Two cohorts of consecutive patients referred for diagnosis of unexplained palpitations to the outpatient clinic of the arrhythmia unit of a Spanish tertiary hospital were compared: one was prospective after the implementation of a new diagnostic protocol based on the systematic use of ELRs, and the other was a historical cohort of patients studied for the same reason before the implementation of the new diagnostic protocol.

The study was performed according to good clinical practice guidelines and the Declaration of Helsinki and complies with local ethics committee standards.

Adult patients examined due to palpitations of unknown origin, having experienced at least 2 episodes during the last 12 months and without a conclusive diagnosis after an initial medical evaluation including a baseline electrocardiogram and blood test were enrolled. Patients unable to use the device or complete the symptoms diary appropriately, whether due to mental or physical disability, were excluded.

The conventional diagnostic strategy begins with an initial cardiac evaluation involving medical history, a physical examination, a resting electrocardiogram, an echocardiogram, and 24-hour Holter monitoring. According to the European Heart Rhythm Association1 recommendations and, depending on the clinical evaluation (clinical characteristics of palpitations, structural heart disease), other diagnostic tests may be necessary, such as an ergometry test or an electrophysiological study. If the patient is not diagnosed, outpatient visits continue at the center; examinations may be repeated at the clinician's discretion.

The protocol includes an initial medical history evaluation, a physical examination, a resting electrocardiogram, echocardiography, and ambulatory monitoring for 21 days using a new-generation, high storage capacity, and extended-life lithium battery ELR (SpiderFlash-T, Sorin Group SRL, Saluggia, Italy) that was implanted in the same week as the inclusion visit. The device records 2 electrocardiogram leads on a high-capacity memory card and provides 3 modes of recording: a) manual activation in the event of symptoms; b) automatic activation at predefined intervals; and c) automatic activation in the event of specific heart rhythm disorders such as supraventricular tachycardia or ventricular tachycardia. The nursing team trains the patient on how to use the ELR and emphasizes the importance of proper use of the symptoms diary. If the patient is not diagnosed, further diagnostic examinations may be necessary, and the patient follow-up continues, similarly to the procedure for the control group.

The new diagnostic protocol was put into practice at the study center in October 2012. From then on, patients were included prospectively in the experimental group until July 2014. Patients for the control group were included retrospectively starting in June 2009. All patients were prospectively followed up to the date when a diagnosis was made, or until July 2015 if no conclusive diagnosis was reached.

Palpitations are defined as a symptom reported by the patient consisting of the perception of an abnormal heartbeat, which may be hard, fast, and/or irregular, irrespective of its duration.

A diagnosis is confirmed when a correlation between the patient's symptoms and electrocardiographic recordings can be reliably established. Standard definitions of arrhythmia types are used.11 A diagnosis of nonarrhythmic palpitations is made when the patient clearly reports clinical symptoms but there is no significant change in baseline cardiac rhythm on the electrocardiographic recording.

Time to diagnosis is the period from the enrolment visit to the time when a center physician records a confirmed diagnosis in the patient's medical history.

For the cost analysis, we collected information from the patients’ medical histories in terms of outpatient visits and emergency admissions as well as tests performed until a diagnosis was reached. Cost assignment was done according to service delivery rates, applicable at the investigator's hospital (high technology center), published in Diari Oficial de la Generalitat de Catalunya12 (Table 1). The ELR recording cost was not included in the official rates. Twice the cost of a standard 24-hour Holter was considered appropriate, according to data published in other health systems.13

For global cost calculation in both groups, an additive formula was applied including all costs of the different examinations and visits performed until a diagnosis was made or until the end of the follow-up if no confirmed diagnosis was made beforehand.

The incremental cost-effectiveness indicator was calculated using diagnostic yields and costs of conventional and alternative approaches.

where T represents the technology or intervention under evaluation, and C identifies the reference or control technology.

Continuous variables are described as mean ± standard deviation. For mean comparisons, the Student t test was used, and the Mann-Whitney U statistic was used when necessary. Category variables are expressed as the absolute value of cases and the percentage (%). Comparative proportions analysis was performed using the chi-square test, and the Fisher test was used when necessary. A multivariate logistic regression model was performed to estimate the effect of the treatment on diagnostic yield, accounting for possible confounders in baseline covariates (high blood pressure, frequency of symptoms, structural heart disease). The odds ratio with the 95% confidence interval (95%CI) was calculated. For the analysis of cost per diagnosis, a multivariate Poisson regression model was considered. Time-to-event was included as an offset. The estimated cost per day with the 95%CI was calculated. Statistical analysis was performed using the Statistics Package for the Social Sciences software 15.0 (IBM SPSS) for Windows (IBM Corporation, Armonk, New York, United States).

A total of 149 patients were included, 58 in the control group and 91 in the ELR group (Figure 1). The baseline characteristics and clinical characteristics of the palpitations are shown in Table 2; 86% of the patients in the control group and 95.3% of those in the experimental group (ELR group) experienced at least 1 episode of palpitations per month.

Figure 1.

Flow diagram of diagnostic outcome. ELR, external loop recorder.

(0.08MB).

Of the control group patients, 20.7% were diagnosed vs 86.8% of the patients with a confirmed diagnosis in the experimental group (ELR group) (P < .001). The diagnostic percentages are shown in Table 3. Clinically significant arrhythmias were recorded in 17.2% of the control group patients and in 38.5% of those in the ELR group. The patients’ symptoms had no arrhythmic origin in 3.45% of patients in the control group and 48.5% of those in the ELR group. The diagnostic yield of the new protocol remained similar, despite the frequency of symptoms (Table 4). In the multivariate logistic regression model, the odds ratio for diagnosis in the ELR group vs the control group was 26.4 (95%CI, 8.4-82.9).

For diagnosed patients, the mean time to diagnosis was 204 days in the control group and 42 days in the ELR group.

Table 5 summarizes all costs per group. The cost per patient and cost per diagnosis in the control group were €1072.71 and €5184.75, respectively, decreasing to €325.66 and €375.13, respectively, for the ELR group. From the Poisson regression model, the cost per diagnosis was €49.6/d (95%CI, €21.14-116.20/d) in the control group and €3.3/d (95%CI, €1.60-6.93 /d) in the ELR group (P < .001).

The cost-effectiveness study revealed that the systematic use of ELR resulted in a cost reduction of €11.30 for each percentage point of increase in diagnostic yield, which is represented in the cost-effectiveness plane shown in Figure 2.

Figure 2.

Incremental cost-effectiveness plane. When the results are located in quadrant IV, the new intervention should be rejected because it is more expensive and less effective (intervention dominated). Conversely, when the results are in quadrant II, the new intervention is shown to be more effective and less expensive than the alternative. p.p., percentage point.

(0.07MB).

This study has several limitations. First, it was a single center, nonrandomized, open study with a cohort of historical controls that could generate biases. We attempted to correct potential patient selection biases by defining strict inclusion and exclusion criteria and including consecutive patients in both groups. To avoid possible information biases, strict criteria for results were defined. However, due to nonrandomization, some differences were observed between groups, especially in symptom frequency, which could cause a results bias. Nevertheless, after adjustment for confounders, the diagnostic yield remained higher and the cost for diagnosis lower in the ELR group.

Another potential limitation is that the population studied is a selected population, with a low rate of structural heart disease and relatively frequent symptoms, such that the diagnostic yield of the ELR may be lower when used in a less selective group of patients or a group with more infrequent symptoms. It is also known that the diagnostic yield of electrophysiology study is greater in patients with significant structural heart disease.4

Finally, because this is a cohort comparative study and one of the cohorts is a historical cohort, secular trend changes in management of the patients may have influenced the results.