Effects of within-person variability in spot urinary sodium measurements on the associations with blood pressure and risk of cardiovascular disease in 0.5 Million adults in UK Biobank

Abstract

Abstract Background Randomised control trials have demonstrated direct positive and causal associations of 24-hr measurements of urinary sodium excretion on blood pressure. However, prospective studies, which often used spot (not 24-hr) measurements of urinary sodium, have reported J-shaped associations with higher risks of cardiovascular disease (CVD) at sodium intake <4 g/day. The reasons for the discrepant results are not fully understood, but have prompted some to question the World Health Organisation's recommendations to restrict sodium intake to <2.3g/day. Purpose We examined the effects of within-person variability in spot urinary sodium (UNa) measurements on immediate and delayed associations of UNa with blood pressure at baseline and at resurvey, and with incident cardiovascular disease in the UK Biobank (UKB). Methods Baseline spot urine samples were measured in 502,619 adults at baseline and in 20,346 participants who were resurveyed at 4 years after baseline. Linear regression was used to assess associations of baseline UNa measurements with systolic blood pressure (SBP; mmHg) at baseline and at resurvey. Cox regression was used estimate the associations between baseline measures of UNa with incident CVD events (recorded from linkage with hospital records). All analyses were adjusted for confounders and corrected for regression dilution bias. Results After excluding participants with prevalent diseases, the primary analyses involved 386,060 adults who were followed-up for a median of 7.8 years, during which ∼13,000 CVD events occurred. Estimated mean (SD) urinary sodium excretion was 77.4 mmol/L (SD 44.4, IQR = 42.8–103.7 mmol/L), and mean SBP/DBP were 137.5/82.3 (SD 18.5/10.1) mmHg, respectively. Within-person variability in UNa was high, with a self-correlation of 0.35 at 4 years between measurements. After adjustment for confounders and correction for regression dilution bias, a 100 mmol/L higher UNa was associated with an immediate 3.2 mmHg higher SBP (95% confidence interval [CI]: 2.8–3.6) in cross-sectional analyses (Figure 1). However, the corresponding associations of baseline UNa with SBP at resurvey was completely attenuated (p=0.20). The predicted risk of CVD was 1.06 (95% CI 1.06–1.07, p<0.001) for a 3.2 mmHg higher SBP, but the observed risk for a 100 mmol/L higher UNa was 0.95 (95% CI 0.82–1.10, p=0.47) (Figure 1). Conclusions While spot measurements of UNa were strongly associated with immediate effects on SBP, the magnitude of within-person variability in UNa precluded detection of associations with SBP several years after baseline or with risk of CVD. The extreme within-person variability in spot UNa may explain the discrepant results of the trials and observational studies of sodium and blood pressure. Figure 1. Spot UNa with SBP and CVD in UK Biobank Funding Acknowledgement Type of funding source: Public grant(s) – National budget only. Main funding source(s): Core funding from the Medical Research Council-Population Health Research Unit, British Heart Foundation