Prediction of raised intracranial pressure complicating severe traumatic brain injury in children: Implications for trial design

Objectives. To describe current patterns of management of raised intracranial pressure (ICP) in traumatic brain injury relevant to clinician buy-in to possible randomized controlled trials of treatments of raised ICP. To examine the feasibility of early identification of children at sufficient risk of developing raised ICP to permit a uniform approach between centers to the initiation of ICP monitoring. This would permit quantification of ICP elevation and enrollment as appropriate to randomized controlled trials of raised ICP interventions. Design: Logistic regression modeling of death before pediatric intensive care unit discharge and decision tree and logistic regression of development of raised ICP through analysis of a prospectively collected, standardized, national data set. Setting: Pediatric intensive care units in the United Kingdom and Eire. Patients: Patients were 501 children < 16 yrs of age primarily admitted to intensive care unit for management of traumatic brain injury in the United Kingdom and Eire between February 2001 and August 2003. Measurements and Main Results. The data analyzed included demographic, acute physiologic, and cranial imaging variables. Death was associated with both raised ICP and the nonmeasurement of ICP. In a subset of 199 patients, an empirically derived decision rule predicted the development of raised ICP at any point during ICU admission with sensitivity of 73% and specificity of 74% (positive predictive value 82% and negative predictive value 63%). Logistic regression modeling performed comparably. The decision rule also predicted raised ICP in 20% of children not undergoing ICP monitoring. Conclusions. Simple models based on early clinical data may predict the development of raised ICP sufficiently well to encourage a consistent approach between centers to initiation of ICP monitoring. We estimate studies designed to detect reductions in ICU mortality will require > 320 children per arm, although this figure may be higher if more conservative assumptions are made.