Resting heart rate does not refine risk stratification of sudden cardiac death in the general population

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – EU funding. Main funding source(s): Horizon 2020 Background Sudden cardiac death (SCD) is a major public health issue in which most events occur in the general population. Although resting heart rate is associated with SCD, its added value regarding prediction is unknown. Methods for risk stratification and prediction are needed to help prevent future SCD. Purpose The aim of this study was to assess the association between resting heart rate (RHR) and SCD in the general population. Methods Participants from the Copenhagen City Heart Study in sinus rhythm with baseline RHR between 40 and 120 bpm were followed from 1993 to 2016. Deaths were adjudicated, by two cardiologists, using death certificates and autopsy reports. Associations between RHR and SCD was investigated using cause specific cox models. The added value of RHR in predicting SCD was investigated by adding RHR to an already published prediction model (including age, sex, cholesterol, statin use, hypertension, systolic and diastolic blood pressure, current smoking status, diabetes, and body mass index (Bogle 2018)). RHR was evaluated as a continuous covariate with a linear effect. Prediction performance measures were evaluated using bootstrap cross validation. Results The study included 10,099 participants of which 8,986 fulfilled inclusion criteria (median age 60 (25%-75%: 47-70), 57% female). During 24 years of follow-up 4,668 deaths including 656 SCD were observed. Mean RHR was 73 bpm (SD 12.4). Increasing HR was associated with incidence of SCD (HR 1.18 (95% CI: 1.12-1.26) per 10 bmp), however, almost to the same extend with death from other causes (HR 1.14 (1.11-1.17). History of AMI and baseline hypertension were differentially associated with the two outcomes, but not RHR (figure 1). We find no added value when including RHR in a prediction model (AUC at 10 years 80.6 vs 80.4 (figure 2), Brier score 3.1 vs 3.1). Allowing for a more flexible RHR-fit (non-linear effect and interaction) did not improve predictions. All results were consistent in sensitivity analyses (only individuals <70 years at entry, censoring after 10 years of follow up, only the most certain SCD diagnoses). Conclusion In the general population, even if RHR was associated with SCD, it did not refine risk stratification when added to a model previously published.