Children are participating in organized sports at earlier ages and in increasing numbers. This trend is particularly evident in women's gymnastics. According to one source, (10) this sport experienced an increase of 461% in interscholastic participants in the United States between 1974 and 1980. Since 1980, however, the number of interscholastic participants has been decreasing while the number of clubs, and consequently younger participants, has been increasing dramatically (14). Early participation has become characteristic of women's gymnastics; the average age of champions and Olympic gold medal winners has decreased markedly over the past 2 decades.

The trends of earlier participation and increased numbers of participants have been accompanied by increased levels of training. For example, the junior elite female gymnast (i.e., 10 to 14 years of age) is reported to train, on the average, 5.36 days per week and 5.04 hours per day (25). Unfortunately, a growing body of evidence suggests that the processes of growth, particularly those associated with periods of rapid growth, render the young female gymnast more susceptible to injury than the postpubescent who characterized the sport less than 2 decades ago (6).

Increased involvement at an early age, with the extreme training intensity required, strongly suggests the possibility of a concomitant increase in the risk of injury. Given the uncertain long-term consequences of pediatric sports injuries and the rising costs of medical care, injury prevention would seem to be of utmost importance. Therefore, reasonable questions to ask include the following: 1) What are the nature and rate of injuries affecting young female gymnasts, and 2) to what extent are such injuries preventable?

The accumulating reports of injuries to young female gymnasts against a background of their apparent vulnerability to such insult indicate an active concern for their welfare. However, an extensive review of the literature pertaining to the epidemiology of gymnastics injuries reveals more questions than answers (19). The purpose of this investigation was to determine the following: 1) What are the nature and rate of injuries that affect young competitive female gymnasts? 2) Do particular host and environmental characteristics predispose the young female gymnast to a higher risk of injury? 3) To what extent can group classification of high and low injury risk gymnasts be predicted from selected host and environmental characteristics? Finding answers to these questions is basic to the establishment of intervention techniques designed to reduce the incidence of injury to young female gymnasts.

Fifty young female gymnasts (mean age, 12.6 years) representing two nationally recognized gymnastics clubs participated in this study. The subjects were all competitive female gymnasts from Elite and I, II, and III levels of competition and were training 20 to 27 hours, 5 to 6 days per week, and up to 12 months of the year. The breakdown of subjects per level of competition and training, compared to United States Gymnastics Federation (USGF) figures, is shown in Table 1. It is evident that the study population is somewhat different from what one would see nationally. At all levels, with the exception of Elite, the study subjects trained longer hours. This was particularly dramatic at the Class III level, where they trained at least twice as many hours each week. An insight into the elitism of the two clubs is perhaps also evident from the fact that 16 of the 50 girls were living away from home.

_______________________________________________________________________________________

Distribution of female gymnasts by competitive level and training

hours per week--comparison with USGF figures

_______________________________________________________________________________________

_______________________________________________________________________________________

All subjects were participants in a longitudinal study of young male and female gymnasts sponsored by the International Institute for Sport and Human Performance at the University of Oregon and by the United States Gymnastics Federation. The present investigation was a component of this longitudinal study and received approval from the Committee for the Protection of Human Subjects at the University of Oregon. Informed consent for participation in the study was obtained from the parents of all gymnasts.

A pilot study was conducted involving gymnasts from both clubs during the period of January through June of 1986. This enabled testing and refinement of the proposed injury surveillance system. Notably, 31 female gymnasts sustained 48 injuries during this 6 month period. An injury was defined as any gymnastics related incident that resulted in a gymnast missing any portion of a workout or competitive event.

The study followed a prospective epidemiologic design. During the summer of 1986, baseline information was collected on the gymnasts who had enrolled for the forthcoming season. These included 19 of those included in the pilot study and an additional 31 gymnasts who had moved up from Class IV and/or had come from outside Eugene to train at one of the gymnastic facilities. Baseline procedures included determination of injury status, maturity assessment, and administration of selected anthropometric measures necessary to the calculation of somatotype. Injury surveillance began in the beginning of September and continued for 1 year.

Musculoskeletal screening was performed by a registered physical therapist before the injury surveillance period. This procedure was supplemented by three additional sources of information: completion of a health history form based on an interview with each gymnast, previous injury surveillance (gymnasts were followed from 2 to 8 months before taking health history), and access to medical records of previously injured gymnasts. In sum, these procedures provided an injury profile for each gymnast involved in the study.

Maturity assessment of the gymnasts was performed by registered nurses before and midway through the injury surveillance period. The "Physician's Guide to Sexual Maturity" (9) was used by the nurses to determine the stage of breast development for each of the female gymnasts. This score, in conjunction with menarcheal information, was used to determine maturation rate, i.e., the speed at which an individual achieves adult height. Tanner Stages 2 and 3 are consistent with the period of most rapid growth (18).

Individuals were classified as experiencing stable growth if either one of the following two conditions existed: 1) breast Stage 1 both before and midway through the injury surveillance period, or 2) onset of menarche before the second circumpubertal rating. Almost invariably, menarche occurs after peak height velocity (31). Individuals not characterized by these conditions were classified as experiencing rapid growth during the majority of the injury surveillance period, i.e., Stages 2 and 3 breast development.

The majority of evidence indicating that individuals are somewhat more susceptible to injury during the adolescent growth spurt arises from case series studies and research with infrahuman species (6). The determination of the relationship between maturation rate and injury status was therefore considered to be an important component of this study.

For the purpose of determining somatotype, the following measurements were taken for all gymnasts using standardized anthropometric instruments: humerus and femur width, tense arm girth, calf girth, skinfolds (triceps, subscapular, suprailiac, calf), height, and weight. The techniques for measurement and reliability determination were consistent with those recommended by Carter (7).

Investigation into the somatotype of athletes indicates selection between and within sports for particular physical characteristics (3,·8). Logic would suggest that the girl with physical characteristics most congruent to the demands of the sport would also be most protected from injury (l). Investigation involving gymnasts (28) and college level athletes (11) offers support for this supposition. However, the question of how often young female gymnasts realize their physical unsuitability to the sport by way of injury has not been answered and therefore was felt important to address in the present study.

Injury surveillance involved a minimum of one visit every 2 weeks by the primary investigator to each of the competitive groups of gymnasts (Levels III, II, I, and Elite). At that time, injury information was gleaned from one or more of several possible sources. First, gymnasts kept a training diary which included, when applicable, an injury component for reporting details of any injury or unusual discomfort. Coaches helped to ensure the gymnasts' regular completion of their diaries. Diaries were reviewed by the primary investigator on each visit. A second source of information was the verbal report of injury from the gymnast or coach. Some coaches would complete the injury report form with the injured gymnast. Finally, the primary investigator occasionally observed an otherwise unreported injury. For example, a girl may have favored an injured part or, more commonly, was missing some or all of practice. These situations alerted the investigator to the possible existence of injury, which he was then able to follow up. Injury was defined as any gymnastics-related incident that resulted in the gymnast missing any portion of a workout or competitive event.

After it had been determined that an injury had occurred, an interview was arranged with the gymnast. During the interview, an injury report form was completed. The form was consistent with that used by the National Collegiate Athletic Association Injury Surveillance System. This form was used to record information on the nature and circumstances of the injury and the number of practice days before the gymnast returned to her former level of training and competition. An injured gymnast was interviewed on each visit until she returned to full training.

Following the 1 year injury surveillance period, the injury report information was supplemented by data from the medical files of injured gymnasts. Information on the gymnasts' exposure to the risk of injury was obtained from their attendance records. This information was recorded in both total hours and days of training so that individual and group rates for injuries and proportional time loss could be determined.

Epidemiology has been defined as the study of the distribution of injuries and their causative agents (22). The study of distribution of injuries is descriptive epidemiology, and the search for determinants of exhibited distribution is analytic epidemiology. Our study included both descriptive and analytical components.

As far as the descriptive component, our purpose was to determine the nature and rate of injuries affecting the young female gymnasts over a 1 year period. Information on injury nature (i.e., location) was calculated in terms of numbers and relative percentage of injuries. The rate of injury was calculated by dividing the total number of injuries sustained by the gymnasts by the total number of hours trained, then multiplying this figure by 1000. This approach yielded an injury rate per 1000 hours exposure. The frequency procedure in SAS was used to determine the number, percentage, and rate of injuries.

There were two related purposes in the analytical component of the study. The first purpose was to identify the nature of the relationship between injury status and selected host and environmental variables. The injury status of each gymnast was determined with reference to two criterion variables: individual injury rate and individual proportional time loss (as a result of injury). It was felt important to consider both of these variables since each reflects a potentially different dimension of injury status. The criterion variables were tested against the following host and environmental variables: maturation rate, endomorphy, mesomorphy, ectomorphy, exposure time, and competitive level. The SAS CANCORR procedure was used to determine the nature of this relationship.

The second purpose in the analytical component of the study was to determine the extent to which group classification into high and low injury risk groups could be predicted from the same predictor variables described above. The SAS STEPDISC procedure was employed using the upper and lower quartiles of each criterion variable (i.e., two separate analyses) as the extreme groups. The SAS DISCRIM procedure was used to determine the extent to which group membership in high and low injury risk groups could be correctly classified.

The mean age of the gymnasts was 12.6 years. On average, these girls trained 803 hours or 187 days throughout the year (Table 2). The mean somatotype of the sample was 2.2-4.14-3.48, which is slightly higher for all components than reported elsewhere in the literature (4). However, the relative dominance of the components is the same; mesomorphy is the dominant component and ectomorphy exceeds endomorphy.

_______________________________________________________________________________________

TABLE 2

Comparison of dropouts and nondropouts

_______________________________________________________________________________________

_______________________________________________________________________________________

The results of the baseline procedures to determine injury status revealed that the 50 gymnasts had amassed 60 injuries during the previous year of participation in gymnastics. Thirty-eight of the 60 injuries resulted in physician consultation. The most frequently occurring injuries were strains (N = 18), nonspecific pain that included injuries diagnosed as stress reaction or overuse types (N = 13), sprain (N = 11), and inflammation (N = 7). The most frequently involved body part was the lower back (N = 11), followed by the wrist (N = 9), ankle (N = 9), and knee (N = 7). These results may be deflated, however, because of the potential recall bias of subjects for the 4 to 10 months they were not followed prospectively.

All but six of the gymnasts were at full training (i.e., uninjured) at the beginning of the injury surveillance period. However, the result of the musculoskeletal screening and health history interview indicated that 38 of the 44 uninjured gymnasts reported experiencing some pain/discomfort associated with training during the weeks just before the start of injury surveillance. The body part most frequently identified was the wrist, followed by the lower back, ankle, and knee. The six injured gymnasts missed an average of 20 days from the beginning of injury surveillance before returning to full practice.

Of the original 50 gymnasts, only 29 remained at the end of the injury surveillance period. However, the availability of attendance records for the gymnasts permitted calculation of individual and group rates for injury and proportional time loss. Notably, 11 of the 21 girls who dropped out from training were injured by the time they withdrew from participation. In addition, 11 of the dropouts, including 5 of those who were injured when they withdrew, were from out of state. A comparison of the dropouts with those who remained, with reference to descriptive findings, is shown in Table 2. Table 3 indicates that the dropout time was relatively evenly distributed over the months of the year, beginning in December. Table 4 shows injury frequency. Notably, seven of the gymnasts remained uninjured throughout the surveillance period. These girls were all Class III, and all but two were prepubescent. They had a mean exposure time of 709 training hours and a mean age of 10.54 years.

_______________________________________________________________________________________

TABLE 3

Injury rates by month of the year

_______________________________________________________________________________________

_______________________________________________________________________________________

_______________________________________________________________________________________

TABLE 4

Injury frequency

_______________________________________________________________________________________

_______________________________________________________________________________________

During the injury surveillance period, the 50 female gymnasts amassed 40,127 hours of training and sustained 147 injuries, for an overall injury rate of 3.66 (147/40,127) injuries per 1000 hours of participation. The 147 injuries were distributed among 43 gymnasts. As shown in Table 5, the Class II girls (N= 7) experienced the highest injury rate followed by the combined Elite and Class I and the Class III gymnasts. When injury rates were calculated by month of the year (Table 3), it was apparent that the first several months of the season were characterized by particularly high injury rates.

_______________________________________________________________________________________

TABLE 5

Injury rate by competitive level

_______________________________________________________________________________________

_______________________________________________________________________________________

As shown in Table 6, the injuries were well distributed throughout the body; few anatomical areas were spared. Most noticeable was the preponderance of injuries involving the lower extremities (63.7%), particularly the knee (14.3%) and ankle (21.1%). The wrist was the most frequently injured body part (9.5%) in the upper extremity, whereas the lower back sustained most of the injuries (12.2%) in the spine and trunk.

_______________________________________________________________________________________

TABLE 6

Injury location and nature of onset

_______________________________________________________________________________________

_______________________________________________________________________________________

Injury types are shown in Table 7. The most common type of injury was nonspecific pain, which accounted for 59 cases (40.1%). Common diagnoses in this category included patellofemoral pain, ankle arthralgia, general pain (e.g., ankle, wrist, elbow), and overuse syndrome (e.g., ankle, knee, wrist). The most frequently injured body part in this category was the lower back (N = 15), followed by the knee (N = II), ankle (N = 11), and wrist (N = 9).

_______________________________________________________________________________________

TABLE 7

Injury type

_______________________________________________________________________________________

_______________________________________________________________________________________

Sprains (19%) were the next most frequent injury type and were followed by strains (17.7%). The most frequently sprained body part was the ankle. The most often strained body part was the lower leg. The majority of strains occurred during the first hour of practice (61.5%). Slightly more than a third of the sprains (35.7%) occurred during the dismount.

It is important to recognize differences in the nature of the onset of the symptoms (19). In our study, injuries reported to have occurred suddenly were considered traumatic injuries. Injuries characterized by a gradual onset were considered to have resulted from overuse. A review of the distribution of injuries by onset (see Table 6) revealed that 65 injuries (44.2%) were of a sudden nature, whereas 82 (55.8%) were characterized by a gradual onset. The body parts most frequently characterized by gradual onset injury included the wrist, knee, and the lower back. More than half of the ankle injuries, however, were characterized by a sudden onset. Cross-tabulation of injury onset with frequent injury type reveals that the vast majority of nonspecific injuries were characterized by gradual onset (93.2%). Almost all sprains (96.4%) and the majority of strains (69.4%), on the other hand, were sudden onset injuries.

Separate analyses of sudden onset injuries by event at the time of injury revealed that floor exercise was the event most often associated with injury (35.4%), followed by the balance beam (23.1%) and the uneven bars (20%). These findings are shown in Table 8. Cross-tabulation of sudden onset injuries with hour of occurrence showed that 32.3% of these injuries occurred during the 1st hour of practice (Table 9).

_______________________________________________________________________________________

TABLE 8

Sudden onset injuries by event

_______________________________________________________________________________________

_______________________________________________________________________________________

_______________________________________________________________________________________

TABLE 9

Hour of practice when sudden onset injuries occurred

_______________________________________________________________________________________

_______________________________________________________________________________________

When the criteria for classifying injuries as reinjuries included same body part, body size, type, nature of onset, and history of injury the previous year, the rate of reinjury was 32.7%. This rate was even higher when particular body parts were isolated. The lower back, for example, was characterized by a 72.2% rate of reinjury. Of the 48 injuries classified as reinjuries, 8 (16.7%) were traumatic and 40 (83.3%) were characterized by a gradual onset, suggesting the frequent reoccurrence of chronic type injuries (Table 10). Overall, 24 of the gymnasts experienced reinjury and 16 of these girls experienced more than one reinjury (Table 11).

_______________________________________________________________________________________

TABLE 10

Cross-tabulation of nature of injury onset and reinjury

_______________________________________________________________________________________

_______________________________________________________________________________________

_______________________________________________________________________________________

TABLE 11

Reinjury frequency

_______________________________________________________________________________________

_______________________________________________________________________________________

A useful measure of injury severity is the duration of restriction from athletic performance (29). Injury severity was classified according to time loss in days (Table 12). This classification approach is consistent with that recommended by the National Athletic Injury/Illness Reporting System, with the following exception (2). Pursuant to the recommendations of the American Orthopaedic Society for Sports Medicine, the terms "mild," "moderate," and "severe" have been avoided (29).

_______________________________________________________________________________________

TABLE 12

Frequency and percentage of injuries by injury severity

_______________________________________________________________________________________

_______________________________________________________________________________________

Overall, 40.8% of the injuries resulted in 7 days or less of time loss before return to former practice levels. A third of the injuries required from 8 to 21 days of time loss, and a little more than a quarter of the injuries (25.9%) involved a time loss of more than 21 days. Separate analyses of injuries requiring 7 days or less of time loss reveals that a little more than a fifth of the total number of injuries required no more than 2 days of time loss.

During the injury surveillance period, the 50 female gymnasts accumulated 2703 days of reduced training from a possible 9338 days as a result of injury. This resulted in a proportional time loss of 0.289 (i.e., 2703/9338) and implies that, as a group, the gymnasts spent 28.9% of the season at less than full practice as a result of injury. This reflects in many cases multiple injuries per person. The mean time loss per injury was 22 days, and the mean time loss per individual was 54 days.

During the injury surveillance period, 36 of the 50 gymnasts consulted with a physician for 59 injuries. During the previous year, 23 of these athletes sought medical advice for 38 injuries. A review of their medical records revealed a variety of diagnoses (Table 13). With reference to the injury surveillance period, the following observations are made. Most noticeable among the diagnoses was the prevalence of stress-related injuries. Almost two-thirds of the injuries (64.4%) were characterized by a gradual onset. The most frequently injured body parts were the wrist and ankle and the most frequent injury type was nonspecific pain. Cross-tabulation with injury severity revealed that 30 of the 38 injuries causing 21 or more days of time loss resulted in physician consultation. Overall, only four of the injuries (2.7%) required surgery.

_______________________________________________________________________________________

TABLE 13

Diagnoses and incidence of physician-seen injuries

_______________________________________________________________________________________

_______________________________________________________________________________________

Canonical correlation analysis was used to determine the multivariate relationship between the injury and predictor variables. The analysis resulted in a nonsignificant effect (Wilks' Lambda = 0.7, F(10,86) = 1.65, P < 0.11). Separate discriminant and subsequent classification analyses were run on each of the criterion variables (i.e., individual proportional time loss and individual injury rate) to determine the extent to which high and low risk injury groups could be predicted from the battery of host and environmental variables.

The discriminant analyses involving individual proportional time loss resulted in a significant effect (Wilks' Lambda = 0.73, F(2,23) = 4.28, P < 0.03), suggesting that the high and low injury risk groups were statistically different and could best be distinguished as a result of the contributions of competitive level (P < 0.05) and maturation rate (P < 0.06). The average squared canonical correlation was 0.271, indicating that 27.1% of the variance was accounted for by these variables. Table 14 shows that 84.6% of the high injury risk gymnasts and 69.3% of the low injury risk gymnasts could be correctly classified. Overall, 20 of the 26 (77%) gymnasts were correctly classified into high and low injury risk groups. However, generalization of this finding to other samples awaits cross-validation.

_______________________________________________________________________________________

TABLE 14

Classification analysis number of observations and percents classified into high and low injury groups criterion variable: proportion time loss

_______________________________________________________________________________________

_______________________________________________________________________________________

The discriminant analysis involving individual injury rate resulted in a nonsignificant effect (Wilks' Lambda = 0.82, F(2,23) = 2.57, P < 0.1). However, it may be noteworthy that the best discriminators of the high and low injury risk groups (although insignificant) were also competitive level (P < 0.11) and maturation rate (P < 0.15).

The comparison of dropouts and nondropouts illustrates the rather substantial contribution of the dropouts to the injury data. This contribution underscores the importance, suggested by Hunter and Levy (12), of not excluding dropouts from the analysis of data. The fact that a greater proportion of the dropouts (Table 2) represented Elite and Class I levels perhaps reflects the relatively high injury rates experienced at these levels. One would normally expect a higher proportion of dropouts at the lower competitive levels, i.e., selective screening. The larger proportion of dropouts who were injured (11/21) or from out of state (11/21), or both (5/21) suggests the role of injury and absence from home as potential sources of motivation for dropping out.

The relatively high group injury rates recorded during the first several months of the surveillance period (Table 3) perhaps reflect the finding that 38 of the 44 initially uninjured gymnasts (at the start of the study) reported experiencing some pain or discomfort associated with training. The high injury rate in January is perhaps explained by an increased training intensity in preparation for the winter competitions.

The overall injury rate of 3.66 injuries per 1000 hours of participation cannot be compared to injury rates derived from previous injury studies involving young female gymnasts, since these studies reported injuries as a function of the number of participants (l0, 16, 23, 27, 30). For the purposes of comparison, the injury rate in this study as well as that of the pilot study have also been determined with reference to 100 participants per season (Table 15). However, this comparison should be interpreted with caution since there is no way of knowing to what extent the gymnasts in these studies were similarly exposed to injury. In addition, we do not know if the injury data from the previous studies included dropouts. When number of participants is used as the denominator in our study, it is apparent that our injury rate was quite high compared with previous studies. This was also true of our injury rate during the 6 month pilot study. These findings perhaps reflect the greater exposure to the risk of injury for the Class III, II, and I subjects in our study. They may also reflect the prospective study design combined with the data collection methods used in the present study. Questionnaires tend to underreport the frequency of injuries as a result of low response rates, short-term or long-term recall bias, and response motivation bias (29).

_______________________________________________________________________________________

TABLE 15

A comparison of gymnastics injuries studies involving young competitive club level female gymnasts

_______________________________________________________________________________________

_______________________________________________________________________________________