Background: Serum sodium has been shown to predict mortality in heart and liver failure.
Aim: To determine whether serum sodium independently predicts in-hospital mortality during any emergency medical admission.
Design: An analysis was performed of all emergency medical patients admitted to St James's Hospital (SJH), Dublin between 1 January 2002 and 31 December 2006, using the hospital inpatient enquiry (HIPE) system, linked to the patient administration system and laboratory datasets. Hospital mortality was obtained from a database of 20 deaths occurring during the same period under physicians participating in the 'on call' roster.
Methods: The serum sodium was determined at admission in all cases where it was deemed clinically necessary. Logistic regression was used to calculate crude and 25 adjusted odds ratios (ORs). Factors adjusted for included age, illness severity score (Modified Apache II score), major disease category, ICU stay, year effect, blood transfusion, gender and sepsis.
Results: A total of 14 239 patients (47.5% male) were included in the analysis. Mortality had a U-shaped distribution and was highest in patients whose sodium level was <125 or >140 mmol/l. The unadjusted OR of death within 30 days of admission was 3.36 (95% CI 2.59-4.36) and 4.07 (95% CI 2.95-5.63) with sodium level <125 and >140 mmol/l, respectively. Adjustment for all of the factors above reduced the mortality odds in all hyponatraemia groups but all remained significant predictors of mortality. After adjustment for illness severity score the OR ratio for death in the >140 mmol/l group fell to 1.41 (95% CI 0.97-2.07).
Discussion: The serum sodium is a powerful initial marker of likely mortality in unselected general medical patients. The increased death rate in hyponatraemic patients is independent of other clinical variables, whereas mortality in the hypernatraemic group is primarily a factor of illness severity.