Management of central diabetes insipidus in a paediatric intensive care unit
Amir Babiker, Carlo Acerini & Robert Tasker
Background: Central diabetes insipidus (CDI) is rare in infants and children. Up to 30 percent of cases are idiopathic and its clinical presentation is poorly defined. In critically ill paediatric intensive care unit (PICU) patient with CDI, there is increased risk of brain damage and death due to severe hyperosmolarity, hypovolaemic shock, hypernatraemic seizures or complications of treatment.
Aim: To assess the causes, management and outcome of CDI in patients admitted to a tertiary PICU with a large neurosurgery and oncology service.
Method: Retrospective data were collected in PICU patients who required assessment of serum and urine osmolalities over a period of 3 years (20068).
Results: A total of 58 patients, 7/58 (12%) had CDI, 1/58 (1.7%) had nephrogenic DI and 10/58 (17.2%) had incomplete DI or solute diuresis. Brain tumours were the commonest cause of CDI. All patients with CDI received full maintenance fluids of: 0.9% sodium chloride (n=5), 0.45% sodium chloride/5% dextrose (n=1), or oral feeds (n=1). Only 2 patients required replacement of fluid losses which were estimated as urine output plus 10% of body weight. An initial dose of intravenous DDAVP was required in 6 patients. One patient was treated with oral DDAVP and had developed SIADH during treatment. He had a hyponatraemic seizure and was treated with 3% sodium chloride infusion and 50% fluid restriction. All patients were monitored by hourly urine output and 4-hourly blood gases for electrolytes level. Repeated doses of DDAVP were required in 4 patients following breakthrough episodes. We discharged 3 patients home on regular DDAVP, transferred 1 to another hospital and 3 died because of their primary illness.
Conclusion: PICU patients with CDI need prompt fluid management, close monitoring of urine output and electrolytes as well as appropriate DDAVP doses to avoid serious complications of the disorder and management pitfalls.