of ≥30 s duration recorded by cECG were 10 (1–428) and 1030 min (≤1–10 020), respectively. An arrhythmia was recorded in 42 cECGs (42%) versus 37 tECGs (37%, p=0.56). Triggered ECG failed to record an arrhythmia in cECG positive cases because of interrupted monitoring in four cases and because of recording failure in one case. Sensitivity, specificity, and positive and negative predictive values of tECG therefore were 88%, 100%, 100%, and 92%, respectively. Quantitative cECG analysis required a median of 20 min (3–205 min) and qualitative tECG analysis 4 min (1–20 min; p<0.001). Skin irritation was a frequent side effect (42%) resulting in premature removal of devices in 16% of patients.

Conclusion Sensitivity of tECG for AF screening as compared to cECG is lower, mainly because of shorter effective monitoring duration. Qualitative tECG analysis is less time consuming than quantitative cECG analysis. Skin irritation is a frequent side effect and reason for premature device removal.

  • Atrial fibrillation
  • atrial flutter
  • Holter ECG
  • implantable cardioverter defibrillator (ICD)
  • sudden adult death syndrome
  • pacemakers
  • arrhythmias
  • invasive electrophysiology
  • atrial arrhythmias
  • rhythms
  • radiofrequency catheter ablation
  • ventricular tachycardia
  • atrial fibrillation, syncope
  • atrioventricular block
View Full Text

Statistics from Altmetric.com

Introduction

Atrial fibrillation (AF) is the most frequent arrhythmia and is associated with significant morbidity and mortality.1 Catheter ablation of AF is increasingly being performed worldwide, and screening for arrhythmia recurrence during follow-up has important therapeutic implications.2 AF is also a major cause of ischaemic stroke and further strokes can effectively be prevented by anticoagulants if AF is diagnosed. Therefore, reliable devices for AF screening are important.

Duration of monitoring has the biggest impact on sensitivity of devices used for AF screening. Implanted devices allowing continuous rhythm surveillance are the most sensitive tools.3 4 Because of high costs and the need for an invasive procedure they are not routinely used.5

Seven day continuous Holter ECG recording systems (cECG) are a reasonable compromise. They allow for prolonged recording of up to 168 h of ECG and are superior to shorter duration Holter ECG, whereas sensitivity is similar compared to daily transtelephonic ECG.6 7 Alternatively, event triggered ECG recording systems (tECG) can be employed.8 Recordings of tECG are faster to analyse, but an ECG is stored for only limited duration and sensitivity therefore relies on trigger settings and software performance of the device. The purpose of this study was to investigate whether tECG is equivalent to cECG for AF screening.

Methods

In this prospective, investigator initiated, single centre study we compared a cECG with a tECG for AF screening. Both devices were simultaneously worn by every patient for 7 days. Patients were allowed to take off the devices for short periods of time—for example, to take a shower—but were asked not to selectively wear only one device. They were instructed how to put the devices back on and were provided with enough spare electrodes for 7 days. The same electrode type was used for both devices. A log sheet was maintained by every patient with a description of their symptoms. After 7 days, patients sent back both devices and the log sheet by mail. Thereafter, patients were questioned by telephone about side effects, whether they had removed the devices prematurely, and which device they preferred. Two experienced electrophysiologists analysed the device recordings. The primary end point was recording of an arrhythmia by either device. The term arrhythmia is used in this article to subsume AF, atrial flutter and atrial tachycardia of more than 30 s duration. Arrhythmias were defined according to standard definitions.1 9 The study was approved by the locally appointed ethics committee, and all subjects gave written, informed consent.

Study population

Assessment of arrhythmia burden before catheter ablation of paroxysmal AF and screening for AF recurrence 3, 6, and 12 months after catheter ablation is routinely performed at our institution. These patients were eligible for the study. Also included were patients with symptoms very suggestive of AF (episodes of fast and irregular palpitations) but without documentation of AF so far. Patients with persistent AF and patients unable to handle the devices independently were excluded. Between November 2007 and January 2011 a total of 88 patients were included in the study, of whom 12 patients undergoing catheter ablation of AF were included twice. Therefore, 100 datasets with corresponding recordings of cECG and tECG were available for analysis. Patient characteristics are shown in table 1.

Table 1

Patient characteristics (n=88)

cECG recorder

For cECG the Lifecard CF (Spacelabs Healthcare, Issaquah, Washington, USA) was used. This system allows continuous recording of two ECG channels for 7 days. Three ECG electrodes were applied to each patient: one right to the upper border of the sternum (electrode 1); one on the right mid-clavicular line at the lower right border of the ribcage (electrode 2); and one on the left anterior axillary line at the lower left border of the ribcage (electrode 3). ECGs were derived from between electrodes 1 to 3 and 2 to 3.

Analysis of cECG

Recordings were analysed with the software Lifescreen (Spacelabs Healthcare). Complete recordings were screened for arrhythmias. This was done by plotting RR intervals against time and visually analysing intervals of 10–15 min. With this strategy, arrhythmias can easily be recognised because of their typical patterns and it is directly evident whether their duration is more than 30 s (for examples, see supplemental figure 1). The duration of effective recording was calculated as the total time with at least one channel recording an ECG of sufficient quality for rhythm analysis. The number and total duration of episodes as well as duration of longest episode and maximal heart rate during arrhythmia were evaluated for AF, atrial flutter and atrial tachycardia. The time needed for cECG analysis was measured.

tECG recorder

The tECG used in this study was the R.Test Evolution 3 (Novacor, Rueil-Malmaison, France). This system monitors and displays the heart rate and summarises the number of atrial and ventricular premature beats as well as supraventricular and ventricular tachycardias during up to 8 days, but without recording a continuous ECG. It can store one ECG channel for a total duration of 20 min. Triggers for recording an ECG stripe can be programmed individually as well as the recording window before and after each trigger and the maximum possible number of recordings for each trigger. Once the maximum number of recordings for a trigger is attained, only events better fulfilling triggering criteria than already recorded events (eg, longer pauses) will be recorded and replace less severe recordings. For this study, the triggers for recording an ECG by the tECG were programmed as displayed in table 2. Two electrodes were applied to each patient, one on the upper part of the sternum and one on the left anterior axillary line at the lower left border of the ribcage. The ECG was derived from between the two electrodes.

Table 2

Settings of triggered ECG

Analysis of tECG

With the software RTSoft (Novacor) all recorded events as well as the 7-day heart rate histogram and arrhythmia summary were printed for analysis. The heart rate histogram in this device is only displayed at times when signal quality is suitable for automatic signal analysis, otherwise gaps are displayed. The duration of effective monitoring was calculated from the heart rate histogram and represents the total time with monitoring of heart rate (ie, signal suitable for automatic rhythm analysis). Heart rhythm of all recorded events was diagnosed. In case of a recording triggered by an artefact and showing sinus rhythm, the recorded event was labelled an artefact. The time needed for tECG analysis was measured.

Statistical analysis

Categorical variables are expressed as numbers and percentages, continuous variables as mean and SD or median and range. Categorical variables were compared with the χ2 test, and continuous variables with the unpaired t test. A value of p<0.05 was considered statistically significant. Analyses were performed using SPSS V.17.0 (SPSS Inc).

Results

Arrhythmias recorded by cECG

An arrhythmia was recorded by cECG in 42 recordings (42%, for details see supplemental table 1). In 13 recordings (13%) only one single arrhythmia was recorded. Median duration of arrhythmia in these 13 recordings was 1020 min (range ≤1–10 020 min) and duration was ≤1 min in three recordings. There was another patient with a total arrhythmia duration of 4 min, whereas all the remaining patients had a total arrhythmia duration >25 min. Non-sustained atrial tachycardias were found in 81 recordings (81%). A log sheet was available for 99 recordings with description of symptoms in 44 log sheets (44%). Symptoms correlated only to sinus rhythm in 15 recordings (15%), only to arrhythmia episodes in 18 recordings (18%), and to both sinus rhythm and arrhythmia episodes in 11 recordings (11%). Of 41 log sheets corresponding to recordings with documented arrhythmia, a symptomatic arrhythmia was described in 29 log sheets (71%), with a median of 2 (range 1–428) symptomatic arrhythmias reported per log sheet.

Arrhythmias recorded by tECG

An arrhythmia was recorded by tECG in 37 recordings (37%, for details see supplemental table 2). A total of 1258 ECG stripes were recorded by the 100 tECG recordings (mean 12.6±4.1; median 12, range 0–21).

Comparison of cECG versus tECG

The cumulative duration of effective monitoring of all 100 recordings was 15 373 h for cECG and 11 803 h for tECG (see also supplemental figure 2). The duration of effective monitoring was significantly longer with cECG, whereas the time needed for analysis was significantly shorter with tECG (table 3). tECG failed to document an arrhythmia in five recordings in which an arrhythmia was recorded by the corresponding cECG (table 4). In four of these cases tECG did not monitor heart rhythm during all arrhythmia episodes, probably because of unsuitable signal quality for analysis, resulting in short total effective monitoring duration (no monitoring) in all four cases. In a case with one single short arrhythmia episode in cECG, no arrhythmia was recorded by tECG despite suitable signal quality for automatic rhythm analysis during the missed episode as evident from the heart rate histogram (no recording, table 4 and supplemental figure 3). In this case, non-sustained arrhythmia episodes were recorded by the tECG after the missed arrhythmia episode. Whether the missed arrhythmia episode was overwritten by subsequent arrhythmia episodes or failed to trigger a recording cannot be assessed.

Table 3

Comparison of continuous (cECG) versus triggered (tECG) ECG

Table 4

Recordings with missed arrhythmia

Continuous ECG did document an arrhythmia in every patient in which an arrhythmia was found in the corresponding tECG. The sensitivity, specificity, and positive and negative predictive values of tECG to record an arrhythmia in cases with cECG positive recordings were 88%, 100%, 100%, and 92%, respectively.

Side effects and patient preference

Telephone questioning was possible in 83 patients (94%). Skin irritation was the only side effect and was reported by 35 patients (42%). Thirteen patients (16%) removed the devices prematurely because of skin irritation. Patient preference was as follows: tECG 42%; cECG 36%; no preference 22%.

Discussion

The main findings of this study are:

  1. Sensitivity of tECG for arrhythmia screening as compared to cECG is lower (88%), mainly because of shorter effective monitoring duration and only rarely because of failure to record an arrhythmia episode.

  2. Quantitative cECG analysis is significantly more time consuming than qualitative tECG analysis.

  3. Skin irritation is a frequent side effect resulting in premature device removal in 16% of patients.

When screening patients for AF, monitoring duration is the variable which most affects the sensitivity of the devices in use. Even when an arrhythmia is recorded by a screening tool, it also has to be recognised as such. In case of the cECG used in this study, analysis was done manually by reviewing complete recordings for the presence of arrhythmia. Diagnosis of a recorded arrhythmia is therefore operator dependent. With the strategy employed in this study for cECG analysis it is unlikely that a recorded arrhythmia of >30 s duration would be unnoticed. Furthermore, signal quality can also be assessed throughout recording, and because two ECG channels are recorded simultaneously the probability of a suitable signal for rhythm analysis increases.

With the tECG, on the other hand, the device software decides what is recorded according to trigger settings. Performance of device software therefore relies on parameter settings and device ability to identify RR intervals successfully. The latter depends on signal quality as well. The device will interrupt rhythm analysis when signal quality is unsuitable for automatic rhythm analysis, represented by gaps in the heart rate histogram. As the tECG monitors only one ECG channel, the probability of a suitable signal for rhythm analysis is lower than for cECG.

These differences in device function are responsible for the pronounced differences in effective monitoring duration between the two devices compared in this study, and also for the lower performance of tECG compared with cECG. Overall, effective monitoring duration was 30% longer with cECG than with tECG. All patients were asked to wear both devices simultaneously throughout the 7 days. It is unlikely that differences in effective monitoring duration are a result of patients preferring to wear the cECG instead of the tECG. Most likely, these differences occurred because the tECG monitors only one ECG channel, which results in a higher probability of unsuitable signal quality for automatic rhythm analysis compared with the two ECG channels recorded by the cECG.

Accordingly, in four of the five cases in which tECG failed to record an arrhythmia, tECG did not monitor heart rhythm during missed arrhythmias. In fact, effective monitoring duration by tECG was considerably shorter than by cECG in these four cases. It is important to recognise short effective monitoring duration despite adequate wearing duration when interpreting a tECG, as this will decrease sensitivity to detect an arrhythmia. With the tECG used in this study short effective monitoring duration is evident from the heart rate histogram. In such cases, the tECG can simply be repeated or a cECG offered to the patient to increase total monitoring duration and sensitivity.

tECG failed to record an arrhythmia in a case with one single short arrhythmia episode in cECG, despite monitoring of the heart rhythm by tECG during the missed episode. Although we aimed to include patients with rare and short episodes of arrhythmia, only three patients had one single episode lasting <1 min. It can therefore only be speculated about the performance of tECG in cases of rare and short arrhythmia episodes. This is nevertheless important, as studies with implantable devices have already demonstrated an increased risk of stroke with atrial high rate episodes of short duration.10

Trigger settings for AF screening as used in our study rely on sudden changes in RR interval during AF or pauses as they can occur after arrhythmia termination. It is imaginable that short arrhythmia episodes can present with limited changes of RR intervals compared to sinus rhythm and therefore will not trigger a recording. Different trigger settings or the use of complex algorithms for rhythm analysis will certainly have an impact on the performance of devices used. The longer the monitoring duration, the more sophisticated the algorithms have to be in order to manage incoming data volume.5 11 As in the case of the implantable loop recorder, mainly RR intervals are analysed by algorithms with only short-time ECG recording and arrhythmia burden is readily displayed.4 Therefore, these devices function like a black box, and one can never be sure that short arrhythmia episodes have not been missed.

In our study, analysis of cECG was significantly more time consuming than analysis of tECG. Nevertheless, arrhythmia burden can also be assessed in cECG while in tECG only a qualitative analysis is possible. We did not assess the time needed for qualitative analysis of cECG. In recordings with arrhythmia episodes, qualitative analysis of cECG will certainly be faster than quantitative analysis, whereas qualitative analysis of cECG in recordings without arrhythmia will not be different as complete recordings have to be screened for arrhythmia episodes. The more timesaving analysis of tECG may be an important advantage in the current era of rising health expenditures and may compensate for the reduced sensitivity and higher need for repeat tECG in cases of short effective monitoring duration.

Many patients with AF are asymptomatic or have both symptomatic and asymptomatic AF episodes.8 12 13 We confirm a high rate of only asymptomatic arrhythmia episodes in our patient population.

The main drawback of any kind of external ECG recorder is the increased risk of cutaneous side effects with prolonged wearing duration, resulting in premature device removal as occurred in our study. Although the risk of this side effect might have been higher in the present study because patients had to wear five instead of two or three ECG electrodes, it illustrates the limits of prolonged external ECG recording for AF screening. New screening methods for AF, allowing for prolonged monitoring duration without side effects and high sensitivity, are still required.

Limitations

Median effective recording time with cECG was 165 h. Nevertheless, in some patients effective monitoring duration with cECG was shorter and arrhythmia episodes could also have been missed by cECG. Different parameter settings and software algorithms of tECG may alter performance of the devices and the results can therefore not be generalised to all settings and types of tECG.

Conclusions

Sensitivity of tECG for AF screening as compared to cECG is lower, mainly because of shorter effective monitoring duration. It is important to recognise short effective monitoring duration by tECG and to repeat tECG until adequate effective monitoring duration has been obtained. Single short arrhythmia episodes may be missed by tECG. Qualitative tECG analysis is less time consuming than quantitative cECG analysis. Skin irritation is a frequent side effect of 7-day Holter ECG.

References

View Abstract

Supplementary materials

Footnotes

  • Funding Dr Tanner was supported by a grant from the Swiss Foundation for Pacemaker and Electrophysiology.

  • Competing interests None.

  • Ethics approval Ethics approval was approved by Kantonale Ethikkommission Bern (http://www.kek-bern.ch).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.