Background: Up to 50% of patients with minor traumatic brain injury (mTBI) develop postconcussion syndrome (PCS). A decision rule to stratify risk for PCS is needed.
Objective: To identify mTBI patients at low and high risk of PCS by comparing the predictive values of variables generated by logistic regression (LR) and recursive partitioning (RP).
Methods: This was a prospective, observational study of 69 mTBI patients aged >16 years presenting to the emergency department of a university teaching hospital. Minor TBI was defined as loss of consciousness <10 minutes or amnesia, Glasgow Coma Scale score (GCS) of 15, no skull fracture on physical examination, nonfocal neurologic exam, and no brain injury on computed tomography if one was done. Clinical/demographic data and the results of a brief neurobehavioral test battery were collected for all patients. The presence of PCS was determined by a validated telephone questionnaire at one month after initial presentation. All variables were subjected to both LR and RP.
Results: Fifty-eight percent had PCS at one month after initial presentation. Low risk: PCS occurred in 9% of men scoring >24 on the Hopkins Verbal Learning A (HVLA) (by LR) and in 9% of those injured in sports scoring >22 on HVLA (RP). High risk: PCS occurred in 89% of women scoring <9 on the Digit Span test (LR) and in 92% of those injured via falls or motor vehicle collision scoring <11.5 on HVLB2 (RP).
Conclusions: Despite the high incidence of PCS, we were able to identify a low-risk subgroup with an average PCS risk of <10% and a high-risk subgroup with a PCS risk of approximately 90%. Combining results from LR and RP expanded the number of patients able to be classified as high/low risk. Prospective validation is necessary.