Research Letter: Surveying Clinician Variability in Neuro-Prognostication for Patients With Severe Traumatic Brain Injury.

Objective: To describe the approach to, accuracy, and variability of neuro-prognostication for patients with severe traumatic brain injury (TBI) among clinicians caring for these patients.

Setting: Data were collected via an anonymous electronic survey.

Participants: The survey was emailed to all clinicians who care for patients with TBI at a Level 1 Trauma Center.

Design: In this observational survey study, respondents were provided vignettes of 3 real patients with TBI and asked to predict 2-year Glasgow Outcome Scale-Extended (GOS-E) scores and 6-month risk of mortality and unfavorable outcome (GOS-E <4). Medical management decisions and approach to goals of care discussions were queried. Results were collected anonymously.

Main measures: Primary outcome measure was prognostic variability. Secondary outcome measures included overall prognostic accuracy and incidence of management recommendations overall and by clinical role.

Results: Of the 55 respondents, 16 (29%) were attending physicians, 19 (35%) residents or fellows, 6 (11%) advanced practice providers, and 14 (25%) nurses. The accuracy of prognostication based on initial patient presentation was 15%. The range of responses for predicted outcome for each patient spanned all GOS-E scores (1-8), and responses for predicted probability of 6-month mortality ranged from 0% to 99% for each patient.

Conclusion: Prognostication accuracy was low, and predictions were highly variable, including all possible neurologic outcomes and mortality ranges. Decreasing the variability in neuro-prognostication for patients with severe TBI is critical.