Electrocardiographic Measures and Prediction of Cardiovascular and Noncardiovascular Death in CKD

Abstract

Limited studies have assessed the resting 12-lead electrocardiogram (ECG) as a screening test in intermediate risk populations. We evaluated whether a panel of common ECG parameters are independent predictors of mortality risk in a prospective cohort of participants with CKD. The Chronic Renal Insufficiency Cohort (CRIC) study enrolled 3939 participants with eGFR<70 ml/min per 1.73 m(2) from June 2003 to September 2008. Over a median follow-up of 7.5 years, 750 participants died. After adjudicating the initial 497 deaths, we identified 256 cardiovascular and 241 noncardiovascular deaths. ECG metrics were independent risk markers for cardiovascular death (hazard ratio, 95% confidence interval): PR interval ≥200 ms (1.62, 1.19-2.19); QRS interval 100-119 ms (1.64, 1.20-2.25) and ≥120 ms (1.75, 1.17-2.62); corrected QT (QTc) interval ≥450 ms in men or ≥460 ms in women (1.72, 1.19-2.49); and heart rate 60-90 beats per minute (1.21, 0.89-1.63) and ≥90 beats per minute (2.35, 1.03-5.33). Most ECG measures were stronger markers of risk for cardiovascular death than for all-cause mortality or noncardiovascular death. Adding these intervals to a comprehensive model of cardiorenal risk factors increased the C-statistic for cardiovascular death from 0.77 to 0.81 (P<0.001). Furthermore, adding ECG metrics to the model adjusted for standard risk factors resulted in a net reclassification of 12.1% (95% confidence interval 8.1%-16.0%). These data suggest common ECG metrics are independent risk factors for cardiovascular death and enhance the ability to predict death events in a population with CKD.

Keywords: chronic kidney disease; electrophysiology; epidemiology and outcomes; mortality risk; risk factors.

Copyright © 2016 by the American Society of Nephrology.

Figures

Figure 1.

Mortality rates across ECG intervals. ECG measures and (A) rates of all-cause mortality; (B) rates of cardiovascular death; (C) rates of noncardiovascular death.

Figure 2.

C-statistics for cardiovascular death. The plot depicts the C-statistics for cardiovascular death across the entire study population followed by subgroups of race and sex. The baseline model includes age, sex, race, history of cardiovascular disease, smoking, diabetes, systolic and diastolic blood pressure, BMI, total cholesterol, HDL, eGFR, and proteinuria (24 hour collection). The baseline plus ECG model includes all the variables above plus the PR, QRS, QTc, and RR intervals, and Cornell voltage.