Cadet characteristics
Starting in August 2014 and ending June 2017, three Service Academies contributed data from 10,604 (n = 2421, 22.83% female) cadets, with 4581 (43.20; n = 934, 8.81% female) at site 1, 5066 (47.78%; n = 1239, 11.68% female) at site 2, and 953 (8.99%; n = 348, 3.28% female) at site 3.
Concussion characteristics
Out of 800 injuries during the study period an injury setting designation was not assigned in 81 cases. There was no association between sex or site with the likelihood of missing injury activity data. School year was significantly associated with missing injury setting data (X2 (2) = 178.07; p < 0.001), the greatest proportion of which occurring during the 2014–2015 academic year, with decreasing missingness each subsequent study year. The larger proportion of missing data in the 2014–2015 academic year was attributed to the rapid start of the study. Also, sport level (varsity, club, intramural) was associated with missing injury setting data (X2 (3) = 41.98; p < 0.001). Club sport level was less likely to have missing data than varsity (X2 (1) = 6.25; p = 0.01) or intramural (X2 (1) = 8.86; p = 0.003). There was no difference between varsity and intramural. Injuries among cadets who skipped the sport level question were most likely to have missing data for injury setting.
A total of 800 (n = 301 female) concussions occurred among 738 (6.96%) cadets. There were 679 (n = 253, 37.26% female), 56 (n = 24, 42.86% female), and 3 (n = 0 female) cadets who experienced one, two, or three concussions respectively during the study period. There were 33 cadets (n = 11, 33.33% female) that had a repeat concussion within the same academic year (66 injuries total injuries). Of the same-season injuries, three occurred within fourteen days of the first injury, and of these, two occurred within seven days of the first injury.
Among all concussions, injuries occurred during the following settings: 24.31% (n = 175) varsity sports, 17.66% (n = 127) club sports, 6.12% (n = 44) intramural sports, 22.81% (n = 164) physical education class, 12.52% (n = 90) academy training, and 16.55% (n = 119) during free time (Fig. 1a). Injury setting across sexes are presented in Fig. 1b.
Univariate asssesment of concussion risk
Univariate concussion risk factors were estimated for any concussion, sport-related, academy training-related, and free-time related concussions. The univariate assessment of concussion risk factors is summarized in Table 1. For any incident concussion, sport level, sex, freshman status, previous concussion, headache, diagnosed migraine, diagnosed ADD/ADHD, diagnosed depression, SCAT symptoms, and BSI somatization score were all significantly associated with any incident concussion (all p’s < 0.01; Additional file 1: Table S2). However, only freshman status had a medium or larger effect size, with over three times the risk of any concussion compared to non-freshman (OR = 3.05; 95% CI: 2.62–3.56) (Table 1).
Sport-related concussions were significantly associated with sport level, sex, freshman status, previous concussion, headaches, and BSI somatization scores (all p’s < 0.05; Additional file 1: Table S5). Contact sport level had the largest effect size (Φ = 1.22), with all other predictors having small effect sizes. Relative to limited-contact sport cadets, contact sport cadets had 2.22 greater risk for any sport-related concussion (OR = 2.22; 95% CI: 1.30–4.03). Risk for sport-related concussion was greater for both contact (OR = 5.60; 95% CI: 3.11–10.92) and limited-contact sport cadets (OR = 2.52; 95% CI: 1.12–5.76) when compared to non-contact sport cadets.
Academy training-related concussions were associated with sport level, sex, freshman status, previous concussion, headache, diagnosed ADD/ADHD, diagnosed depression, SCAT symptoms (number and severity) at baseline, and all BSI symptom (all p < 0.05; Additional file 1: Table S8). Freshman status and medical comorbidities (headache, diagnosed ADD/ADHD, and diagnosed depression) had medium to large effect sizes. Freshman had more than a nine-fold increased risk for incident academy training-related concussion relative to non-freshman (OR = 9.11; 95% CI: 6.76–12.27). Cadets who reported having headaches in the three months before their baseline had 2.70 times increased training-related concussion risk than those who did not have headaches (OR = 2.70; 95% CI: 2.08–3.51). Similarly, cadets with diagnosed ADD/ADHD or depression had increased risk of academy training-related concussions (Additional file 1: Table S8).
Sex, previous concussion, and BSI depression scores were significantly associated with free time concussions (all p’s < 0.05; Additional file 1: Table S11). However, only sex had at least a medium effect size. Females had 2.82 times the increased risk for free time concussions compared to males (OR = 2.82; 95% CI: 1.95–4.07).
Univariate tests evaluated sex as a possible moderator. Thus, concussion predictors were assessed for their association with sex (Additional file 1: Table S1). All variables were significantly associated with sex (all p’s < 0.05). The largest effect sizes were observed with sport level, contact level among varsity cadets, diagnosed depression, and BSI scores. Interactions between sex and these variables were assessed in the mixed models.
Multivariate assessment of concussion risk
Service Academy was a significant random effect for each mixed model. Each subsequent model included site as a random effect. All models are included in the additional file and the final models are interpreted below.
Females, freshman status, and varsity and club cadets were characteristics associated with increased risk for any concussion (Fig. 2, Additional file 1: Table S4). Additionally, medical history including previous concussion, headache in the past 3 months, and diagnosed ADD/ADHD all increased concussion risk. Finally, increasing BSI somatization symptom scores were associated with increased concussion risk (Fig. 2, Additional file 1: Table S4). Each factor was significant after controlling for all other covariates. There were no significant interactions between sport level, diagnosed depression or BSI symptom scores with sex (all p > 0.05).
As observed with concussions of any setting, female sex, varsity and club status, and freshman status were characteristics associated with increased risk for sport-related concussion (Fig. 2, Additional file 1: Table S7). Varsity cadets had the greatest risk for sport-related concussion compared to club (OR = 1.38; 95% CI: 1.03–1.83) and intramural cadets (OR = 5.33; 95% CI: 3.93–7.22). Previous concussion and headache were associated with two and 1.5 times greater risk for sport-related concussion, respectively (Fig. 2, Additional file 1: Table S7). Finally, higher BSI somatization symptom scores were associated with increased sport-related concussion risk (Fig. 2, Additional file 1: Table S7).
Among academy training-related concussions, sex significantly interacted with BSI somatization symptoms. With increasing BSI somatization symptoms, the risk for academy training-related concussion increased, but at a greater rate for males compared to females (Fig. 2, Additional file 1: Table S10). Varsity cadets had a 44% lower risk for academy training-related concussions compared to intramural cadets (OR = 0.56; 95% CI: 0.40–0.78). There were no significant differences between varsity and club, or club and intramural cadets. Freshman had eight times greater risk for an academy training-related concussion compared to non-freshman (OR = 8.17; 95% CI: 5.87–11.37). Previous concussion, headache, and diagnosed ADD/ADHD also significantly increased the risk of an academy-training-related concussion (Fig. 2, Additional file 1: Table S10).
Risk for free time-related concussions was significantly estimated by sex, freshman status, and previous concussion. Females, freshman, and cadets with a prior concussion were all at significantly increased risk for a free time-related concussion (Fig. 2, Additional file 1: Table S13).
Within varsity athlete cadets there was a significant sex and BSI anxiety symptom score interaction. With increasing BSI anxiety symptom scores, concussion risk increased at a greater rate for males than females (Additional file 1: Table S15). Contact sport cadets were at greater risk for any concussion compared with non-contact cadets (OR = 2.43; 95% CI: 1.60–3.68). There was no significant difference in risk between contact and limited-contact or limited-contact and non-contact cadets (both p’s > 0.05). As observed within the entire cadet sample, varsity females, freshman, cadets with a previous concussion, or cadets with headaches in the 3 months prior to baseline were all at increased risk for concussion (Additional file 1: Table S15).
Within varsity-sport injuries, females had 1.71 times the risk for a varsity sport concussion compared to males (Additional file 1: Table S17). Contact and limited-contact sport cadets had a greater risk of a sport-related concussion than non-contact sport cadets. Contact sport cadets had 2.15 times greater risk than limited-contact cadets for sport-related concussion (95% CI: 1.18–3.92). Previous concussion and increasing BSI somatization symptom scores also increased sport-related concussion risk (Additional file 1: Table S17).
Varsity cadets who were female (OR = 1.96; 95% CI: 1.08–3.56), freshman (OR = 15.82; 95% CI: 6.70–37.38), and who had a previous concussion (OR = 1.93; 95% CI: 1.06–3.16) had increased risk for an academy training-related concussion. For free time-related concussions, sex was the only significant predictor. Females had 2.89 (95% CI: 1.35–6.19) times greater risk for a free time-related concussion than males.