Diagnostic predictive values for sport-related concussions: a systematic review and diagnostic meta-analysis

OBJECTIVE Sport-related concussions (SRCs) can cause significant neurological symptoms, and approximately 10%–15% of athletes with SRC experience a prolonged recovery. Given the lack of visible injury on brain imaging and their varied presentations, concussions can be difficult to diagnose. A variety of tests and examination methods have been used to elicit a concussion diagnosis; however, the sensitivity and specificity of these tests are variable. The authors performed a systematic review and meta-analysis to evaluate the sensitivity and specificity of standardized tests and visible signs like balance and vision changes in the diagnosis of SRC. METHODS A PRISMA-adherent systematic review of concussion diagnostic examinations was performed using the PubMed, MEDLINE, Scopus, Cochrane, Web of Science, and Google Scholar databases on December 1, 2022. Search terms included “concussion,” “traumatic brain injury,” “diagnosis,” “sensitivity,” and “specificity.” Each method of examination was categorized into larger group-based symptomatologic presentations or standardized tools. The primary outcome was the diagnosis of concussion. Pooled specificity and sensitivity for each method were calculated using a meta-analysis of proportion and were hierarchically ranked using P-scores calculated from a diagnostic frequentist network meta-analysis. RESULTS Thirty full-length articles were identified for inclusion, 13 of which evaluated grouped symptomology examinations (balance and overall clinical presentation) and 17 of which evaluated established formalized tools (ImPACT, King-Devick [K-D] Test, Sport Concussion Assessment Tool [SCAT]). The pooled specificity of the examination methods differed minimally (0.8–0.85), whereas the sensitivity varied to a larger degree (0.5–0.88). In a random effects model, the SCAT had the greatest diagnostic yield (diagnostic OR 31.65, 95% CI 11.06–90.57). Additionally, P-score hierarchical ranking revealed SCAT as having the greatest diagnostic utility (p = 0.9733), followed sequentially by ImPACT, clinical presentation, K-D, and balance. CONCLUSIONS In deciphering which concussion symptom-focused examinations and standardized tools are most accurate in making a concussion diagnosis, the authors found that the SCAT examination has the greatest diagnostic yield, followed by ImPACT, clinical presentation, and K-D, which have comparable value for diagnosis. Given the indirect nature of this analysis, however, further comparative studies are needed to validate the findings.