Head injury is a non-rare cause of mortality and morbidity in young adults. It is difficult and often uncertain to assess the long-term neurological prognosis in these patients. However, this element is of major importance because it allows the clinician to determine the intensity of the therapies used and to adapt the information given to the patient and his relatives.
Numerous studies have been carried out to study the prognosis of these patients, of which two are the most important: CRASH-study and IMPACT-study. However, of all these studies, none of them had the objective of comparing whether the experience of the clinician influenced the accuracy of his prognosis. Similarly, the PREDICT-TBI study conducted at our center, which aimed to compare the clinician's assessment with a prognostic score already in use (IMPACT score) for determining the prognosis of the brain injured person, did not study the influence of the clinician's experience on the accuracy of the prognosis.
Assessment of clinicians' judgment of neurological outcome in 20 patient charts.
20 records will be randomly selected from the N patients included in the PREDICT-TBI study.
These 20 pseudonymities records will be presented, including admission scans and pre-hospital clinical data.
For each clinician included, a questionnaire of the assumed neurological outcome of each patient will be completed and compared with the true neurological outcome, the main variable being the GOSE neurological outcome at 6 months (already collected in the PREDICT-TBI study).
Our hypothesis is that clinicians have a reliable judgment of the neurological outcome of the patient by estimating the probability of error of 30% in senior physicians and 40% in junior physicians. The reliability of this judgment will be measured by assessing the number of errors each clinician will make on the 20 medical records.
The main issue of the study is to better understand the prognosis at the initial phase in order to adapt the nature of the information given to the families in terms of prognosis.
The number of subjects needed is calculated by simulating two-sided tests of a generalized binomial model that compares the "senior" group to the "junior" group on 10,000 samples made up of n \* 2 binomial variables (n = 20, probability of error in seniors p1 = 0.3 (PREDCIT-TBI) and p2 = 0.4 for juniors). With an alpha risk of 5%, a power of 80%, we calculate the n = 18 physicians per group.
Inclusion and consent: After obtaining their consent we will include 18 doctors per group spread over several establishments.
Methodology: We will randomly select 20 files of moderate or severe traumatic brain injury patients treated in the neuro-ICU department of the Pitié-Salpêtrière hospital between March 2019 and December 2020 and included in the PREDICT-TBI study (CPP no. 19.01.21.68040) after having obtained their non-opposition from themselves or from their trusted person and not having been the subject of a limitation of active therapies during the first 24 hours. The patient was also not the victim of a lack of care during the first 24 hours. The prognostic data recognized in the literature for the first 24 hours of treatment are found in the computerized medical records of METAVISION and Carestream. This data, after being anonymized, will be submitted to the clinician for evaluation during an interview of a predefined duration. The prognosis will be assessed by means of a questionnaire submitted to all physicians detailed in the appendix. Then, it will be compared to the effective neurological status 6 months after the trauma assessed by a phone interview with the patient himself or his relatives.
