3.4 Predictors of pacing-induced cardiomyopathy
On univariate analysis, significant predictors of PICM were lower baseline LVEF (OR 0.85 per 1% decrease, 95% CI 0.76 - 0.95, p-value 0.004), paced QRS duration (OR 1.34 per 10 msec increase, 95% CI 1.22 - 1.48, p-value < 0.001), and RV pacing burden (Table. 2). Patients in the highest tertile of RV pacing (> 98.0% RV pacing) were at highest risk (OR 4.31, 95% CI 2.02 - 9.18, p-value < 0.001) compared to those in the lowest tertile (reference group; < 34.0% RV pacing). Patients in the middle tertile (34.0 - 98.0 % pacing) had intermediate risk (OR 3.67, 95% CI 1.70 - 7.92, p-value < 0.001).
Multivariable analysis was performed by adding the variables that reached statistical significance on univariate analysis and those which were considered potential confounders i.e. age, sex, presence of diabetes, hypertension, beta-blocker, and angiotensin converting enzyme (ACE) inhibitor use, to the model (table.2). After adjustment for other variables, baseline LVEF (adjusted OR 0.87 per 1% increase, 95% CI 0.77 - 0.97, p-value 0.016), paced QRS duration (adjusted OR 1.22 per 10msec increase, 95% CI 1.10 - 1.48, p-value < 0.001), and RV pacing burden remained significant (Table 2).