PURPOSE
The incidence of hyponatremia is unknown, their causes are multiple. The higher mortality, especially in intensive care units, is currently unexplained. The objective of this article is to evaluate the incidence of hyponatremia, to assess their causes and to identify predictors of prognosis in intensive care units.
METHODS
We included retrospectively all patients admitted at department of medical intensive care unit between January 1996 and February 2001, who presented at the admission, an hyponatremia (< 130 mmol/l). We excluded all patients who presented a hospital acquired hyponatremia, or hyponatremia associated with hyperglycemia > 13 mmol/l or with mannitol administration. Data were analysed by univariate methods, then by multivariate analysis.
RESULTS
During the study period, 300 patients were identified among 2188: the incidence was 13.7% with 95% confidence interval (95% CI) between 9.8 % and 16.7%. Hypovolemic hyponatremia was observed in 25.7%, hypervolemic in 23.7% and normovolemic in 50.6%. In-hospital mortality was 37.7% (95% CI: 31.8% - 42.3%). Nine data were significantly associated with higher mortality in univariate analysis, but only 5 were identified as independant predictors of hospital mortality in multivariate analysis: hyponatremia < 125 mmol/l with a significant relative risk (RR) (RR = 2.10; 95% CI: 1.43-3.08; p < 0.001), Glasgow score < 9 (RR = 2.66; 95% CI: 1.25-5.66; p = 0.01), Glasgow score between 9 and 14 (RR = 1.94; 95% CI: 1.31-2.88; p < 0.001), shock (RR = 1.80; 95% CI: 1.10-3.05; p = 0.02) and blood urea concentration > 10 mmol/l (RR = 1.59; 95% CI : 1.08-2.34; p = 0.02).
CONCLUSION
The frequency of hyponatremia is high; the normovolemic type represented 50%. Mortality is linked, in greater part, to organs dysfunction, but the severity of hyponatremia remained a significant predictor of mortality.