Sodium fluctuation as a parameter in predicting mortality in general hospitalized patients

Abstract

Background: Dysnatremia is the most common electrolyte disorder in hospitalized patients. Sodium fluctuation level may be a better parameter in dysnatremia management. We aimed to examine the association between sodium fluctuation level during hospitalization and mortality and to evaluate its value in predicting poor prognosis among general hospitalized patients.

Methods: Data were collected from patients admitted to Peking Union Medical College Hospital. The generalized estimated equation (GEE) was used to examine the relationship between sodium fluctuation level and mortality. Receiver-operating characteristic (ROC) curve analysis was performed to calculate the optimal cutoff value and the area under the ROC curve (AUC).

Results: Sodium fluctuation level showed a dose-dependent association with increased mortality in general hospitalized patients. After adjusting age, sex, length of hospital stay, and Charlson comorbidity index, the ORs of group G2 to G6 were 5.92 (95% CI 5.16-6.79), 26.45 (95% CI 22.68-30.86), 50.71 (95% CI 41.78-61.55), 104.38 (95% CI 81.57-133.58), and 157.64 (95% CI 112.83-220.24), respectively, p trend <0.001. Both normonatremia and dysnatremia patients on admission had the dose-dependent associations similar to general hospitalized patients. The AUC of sodium fluctuation level was 0.868 (95% CI 0.859-0.877) in general hospitalized patients, with an optimal cutoff point of 7.5 mmol/L, a sensitivity of 76.5% and a specificity of 84.2%.

Conclusion: We determined that sodium fluctuation level had a dose-dependent association with increased mortality in general hospitalized patients. Sodium fluctuation level could be used to develop a single parameter system in predicting mortality in general hospitalized patients with acceptable accuracy, sensitivity, and specificity.

Keywords: dysnatremia; mortality; prediction; single parameter system; sodium fluctuation level.

Figure 1

Flow chart of inclusion and exclusion of study cohort.

Figure 2

The association between serum sodium fluctuation level during hospitalization (cutoff = 6 mmol/L) and mortality. The Odds Ratios (OR) and 95% Confidence Intervals (CI) were generated based on generalized estimated equations. (A,C,E) Showed the OR and 95% CI of sodium fluctuation level during hospitalization (cutoff = 6 mmol/L) after adjustment of age, sex, length of hospital stays, and Charlson Comorbidities Index in general hospitalized patients, normonatremia patients on admission, and dysnatremia patients on admission. (B,D,F) Showed the OR and 95% CI of sodium fluctuation level during hospitalization (cutoff = 6 mmol/L) after adjustment of age, sex, length of hospital stays, myocardial infarction, chronic lung disease, moderate-to-severe liver failure, moderate-to-severe kidney failure, metastatic solid tumor, serum sodium level on admission and average serum sodium level during hospitalization in general hospitalized patients, normonatremia patients on admission, and dysnatremia patients on admission. CI, confidence interval.

Figure 3

The receiver operating character curve of minimum serum sodium level (blue curve), maximum serum sodium level (green curve), and sodium fluctuation level during hospitalization (red curve) in predicting mortality. The cutoffs of minimum serum sodium level, maximum serum sodium level, and sodium fluctuation level during hospitalization were determined based on the optimal cutoffs in general hospitalized patients. (A–C) Showed the Receiver Operating Character Curves in general hospitalized patients, normonatremia patients on admission, and dysnatremia patients on admission, respectively. AUC, area under curve; CI, confidence internal; Sen, sensitivity; Spe, specificity.