Injury History and Perceived Knee Function as Risk Factors for Knee Injury in Youth Team-Sports Athletes

Background: The identification of risk factors for sports injuries is essential before injury prevention strategies can be planned. Hypothesis: Previous acute knee injury and lower perceived knee function measured by Knee injury and Osteoarthritis Outcome Score (KOOS) will increase the risk of acute knee injury in youth team-sports athletes. Study Design: Prospective cohort study. Level of Evidence: Level 3. Methods: At baseline, youth (≤21 years old) male and female basketball and floorball athletes completed a questionnaire on previous acute knee injuries and perceived knee function (KOOS). A total of 211 male and 183 female athletes were followed for an acute knee injury up to 3 years. Unadjusted and adjusted Cox regression models were used in risk factor analyses. Results: In male athletes, previous acute knee injury and lower KOOS Pain, Activities of Daily Living, Sport and Recreation, and knee-related Quality of Life subscale scores increased the risk of acute knee injury in the unadjusted analyses. Adjusted analyses for male injuries were not performed because of low number of acute knee injuries (n = 18). In female athletes, previous acute knee injury increased the risk of acute knee injury when adjusted for athletes’ age and body mass index (hazard ratio, 2.6 [95% CI, 1.3-5.2]). In female athletes, none of the KOOS subscale scores were associated with the increased risk of acute knee injury in the adjusted analyses. Conclusion: Previous acute knee injury was associated with the risk of new acute knee injury in youth male and female athletes. In youth male athletes, additionally, lower perceived knee function in 4 out of 5 KOOS subscale scores were associated with the increased risk of new acute knee injury. Clinical Relevance: The treatment and rehabilitation of the present acute knee injury and secondary prevention of reinjury should be emphasized in youth team-sports athletes.

B asketball and floorball are both fast-paced indoor teamsports with the similar incidence of acute knee injury in youth athletes. 28,29 These injuries are more common in youth female than in male athletes. 6,24,33,39 The majority of acute knee injuries occur in noncontact situations and are often severe causing a long absence from sports. 2,11,29 Furthermore, severe knee injury is a common cause of early osteoarthritis. 34 To prevent sports injuries, knowledge of injury risk factors is essential. 40 After identifying injury risk factors, an attempt to ameliorate the effects of these risk factors can be done by introducing appropriate injury prevention strategies. 23 For example, the use of proprioceptive balance board program have been shown to reduce the risk of new ankle sprains in volleyball athletes with a history of previous ankle sprains. 41 1065443S PHXXX10.1177/19417381211065443Hietamo et alSPORTS HEALTH

research-article2022
Injury History and Perceived Knee Function as Risk Factors for Knee Injury in Youth Team-Sports Athletes Sports injuries are thought to be caused by complex interactions of many risk factors. 23 However, the measurement of non-modifiable risk factors such as anatomy 37 or modifiable risk factors such as muscle strength 18 and sport-specific skills 21 requires health professionals and clinical testing equipment, which are rarely available especially for nonprofessional and youth athletes. Therefore, only less sophisticated injury screening tools might be available for these athletes. If questionnaires could identify athletes who are at increased injury risk, they would be simple, feasible, and time-saving instruments.
Previous knee injury and lower scores in one or more Knee injury and Osteoarthritis Outcome Score (KOOS) 35 subscales have been found to increase the risk of any type of knee injury in youth soccer athletes. 8,19,38 However, in the studies of Steffen et al 38 and Kucera et al, 19 athletes' exposure time in trainings and matches was collected on a team level. Exposure time is recommended to be collected on an individual level in studies investigating relationship between injuries and individual risk factors. 16 Clausen et al 8 registered individual exposure times but included also overuse knee injuries in the analyses. Considering the fact that time-loss based injury definition substantially tends to underestimate the number of overuse injuries 4 it is possible that the total number of new knee injuries have been underestimated in relation to the previous knee injuries. According to Clarsen et al, 7 time-loss based injury definition is better applicable for registering acute injuries. In addition, to our knowledge, there are no previous studies investigating the association between KOOS subscale scores and knee injury risk in youth male athletes.
The purpose of this 3-year prospective study was thus to investigate previous acute knee injury and perceived knee function as potential risk factors for an acute knee injury in youth team-sport athletes. We hypothesized that previous acute knee injury and lower perceived knee function will increase the risk of new acute knee injury.

Study Design and Participants
This study is part of the Predictors of Lower Extremity Injuries in Team Sports study. 31 The study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee (ETL-code R10169). The participants signed a written informed consent before entering the study (including parental consent for participants younger than 18 years).
Junior-aged (≤21 years) basketball and floorball athletes were recruited from 9 basketball and 9 floorball teams from 6 sports clubs from Tampere city district. All athletes played at the 2 highest junior league levels. The following were the inclusion criteria: 21 years of age or younger and official team member. Altogether 214 male (102 basketball and 112 floorball) and 189 female (107 basketball and 82 floorball) athletes entered the study during the preseason (April-May) in 2011, 2012, or 2013. Each athlete completed a baseline questionnaire, including questions about age, sex, playing experience, playing level, knee injury history, and knee function (KOOS). Standing height (cm) and weight (kg) were recorded. After baseline, prospective injury registration continued until the end of April 2014. The complete data were obtained from a total of 211 male and 184 female athletes. One female athlete did not participate in the follow-up, leading to a total of 211 male and 183 female athletes in the final analysis ( Figure 1).

Previous Acute Knee Injury and KOOS Questionnaire
A previous acute knee injury was recorded if an athlete had ever sustained 1 or more sport-related knee injuries resulting in a specific identifiable event leading to absence from trainings and matches for at least the subsequent 24 hours. Acute knee injuries during the previous 12 months as well as previous anterior cruciate ligament (ACL) injuries were also recorded. The KOOS is a self-administered knee-specific questionnaire comprising of 5 subscales: Pain, Symptoms, Activities of Daily Living (ADL), Sport and Recreation (Sport/Rec), and kneerelated Quality of Life (QOL). Each item is scored from 0 (no problems) to 4 (extreme problems) using a Likert-type scale. A normalized score from 0 to 100 is then calculated for each subscale where 0 indicates extreme problems and 100 no problems. 35 KOOS subscale scores were recorded for both knees separately. 13,38 Missing data were handled according to the recommendations from the KOOS Users Guide. 1 The Finnish-translated KOOS version has demonstrated good validity according to Spearman correlation coefficients 0.48 to 0.81 between KOOS subscales and subscales from other selfadministered knee-specific outcome measures and good to  excellent test-retest reliability with intraclass correlation coeffients 0.73 to 0.86 for all KOOS subscales. 25

Injury and Exposure Registration
During a follow-up period (May 2011-April 2014), all acute knee injuries were registered. Two study physicians contacted the teams once a week to check possible new injuries, and after each injury was reported, the injured athlete was interviewed by telephone using the structured questionnaire, including, for example, questions about injury date, injury situation, injured body part, and injury diagnosis given by a physician. Injury definition was adapted from definition by Fuller et al. 16 An injury was recorded if the athlete was unable to fully participate in matches or training for at least the subsequent 24 hours regardless of the injury diagnosis or given medical treatment.
Only time-loss injuries and injuries that occurred in teams' scheduled training sessions or matches were included in this study. The injuries were classified as contact (ie, direct contact or strike to the involved knee) or noncontact (ie, no direct contact to the involved knee). If an injury had not been diagnosed by the physician, the study physician made the diagnosis by telephone without seeing the injured athlete. All ACL, posterior cruciate ligament, and meniscal injuries were verified by magnetic resonance imaging. During the follow-up, the coach of each team recorded athletes' participation in trainings and matches. Athlete attendance in a training session (yes/no), duration of a training session (hours), and attendance in each period of a match (yes/ no) were recorded individually on a team diary. The diaries were returned after each follow-up month and the individual monthly exposure time (hours) was registered for all athletes. If an acute knee injury occurred, the total exposure time from the beginning of the follow-up to the injury date was calculated. Injury incidences were calculated as the number of injuries per 1000 player-hours and reported with 95% CIs: ([Incidence rate -1.96 × SE of incidence rate] × 1000 hours) to ([Incidence rate + 1.96 × SE of incidence rate] × 1000 hours). Recurrent injuries were included in incidence calculations.

Sample Size
According to Bahr and Holme, 5 the sample size needs to be 20 to 50 injuries to detect moderate to strong associations between risk factors and injury risk. Estimates based on previous studies suggest that 0.1 to 0.2 acute knee injuries occur per athlete per year in basketball and floorball. 24,30 Therefore, we estimated that if we recruited 400 athletes during the 3-year study and if the mean follow-up period in athletes would be 1 year, 40 to 80 acute knee injuries will appear among athletes.

Statistical Analysis
Descriptive data are presented as the mean ± SD or the median and interquartile range. An independent-samples t test was used to compare group differences between sports for normally distributed variables and the Mann-Whitney U test for nonnormally distributed variables. Fisher exact test was used to compare group differences for categorical variables. Relative risks (RR) with 95% CIs 10 were calculated to measure the association between previous and new acute knee injury. Unadjusted mean differences for the KOOS subscale scores between previously injured and uninjured knees were analyzed according to the mixed linear models (gamma distribution). To correct potential dependence between right and left knee, an athlete was considered as a cluster with 2 knees.
Because athletes' individual playing and training times were recorded, Cox mixed-effect regression models were calculated for risk factors. 5 The primary outcome was a new acute (contact or noncontact) knee injury and a secondary outcome a new acute noncontact knee injury. The athlete was a unit of analysis in the models analyzing previous acute knee injury. In the models analyzing KOOS subscale scores, the knee was the unit of analysis. 13,38 In all models, sports club was considered as a cluster, 21 because playing and training styles may differ between the clubs. In the models analyzing KOOS subscale scores, the athlete was considered as a cluster also. Unadjusted and adjusted models were made separately for male and female athletes, respectively. 40 The adjustment factors that might mostly influence the risk of injury based on previous studies 9, 26 were chosen and included in the models according to the number of injuries in each model, following the recommendation of at least 10 injuries needed per included variable. 32 The adjustment factors were age and body mass index (BMI) in the models analyzing previous acute knee injury and previous acute knee injury, age, and BMI in the models analyzing KOOS subscale scores.
Cox hazard ratios (HRs) with 95% CIs were calculated for risk factors. P values <0.05 were considered significant. A receiver operating characteristic (ROC) curve analysis was calculated to assess the combined sensitivity and specificity of a risk factor in cases where significant associations between the risk factor and the outcome were found. The combined sensitivity and specificity were defined as "excellent" (0.90-1.00), "good" (0.80-0.89), "fair" (0.70-0.79), "poor" (0.60-0.69), and "fail" (0.50-0.59). 36 Statistical analyses were conducted in Statistical Package for the Social Sciences (Version 20.0.0; IBM Corp), except the regression models, which were conducted in R (Version 3.1.2; R Foundation for Statistical Computing).

Cohort Baseline Characteristics
Complete data were obtained from 211 (99%) male and 183 (97%) female athletes. The median follow-up period was 1.0 (0) and 1.0 (1.0) years in male and female athletes, respectively. As seen in Table 1, significant group differences between basketball and floorball athletes were observed in age and match exposure in both sexes, in playing experience and KOOS Symptoms subscale score in male participants as well as in height, training exposure, and total exposure in female participants.

Previous Injury Characteristics
A total of 53 male and 46 female athletes reported or more previous acute knee injuries. Both knees had been injured in 18 male and in 13 female athletes. A total of 29 male and 23 female athletes reported that the last acute knee injury had occurred during the previous 12 months. Previous ACL injury had occurred in 7 male and in 6 female athletes. As shown in Table 2, all 5 KOOS subscale scores were significantly lower in previously injured knees compared with uninjured knees in both sexes at baseline. The highest mean KOOS subscale scores were observed in ADL and the lowest in Symptoms subscales in both previously injured and uninjured knees at baseline regardless of sex (Table 2).

Injury Characteristics
A total of 18 male and 32 female athletes had a new acute knee injury during the follow-up. Five female athletes had both knees injured (Table 3). In addition, 2 male and 3 female athletes had 1 reinjury to the same knee. In total, 50% of acute knee injuries in male participants and 32% in female participants were diagnosed by the study physician in telephone.

Risk Factor Analysis: Previous Acute Knee Injury
Both male and female athletes with previous acute knee injury were more likely to sustain a new acute (any type or noncontact) injury compared with previously uninjured athletes (Figures 2 and 3). In male athletes, previous acute knee injury was associated with 5.  Table 4). Because of the low number of acute knee injuries in male athletes (n = 18), adjusted analyses were not performed. In female athletes, previous acute knee injury increased the risk of any type of acute knee injury by 2.6-fold (HR, 2.6 [95% CI, 1.3-5.2) in the adjusted risk factor analyses (Table 4). In female athletes with a previous acute knee injury, the probability of having a new acute knee injury was 30.4% (95% CI, 21.0%-41.9%). Correspondingly, in those who did not have a previous injury, the probability was 13.1% (95% CI, 9.9%-17.2%). However, ROC curve analysis showed an area under the curve of 0.61, indicating "poor" combined sensitivity and specificity of previous acute knee injury. In female athletes, previous acute knee injury also increased the risk of acute noncontact knee injury (HR, 2.4 [95% CI, 1.1-5.0]) in the adjusted analysis (Table  4). In addition, acute knee injury during the previous 12 months was associated with the increased risk of any type of acute knee injury in female athletes (HR, 2.6 [95% CI, 1.1-6.1]) in the adjusted analysis, but no associations were found between previous ACL injury and new acute knee injury (Table 4).

Risk Factor Analysis: KOOS
Significantly lower scores in injured compared with uninjured knees were observed in KOOS Pain, ADL, Sport/Rec, and QOL subscales in male and in Pain, ADL, and QOL subscales in female athletes with any type of acute knee injury (Appendices 1 and 2, available in the online version of this article). In addition, male athletes with acute noncontact knee injury had significantly lower scores in all KOOS subscales in their injured knees compared with uninjured knees (Appendix 1, available online).     In male athletes, lower KOOS Pain, ADL, Sport/Rec, and QOL subscale scores increased the risk of any type of acute knee injury in the unadjusted risk factor analysis. The same risk factors were also associated with the increased risk of acute noncontact knee injury in the unadjusted analyses (Table 4). In female athletes, lower KOOS ADL subscale scores increased the risk of any type of acute knee injury in the unadjusted analysis. The trend was similar in the adjusted analysis, but the observed HR was not statistically significant (Table 4).

discussion
The main finding of this study was that previous acute knee injury increased the risk of sustaining a new acute knee injury by 2.6-fold in youth female and 5.8-fold in youth male athletes. Second, in youth male athletes, the unadjusted risk factor analyses showed that lower KOOS Pain, ADL, Sport/Rec, and QOL subscale scores increased the risk of acute knee injury, but in their female counterparts, none of the KOOS subscale scores was associated with the increased risk of acute knee injury in the adjusted risk factor analyses.

The Risk Factors and Male Injuries
In the present study, previous acute knee injury was associated with the increased the risk of new acute knee injury in youth male athletes. Previous knee injury has also been shown be associated with an increased risk of any type of knee injury in intercollegiate male basketball (RR, 4.23 [95% CI, 2.07-8.67]) 22 and knee sprain in elite male soccer athletes (odds ratio, 4.6 [95% CI, 1.6-13.4]). 3 Hägglund et al 17 found that elite male athletes with an acute knee injury had a 3-fold (HR of 3.1 [95% CI, 1.3-7.6] in the unadjusted analysis) increase in risk of new acute knee injury in the subsequent season. Considering that Hägglund et al registered only acute knee injuries, the injury risk associated with having a previous injury was considerably higher in the present study. This may be due to the recall bias related to the athletes' retrospective reporting of previous acute knee injuries in our study. Kucera et al 19 studied youth (<18 years) soccer athletes and presented a nearly 6-fold increased risk for acute knee injury in previously knee-injured athletes compared with uninjured athletes. Although the study group in the Kucera et al study included both sexes, the study supports our finding of high reinjury risk of acute knee injury in youth male athletes. Consistently, young age has previously been found to be a risk factor for secondary ACL injury, especially in male participants. 43,44 We found that previous ACL injury was also associated with the increased risk of acute knee injury in youth male athletes. This finding is in line with the previous study in Swedish elite soccer athletes. 42 Waldén et al 42 reported a 2.7-fold increase in risk of acute knee injury in athletes with a history of ACL injury compared with athletes without a history of ACL injury.
In the present study, 4 out of 5 KOOS subscale scores were associated with the increased risk of new acute knee injury. In contrast to our study, Engebretsen et al 13 reported that only a lower KOOS Pain subscale score was associated with the future acute knee injury in the unadjusted risk factor analyses, and no associations were found in the adjusted analyses. However, the athletes in their study were considerably older (mean age 24 years) compared with our study. It is likely that adult athletes with longer sports careers are more likely to have sustained previous knee injuries compared with younger athletes. In addition, rehabilitation of previous injuries may also be more successful in experienced adult athletes and because of dropout of severe cases, they may have fewer knee problems and thus higher KOOS subscale scores in previously injured knees. 3,17,22 The Risk Factors and Female Injuries The findings concerning injury history correspond with previous findings from 2 Scandinavian studies in youth female soccer athletes. 8,38 Steffen et al 38 found that history of previous knee injury increased the risk of any type of injury to the same knee by 40%. Clausen et al 8 reported an over 3-fold increase in risk of sustaining any type of knee injury in previously injured compared with uninjured athletes. In contrast to our study, Faude et al 14 studied a cohort of elite female soccer athletes (mean age 22 years) and found that athletes with previous knee sprain did not have a significantly higher risk of the same injury. This may be due to the dropout from sports of those with severe or multiple injuries in younger age.
In contrast to youth male athletes, we found no association between previous ACL injury and the risk of new acute knee injury in youth female athletes. However, this finding should be interpreted with caution, because the mean age of female athletes in the present study was 15 ± 2.0 years and only 6 athletes had previous ACL injury.
In the present study, none of the KOOS subscale scores was associated with the increased risk of acute knee injury in youth female athletes. Steffen et al 38 found that lower scores in all KOOS subscales-and Clausen et al 8 in Sport/Rec, QOL, and ADL subscales-were associated with the increased risk of any type of knee injury. However, in contrast to our study, previous injury was not treated as a confounder in the analyses in these studies.
As presented in our study, strong associations exist between previous acute knee injury and all KOOS subscale scores in both sexes. Surprisingly, only a lower KOOS ADL subscale score was associated with the increased risk of acute knee injury in youth female athletes in the unadjusted risk factor analysis in the present study. Clausen et al reported also the high risk for future knee injury in athletes with a lower KOOS ADL subscale score (RR, 5.38 [95% CI, 1.73-7.46] for score <80 compared with score ≥80). However, we found a mean difference of only 1.5 in the ADL subscale scores between injured and uninjured knees in youth female athletes, limiting the clinical relevance of the finding. In addition, the mean ADL subscale scores in the present study were more than 95 in both previously injured and uninjured knees in both sexes, indicating that difficulties with ADL function are typically mild in adolescence. 15 We found that in female athletes, as in male athletes, mean scores in the KOOS Symptoms subscale were considerably lower compared with other KOOS subscales in both previously injured and uninjured knees. Steffen et al 38 reported the mean KOOS Symptoms subscale scores of 58.6 ± 12.9 and 67.1 ± 10.1 in previously injured and uninjured knees, respectively. The reasons for low scores in the KOOS Symptoms subscale compared with other KOOS subscales in youth athletes are unclear. In contrast to youth male athletes, apart from the ADL subscale, we found no associations between lower KOOS subscale scores and the risk of new acute knee injury in youth female athletes even in the unadjusted analyses. However, it should be noted that we investigated only acute knee injuries. The incidence of overuse knee injuries in youth basketball and floorball athletes is remarkable high, especially in female athletes, 20 and these injuries might have also affected the KOOS scores.

Clinical Implications
The findings of the current study, especially with regard to athletes with previous acute knee injuries, suggest the need for neuromuscular training to prevent reinjury. Neuromuscular injury prevention programs have been shown to be effective in the prevention of acute knee injuries in youth athletes 12,27 and are recommended to be included in regular training. Considering the findings of the present study, especially athletes with previous acute knee injuries, should be motivated for neuromuscular training to prevent reinjuries. This study also gives evidence that KOOS Pain, ADL, Sport/Rec, and QOL subscale scores may be useful when identifying youth male athletes with increased risk of acute knee injury, but further studies are needed.
Regardless of significant association between previous and new acute knee injury in youth female athletes in the adjusted analyses, the combined sensitivity and specificity of previous acute knee injury in predicting future acute knee injury was classified as "poor" according to ROC curve analysis. We found that history of previous acute knee injury can correctly classify only 61% of injured and uninjured female athletes. Therefore, in clinical practice, previous acute knee injury alone cannot be recommended to predict the future acute knee injury in youth female athletes. However, while we cannot predict future injury, it may be still useful to determine youth athletes who might be at increased risk for future acute knee injury by recording previous injuries and focused neuromuscular training, especially for athletes with previous acute knee injuries.

Study Strengths and Limitations
This study had several strengths including the relatively long follow-up, large sample size, and low dropout rate. Also, prospectively collected injury and exposure data enabled the use of Cox regression models. In addition, risk factors presented in our study are easily available, simple, and fast to use in clinical practice.
This study also had limitations. Originally, KOOS has been developed for studies concerning treatment of knee injuries and for long-term follow-up of patients with osteoarthritis. 35 Therefore, it may have limited value for the assessment of knee function in youth and mainly knee-healthy athletes. Also, self-reported injury history relies on athlete recall, and thus some previous injuries may have gone unreported. However, we believe that using only previous acute knee injury in the analyses minimized the risk for recall bias. In addition, despite the 3-year follow-up, the incidence of acute knee injury especially in male athletes was relatively low, thus limiting the statistical power of the study. Thus, small group differences and risk estimates might not have been detected 5 and adjusted risk factor analyses for male injuries were not able to be performed.

conclusion
Our prospective study showed that previous acute knee injury increased the risk of new acute knee injury in youth female and male athletes. In youth male athletes, lower perceived knee function measured by KOOS Pain, ADL, Sport/Rec, and QOL subscale scores increased the risk of new acute knee injury. In female patients, none of the KOOS subscale scores were associated with the increased risk of acute knee injury.