Effects of daily hemodialysis on heart rate variability: results from the Frequent Hemodialysis Network (FHN) Daily Trial

Abstract

Background: End-stage renal disease is associated with reduced heart rate variability (HRV), components of which generally are associated with advanced age, diabetes mellitus and left ventricular hypertrophy. We hypothesized that daily in-center hemodialysis (HD) would increase HRV.

Methods: The Frequent Hemodialysis Network (FHN) Daily Trial randomized 245 patients to receive 12 months of six versus three times per week in-center HD. Two hundred and seven patients had baseline Holter recordings. HRV measures were calculated from 24-h Holter electrocardiograms at both baseline and 12 months in 131 patients and included low-frequency power (LF, a measure of sympathetic modulation), high-frequency power (HF, a measure of parasympathetic modulation) and standard deviation (SD) of the R-R interval (SDNN, a measure of beat-to-beat variation).

Results: Baseline to Month 12 change in LF was augmented by 50% [95% confidence interval (95% CI) 6.1-112%, P =0.022] and LF + HF was augmented by 40% (95% CI 3.3-88.4%, P = 0.03) in patients assigned to daily hemodialysis (DHD) compared with conventional HD. Changes in HF and SDNN were similar between the randomized groups. The effects of DHD on LF were attenuated by advanced age and diabetes mellitus (predefined subgroups). Changes in HF (r = -0.20, P = 0.02) and SDNN (r = -0.18, P = 0.04) were inversely associated with changes in left ventricular mass (LVM).

Conclusions: DHD increased the LF component of HRV. Reduction of LVM by DHD was associated with increased vagal modulation of heart rate (HF) and with increased beat-to-beat heart rate variation (SDNN), suggesting an important functional correlate to the structural effects of DHD on the heart in uremia.

Keywords: daily hemodialysis; end-stage renal disease; frequent hemodialysis network; heart rate variability; left ventricular mass.

Figures

FIGURE 1.

Shown are estimated percent differences in geometric mean changes between the frequent and conventional HD groups, with 95% CIs. The P-values refer to tests of interactions of the treatment with age (as a continuous variable) and diabetic status. The P-values were not adjusted for multiple testing.

FIGURE 2.

Shown are scatter plots relating change from baseline to 12 months in left ventricular mass and in left ventricular EDV to the change in the log transformed high-frequency component within the two treatment groups. Local regression curves are displayed.

FIGURE 3.

Shown are scatter plots relating change from baseline to 12 months in left ventricular mass and in left ventricular EDV to the change in the SD of RR intervals within the two treatment groups. Local regression curves are displayed.