Table 3

Association between potential risk factors and game-related injury, and with concussion in Bantam ice hockey players in Alberta and British Columbia (2014–2015 and 2015–2016) in the lower 60% by division of play (with MICE imputation)

Risk factorAll injury adjusted*Concussion adjusted†
Body checking groupIRRs (95% CI)IRRs (95% CI)
Body checking1 (Reference)1 (Reference)
Non-body checking0.44 (0.27 to 0.74)0.60 (0.31 to 1.18) ‡
Year of play
First1 (Reference)1 (Reference)
Second0.96 (0.73 to 1.27)1.06 (0.74 to 1.52)
Player size§
20<40 kg1.15 (0.58 to 2.28)20<70 Kg1 (Reference)
40<60 kg1.02 (0.64 to 0.61)70 to 109 Kg1.85 (0.91 to 3.76)
60<80 kg1 (Reference)
80–109 kg2.64 (1.03 to 6.79)
Previous injury in the last year (including concussion)
No1 (Reference)NA
Yes1.42 (0.98 to 2.06)NA
Previous concussion (any concussion, no date limit)
NoNA1 (Reference)
YesNA1.53 (0.94 to 2.48)
Defence1 (Reference)1 (Reference)
Goalie0.24 (0.07 to 0.77)0.19 (0.03 to 1.43)
Forward1.01 (0.69 to 1.48)1.01 (0.6 to 1.7)
  • *IRRs based on multilevel Poisson regression analysis with two random effects: one at a team level and one at a subject level, offset for exposure hours and adjusted for covariates (body checking, year of play, previous injury/concussion in the last year, player size and position).

  • †IRRs based on multilevel Poisson regressions (two covariates included in each model: body checking group included in all models, plus one other covariate from the table) with one random effect at a team level, owing to fewer injuries, offset for exposure hours.

  • ‡IRR based on multilevel Poisson regression with one random effect at a team level and adjusted for previous history of concussion. IRRs from the other models were similar.

  • §Categories for player size were based on the linear relation between the outcome and player size.

  • IRRs, incidence rate ratios; MICE, multiple imputation by chained equations.