To study the effect of formulas on the estimation of dietary sodium intake (sodium intake) and its association with mortality, we analyzed the TOHP (Trials of Hypertension Prevention) follow-up data. Sodium intake was assessed by measured 24-hour urinary sodium excretion and estimations from sodium concentration using the Kawasaki, Tanaka, and INTERSALT (International Cooperative Study on Salt, Other Factors, and Blood Pressure) formulas. We used both the average of 3 to 7 urinary measurements during the trial period and the first measurement at the beginning of each trial. Additionally, we kept sodium concentration constant to test whether the formulas were independently associated with mortality. We included 2974 individuals aged 30 to 54 years with prehypertension, not assigned to sodium intervention. During a median 24-year follow-up, 272 deaths occurred. The average measured sodium intake was 3766±1290 mg/d. All estimated values, including those with constant sodium concentration, were systematically biased with overestimation at lower levels and underestimation at higher levels. There was a significant linear association between the average measured sodium intake (ie, gold standard method) and mortality. This relationship was altered by using the estimated sodium intakes. There appeared to be a J- or U-shaped relationship for the average estimated sodium by all formulas. Despite variations in the sodium-mortality relationship among various formulas, a common pattern was that all estimated values including those with constant sodium appeared to be inversely related to mortality at lower levels of sodium intake. These results demonstrate that inaccurate estimates of sodium cannot be used in association studies, particularly as the formulas per se seem to be related to mortality independent of sodium.
Keywords: cohort studies; follow-up studies; humans; mortality; sodium, dietary.