Explaining the inconsistent associations of PR interval with mortality: The role of P-duration contribution to the length of PR interval

Abstract

There is a strong interest in PR interval as a predictor for adverse outcomes. However, inconsistent reports have emerged. The purpose of this study was to test the hypothesis that the significance of PR interval as a predictor depends on the level of contribution of P duration to its length, a contribution that varies across populations. We tested our hypothesis in 7501 participants from the Third National Health and Nutrition Examination Survey (NHANES III). Participants were divided into two subgroups based on the median P-duration contribution to PR interval (P duration/PR interval * 100). The risk of mortality associated with prolonged (>200 ms) and short (<120 ms) PR interval compared with normal PR interval was examined in all participants and each subgroup. P-duration contribution to the length of PR interval ranged from 30% to 90% (median 70%). During median follow-up of 13.8 years, 2541 deaths occurred. In a multivariable adjusted model, short but not prolonged PR interval was associated with mortality (hazard ratio [HR], (95% confidence interval [CI]): 1.54 (1.18, 2.00) and 1.02 (0.90, 1.16), respectively). However, in a stratified analysis by P-duration contribution to PR interval, both short and prolonged PR interval were associated with mortality in participants with high P-duration contribution (HR (95% CI):1.46 (1.10, 1.94) and 2.00 (1.34, 2.99), respectively) but not in participants with low P-duration contribution (HR (95% CI):1.53 (0.68, 3.41) and 0.99 (0.87, 1.13), respectively); interaction P = .008. PR-interval associations with outcomes are dictated by the level of contribution of P duration to its length, a contribution that has a wide range and is expected to vary across populations. These findings could explain the inconsistent reports of PR-interval associations in different studies and call for caution when using PR interval in risk prediction models.