Epidemiology, clinical and economic outcomes of admission hyponatremia among hospitalized patients

Abstract

Background: Hyponatremia, the most frequent electrolyte derangement identified among hospitalized patients, is associated with worsened outcomes in patients with pneumonia, heart failure and other disorders.

Research design and methods: We performed a retrospective cohort study of hospitalized patients to quantify the attributable influence of admission hyponatremia on hospital costs and outcomes. Data were derived from a large administrative database with laboratory components, representing 198,281 discharges from 39 US hospitals from January 2004 to December 2005. Hyponatremia was defined as admission serum [Na(+)]<135 mEq/L.

Results: The incidence of hyponatremia at admission was 5.5% (n=10,899). Patients with hyponatremia were older (65.7+/-19.6 vs. 61.5+/-21.8, p<0.001) and had a higher Deyo-Charlson Comorbidity Index score (1.8+/-2.1 vs. 1.3+/-1.8, p<0.001) than those with normal [Na(+)]. A higher proportion of hyponatremic patients required intensive care unit (ICU) (17.3% vs. 10.9%, p<0.001) and mechanical ventilation (MV) (5.0% vs. 2.8%, p<0.001) within 48 hours of hospitalization. Hospital mortality (5.9% vs. 3.0%, p<0.001), mean length of stay (HLOS, 8.6+/-8.0 vs. 7.2+/-8.2 days, p<0.001) and costs ($16,502+/-$28,984 vs. $13,558+/-$24,640, p<0.001) were significantly greater among patients with hyponatremia than those without. After adjusting for confounders, hyponatremia was independently associated with an increased need for ICU (OR 1.64, 95% CI 1.56-1.73) and MV (OR 1.68, 95% CI 1.53-1.84), and higher hospital mortality (OR 1.55, 95% CI 1.42-1.69). Hyponatremia also contributed an increase in HLOS of 1.0 day and total hospital costs of $2,289.

Conclusions: Hyponatremia is common at admission among hospitalized patients and is independently associated with a 55% increase in the risk of death, substantial hospital resource utilization and costs. Potential for bias inherent in the retrospective cohort design is the main limitation of our study. Studies are warranted to explore how prompt normalization of [Na(+)] may impact these outcomes.