Data Collection
Injury data used for the analysis stem from the Austrian Ski Federation’s ISS, which was already established in 1993. Until 2009, data collection was done by using a paper-and-pencil questionnaire. Since then, an adapted online version of this instrument has been applied and continuously developed further (e.g. integration of an atlas of anatomy). To reduce a potential recall bias, all team members were advised to record injuries immediately after occurrence. At the end of each season records were checked (for more details see Barth et al. [2]). Due to methodological issues, data from the seasons 1993 to 1996 as well as 1999 had to be discarded; thus, data from 23 seasons over a time span of 24 seasons were analysed. A season lasted from 1 May to 30 April of the following year.
Between the seasons 1997 and 2020, 2333 skier seasons (1291 of male athletes; 1042 of female athletes) were recorded for the Austrian Ski Team and can be assigned to its four different squads: Team National (n = 501), Team A (n = 462), Team B (n = 732), and Team C (n = 638).
The criteria used to ascribe athletes to Team National, Team A, and Team B were primarily based on the World Cup Start List (WCSL), i.e. the points systems used to determine the start positions at races. Team National consists of the best racers (e.g. WCSL top 10, medal winner at Olympic Games or World Championships). A WCSL position from 11 to 15 qualifies a racer for Team A. For Team B not only the WCSL position, but also age is considered as qualification criterion. The qualification criteria for Team C are also based on competition results. Although further criteria (e.g. EC ranking, FIS ranking, coach judgment) are used and criteria were slightly adjusted between 1997 and 2020, it seems justified to conclude that Team National, Team A, and Team B racers represent some of the best ski racers in the World and in Europe; Team C comprises some of the best junior racers.
Injury events were classified as severe when the absence from training and/or competition lasted longer than five weeks. Since data on the number of training runs were not available for all seasons, we had to restrict our analysis to severe injury events occurring during FIS competitions, including the official downhill training runs (hereinafter: SIEcomp). Injury data from all seasons from 1997 to 2020 except 1999 were available for the analysis. Of the 371 severe injury events that occurred, 169 were classified as SIEcomp.
Information on every athlete’s exposure to risk in terms of participation in FIS competitions including the official downhill training runs was obtained from the official FIS website [7]. We counted the number of runs (and not the number of competitions), which means that giant slalom as well as slalom competitions counted as two runs if an athlete managed to qualify for and started in the second run. When an athlete did not finish a run or was disqualified, we counted the respective run as a full run. In combined competitions, we assigned the runs as well as injuries that occurred to the respective disciplines. Until the season 2007/08 (hereinafter: 2008), official downhill training runs were not included in the FIS database. Thus, we calculated sex- and team-specific ratios of competitions and official trainings from the season 2008 onwards to obtain multiplicators for the seasons before 2008. The respective multipliers were then used to calculate an athlete’s downhill exposure—i.e. the number of runs in competitions and official training an athlete did.
In sum, Austrian Ski Team athletes participated in 114,531 runs (male athletes: 64,395; female athletes: 50,137; Team National: 18,153; Team A: 17,978; Team B: 39,226; Team C: 39,174). For further analyses, we pooled Team National and Team A as Team World Cup (Team WC). Team B—consisting largely of racers competing in the EC—is hereinafter referred to as Team Europa Cup (Team EC). Team C comprised junior alpine ski racers and is hereinafter called Team National Junior (Team NJ).
Data Analysis
Following the approach of the previous empirical studies (see Tables 1, 2, 3), we expressed injury incidence as the number of SIEcomp per 100 skier seasons or per 1000 runs. We applied Generalised Estimating Equation for Poisson Regression (package gee, version 4.13–20 [8]), which allowed us to take into account the dependence within clusters—i.e. the fact that athletes are commonly members of the Austrian Ski Team for several years and therefore the measurements for one and the same athlete were dependent. For an easier interpretation, estimated coefficients and their respective 95% (Wald) confidence intervals (CIs) were exponentiated.
To describe the change in the SIEcomp incidence throughout the time span of 24 seasons (using data from 23 seasons) as well as to assess the models’ adequacy [9] we made two graphical representations of the predictions, their 95% CIs (package stats, version 4.0.2 [10]), and the observations concerning the SIEcomp incidence (expressed as SIEcomp per 100 skier seasons and 1000 runs) in each season for all Austrian Ski Team racers as well as for the subgroup Team WC.
The value of the exponentiated coefficients corresponded to the sex-specific RR, which was reported as female/male ratio. Incidence and RR were presented with their 95% CIs. A two-tailed p ≤ 0.05 was considered significant. Implementation of all analysis was done with R (version 4.0.2).