A preliminary clinical prediction model for upper-extremity injury in collegiate baseball: a single-center retrospective study

Context: Evidence-based practice (EBP) has become a point of emphasis in athletic training education and clinical practice. One approach to EBP is utilizing clinical prediction models (CPM) to assist clinicians in the screening, diagnosis, prognosis, and treatment of injury. A number of isolated risk factors, which can be categorized as: (a) self-report outcome scores, (b) sport performance factors, or (c) functional performance measures (FPMs), have been identified and suggested to be causal with regard to upper-extremity (UE) sports related injury (SRI) in baseball athletes. Objective: to develop a preliminary CPM for UE SRI derived from multiple factors specific to self-reported outcome measures, sport performance risk factors, and FPMs. Design: Retrospective cohort study. Setting: National Collegiate Athletic Association Division I baseball program. Subjects: Thirty-six athletes who completed the preseason pre-participation examination, the Kerlan-Jobe Orthopaedic Clinic overhead athlete shoulder and elbow score (KJOC-SES,) and the Targeted Enhanced Athletic Movement Screen (TEAM-S). Main Outcome Measures: Independent variables for this study included the self-reported outcome measures (KJOC-SES), sport performance risk factors, and FPMs (TEAM-S). The dependent variable for this study was self-reported history of UE SRI. Results: Univariate analyses identified nine predictor variables that differed between injured and non-injured athletes (p ≤ 0.10): KJOC-SES, playing position (pitcher), single leg squat stride foot, single leg squat balance foot, shoulder mobility test throwing arm, shoulder mobility test non-throwing arm, CKCUEST (TEAM-S score), and CKCUEST (absolute score). Forward step-wise logistic regression yielded a resultant two-factor clinical prediction model consisting of playing position and KJOC-SES. The two-factor CPM based on KJOC-SES scores ≤ 86 and playing position (pitcher) yielded diagnostic utility measures as follows: Sensitivity of 83% (95% CI: 0.55 to 95); Specificity of 95% (95% CI: 79 to 99); Positive likelihood ratio of 22.0 (95% CI: 2.88 to 138.5); and negative likelihood ratio of 0.17 (95% CI: 0.04 to 0.61). Conclusion: A preliminary two-factor CPM comprised of KJOC-SES (≤ 86) and playing position (pitcher) retrospectively predicted UE SRI in a cohort of baseball players with strong diagnostic utility