In 1995 Arendt and Dick reviewed injury data over a 5-yr time period using the National Collegiate Athletic Association (NCAA) injury surveillance system (ISS) (1). They reported a statistically significant difference in the rate of anterior cruciate ligament (ACL) injuries between male and female soccer players and male and female basketball players. Potential causative factors hypothesized by the authors to explain this difference were extrinsic variables including: mechanism of injury, years of experience, muscle strength; and intrinsic variables including: joint laxity, Q-angle, and notch size.
In hopes of better identifying potential risk factors Arendt et al. (2) began collecting those variables that could be readily obtained on athletes identified post-ACL injury that reported a noncontact mechanism of injury. ACL injury is defined, following NCAA methodology, as any ACL injury partial or complete. Verification of these injuries can come from a certified athletic trainer or a physician. The variables investigated included lower extremity (hip and knee range of motion) flexibility measurements, hormonal data including menstrual cycle regularity and timing of injury relative to the menstrual cycle, and historical data that included years of participation in the sport in which the injury occurred. The goal was to develop a “profile” of the noncontact injured athlete in hopes of better defining the common factors in those athletes who sustained a noncontact ACL injury (2). The population was limited to noncontact ACL injuries because this was the mechanism of injury most commonly reported and potentially open to intervention. A subsequent consensus conference (8) was held, which included a group of interested orthopedists, primary care physicians, certified athletic trainers, and biomechanists who sorted potential risk factors into anatomic, hormonal, and neuromuscular categories. Despite significant energy by these and numerous other investigators, the risk factors responsible for this gender difference in ACL injury rates remain speculative.
The occurrence of noncontact ACL injuries in basketball and soccer remains small. In order to improve the sensitivity of the study, the injury definition was expanded to include ankle sprains theorizing that any change or pattern in hormonal levels affecting ACL integrity should affect other ligaments as well. Inversion injury to the ankle was chosen as a frequently occurring and readily diagnosable noncontact injury to help test this theory (10). This expanded data set allowed for the determination of the rate of noncontact ankle and noncontact ACL injuries in NCAA basketball and soccer for female athletes. It allowed us to: a) test the earlier recall-based periodicity results demonstrating that the highest number of noncontact ACL injuries occurred around days 7–9 for both “on pill” and “off pill” groups, and b) determine if there is a protective effect from the use of oral contraceptives (3).
All NCAA schools were invited to participate in this IRB-approved project, and signed consent forms were obtained from all participants during three seasons (basketball 2000–2001, basketball 2001–2002, and soccer 2001–2002). The certified athletic trainer at each participating school agreed to provide a roster at the beginning of the season with the hormonal therapy status of each athlete. Any athlete reporting oral contraceptive use had their contraceptive classified as mono or triphasic according to the Physician's Desk Reference (PDR) (13). Those athletes using injectable contraceptives, medications that could not be identified, who changed their contraceptive status over the season, or quit the team during the season were listed as “other.”
A noncontact ACL injury was defined as any injury assessed by the athlete or certified athletic trainer to be the result of a noncontact mechanism of injury and documented by clinical or imaging exam. A noncontact ankle sprain was defined as an inversion injury to the ankle diagnosed by clinical examination and assessed by the athlete or certified athlete trainer to be the result of a noncontact mechanism of injury and resulting in restricted activity of greater than 48 h. Any athlete who sustained either a noncontact ACL injury or ankle sprain was then further queried by their certified athletic trainer to determine the regularity of their menstrual cycle over the course of the last year, date of prior onset of menses and date of onset of next menses following the injury. Weekly e-mails were sent to remind each participating institution about this project and for data clarifications as needed. Each certified athletic trainer was asked to reverify his or her starting roster and the athletes' contraceptive status at the end of the season. Only injuries sustained during the traditional NCAA season for each sport were considered eligible for this study. Injury identification was made by either the certified athletic trainer or a team physician. Partial and complete ACL injuries were included in accordance with prior NCAA methodology. Reinjuries to the ACL or ankle were excluded.
Data for the two seasons of basketball were compared using chi-square 2 × 2 tables to determine if there were any differences in injury rates between the 2 yr for overall injury rates as well as for ankle and noncontact ACL injuries independently. Rates of soccer injuries were then compared to the combined basketball data in order to determine any differences in injury rates between the sports.
In order to determine if the newer data would substantiate the earlier periodicity results, this data was applied to the earlier analysis on both the retrospective recall data and the prospective collection data. In order to allow for standardization with those athletes using hormonal therapy, anyone who sustained an injury greater than 28 d after the onset of their last menstrual cycle or was classified as other (not using oral contraceptives as a method of contraception) was deleted from the analysis. Centered moving average with a span of four days, as well as nonlinear trigonometric regression analysis with bootstrap estimation of the confidence intervals, were used to determine the periodicity of injury.
Performing a power analysis using the earlier data suggests that a sample size of 4118 athletes would be needed with 1113 athletes using hormonal therapy and 3005 not using hormonal therapy. This sample size would be sufficient to determine if there is a protective effect from injury in the “on pill” group. From the earlier data, the “on pill” group was expected to have an overall injury rate 4% less than those athletes in the “off pill” group. These calculations are based on an alpha = 0.05 and power = 0.80. Our total study population was 3150.
All data calculations were done using SPSS® (SPSS, Inc., Chicago, IL) and Power and Sample Size Calculations (7).
One thousand sixteen schools sponsor basketball and 878 schools sponsor soccer under NCAA rules. Two hundred nine schools across the three seasons agreed to participate. Ninety-five (9.4%) schools participated in basketball during 2000–2001 (41 Division I, 22 Division II, 32 Division III), 63 (6.2%) schools participated in basketball during 2001–2002 (26 Division I, 8 Division II, 29 Division III), and 51 (5.8%) participated in soccer during 2001–2002 (28 Division I, 8 Division II, 15 Division III). Participating athletes in these three collection periods yielded 3229 athletes eligible for participation in this project. Seventy-nine athletes refused participation within those schools, leaving a study population of 3150. Two thousand twenty-six reported using no hormonal therapy, and 1024 reported using hormonal therapy. There were 45 ACL injuries and 116 ankle sprains in this population.
To verify the accuracy of method of data collection used for determining hormonal level, 67 athletes reported date of onset of menses using both retrospective recall and prospective collection. There were 22 ACL injuries and 45 ankle sprains in this group of athletes. It was anticipated that the majority of athletes who reported data using both methods of data collection would report cycle lengths equal to 28 ± 4 d, especially if they reported having regular (12 per year) menstrual cycles (Fig. 1). In particular, it was anticipated that athletes using oral contraceptives would report time differences between the onset of their prior menstrual cycle and their next menstrual cycle equal to 28 d. Athletes on oral contraceptives reported times spans from 19 to 50 d. The rest of the athletes reported greater time spans, thus it was concluded that the two methods of data collection are not equivalent.
If an athlete reported 28 d or less between either time of injury and prior onset of menses or time of injury and onset of next menses, they were then eligible for the periodicity analysis. There were 30 athletes (12 ACL and 18 ankle) who sustained either a noncontact ACL injury or ankle sprain who reported not using any hormonal therapy and 26 athletes (8 ACL and 18 ankle) who reported using oral contraceptives in this group.
There were 45 noncontact ACL injuries and 116 noncontact ankle sprains out of 3150 athletes across the three seasons. Table 1 shows that the rate of noncontact ACL injury in basketball between 2000–2001 and 2001–2002 was almost the same (0.017 vs 0.014, P > 0.05). The rate of noncontact ankle injury in basketball between 2000 and 2001 was also similar (0.04 vs 0.04, P > 0.05). Table 2 shows the rate of injuries of the two basketball seasons combined compared to the one soccer season. The rate of noncontact ACL injury in basketball was twice as high as in soccer (0.02 vs 0.01, P < 0.001). The rate of noncontact ankle injury was twice as high in basketball as in soccer (0.04 vs 0.02, P < 0.001). The overall rate of injury in basketball for these two injuries compared with soccer was significantly higher (P < 0.001)
Hormonal therapy use and protective effect.
Rates of hormonal therapy usage remained stable at 42% during the 2 yr of basketball, whereas in soccer over 70% of the athletes reported using hormonal therapy. This difference between the two sports was statistically significantly. There was no difference in rate of injuries between those athletes using hormonal therapy and those athletes not using hormonal therapy.
For those athletes not using hormonal contraceptives, the recall method of data collection demonstrated periodicity (Fig. 2). Periodicity is defined as a pattern that can be identified over a given time frame and repeats over several time frames. The peak occurrence of injury was between days 7 and 9 after onset of menses. For those athletes using hormonal contraceptives, regardless of whether timing of injury was calculated using prospective date of onset of next menses or retrospective date of onset of last menses, no periodicity was present. (Figs. 4 and 5) No periodicity was present for those athletes not using hormonal contraceptives reported via the prospective data either (Fig. 3).
The final power for our sample combining both ACL and ankle injuries is 0.06. This data allows for a detection of five times or greater difference in injury rates between the “on hormonal therapy” and “off hormonal therapy” groups. It was not possible to determine if there is a difference in injury rates between those using monophasic oral contraceptives and those using triphasic oral contraceptives with the current sample size. Based on the injury rates with this data we would have needed 15,500 people in the “on hormonal therapy” group and 31,000 in the “off hormonal therapy” group in order to have adequate power.
Rates of injury in basketball remained constant during the 2 yr and are comparable with other NCAA data indicating that these rates are reliable. Because the schools reporting during the 2 yr were not the same schools, it is reasonable to assume that reporting was complete and that this is a representative sample.
Because the hormonal milieu of the female athlete is notably different from the male athlete, direct or indirect effects of hormones as a risk factor has been hypothesized as the cause of the increased rate of noncontact ACL injuries in females. Moller-Nielsen et al. (1989) (11) reported on 86 First, Second, and Third Swedish Football League soccer players followed over the 1984 season. These athletes kept a 1-yr calendar of menstrual cycle and contraceptive status. During this time, they sustained 62 traumatic injuries. The most common injury was ankle sprain, followed by knee sprain. Most injuries occurred either just before or at the onset of menses. Myklebust et al. (12), using a menstrual cycle questionnaire, reported on 23 noncontact ACL injuries that occurred to female elite handball players during a 3-yr period. The majority of these injuries occurred just before or just after menses. Wojyts et al. (16,17) reported that in 40 athletes, the majority of injuries occurred during the ovulatory phase, defined as cycle days 10–14. Slauterbeck et al. correlated saliva and serum estrogen and progesterone in 37 athletes at the time of ACL injury. He found that a greater number of ACL injuries occurred on days 1–2 (14).
Looking at seven active females, Heitz et al. (9) demonstrated that knee ligamentous laxity as measured by a knee arthrometer increased significantly throughout the menstrual cycle with peak knee laxity occurring at time of peak estrogen and peak progesterone levels (as measured by blood levels). The authors hypothesized that ligamentous laxity could be a factor in the ACL knee injury risk equation. Deie et al. (6) in 2002 also demonstrated significant changes in laxity using a knee arthrotometer in 16 women across the menstrual cycle with peak laxity in the luteal phase.
Arendt et al. (3) modified their original data collection methodology to obtain more detailed hormonal data on female basketball athletes for the 1996–1998 seasons. In the original data collection method athletes were asked to recall the date of onset of their menses before their injury as well as the regularity of their menstrual cycle and their use of oral contraceptives. This data allowed us to demonstrate that in a population of 83 female collegiate athletes the majority of noncontact ACL injuries occurred just after the onset of menses. The limitation of this data was whether or not there was a protective effect from the use of oral contraceptives and what the rate of injury was because of no data on the uninjured population.
To answer these questions, for the 2000–2001 basketball season, the variables were expanded in the data set from the prior study to include information on all the female athletes of the participating schools playing basketball or soccer. Specifically the contraceptive status of all team members was added to the data set. In 2001–2002 for basketball and soccer, the date of onset of the next menstrual cycle for the injured athlete was also added. This additional data allowed for the calculation of rates of injury as well as verify the accuracy of the beginning of the previous menses as an accurate way to assess cycle length via recall and obtain a more accurate assessment of cycle/hormonal state. By obtaining the date of onset of the next menstrual cycle, the hormonal phase of the injured athlete could be identified more accurately, because it is known that ovulation occurs on average 14 d before onset of menses (15). Therefore the luteal phase is the most consistent phase in length of days. If a menstrual cycle is greater than 28 d, the cycle length tends to elongate in the follicular phase, keeping the luteal phase (14–15 d before menstruation) constant. Whether use or type of hormonal therapy was a factor in the rate or timing of the injury, could also be determined by this data set.
By calculating the total number of days between the onset of the last menstrual cycle before injury and the onset of the first menstrual cycle after injury and comparing this to the athletes' reported number of menstrual cycles a year, it could be determined if there was consistency between the two methods of data collection in determining where in the hormonal cycle the athletes' injury occurred.
This data demonstrated that the risk of noncontact ACL or ankle injury is statistically significantly higher in basketball than in soccer (P < 0.001). The two sports are felt to be similar in terms of running, cutting, and jumping activities. Primary differences between the two sports are the surface of play and type of shoes, and possibly the speed at which changes in direction are made, ball in hand, ball at foot. All of these soccer injuries occurred during outdoor participation, but no information was collected on the surface during which the injury occurred. No information was collected on the type of footwear involved at the time of injury. The rate of noncontact ACL injury is still very small, however, half that of noncontact ankle injuries.
Time loss from noncontact ACL injuries is typically longer than with a noncontact ankle sprain, and theoretically the well-documented potential long-term consequences of knee pain, loss of motion, and arthritis are more devastating than the less well-documented, but possibly similar, long-term consequences of a noncontact ankle sprain.
These data do not support the periodicity found in the earlier work. In this study, only the group of athletes with recall data collection and not using hormonal therapy demonstrated periodicity. All the other groups analyzed demonstrated no pattern of injury. There was clear discrepancy in the information provided between the recall and prospective data collection methods. One would assume that the prospective method of data collection, following the athlete daily after injury to determine the onset of menses, would provide the more accurate information as to hormonal status at the time of injury. Therefore the lack of periodicity in three of the four analyses in this data set would support the fact that there is no periodicity in the occurrence of noncontact ACL injury or ankle sprain.
Other injuries or illness were not accounted for that may have affected the athletes' participation and thus their risk of injury status. Everyone did not participate every day and every athlete did not participate at an equivalent level of competition. There are also theories that stress affects the onset of menses and injury may constitute sufficient stress to change an athlete's cycle (4,5). In order to detect a 1–2% difference in injury rates, approximately 45,000 athletes would be needed: 15,618 athletes “on hormonal therapy,” and 31,000 athletes “not using hormonal therapy.”
Noncontact ACL injuries and ankle sprains occurred at significantly higher rates in basketball than in soccer. There was no difference in rate of injury between those athletes who used hormonal therapies and those who did not. Retrospective recall is not equivalent to prospective recall in determining menstrual phase. There is no definable pattern to the occurrence of these injuries. The overall number of these injuries sustained is very low in relation to the number of athletes participating in these sports.
1. Arendt, E., and R. Dick. Knee injury patterns among men and women in collegiate basketball and soccer: NCAA data and review of literature. Am. J. Sports Med.
2. Arendt, E. A., J. Agel, and R. Dick. Anterior cruciate ligament injury patterns among collegiate men and women. J. Athl. Train.
3. Arendt, E. A., B. Bershadsky, and J. A. Agel. Periodicity of non-contact anterior cruciate ligament injuries during the menstrual cycle. J. Gend. Spec. Med.
4. Blackwell, B, and P. McCullagh. The relationiship of athletic injury to life stress, competitive anxiety and coping resources. Athl. Train.
5. Cryan, P. D., and W. F. Alles. The relationship between stress and college football injuries. J. Sports Med. Phys. Fitness.
6. Deie, M., Y. Sakamaki, Y. Sumen, Y. Urabe, and Y. Ikuta. Anterior knee laxity in young women varies with their menstrual cycle. Int. Orthop.
7. Dupont, W. D., and W. D. Plummer. Power and sample size calculations: A review and computer program. Cont. Clin. Trials.
8. Griffin, L. Y. Prevention of Noncontact ACL Injuries. Rosemont, IL: American Academy of Orthopaedic Surgeons, 2001, pp. 1–122.
9. Heitz, N. A., P. A. Eisenman, C. L. Beck, and J. A. Walker. Hormonal changes throughout the menstrual cycle and increased anterior cruciate ligament laxity in females. J. Athl. Train.
10. Krips, R., N. van Dijk, H. Lehtonen, T. Halasi, B. Moyen, and J. Karlsson. Sports activity level after surgical treatment for chronic anterolateral ankle instability: A multi-center study. Am. J. Sports Med.
11. Moller-Nielson, J., and M. Hammer. Women's soccer injuries in relation to the menstrual cycle and oral contraceptive use. Med. Sci. Sports Exerc.
12. Myklebust, G., S. Maehlum, I. Holm, and R. Bahr. A prospective cohort study of anterior cruciate ligament injuries in elite Norwegian team handball. Scan. J. Med. Sci. Sports
13. PDR. Physicians' Desk Reference. Montvale, NJ: Medical Economics Co., 2000, pp. 1–3355.
14. Slauterbeck, J. R., S. F. Fuzie, M. P. Smith, et al. The menstrual cycle, sex hormones, and anterior cruciate ligament injury. J. Athl. Train.
15. Vollman, R. F. The menstrual cycle. In: Major Problems in Obstetrics & Gynecology
, Vol. 7. Philadelphia: W. B. Saunders, 1977.
16. Wojtys, E. M., and L. J. Huston. Longitudinal effects of anterior cruciate ligament injury and patellar tendon autograft reconstruction on neuromuscular performance. Am. J. Sports Med.
17. Wojtys, E. M., L. J. Huston, T. N. Lindenfeld, T. E. Hewett, and M. L. V. H. Greenfield. Association between the menstrual cycle and anterior cruciate ligament injuries in female athletes. Am. J. Sports Med.