Automated Quantification of QT-Intervals by an Algorithm: A Validation Study in Patients with Chronic Obstructive...

To assess the diagnostic accuracy of a purpose-designed QTc-scoring algorithm versus the established hand-scoring in patients with chronic obstructive pulmonary disease (COPD) undergoing sleep studies.

We collected 62 overnight electrocardiogram (ECG) recordings in 28 COPD patients. QT-intervals corrected for heart rate (QTc, Bazett) were averaged over 1-min periods and quantified, both by the algorithm and by cursor-assisted hand-scoring. Hand-scoring was done blinded to the algorithm-derived results. Bland-Altman statistics and confusion matrixes for three thresholds (460, 480, and 500ms) were calculated.

A total of 32944 1-min periods and corresponding mean QTc-intervals were analysed manually and by computer. Mean difference between manual and algorithm-based QTc-intervals was -1ms, with limits of agreement of -18 to 16ms. Overall, 2587 (8%), 357 (1%), and 0 QTc-intervals exceeding the threshold 460, 480, and 500ms, respectively, were identified by hand-scoring. Of these, 2516, 357, and 0 were consistently identified by the algorithm. This resulted in a diagnostic classification accuracy of 0.98 (95% CI 0.98/0.98), 1.00 (1.00/1.00), and 1.00 (1.00/1.00) for 460, 480, and 500ms, respectively. Sensitivity was 0.97, 1.00, and NA for 460, 480, and 500ms, respectively. Specificity was 0.98, 1.00, and 1.00 for 460, 480, and 500ms, respectively.

Overall, 8% of nocturnal 1-min periods showed clinically relevant QTc prolongations in patients with stable COPD. The automated QTc-algorithm accurately identified clinically relevant QTc-prolongations with a very high sensitivity and specificity. Using this tool, hospital sleep laboratories may identify asymptomatic patients with QTc-prolongations at risk for malignant arrhythmia, allowing them to consult a cardiologist before an eventual cardiac event.