Faculty of Health and Environmental Science, National Institute of Stroke and Applied Neuroscience (NISAN), Auckland University of Technology, Auckland, New Zealand.
OBJECTIVES: To identify which aspects of initial clinical assessment for sport-related mild traumatic brain injury (SR-mTBI) predict whether an athlete achieves symptom resolution within 14 days of the injury. RESEARCH DESIGN: Retrospective cohort study using prospectively collected data.
METHODS: Clinical assessment data were collected from 568 patients diagnosed with SR-mTBI at a single medical clinic between February 2017 and December 2018. Demographic data, medical history, SCAT-5 testing, and physician notes were included in the data set. Data were processed and analysed to identify a shortlist of predictor variables to develop a logistic regression model to discriminate between SR-mTBI symptom resolution that occurred in ≤ 14-days or > 14-days. The data were randomly divided into model development and validation subsamples. The top 15 models were analysed to determine the predictor variables to be included in the final logistic regression model. The final model was then applied to the validation subsample.
RESULTS: Half of the athlete participants in this study experienced > 14-day symptom resolution. The final logistic regression model included sex, symptom reporting at initial assessment and presentation with a physiological predominant symptom cluster. The model accounted for 0.90 and 0.85 of the area under the curve and predicted recovery trajectory with 81% and 76% accuracy for the training and validation subsamples, respectively.
CONCLUSIONS: Being female, reporting a higher Positive Symptom Total at initial assessment, and being less likely to have a physiological predominant symptom cluster at initial assessment predicted > 14 versus ≤ 14-day SR-mTBI symptom resolution with a high level of accuracy.