School-level determinants of incidence of sports-related concussion: Findings from the CARE Consortium

OBJECTIVE: Epidemiologic research on sports-related concussion (SRC) has focused on individual risk factors, with limited research on institutional risk factors and variability in concussion rates. METHODS: This study used data from 53,822 athletes-seasons collected at 30 United States sites (26 civilian institutions and 4 military service academies), from 2014/15 to 2018/19 academic years, by the Concussion Assessment, Research, and Education Consortium. School-level risk factors included competitive division (DI, DII, DIII), school type (military/civilian) and a Sport Risk Index (SRI; Low, Medium, High). For comparability between civilian institutions and military academies, only NCAA athletes and concussions in sports games and practices were included. Random intercepts log-binomial regression was used to estimate Risk Ratios (RRs) and model variability in SRC risk. RESULTS: A total of 2,503 SRCs were observed during the study period, including 829 competition SRCs (33%) and 1,674 practice SRCs (67%). Most variability in SRC risk was at the level of athlete or team (within-school), rather than at the school-level. Specifically, across the three SRC outcomes (all [competition and practice combined], competition-only, and practice-only), within-school variability was 5 to 7 times greater than between-school variability. Three school-level risk factors (Division, School Type, and SRI) accounted for over one-third (36%) of between-school variability. SRI was the strongest school-level predictor of SRC risk (RR = 5.7; 95%CI: 4.2, 7.6 for High vs. Low). SRC risk was higher for Division I compared to Divisions II/III (RR = 1.6; 95%CI: 0.9, 2.9 for DI vs. DIII), and military academies had a moderately elevated risk of SRC (RR = 1.4; 95%CI: 0.7, 2.7). CONCLUSION: A large portion of the apparent variability between schools was attributable to structural factors (sport risk and competitive level), suggesting that there were minimal systemic differences in concussion identification between schools. While most variability is within-school, understanding school-level determinants of concussion risk may still be important in providing the implementation science context for individual-level interventions.