Can we really predict the change in serum sodium levels? An analysis of currently proposed formulae in hypernatraemic patients.

BACKGROUND: Hypernatraemia is common in intensive care patients and may present an independent risk factor of mortality. Several formulae have been proposed to guide infusion therapy for correction of serum sodium. Unfortunately, these formulae have never been validated comparatively. We assessed the predictive potential of four different formulae (Adrogué-Madias, Barsoum-Levine, Kurtz-Nguyen and a simple formula based on electrolyte-free water clearance) in correction and maintenance of serum sodium in 66 hyper- and normonatraemic ICU patients.

METHODS: With daily measurements of sodium/potassium and fluid/electrolyte balances, a day-to-day prediction of serum sodium levels was calculated using the four formulae. This was compared to the measured changes in serum sodium.

RESULTS: Six hundred and eighty-one patient-days (194 hypernatraemic) in 66 patients were available for calculations. Prediction of serum sodium levels using all four formulae correlated significantly (P < 0.05) with measured changes in serum sodium. Individual variations were extreme, and the mean differences (+/-SD) for predicted versus measured serum sodium were within the range of 3.4-4.5 (+/-4.4-4.7) mmol/l similar for the Adrogué-Madias, Barsoum-Levine and Nguyen-Kurtz formulae. In comparison, our proposed formula underestimated the changes of serum sodium (mean +/- SD -1.5 +/- 5.3). During hypernatraemia, the differences between predicted and measured values were even greater (mean +/- SD 5.0-6.7 +/- 3.9-4.3) using the published formulae compared to our formula (mean +/- SD 0.2 +/- 4.0).

CONCLUSIONS: Currently available formulae to guide infusion therapy in hyper- and normonatraemic states do not accurately predict changes of serum sodium in the individual ICU patient. In clinical practice, infusion therapy should be based on the reasons for hypernatraemia and serial measurements of serum sodium to avoid evolution of derangements.