QT variability during initial exposure to sotalol: experience based on a large electronic medical record

Abstract

Aims: A prolonged QT interval is associated with increased risk of Torsades de pointes (TdP) and may be fatal. We sought to investigate the extent to which clinical covariates affect the change in QT interval among 'real-world' patients treated with sotalol and followed in an electronic medical record (EMR) system.

Methods and results: We used clinical alerts in our EMR system to identify all patients in whom a new prescription for sotalol was written (2001-11). Rate-corrected QT (QTc) was calculated by Bazett's formula. Correlates of sotalol-induced change in the QTc interval and sotalol discontinuation were examined using linear and logistic regression, respectively. Overall, 541 sotalol-exposed patients were identified (n = 200 women, 37%). The mean first sotalol dose was 86 ± 39 mg, age 64 ± 13 years, and BMI 30 ± 7 kg/m(2). Atrial fibrillation/flutter was the predominant indication (92.2%). After initial exposure, the change in the QTc interval from baseline was highly variable: ΔQTc after 2 h = 3 ± 42 ms (P = 0.17) and 11 ± 37 ms after ≥48 h (P < 0.001). Multivariable linear regression analysis identified female gender and age, reduced left ventricular ejection fraction, high sotalol dose, hypertrophic cardiomyopathy, and loop diuretic co-administration as correlates of increased ΔQTc at ≥48 h (P < 0.05 for all). Within 3 days of initiation, 12% discontinued sotalol of which 31% were because of exaggerated QTc prolongation. One percent developed TdP.

Conclusion: In this EMR-based cohort, the increase in QTc with sotalol initiation was highly variable, and multiple clinical factors contributed. These data represent an important step in ongoing work to identify real-world patients likely to tolerate long-term therapy and reinforces the utility of EMR-based cohorts as research tools.

Keywords: Arrhythmia; Atrial fibrillation; Beta-blocker; Electronic medical records; Long QT syndrome; Torsades de pointes.

Figures

Figure 1

Flow chart of study population. Legend: Patients refusing consent comprise an estimated

Figure 2

Changes in QTc from baseline after sotalol administration (ΔQTc) at 2 and ≥48 h. Legend: (A) 2 h, n= 382; (B) ≥48 h, n = 511.

Figure 3

Estimated effect of clinical variables on QTc changes following sotalol administration according to multivariable linear regression analyses. Legend: ΔQTc = (≥48 h sotalol QTc−baseline QTc); patients who contributed to the analysis: *n = 337; model controlled for baseline QTc interval; For dichotomous variables N denotes the number of patients with variables of interest and for continuous variables N denotes the number of patients with non-missing values.

Figure 4

Likelihood of sotalol discontinuation within 3 days according to multivariable logistic regression.