Ayus and Arieff's letter deserves clarification on three points.

First, can the plasma sodium level decrease to 120 mmol/L if only isotonic saline was infused? It is universally agreed that severe hyponatremia must be avoided immediately after surgery. Anything that adds electrolyte-free water (for example, hypotonic infusions or desalination of isotonic saline) when antidiuretic hormone acts contributes to the degree of hyponatremia and should be minimized. Moreover, the amount given relative to body size is important. A calculation might help here. Almost 15% of total body water must be retained as electrolyte-free water to cause the plasma sodium level to decrease to 120 mmol/L. If isotonic saline were the sole fluid given to a patient with 25 L of total body water (50 kg, 50% water), one must "desalinate" 8 L (the volume infused in our patients was as high as 10 L); much less is needed with a smaller volume of total body water. Finally, a wrong impression is gleaned from an analysis of only mean values: Two patients had a plasma sodium level of 131 mmol/L, but no one had expected this decrease because no electrolyte-free water had been given.

The second point concerns the relation between sex and incidence of fatal outcome from our data. Because the index cases of acute encephalopathy represent highly selected data, we would rather not draw conclusions about the association between poor outcome and sex from such data.

The third point concerns the degree of hyponatremia needed to cause death. With the index cases, one must be cautious when interpreting the degree of hyponatremia needed to induce fatal hyponatremic encephalopathy. We do not know whether the measured plasma sodium level represented the nadir of the hyponatremia in these patients. Other processes could cause a sudden increase in natremia. For example, the plasma sodium level can increase abruptly with a seizure [1], and central diabetes insipidus can develop with brain herniation (time with the true nadir of plasma sodium level causing large water diuresis. If blood sampling is delayed, one might now correlate the syndrome with an artificially higher plasma sodium level. Accordingly, we were not certain of the nadir value for natremia in these patients.

In summary, although we agree that routine use of hypotonic fluids in the postoperative period should be avoided, it is important to be aware that a significant degree of hyponatremia can and does occur in selected patients treated with isotonic saline in this setting.