Evidence for a curvilinear relation between blood pressure and urinary sodium in men

With the use of a linear model, the relation between urinary sodium and blood pressure has been reported to be positive, non-significant, or negative. The hypothesis that this relationship is more complex than linear was investigated in two different study populations, which were independently recruited and examined by different observers. In 1,071 men randomly selected from the general population and in an unrelated sample of 1,209 military men, systolic and diastolic blood pressure were correlated with urinary sodium following a model, which included both the linear and quadratic terms of urinary sodium. In both groups of men, these second order models, adjusted for age and body mass index, provided a better fit (P less than 0.05) than the relationships with only the linear term of urinary sodium. The quadratic models explained from 0.35 to 1.10% of the blood pressure variance. Third order models, which in addition included the cubic term of urinary sodium, did not further improve the correlations between systolic and diastolic blood pressure and urinary sodium in men. In 1,010 women drawn from the general population and in 499 military women, neither the first nor the second order correlations between systolic and diastolic blood pressure and urinary sodium were statistically significant. In conclusion, the present results, reproducible in two different study populations, suggest that a second order model is more appropriate than a simple linear correlation to describe the weak relationship between blood pressure and urinary sodium in men. However, recommendations for the prevention of hypertension must not be changed, until the present findings are confirmed by intervention studies.