Share this article on:

Lower limb morphology and risk of overuse injury among male infantry trainees


Medicine & Science in Sports & Exercise: August 1996 - Volume 28 - Issue 8 - p 945-952
Clinical Sciences: Clinical Investigations

The effect of anatomic variation on the risk of overuse injuries has not been adequately evaluated. To determine the association of several common anatomic characteristics (genu varum, genu valgum, genu recurvatum, and lower limb length differences) with risk of overuse injury, we made prospective morphologic measurements of young men prior to beginning 12 wk of Army infantry training. The training included frequent running, marching, calisthenics, and other vigorous activities. Lower extremity anatomic landmarks were highlighted, and front- and side-view photographic slides were taken of the 294 study volunteers. The slides were computer digitized, and the following measures calculated: pelvic width to knee width ratio (to assess genu valgum/varum), quadriceps angle (Q-angle), knee angle at full extension, and lower limb length differences. The cumulative incidence of lower limb overuse injury was 30%. Relative risk (RR) of overuse injury was significantly higher among participants with the most valgus knees (RR = 1.9). Those with Q-angle of more than 15° had significantly increased risk specifically for stress fractures (RR = 5.4). Anatomic characteristics were associated with several other types of injuries, including pain and nonacute muscle strain due to overuse. This pilot study provides evidence that some lower limb morphologic characteristics may place individuals at increased risk of overuse injuries.

Walter Reed Army Institute of Research, Washington, DC; U.S. Army Center for Health Promotion and Preventive Medicine, Aberdeen Proving Ground, MD; Army Research Institute of Environmental Medicine, Natick, MA; and Walter Reed Army Medical Center, Washington, DC

Submitted for publication September 1995.

Accepted for publication April 1996.

The authors thank Wendy Robinson Weringo, formerly of ARIEM, for valuable assistance with all aspects of conducting this study and Drs. Robert Tuthill and Anne Stoddard of the University of Massachusetts School of Public Health for guidance in study design and analysis. Both Dr. Wayne Leadbetter of the Orthopaedic Center, Rockville, MD, and Dr. William J. Hopkinson, Associate Professor in the Department of Orthopaedic Surgery, Loyola University Stritch School of Medicine, provided careful review and helpful comments. We also thank the soldiers who volunteered their participation, without whose cooperation this study would not have been possible.

Address for correspondence: David N. Cowan, SRA Technologies, Inc., 8110 Gatehouse Road, Suite 600W, Falls Church, VA 22042.

Despite considerable speculation on the issue, the actual effect of lower limb anatomic variation on the risk of injuries in active populations has not been studied adequately in well-designed epidemiologic studies(17,22). Based on clinical impressions(3,8,20) and case series(4,12), several anatomic risk factors have been postulated. These include genu valgum (“knock knee”) and genu varum (“bow legs”)(4,8,12,20), excessive quadriceps angle(Q-angle) (3,4,8,12), genu recurvatum(“back knee,” or hyperextension of the knee)(4), and unequal lower limb lengths(3,4,12,20). Notwithstanding these reports, in their review of the literature Powell et al.(22) stated that “none of the epidemiologic studies evaluated the role of anatomic factors in running injuries,” that“case studies are unable to establish causality,” and that“careful, abnormality-specific studies should be a top priority for future research.”

The epidemiologic approach to assessment of exercise and physical training-related injuries has recently been successfully applied to identify a number of risk factors among military trainees. These include sedentary lifestyle, i.e., inactivity (7,15), low levels of physical fitness (14,15), high and low levels of flexibility (15), excessive amounts of physical training(16), cigarette smoking(15,24), and foot morphology(5).

To better understand the possible associations between lower extremity overuse injuries and variations in lower limb morphology and to generate quantified hypotheses for further research, data from a prospective study of U.S. Army infantry trainees were analyzed. We examined the associations of overuse injuries in general and several specific overuse injuries with four measures of lower limb morphology: pelvic-patellar ratio (to assess genu valgum and varum), Q-angle, knee angle at full extension, and lower limb length differences.

Back to Top | Article Outline



We prospectively evaluated and followed a cohort of recruits entering U.S. Army basic infantry training. The population has been described in detail elsewhere (15). The cohort was followed throughout the duration of basic training (12 wk). During this initial entry training, trainees engaged in physical training 5-6 d·wk-1. Each day they ran or marched, on average, 40 min (for a total of about 200 miles over the 12 wk of training), did 30 min of stretching and calisthenics, and did 20 min of drill and ceremony. In addition to these routine exercises, they episodically maneuvered over obstacle courses, practiced hand-to-hand and rifle bayonet combat skills, and performed other vigorous training activities as well as walked from one training area to the next.

Back to Top | Article Outline

Photographs and Digitization

After being informed of the nature of this study, 294 volunteers (97% of the 303 eligible to participate) had front- and left-side-view photographic slides taken of lower extremity anatomic landmarks. Skeletal landmarks were highlighted with adhesive fluorescent markers and hemispheres by a physician(BHJ or DWP) prior to photography using guidelines from Hoppenfeld(9,10,11). The landmarks highlighted and utilized in this report were the medial and lateral malleoli, head of the fibula, tibial tuberosity, midpoint of the patella, greater trochanter, and anterior superior iliac spine (ASIS). The anatomic characteristics evaluated in this report were genu valgum (“knock knees”), genu varum(“bow legs”), Q-angle, knee angle at full extension, and lower limb length differences.

For the photographs participants were positioned standing on a bench facing the camera. Each subject's heels were spaced by aligning them against a 3 inch wide block with the forefeet further apart than the heels so that the medial border of the feet formed an angle of 60° (each foot was 30° divergent from the photographic line of sight). A mirror was mounted on the left side of the subject, placed at a 45° angle to the photographic line of site, to provide an image of the lateral aspect of the left lower limb. In order to provide a known distance in the photographic image that could be used to convert distances digitized between points into actual distances, a rod of precise length (0.75 m) was mounted to the right of the subject, perpendicular to the floor.

Photographs were taken with a 35 mm single lens reflex camera equipped with a 200 mm telephoto lens using color transparency film. Lighting was provided with four photo-flood lamps mounted 3 m from the subject and angled in toward the subject at 45° from the photographic line of sight, providing maximum lighting with minimum reflected glare. The camera was mounted on a tripod 15 m from the subject, with the optical line of sight directed to the knee area.

The photographs were digitized using an Altek (Silver Spring, MD) model ACT23 projection digitizing table with resolution of 0.01 mm. The ends of the reference rod were digitized, as were all of the highlighted points. For each photograph actual distances were calculated using a factor determined from the reference rod. The x,y table coordinates for each digitized site were stored in the central processing unit of the digitizer. Angles in degrees were calculated with formulae involving trigonometric functions. All data for inclusion in this study were digitized by one investigator (RMR). While photographs were taken of 294 subjects, errors occurred in the digitization process for Q-angle and pelvic-patellar ratio for one subject (0.3% of total) and for leg length difference among eight subjects (2.7% of total). These individuals were excluded from analysis only for those measures found to be in error.

To assess reliability of measurements of pelvic width, interknee width, and lower limb lengths, two investigators independently conducted two trials on eight individuals (selected from available research volunteers), placing landmarks, photographing, and digitizing the slides. Intrarater correlations ranged from 0.90 to 0.99 for these three measures, while interrater correlations ranged from 0.93 to 0.99.

As shown in Figure 1, the measure of genu varum and valgum utilized in this report is the pelvic width (H) divided by the distance between the midpoints of the patellas (K), or the pelvic-patellar ratio. This measure is based on the concept that the pelvic width is small relative to the interknee distance in those with genu varum and larger in those with genu valgum. For genu varum and valgum and knee angle, subjects were divided into five equal size groups (quintiles), from low to high, and the mid-20% was used as the referent or comparison group. By definition, the midquintile of the observed distribution was considered “normal.”

The knee angle at full extension was measured from the side-view photograph and defined as the angle (R) formed by a line from the lateral malleolus to the head of the fibula and from the head of the fibula to the greater trochanter, as also shown in Figure 1. Angles less than 180° indicate knee flexion, while angles greater than 180° indicate knee hyperextension. Subjects were also divided into five equal size groups(quintiles), from low to high, for knee angle, and the mid-20% was used as the referent or comparison group.

The Q-angle was measured as the acute angle (Q) formed by a line from the tibial tuberosity through the midpoint of the patella and a line from the ASIS through the midpoint of the patella (3,8), as shown in Figure 1. This angle represents the degree of deviation of the patellar tendon from the line of pull by the quadriceps muscles on the patella. Excessive Q-angle has been described by different authors as being over 10° (8), over 15° (4), and over 20° (3). The Q-angle groups used in this report were Q ≤ 10°, 10° < Q ≤ 15°, and 15° > Q, and the overuse injury risk associated with each of these levels was assessed.

The absolute difference in lower limb lengths, from the ASIS to the medial malleolus, was determined (3). Brody(3) speculated that lower limb length differences of as little as 0.25 inch (0.64 cm) could result in pain or injury, so the lower limb length difference (LLLD) groups used in this report were LLLD ≤ 0.5 cm, 0.5 cm < LLLD ≤ 1.0 cm, 1.0 cm < LLLD ≤ 1.5 cm, and 1.5 cm> LLLD.

Back to Top | Article Outline


All injuries resulting in a sick-call visit were identified and recorded. This was accomplished by a complete (100%) medical records review by a physician (BHJ) of every record of all 294 participants, performed at 2-3 wk intervals. The diagnoses of injuries were made by clinic and hospital medical personnel not associated with the study. Medical staff were aware that a study was under way but had no knowledge of which soldiers were participating in the study. Overuse injuries were defined as conditions suspected of being caused by repetitive microtrauma associated with activities such as running and marching (6). These overuse injuries occur when an“excess of injury exists in exercised tissue”(19) and included specifically diagnosed conditions, such as stress fractures, tendonitis, bursitis, fasciitis, and patello-femoral syndrome, as well as nonacute muscle strains (also called muscle overuse syndrome) and pain that the diagnosing clinicians felt were due to“overuse” or “stress.” Both overall risk of overuse injury and risk of specific types of overuse injury were examined in this report. The anatomic sites of overuse injuries were also noted.

Back to Top | Article Outline


The cumulative incidence of injury (number of events divided by the population at risk) was calculated and expressed as a percentage. Under the conditions of short follow-up time, cumulative incidence is equivalent to risk(25), and the terms are used interchangeably in this report. Risks (cumulative incidences) for any overuse injury, for specific injuries, and for body part injured were calculated. Because an individual could have multiple types of injury as well as multiple body parts injured, the sum of specific injury types and locations was different than the overall risk of overuse injury.

Relative risk was the measure of association comparing the injury risks for different levels or degrees of each morphologic characteristic and injury and is defined as the cumulative incidence of injury in an exposed group divided by the incidence in the referent group. Exact P-values of the relative risks, calculated with chi-square, were reported with 95% confidence intervals about the estimate. Because this was a pilot study with fewer than 300 subjects, we were concerned that we might not identify associations between anatomic characteristics and risks of injury that might be fruitful areas to be examined in larger populations. In order to reduce the likelihood of missing an important association, we calculated the exact P-value of the point estimates in addition to the 95% confidence intervals(26). We interpreted P-values between 0.05 and 0.1 as being suggestive of associations deserving further study.

Chi-square tests and confidence intervals for relative risks were calculated using Epi Info statistical software (Dean, A. D., et al. Epi Info, Version 5. USD, Inc., Stone Mountain, GA.). Exact P-values for comparisons involving small expected numbers or empty cells were calculated with Fisher's exact test using the Epi Info software.

Tests for linear trends in association between selected categories of lower limb measures and risk of injury were conducted with Mantel-Haenszel chi-squares (18), also calculated with Epi Info. If the data indicated there might be some association between levels of measurement and risk of injury but tests of trend were not significant, the data were further explored using partitioned chi-square(1,27). This was done by combining the data in the two lowest strata and comparing them with the combined data from the two highest strata.

Back to Top | Article Outline


The descriptive characteristics of the participants and statistics for pelvic-patellar ratio, knee angle, Q-angle, and lower limb length differences are presented in Table 1. A subject with genu valgum is shown in Figure 2, while a subject with genu varum is shown in Figure 3. A subject from the highest quintile of knee angle at full extension is presented in Figure 4. Excessive Q-angle and lower limb length differences cannot be clearly seen on the photographs, so no figures demonstrating those conditions are presented.

The overall risk of lower limb overuse injury was 30% (involving 88 individual subjects), which was 83% of all 106 lower extremity injuries. Five percent of participants (15 subjects) were treated for knee injuries attributed to overuse (patello-femoral syndrome, chondromalacia, and pain not otherwise specified), 5% (15 subjects) were diagnosed with stress fractures, and another 5% (15 subjects) were assessed as having overuse muscle strains. Nineteen percent of participants (56 subjects) were treated for nonspecific musculoskeletal pain felt to be due to overuse. Other types of overuse injuries also occurred but were too few in number to evaluate. The specific body part injured was available for 86 subjects. The most common sites injured were the foot (9.9% [29 subjects]) and the knee (8.2% [24 subjects]). The ankle and calf were each injured by 13 subjects (4.4%). Further details on the types and locations of injury, both traumatic and overuse, as well as other risk factors identified, have been presented elsewhere(5,15,16). Most injuries resulted in a few(1-3) days of limited duty. More serious injuries, particularly stress fractures, resulted in several weeks to several months of limited duty. There were no hospitalizations due to overuse injuries.

The cumulative incidence of overuse injury (percent) and relative risks for all measures are presented in Table 2. For quintile 1, those defined as having genu valgum, the risk of any overuse injury was 41%(24 subjects), compared with 22% (13 subjects) for those with“normal” knees in the midquintile, for a relative risk of 1.9(P = 0.024). The risk for nonspecific musculoskeletal pain among the most valgus was 26% (30 subjects), compared to 14% (8 subjects) for the midquintile, for a relative risk of 1.9 (P = 0.094). There was a significant linear trend for increasing risk of stress fracture as knees ranged from most varus to most valgus (P = 0.033), with risks for each quintile of 2% (one subject), 2% (one subject), 5% (three subjects), 7%(four subjects), and 9% (five subjects), respectively. None of the specific sites of injury were significantly associated with any quintiles of pelvic-patellar ratio.

For overall risk of overuse injury, there was no significant trend evident across quintiles of knee angle. Seven percent (eight of 117 subjects) of those in the highest two quintiles of knee extension sustained overuse knee injuries, compared to 0% (of 117 subjects) for those in the two lowest quintiles (knee flexed), resulting in a significantly elevated but undefined relative risk (P = 0.003, Fisher's exact test). None of the specific sites of injury were significantly associated with knee angle.

For any overuse injury, those subjects with Q-angles in excess of 15° had a moderate but not statistically significant relative risk of 1.5 (40% [17 subjects] vs 27% [41 subjects]; P = 0.096) compared to those with Q-angle of ≤ 10°; the relative risk was 2.88 (10% [4 subjects] vs 3% [5 subjects]; P = 0.091) for muscle strain; 1.73 (31% [13 subjects] vs 18% [27 subjects]; P = 0.065) for musculoskeletal pain; and a highly significant 5.39 (14% [6 subjects] vs 3% [5 subjects]; P = 0.008) for stress fracture. Those in the highest Q-angle group had a higher risk of injury to the calf, with a relative risk of 4.29 (14% [6 subjects] vs 3% [5 subjects]; P = 0.008), but not for any other site of injury.

Because there was concern that the measures of pelvic-patellar ratio and Q-angle were correlated, the association between them was assessed. There was a low (0.24) but statistically significant (P < 0.01) Spearman coefficient of correlation, indicating that one measure is of little value in predicting the other. Thus, it can be assumed that these two measures are evaluating two different morphologic characteristics.

None of the levels of lower limb length differences had elevated overall risk of injury, and there was no significant linear trend for increasing risk with increasing length difference. The relative risk for musculoskeletal pain among those with a length difference of greater than 1.5 cm was 2.03 (43% [6 subjects] vs 20% [30 subjects]; P = 0.071) compared to those with a difference of 0.5 cm or less. None of the specific sites of injury were significantly associated with differences in lower limb length. The range of leg length discrepancies was small and probably not sufficient to cause pathology.

Back to Top | Article Outline


The key finding of this study is that anatomic variants of the lower extremity (genu valgum, high Q-angle, excessive knee angle) are associated with elevated risk of some types of overuse injury among young men undergoing initial entry Army infantry training. Army recruits with genu valgum experienced more overuse injuries, and those with excessive Q-angle had higher risk of stress fractures. The 40% of the population with the greatest knee extension sustained more overuse knee injuries than the 40% with the greatest degree of flexion. Suggestive associations (P-values from 0.05 to 0.1) were observed between high Q-angle and any overuse injury and between valgus knees, excessive Q-angle, and lower limb length differences and musculoskeletal pain due to overuse. These potentially etiologic associations deserve evaluation in studies of other, larger populations.

Knowledge of anatomic and other risk factors for overuse injuries, such as those evaluated in this study, is important to the Army because these injuries are the major source of morbidity and time loss among military trainees(13). According to Franz (6),“pathologically, overuse syndromes in runners start as microtrauma to localized tissue areas that are under biomechanical stress. As inflammation develops faster than a tissue's capacity to repair, macroscopic irritation and damage culminate in tissue disruption.” Thus, overuse injuries result from repetitive submaximal loading on bone, cartilage, ligament, tendon, and muscle that exceeds the ability of the loaded tissue to recover(2). Army infantry training exposes recruits to an extremely demanding physical environment that frequently far exceeds their prior level of physical activity. Specifically, they walk, march, and run much further distances than those to which their musculoskeletal systems are accustomed. Previous studies have shown that low physical fitness(14,15) and excessive running mileage(16) are risk factors for injuries among basic trainees. It has also been shown that high arches (5) and both very high and very low hamstring and low back flexibility (15) place Army trainees at greater risk of injury. The effect of other anatomic variants on the risk of injury among military trainees has not been evaluated.

The civilian sports medicine literature suggests that the extremes of anatomic variation and malalignment of the lower extremities predispose runners and athletes to injury (6,19,23). On the basis of case series and clinical experience a number of anatomic factors have been hypothesized to be associated with increased risk of injury among civilian athletes and exercise participants(3,4,6,8,12,19,20,23). For example, in their review of 180 patients with injuries, James et al.(12) stated that while “no specific anatomic factor correlates with any specific injury,” excessive Q-angle and genu varum are associated with pain. Clement et al. (4) examined 1,650 patients and speculated that several anatomic problems may be associated with injury risk (e.g., genu valgum, varum, and recurvatum, and excessive Q-angle) but did not specify associations with particular injuries. Brody(3) speculated that a leg length discrepancy of as little as 0.25 inch could cause back pain and that excessive Q-angle as well as genu valgum could cause overuse knee injury but provided no data or estimates of relative risks for these factors. Hoerner (8) stated that excessive Q-angle causes chondromalacia patella and that “bowlegged individuals... have a tendency to produce degenerative joint disease” but offered no risk estimates.

Just as with civilian sports and fitness program participants, there has been concern that the combination of intensive physical training and anatomic deformities might result in excessive risks of injury in some military trainees. This study was designed specifically to determine whether easily measurable anatomic variants of the lower extremity predisposed trainees to injury. The results of this study provide some of the first epidemiologic evidence that anatomic factors such as genu valgum and high Q-angles are associated with greater risk of overuse training-related injuries.

The results of this study are most generalizable to other populations of male Army trainees and other similarly fit, healthy young males, such as high school and college athletes. Because the most extreme degrees of anatomic variation are probably excluded from entry into the Army, we would expect broader ranges of anatomic variability in the general population. Therefore, the most appropriate cutpoints in measures of morphology and exact elevations in risk of injury for other populations cannot be determined from this study population. Also, while this study did not examine women, the sports medicine literature suggests that excessive Q-angle and genu valgum may predispose women to higher risks of exercise- and sports-related injuries(23). Putukian (23) suggests that further study of these and other and other similar anatomic characteristics of women will be important.

Future epidemiologic studies will require methods to quickly and accurately measure anatomic risk factors among large populations of both men and women. This report shows several significant associations between anatomy and risk and describes a number of simple but fruitful methods of assessing lower extremity anatomy suitable for rapidly evaluating large numbers of individuals. The methods are quantitative and easily reproducible so that comparisons with future research results are facilitated. If the findings of this study are confirmed by future research, the methods are simple enough that they may be of value for preseason athletic team screening or clinical evaluations for teams with similar training or activity patterns. The landmarks we utilized are easily identified and well accepted. Furthermore, the measurements made do not need to be digitized in order to be evaluated. Equipment commonly used in sports medicine and orthopedic clinics, such as tape measures, goniometers, and anthropometers, can be employed in place of the digitized slides we used.

Future work may consider dynamic as well as static measures and evaluate anatomy and biomechanical factors (e.g., patellar displacement) during load-bearing activities. However, careful consideration should be given to all aspects of such studies prior to their initiation, as dynamic measures will likely be more difficult, time consuming, and expensive to conduct and the findings more challenging to interpret. The potential number and types of risk factors to consider will be many times greater, requiring larger study populations. The combination of more complex measures requiring larger populations may inhibit the conduct of such studies unless well-formulated hypotheses are available to be tested. Data such as presented in this report provide a foundation for developing more complex studies of the dynamic biomechanical determinants of injury.

Back to Top | Article Outline


Consistent with the clinical sports medicine literature(3,4,6,8,12,19,20), this study identified several anatomic risk factors for overuse injuries associated with vigorous physical training. The risk factors were genu valgum, excessive Q-angle, and genu recurvatum. While these findings are most applicable to male military trainees, the generalities of the findings may also apply to other similarly active populations of young males. The exact degrees of anatomic deformity associated with elevated levels of injury risk will depend on the amount and intensity of training as well as the anatomic and fitness characteristics of the population of interest. The anatomic factors identified here deserve further study in larger military and civilian populations that include both men and women.

Figure 1-Leg morphology measures: pelvic width (H), knee width (K), Q-angle (Q), and knee extension/flexion angle. Pelvic-patellar ratio = H·K-1

Figure 1-Leg morphology measures: pelvic width (H), knee width (K), Q-angle (Q), and knee extension/flexion angle. Pelvic-patellar ratio = H·K-1



Figure 2-Genu valgum (“knock knee”).

Figure 2-Genu valgum (“knock knee”).

Figure 3-Genu varum (“bow legs”).

Figure 3-Genu varum (“bow legs”).

Figure 4-Genu recurvatum (“back knee,” or hyperextension of the knee).

Figure 4-Genu recurvatum (“back knee,” or hyperextension of the knee).



Back to Top | Article Outline


1. Agresti, A. Categorical Data Analysis. New York: John Wiley & Sons, 1990, 50-54.
2. American Academy of Orthopaedic Surgeons.Orthopaedic Knowledge Update 4, J. W. Frymoyer (Ed.). Rosemont, IL: American Academy of Orthopaedic Surgeons, 1993, pp. 91-94.
3. Brody, D. M. Running injuries. Clin. Symp. 32(4):1-36, 1980.
4. Clement, D. B., J. E. Taunton, G. W. Smart, and K. L. McNicol. A survey of overuse running injuries. Phys. Sports Med. 9(5):47-58, 1981.
5. Cowan, D. N., B. H. Jones, and J. R. Robinson. Foot morphology and risk of exercise related injury. Arch. Fam. Med. 2:773-777, 1993.
6. Franz, W. B. Overuse syndromes in runners. In:Office Sports Medicine, Chap. 24, M. B. Mellion (Ed.). Philadelphia: Hanley & Belfus, Inc., 1996, pp. 298-317.
7. Gardner, L. I., J. E. Dziados, B. H. Jones, et al. Prevention of lower extremity stress fractures: a controlled trial of a shock absorbent insole. Am. J. Public Health. 78:1563-1567, 1988.
8. Hoerner, E. F. Injuries of the lower extremities. In:Sports Injuries: The Unthwarted Epidemic, P. F. Vinger and E.F. Hoerner (Eds.). Littleton, MA: PSG Publishing Company, Inc., 1986, pp. 235-249.
9. Hoppenfeld, S. Physical examination of the hip and pelvis. In: Physical Examination of the Spine and Extremities. Norwalk, CT: Appleton-Century-Crofts, 1976, pp. 143-169.
10. Hoppenfeld, S. Physical examination of the knee. In:Physical Examination of the Spine and Extremities. Norwalk, CT: Appleton-Century-Crofts, 1976, pp. 171-196.
11. Hoppenfeld, S. Physical examination of the foot and ankle. In: Physical Examination of the Spine and Extremities. Norwalk, CT: Appleton-Century-Crofts, 1976, pp. 197-235.
12. James, S. L., B. T. Bates, and L. R. Osternig. Injuries to runners. Am. J. Sports Med. 6:40-50, 1978.
13. Jones, B., R. Manikowski, J. Harris, et al. Incidence and risk factors for injury and illness among male and female Army basic trainees. Technical Report T-19-88. Natick, MA: US Army Research Institute of Environmental Medicine, 1988.
14. Jones, B. H., M. W. Bovee, J. M. Harris, et al. Intrinsic risk factors for exercise-related injuries among male and female Army trainees. Am. J. Sports Med. 21:705-710, 1993.
15. Jones, B. H., D. N. Cowan, J. P. Tomlinson, J. R. Robinson, D. W. Polly, and P. Frykman. The epidemiology of injuries associated with physical training among young men in the Army. Med. Sci. Sports Exerc. 25:197-203, 1993.
16. Jones, B. H., D. N. Cowan, and J. J. Knapik. Exercise, training, and injuries. Sports Med. 18:202-214, 1994.
17. Koplan, J. P., D. S. Siscovick, and G. M. Goldbaum. The risks of exercise: a public health view of injuries and hazards. Public Health Rep. 100(2):189-195, 1985.
18. Mantel, N. and W. Haenszel. Statistical aspects of the analysis of data from retrospective studies of disease. J. Natl. Cancer Inst. 22:719-748, 1959.
19. Mellion, M. B. Medical syndromes unique to athletes. In:Office Sports Medicine, Chap. 15, M. B. Mellion (Ed.). Philadelphia: Hanley & Belfus, Inc., 1996, pp. 150-174.
20. Micheli, L. J. Lower extremity overuse injuries.Acta Med. Scand. Suppl. 711:171-177, 1986.
21. Peterson, L. and P. Renstrom. Sports Injuries: Their Prevention and Treatment. Chicago: Year Book Medical Publishers, Inc., 1986, p. 41.
    22. Powell, K. E., H. W. Kohl, C. J. Caspersen, and S. N. Blair, An epidemiological perspective on the causes of running injuries.Phys. Sports. Med. 14(6):100-114, 1986.
    23. Putukian, M. The athletic woman. In: Office Sports Medicine, Chap. 8, M. B. Mellion (Ed.). Philadelphia: Hanley & Belfus, Inc., 1996, pp. 81-101.
    24. Reynolds, K. L., H. A. Heckel, C. E. Witt, et al. Cigarette smoking, physical fitness, and injuries in infantry soldiers.Am. J. Prev. Med. 10:145-150, 1994.
    25. Rothman, K. J. Measures of effect. In: Modern Epidemiology, Chap. 4. Boston/Toronto: Little, Brown and Company, 1986, pp. 35-40.
    26. Selvin, S. The analysis of contingency table data: logistic model I. Statistical Analysis of Epidemiologic Data. Chap. 7. Monographs in Epidemiology and Biostatistics. New York: Oxford University Press, 1991, pp. 177-178.
    27. Volicer, B. J. Advanced Statistical Methods with Nursing Research Applications. Bedford, MA: MERESTAT, 1981, pp. 75-78.


    ©1996The American College of Sports Medicine