PS1-38 VARIATION IN TREATMENT DECISIONS IN TRAUMATIC BRAIN INJURY: PREDICTORS AND ASSOCIATIONS WITH OUTCOME

Sunday, October 18, 2015
Grand Ballroom EH (Hyatt Regency St. Louis at the Arch)
Poster Board # PS1-38

Maryse C. Cnossen, MSc.1, Suzanne Polinder, PhD.1, Teuntje Andriessen, MSc.2, Joukje Naalt, van der, PhD.3, Iain Haitsma, PhD.1, Janneke Horn, PhD.4, Gaby Franschman, PhD.5, Peter Vos, PhD.2, Ewout W. Steyerberg, PhD6 and Hester F. Lingsma, PhD.7, (1)Erasmus University Medical Center, Rotterdam, Netherlands, (2)Radboud University Nijmegen Medical Center, Nijmegen, Netherlands, (3)University Medical Center Groningen, Groningen, Netherlands, (4)Academic Medical Center Amsterdam, Amsterdam, Netherlands, (5)VU University Medical Center, Amsterdam, Netherlands, (6)Erasmus MC, University Medical Center, Department of Public Health, Rotterdam, Netherlands, (7)Erasmus MC, Rotterdam, Netherlands
Purpose: To examine variation in treatment decisions for traumatic brain injury (TBI) and assess factors influencing treatment decisions and the association between treatment decisions and outcome. 

Method: We used the Prospective Observational Cohort Neurotrauma (POCON) study, with consecutively included moderate and severe TBI patients from five level I trauma centers in the Netherlands between 2008-2009. Differences between centers in patient characteristics, treatment decisions, and six-month outcome (the ordinal Glasgow Outcome Scale Extended (GOSE), range 1 to 8) were assessed. We related patient characteristics and center to treatment decisions as described in current TBI guidelines. Centers were divided into “aggressive” and “nonaggressive” based on the frequency of intracranial pressure monitoring (ICP) placement to explore the association between treatment decisions and outcome. Sensitivity analyses were performed to check whether the effect of treatment approach on outcome was similar with other definitions of aggressiveness. 

Result: We studied 503 patients and found significant variation between the five centers in patient- and treatment characteristics, and outcome. Treatment decisions were modestly predicted by patient characteristics (Area Under the Curve ranged between 0.67-0.85). The addition of center significantly improved the performance of six out of seven models predicting treatment, indicating a tendency for differential treatment of similar patients across centers. Patients treated in centers maintaining an aggressive approach to ICP monitoring had a more favorable outcome than patients treated in nonaggressive centers (OR: 1.66; 95% CI 1.17-2.39). Sensitivity analyses with other definitions of aggressiveness, however, revealed inconsistent findings. 

Conclusion: Considerable variation between centers exists in guideline based treatment decisions for patients with brain injury. An aggressive treatment approach may imply better outcome, although further confirmation is required.