Removing confounders from the relationship between mortality risk and systolic blood pressure at low and moderately increased systolic blood pressure

Abstract

Objective To assess the relationship between mortality risk and systolic blood pressure (SBP) at low and moderately increased SBP (less than the 70th percentile) before and after correcting for the regression–dilution bias and J-curve effects. Design Cohort study. Setting: The First National Health and Nutrition Examination Survey Epidemiologic Follow-Up Study. Participants The 6839 individuals who participated in the 1982–1984 survey for whom there were no missing data (age range 34–87 years). Main outcome measure Cardiovascular disease mortality (n = 678) during a 9-year follow-up. Methods Corrections were made for the regression–dilution bias by using average SBP during the decade before baseline as the mortality predictor, and for J-curve effects by excluding individuals who exhibited high age-stratified mortality rates and a decrease in SBP. Cox's regression was used to analyse the follow-up relationship between mortality risk and SBP. Results The corrected relative cardiovascular disease mortality risk was 1.23 (95% confidence interval (CI), 1.16 to 1.31) for a 10 mmHg increase in SBP. The relationship was monotonically positive starting at the lowest SBP category in the analysis (<115 mmHg), and robustly so above about the 32nd percentile (120 mmHg). The equivalent uncorrected result was 1.08 (95% CI, 1.05 to 1.13), and the relationship was J-shaped and became positive above about the 68th percentile (135 mmHg). Below the 70th percentile of SBP, individuals in the corrected analysis were 89% of all individuals. Conclusion These two corrections transformed the relationship between mortality risk and SBP at low and moderately increased SBP from no association to a robustly positive association starting at 120 mmHg, for the majority of individuals.