Central diabetes insipidus in pediatric severe traumatic brain injury.

OBJECTIVES: To determine the occurrence rate of central diabetes insipidus in pediatric patients with severe traumatic brain injury and to describe the clinical, injury, biochemical, imaging, and intervention variables associated with mortality. DESIGN: Retrospective chart and imaging review. SETTING: Children's Hospital, level 1 trauma center. PATIENTS: Severely injured (Injury Severity Score ≥ 12) pediatric trauma patients (>1 month and <18 yr) with severe traumatic brain injury (presedation Glasgow Coma Scale ≤ 8 and head Maximum Abbreviated Injury Scale ≥ 4) that developed acute central diabetes insipidus between January 2000 and December 2011. MEASUREMENTS AND MAIN RESULTS: Of 818 severely injured trauma patients, 180 had severe traumatic brain injury with an overall mortality rate of 27.2%. Thirty-two of the severe traumatic brain injury patients developed acute central diabetes insipidus that responded to desamino-8-D-arginine vasopressin and/or vasopressin infusion, providing an occurrence rate of 18%. At the time of central diabetes insipidus diagnosis, median urine output and serum sodium were 6.8 ml/kg/hr (interquartile range = 5-11) and 154 mmol/L (interquartile range = 149-159), respectively. The mortality rate of central diabetes insipidus patients was 87.5%, with 71.4% declared brain dead after central diabetes insipidus diagnosis. Early central diabetes insipidus onset, within the first 2 days of severe traumatic brain injury, was strongly associated with mortality (p < 0.001), as were a lower presedation Glasgow Coma Scale (p = 0.03), a lower motor Glasgow Coma Scale (p = 0.01), an occurrence of fixed pupils (p = 0.04), and a prolonged partial thromboplastin time (p = 0.04). Cerebral edema on the initial computed tomography, obtained in the first 24 hrs after injury, was the only imaging finding associated with death (p = 0.002). Survivors of central diabetes insipidus were more likely to have intracranial pressure monitoring (p = 0.03), have thiopental administered to induce coma (p = 0.04) and have received a decompressive craniectomy for elevated intracranial pressure (p = 0.04). CONCLUSIONS: The incidence of central diabetes insipidus in pediatric patients with severe traumatic brain injury is 18%. Mortality was associated with early central diabetes insipidus onset and cerebral edema on head computed tomography. Central diabetes insipidus nonsurvivors were less likely to have received intracranial pressure monitoring, thiopental coma and decompressive craniectomy.