This study sought to develop a predictive model for injury in varsity athletes from sports where both upper and lower extremities are used using preseason performance measures. The study accounted for estimated exposure to the risk of injury for male and female basketball, volleyball, and hockey athletes. Overall, the results indicated that female gender is the most predictive factor in determining time to injury in game or practice regardless of preseason performance. Further, volleyball had significantly shorter time to injury than the other sports studied.
Univariate associations suggested that athletes with limited upper extremity endurance as demonstrated by low push up performance had a shorter time to injury. However, when accounting for the relationship between gender and push up performance in the multivariate model, female gender better predicts shorter time to injury. Similarly, Augustsson et al found that males performed significantly more push ups than females and had 44% greater upper body endurance strength . They suggested that females who train upper body endurance may be more likely to avoid injury . In this study, pushups data alone was not sufficient to predict time to injury. However, our results provide some support to investigate the belief that strength and conditioning in athletes may be a good prevention strategy not only for occurrence but also for time to injury.
Vertical jump was the only other pre-season performance measure showing potential to predict time to injury. Athletes who had higher vertical jump scores showed a trend toward being injured earlier. This finding may appear counterintuitive given that greater muscular or aerobic performance scores are typically desirable in sports. However, Martel et al claim that plyometric training of high intensity and high impact to improve vertical jump height increases the possibility for muscle damage and injury . Additionally, Hewett et al identified female athletes with a higher vertical jump performance scores as having higher risk landing profiles and that high jumping females were more likely to incur injuries, especially at the knee . In volleyball and basketball high vertical jumps are desirable, however this research and those previously mentioned [22–24] suggest that this greater performance attribute may be associated with increased risk of injury. Additionally, athletes with the greatest vertical jump height are likely to play more games and thus have greater exposure to injury in competitive situations. Therefore the combination of increased performance and exposure may put athletes in starting positions at higher risk of injury compared to other players.
Of the injuries collected over the 2008/2009 varsity season, injuries occurred in a greater percentage of games (11.8%) than practices (6.9%). This finding is consistent with data from the NCAA where athletes were 3.5 times more likely to sustain an injury during a game than practice . Both our study and the NCAA study found that injuries occurring in games were greatest for athletes involved in contact sports, compared to sports traditionally considered “non-contact” such as volleyball . Contacts are more frequent in game situations .
In the present study, lower extremity injury was more prevalent than upper extremity injury in hockey, volleyball, and basketball which is consistent with previous findings [40–43]. The most common types of injuries in the present study across all sports examined were muscle strain (23%), ligament sprain (20%), and tendonitis (12%). These findings are consistent with other studies reporting muscle sprain/strain as the most prevalent injury [41–43]. However, Hootman et al report ligament sprain (14.8%) as the highest injury prevalence . This discrepancy may be due to the difference in injury definition used as Hootman et al  only accounted for injuries that resulted in limitation of participation in competition or practice, while we defined injury as one that could have resulted in limited participation, or in the athlete seeking medical attention. In our study, we may have captured more injuries as some participants may have incurred an injury and sought medical treatment but did not restrict themselves from practice.
Gender and sport differences
Our study found that females were more likely to be injured and had a shorter mean time to injury than males. These results are consistent to the majority of studies found in the literature [13, 39, 44–46]. For example Murphy et al found a discrepancy between male and female injury incidence among diverse populations . They found females to be at greater risk than males when identifying risk factors for lower extremity injury . On the other hand others have found that gender was not an important factor in sport injuries for athletes involved in volleyball, basketball, soccer, wrestling and running . One reason for this difference may be because men are less likely to report injuries . The increased likelihood to report injuries by females may be a cultural bias that may be considered in future studies relying on self-reporting of injuries. For our results, this bias if true likely influenced only injury prevalence reporting and not time to injury in this study.
To our knowledge, no previous studies examined gender differences with respect to mean time to injury thus our finding that female’s time to injury was shorter than males is novel. We are confident in the finding that mean time to injury is shorter in females because although our exposure estimate may not be exact the error in estimating exposure in our study should be the same for male and females. Others have identified other female risk factors for increased risk of injury.
Our findings demonstrate that both men’s and women’s volleyball had the highest rate of injury and the shortest time to injury. Compared to the longitudinal NCAA data our findings are less than the NCAA women’ volleyball data which reported 4 injuries per 1000 athlete exposures. The differences might exist due sample size and further longitudinal data collection with Canadian varsity athletes might elucidate any real differences compared to the NCAA data. However, it is noteworthy that we cannot technically compare exposure differences because the NCAA exposure data is a frequency measure defined as 1 athlete participating in 1 practice , whereas our exposure is a calculated estimate of exposure in minutes. It has been indicated that a limitation of the NCAA data is the sensitivity of the exposure estimate whereby, “future authors should use a finer level of exposure measurement, such as player-minutes”  In this case our estimate of exposure in minutes allows for time to injury to be investigated, with the finding that volleyball has significantly shorter time to injury than basketball or hockey.
Limitations and strengths
Accounting for and documenting practice and game exposure when predicting injury is a novel aspect of this study. The collection of the measurement of the potential predictors of injury was collected before the exposure. However, injury information was collected retrospectively with a risk of recall bias. For feasibility reason, we based the injury data survey on the NCAA Injury Surveillance System modified so that an athlete could complete the survey retrospectively. A member of the research team was present to answer questions or clarify items when the survey was administered. Furthermore, practice and competition schedules were used to help athletes remember injury dates and to reduce any recall bias that may have occurred. Others have shown that retrospective tracking of injuries is sufficiently accurate in athletes . Our injury survey also ensured that we captured all injuries including those that may not have been reported to athletic therapy staff in a prospective injury tracking system because we were asked the athletes themselves.
We combined upper and lower extremity injuries when examining the relationship between preseason performance measures and injury occurrence. This may have reduced the predictive power of the preseason performance measures used in this study because some preseason measures may only be predictive of specific injuries. However, the intent was to assess the ability of the preseason measures to identify athletes at risk of any injuries in sports involving both upper and lower extremity activities.
Missing data may have influenced the results. Due to a scheduling conflict men’s hockey did not complete the Modified Illinois Agility test. Other missing fitness measures (<3 per measure) were due to either scheduling conflict or exclusion of invalid result due to instructions not being followed.
Future studies should continue to utilize a comprehensive approach to document time exposed to injury. Larger sample sizes would help confirm if preseason measures are significant predictors of a specific injury in multivariate models when still accounting for gender and sport. Alternatively, because we found differences between genders and between sports, we recommend focussing on only one gender and one sport when planning studies on predictions of shorter time to injuries. Males and females practicing different sports may have different backgrounds in terms of training, and previous injuries. Further, different sports include different activities. Poor scores on fitness measures may predict shorter time to injuries when exposed to risky activities more prevalent in a given sport than another. Nevertheless, our results show that the preseason measures selected in the present study were not strong predictors of short time to injuries over and above gender and sport in sports that were selected because they involved both upper and lower extremity use. Our results provide novel evidence that gender and sports are key predictors of time to injury in out selected varsity sports. To our knowledge this information was lacking from the literature as analyses predicting time to injury are rare in the varsity sport literature.
Likely this means that a different predictor set will be required for different sport even though for simplicity varsity programs may have benefited from using a common dataset for all sports and genders.
In the future, it may be beneficial to examine the predictive ability of more performance measures, however, to ensure the feasibility of preseason screening the added measures should have the ability to be administered rapidly, and require little equipment. Identifying more performance measures with the ability to predict injury is of utmost value for varsity athletics. In the future, collecting injury information and documenting exposure in minutes prospectively should be considered. Future research should consider the impact of other conditions that may influence participation in practice or games when estimating exposure. Also, the modified varsity sport injury survey could be compared to prospective injury recording in order to confirm its validity.