Diagnostic Accuracy of a Smartphone-Operated, Single-Lead Electrocardiography Device for Detection of Rhythm and Conduction Abnormalities in Primary Care

Jelle C.L. Himmelreich, MD, MSc; Evert P.M. Karregat, MD, MSc; Wim A.M. Lucassen, MD, PhD; Henk C.P.M. van Weert, MD, PhD; Joris R. de Groot, MD, PhD; M. Louis Handoko, MD, PhD; Robin Nijveldt, MD, PhD; Ralf E. Harskamp, MD, PhD

Disclosures

Ann Fam Med. 2019;17(5):403-411. 

In This Article

Discussion

The diagnostic properties of the KardiaMobile 1L-ECG device as assessed by cardiologists against simultaneously performed 12L-ECG in a primary care population were excellent for AF/AFL. The AF detection algorithm showed high sensitivity and specificity for AF/AFL. Visual assessment of the 1L-ECG recordings by cardiologists resulted in high sensitivity and specificity for rhythm abnormalities and high specificity but low sensitivity for conduction abnormalities. To our knowledge, this is the first study to validate the KardiaMobile device for both AF and common non-AF ECG abnormalities against simultaneously performed 12L-ECG in a primary care population.

Clinical Relevance

Patients who present to their primary care physician with palpitations often no longer have symptoms at the time of consultation or when ECG is performed.[3] When ECG is performed during palpitations, an abnor mal heart rhythm is found in approximately one-half of patients, whereas this drops to approximately one-fifth when symptoms are no longer present at the time of ECG.[3] The findings from the present study are therefore highly relevant for primary care physicians because the smartphone-operated ECG device operates as a point-of-care test and allows for immediate rhythm assessment during a symptomatic episode. Moreover, our findings support patients' use of the device at home as a 1L event recorder, provided that the ECG readings are assessed by a cardiologist. We note that the device is already available on the consumer market for this purpose.

Our stratified analysis by indication for ECG showed that in older patients for whom ECG was not indicated primarily for cardiac symptoms, a negative reading excluded AF with a similarly high degree of certainty as that for symptomatic patients, despite differences in pretest likelihood within our sample. These results may be relevant for primary care physicians because they are encouraged to perform proactive case identification in asymptomatic patients with elevated risk of developing AF (eg, via pulse palpation followed by ECG).[16] Here, the 1L-ECG device could be a valuable point-of-care tool for at-risk patients for whom traveling to the practice for standard 12L-ECG is too cumbersome or for primary care physicians who do not possess a 12L-ECG device.

We added the comparison on any rhythm abnormality because for primary care patients, cardiac symptoms may often be explained by ectopy.[1] We found that the 1L-ECG device can correctly classify instances of ectopic beats, suggesting that it may be useful as a point-of-care diagnostic instrument for this rhythm anomaly.

The 1L-ECG device was less sensitive for conduction abnormalities, which in the present study particularly involved the detection of first-degree AV blocks. For primary care physicians, however, the detection of conduction abnormalities is generally less clinically relevant than the detection of arrhythmias, with the notable exception of decisions regarding the prescribing of QT-prolonging medication.[17] Whereas the QT interval was not scored in the present study, others have reported the KardiaMobile's ability to accurately assess QT intervals.[18]

Strengths and Limitations

Our study had a number of strengths. First, we included consecutive patients who underwent 12L-ECG as part of routine medical practice, resulting in a cohort generalizable to general practice. Second, the study design ensured simultaneous rather than consecutive 1L- and 12L-ECG recordings, as done in prior studies.[10,19,20] This allowed for a comparison on the detection of ectopic beats, which may be a frequent cause for palpitations in primary care.[21] Third, by providing a stratified analysis according to indication for ECG, we were able to show that the 1L-ECG device performed similarly in patients with symptoms vs those who present as part of protocol-driven (secondary) preventive care. Fourth, we ensured standardized interpretation of all recordings by blinded assessment of 1L-and 12L-ECG recordings in random order.

Several limitations should be mentioned. First, this study was not designed to determine to what extent primary care physicians are able to assess the 1L-ECG signal, but rather to describe the test characteristics of the 1L-ECG device in a representative primary care patient sample when analyzed by experts (cardiologists/electrophysiologists). Second, the use of recordings of different durations (10-second 12L-ECG vs 30-second 1L-ECG) may have led us to underestimate the specificity of 1L-ECG in the analysis of any rhythm abnormality, given that ectopic beats might have occurred during the nonoverlapping 20 seconds of the 1L-ECG recording. Third, we presented cardiologists with the PDF file of the 1L-ECG recording instead of having them assess the recording from a smartphone or tablet screen, which is how physicians will often use the device.[20] Fourth, the KardiaMobile application did not provide an automated assessment of conduction intervals in milliseconds, as is done for most 12L-ECGs. This might have affected sensitivity in the analysis on any conduction abnormality. Fifth, the 95% CIs were relatively wide, owing to sample size and prevalence of the studied outcomes among the cohort. Finally, the present study was not designed to study whether the availability of a smartphone 1L-ECG would change ECG use, diagnosis, or patient management.

Previous Work

The good diagnostic properties that we found for the 1L-ECG device for AF/AFL, when assessed by cardiologists or by the smartphone application algorithm, coincide with a number of prior studies[10,11,19,20,22–27] (see Supplemental Table 3, and Supplemental Table 4, http://www.AnnFamMed.org/content/17/5/403/suppl/DC1/ for an overview of prior studies that validated the KardiaMobile 1L-ECG device for rhythm and/or conduction abnormalities). Notable exceptions are 2 studies by Chan et al[8,28] and 1 study by Desteghe et al[29] that reported sensitivities of 71.4%, 66.7%, and 65.9%, respectively, for the KardiaMobile algorithm to detect AF. The authors provide no clear explanation for the AF-detection algorithm's low sensitivity in their respective studies, which were all performed with selected elderly patients.

Although a number of studies have assessed the presence of ectopic beats on 1L-ECG recordings, none have validated 1L-ECG for ectopy alone or as part of a composite outcome.[8,10,11,22–24,28,29] One study validated 1L-ECG against 12L-ECG for conduction abnormalities. That study, by Haberman et al,[19] found high specificity but sensitivities of 77.3% and 72.4%, respectively, for AV block and BBB. The results for AV block contrast with those from our present study, in which none of the AV blocks were detected using the 1L-ECG device (Figure 2). We note that Haberman et al[19] determined automated conduction intervals for 1L-ECG before assessment by electrophysiologists, whereas in our present study, automated intervals for 1L-ECG were absent.

Our present work adds to the literature by validating 1L-ECG against 12L-ECG in a primary care setting of consecutive patients and by validating 1L-ECG for a broad spectrum of cardiac arrhythmias and conduction disturbances including ectopic beats.

Future Work

Further study is required to evaluate the safety and efficacy of the 1L-ECG device in the hands of primary care physicians instead of cardiologists, particularly for detecting AF/AFL. Moreover, future studies are warranted to determine whether the availability of 1L-ECG changes the use of 12L-ECG, diagnosis, and/or patient management. Data should be obtained to study the net benefit as well as impact on cost-effectiveness of adding the KardiaMobile algorithm's or cardiologists' assessment to that of primary care physicians for arrhythmia detection. Findings from such studies might determine whether and how the KardiaMobile 1L-ECG device can be safely and effectively implemented in clinical practice as well as used in future screening programs for detecting AF in at-risk general populations.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....