Improving pericardial effusion detection: evaluating sensitivity and specificity of traditional and novel ECG criteria

None.

Historical studies have demonstrated that conventional low voltage criteria on ECG have high specificity but limited sensitivity in detecting pericardial effusion (PE). Indeed, an ECG showing low voltage can be indicative of several conditions, including PE.

This study aimed to introduce new ECG criteria organized into a score called ARENA, to improve ECG diagnostic performance in detecting PE. Furthermore, we compared the sensitivity and specificity of traditional low voltage ECG criteria with the ARENA score using a test cohort.

This retrospective case-control study was conducted at our public University hospital from January 2022 to May 2023. Inclusion criteria were: age ≥18 years old, an available 12 lead ECG and an echocardiogram performed within 5 days from the recorded ECG. Exclusion criteria were: ventricular paced rhythm, heart transplantation, poor subcostal echocardiographic window. In the cohort of patients were selected: patients with ≥1.0 cm of PE on echocardiography and patients without effusion on echocardiography. ECG parameters were evaluated by two cardiologists blinded for the presence or absence of PE. Echocardiographic images were reviewed by an expert echocardiographer.

Optimal ECG criteria with the highest ROC curve area were selected from the test cohort to develop the ARENA score. Patients were assessed using both traditional low voltage ECG criteria and the novel ARENA score.

Statistical analysis showed that with 96 patients in the test cohort, there would be >90% power to detect a significant area under the curve (AUC) of 0.7 (vs. the null hypothesis of AUC of 0.5) to detect a moderate effect size.

We used the AUC analysis to estimate the predicted performance of the proposed criteria. The McNemar test was used to assess the lack of agreement between the ECG criteria and the gold standard (echocardiography), and the results were reported as percentages with their respective 95% confidence intervals (CI). A p-value < 0.05 was considered statistically significant.

The patient population consisted of 97 patients: 50 patients with PE, and 47 patients without it. The ARENA score showed a sensitivity of 52.0% (95% CI: 39.2% to 64.8%), compared to 14.0% (95% CI: 6.2% to 21.8%) of the traditional criteria. Specificity was 94% (95% CI: 84.1% to 100%) for ARENA Score and 100% (95% CI: 96.3% to 100%) for traditional criteria. Accuracy of the ARENA score was 70.0%, compared to 57% of the traditional criteria. The McNemar test indicated a p-value of <0.001 in the performance comparison. The AUC for the ARENA score was 0.781.

The novel ARENA score demonstrated an improved sensitivity while maintaining high specificity in detecting PE. This new score could become a valuable diagnostic tool for detecting PE by ECG in daily practice.

ROC Curve ARENA Score