Dr. Taylor requests therapeutic details of our patients with hyponatremic encephalopathy [1]. Among 16 treated male patients, the change in serum sodium was 19 mmol after 24 hours and 21 mmol after 48 hours. All seven surviving women were treated with hypertonic NaCl within 1 hour of the diagnosis: Changes in serum sodium were 15 mmol after 24 hours and 18 mmol after 48 hours. Only 7 of the 33 women were treated for their hyponatremia, and therapy was delayed for 5 hours. The serum sodium was then increased by 13 mmol after 24 hours. The only male patient who died had absorbed hypotonic fluid through the bladder. None of the 65 patients had central pontine myelinolysis, and it is now clear that this disorder is not a consequence of the therapy for hyponatremia [2].

Dr. Fillman emphasizes that deaths from postoperative hyponatremic encephalopathy appear to be eminently preventable by the simple expedient of administering isotonic fluids postoperatively. We strongly agree and encourage efforts to further disseminate this information.

We are pleased that Dr. Sterns now agrees that symptomatic postoperative hyponatremia is a life-threatening condition and deserves urgent therapy [3]. However, he is incorrect when he challenges our conclusion that hyponatremia in young women carries a high mortality rate. He may misunderstand our research design, a case–control study [4]. Dr. Sterns states that among the consecutive patients with postoperative hyponatremia, none died or became encephalopathic. In fact, six of these patients developed hyponatremic encephalopathy [5], and two died. However, in keeping with proper case–control design, these patients were excluded from the control group because the control population cannot have the disease [4]. Dr. Sterns also inquires, "What explains the extraordinarily high morbidity and mortality rates in the case patients?" By case–control design, the cases must have the disease and were selected on that basis. The statement that "mortality statistics in a patient sample with these selection criteria are uninterpretable" must be evaluated in light of many important results uncovered by these same selection criteria: associations between smoking and lung cancer; toxic shock and tampons; aspirin and the Reye syndrome [4]. Contrary to the view of Dr. Sterns that "such deaths should be vanishingly rare in university teaching hospitals," Dr. Fillman has confirmed that they unfortunately are not. Last, Dr. Sterns suggests that our data do not reveal the mortality rate from postoperative hyponatremia. In this he is correct: A case–control study is not designed to reveal the incidence of a condition [4].