The patient was an obese 24-year-old woman with panhypopituitarism who was receiving stable hormone replacement therapy with hydrocortisone, thyroxine, desmopressin, and estrogen. Depression was diagnosed, and her psychiatrist prescribed paroxetine, 10 mg daily. Within 1 week, she developed polyuria and polydipsia despite a stable dose of desmopression. Three weeks after beginning paroxetine, the patient had lost 20 pounds; physical examination showed her to be mildly dehydrated but other-wise well. Her fasting and nonfasting blood glucose levels were 345 and 699 mg/dL, respectively. Her serum electrolyte and bicarbonate levels were normal, and no ketones were present. She had no history of diabetes mellitus, and her blood glucose levels had been normal (79 to 111 mg/dL) on several occasions before paroxetine therapy began.

The patient began receiving glyburide, 5 mg daily, but during the next 2 weeks she discontinued therapy with paroxetine and glyburdie. Her blood glucose level was normal 1 and 4 weeks (94 mg/dL and 77 mg/dL, respectively) after paroxetine and glyburide therapies were discontinued. Her hemoglobin A_1c value was 10.3% (normal, 3.7% to 7.1%) 4 weeks after paroxetine therapy was discontinued and decreased to 6.6% during the next 2 months. Her hormone replacement therapy regimen was unchanged during the entire period, and she was taking no other medications. Lorazepam was prescribed for her psychiatric symptoms, and she is currently euglycemic.

The incidence of hyperglycemia in the premarketing assessment of paroxetine was 0.1% to 1%. However, the degree of hyperglycemia was not reported, nor was paroxetine proven to be the cause in those patients [2]. Severe hyperglycemia has not been reported with any of the selective serotonin reuptake inhibitors. The increases and subsequent decreases in blood glucose and hemoglobin A_1c values noted in my patient, however, correlated closely with the use of paroxetine, implicating it as the most likely cause of her transient diabetes mellitus. The mechanism of this action is unknown but presumably was related to effects of the drug on insulin release or action. Physicians should be aware of this potentially severe side effect.