Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Body Composition and Physical Performance 1992. Pp. 141-173. Washington, D.C. National Academy Press 9 Associations Among Body Composition, Physical Fitness, and Injury in Men and Women Army Trainees Bruce H. [ones, Matthew W. Bovee, and Joseph ]. Knapik INTRODUCTION Policies regulating the body composition of men and women in the military service are a matter of ongoing interest to the U.S. Army. Body composition is considered to be a component of a soldier's physical fitness, and in the Army's view, obesity is associated with being unfit and "unsol- dierly." This association is important because physical fitness is an essen- tial component of military readiness for combat. To be prepared for its combat mission, the Army attempts to select individuals with the fitness and stamina to endure the rigors of Army training and combat. Simply selecting fit men and women is not adequate, however, because physical training is necessary to both develop and maintain the fitness of soldiers. For physical training to be effective, however, it must overload cardiovas- cular and musculoskeletal systems. This overloading entails a risk of mus- culoskeletal injury. Thus an understanding of the interactions among body composition, physical fitness, training, and injuries is an essential founda- tion for policies governing both body composition and physical fitness. In the following background material, the links between body composi- tion and physical fitness made in Army regulations and policy will be re- viewed, and components of physical fitness deemed to be essential to the Army's mission will be enumerated. The assumption that body composition reflects an individual's physical fitness will be explored. Also, the interac 141
42 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK lions between fitness, training, and injuries will be examined. Following this background material, the results of two recent Army studies will be presented. Individuals with a wide range in body fat (BF) volunteer to join the Army annually, but not all are accepted for service. Because the Army is concerned about the fitness of enlisters, some volunteers are medically dis- qualified solely on the basis of their height and weight before entering the service, the presumption being that they are not physically fit enough to be enlisted (Fried!, chapter 3~. Army standards for medical fitness are set forth in Army Regulation (AR) 40-501 (1987), which contains tables of accept- able heights and weights for enlistment in the Army. On the basis of these height-weight standards, only about 5 percent of eligible men in the United States would be excluded from service in the Army. In contrast, over 30 percent of otherwise eligible women would be excluded (Friedl et al. 1989~. In addition to height-weight standards, the Army also has a weight control program that defines acceptable percentages of BE for individuals who fail to meet the height-weight standards after enlistment (Army Regu- lation 600-9; U.S. Army, 1986~. The two primary stated purposes of the Army weight control program are to ensure that soldiers are adequately physically fit to accomplish their combat mission and that they present "a trim military appearance." Because of its demand for physically fit soldiers, the Army has a pro- gram of physical fitness, which is defined in Army Regulation 350-15 (AR 350-15; U.S. Army, 1989~. The objectives of this program are to enhance combat readiness by developing and maintaining high levels of physical fitness in all soldiers as measured by cardiorespiratory and muscular endur- ance, muscle strength, flexibility, anaerobic performance, competitive spir- it, self-discipline, and BF composition. The emphasis on physical fitness in both the selection and retention process seems appropriate because soldiers must have enough stamina and strength to perform a wide variety of physi- cally demanding tasks such as marching with loads, digging fox holes, scaling walls, and loading artillery shells. Physical fitness and appropriate body composition are achieved and maintained through physical training. The Army's program of physical fitness training is described in Army Field Manual 21-20 (FM 21-20; U.S. Department of the Army, 1985~. The manual lists five components of fitness the program strives to develop: · cardiorespiratory endurance; · strength; · muscle endurance; · flexibility; and · body composition, which includes lean body mass and fat mass and which is affected by the other components of fitness.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 143 Army doctrine links physical fitness, weight control, body composition, and physical training. Regulations regarding physical fitness and training in- dicate that body composition is simply a subcomponent of fitness (U.S. De- partment of the Army, 1985, 1989~. Even the weight control regulation (AR 600-9) states that its primary objective is physical fitness. For this reason, before policies on the issue are decided, it is important to assess the degree to which body composition influences the other components of physical fitness. Injuries are another important consideration. Soldiers disabled by inju- ry are less able to perform their regular duties even if they are otherwise highly physically fit. In a sense, injury-prone soldiers are less physically fit than those who are able to continuously perform their duties. Training- and activity-related injuries are a matter of concern to the military not only because they limit the function of individual soldiers, but because they impinge on the combat readiness of entire units when their incidence is even moderately high. Existing data indicate that the incidence of training-related injuries is high especially during basic training (Bensel and Kish, 1983; Cowan et al., 1988; Jones et al., 1988; Kowal, 19801. One report (Tomlinson et al., 1987) indicates that training-related injury rates are high among active duty sol- diers as well. The majority of these injuries are overuse conditions of the lower extremities, which arise directly from Army training or sports activi- ties that the Army encourages (Jones, 1983; Tomlinson et al., 19871. His- torical data also indicate that musculoskeletal injuries similar to those seen in training are a common cause of morbidity even during wartime (Reister, 1975~. In exploring body composition as an indicator of fitness, it is important to examine the relationship of body composition not only to components of fitness listed in the Army fitness and training documents but also to injury. Scientific literature on the interrelationships among body composition, physical fitness, training, and injury will be explored next. In these studies body composition is measured by either percent BF or body mass index (BMI). It is well accepted by both military (Jette et al., 1990; Vogel and Friedl, chapter 6) and civilian (Buskirk and Taylor, 1957; Cureton et al., 1979; Miller and Blyth, 1955) authorities that increased BF is associated with decreased weight-bearing endurance performance. Also, performance of other physical activities and exercises are negatively affected by higher levels of BF (Cureton et al., 1979; Jette et al., 1990; Vogel and Friedl, chapter 61. Despite their significance, the correlations between percent BF or BMI and other measures of physical fitness are low. Body composition explains only 5 to 30 percent of the variance in endurance performance measured by maximum oxygen uptake or timed run distance and even less of other factors, such as sit-ups, push-ups, or vertical jumps (Cureton et al., 1979; Jette et al., 1990; Vogel and Friedl, in press).
144 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK It is also well established that there is a dose response relationship between increased training volume and increased risk of injury (Koplan et al., 1985; Powell et al., 19861. Several studies have documented that higher amounts of training-especially higher running mileage are associated with higher injury rates (Blair et al., 1987; Koplan et al., 1982; Macera et al., 1989a,b; Marti et al., 1988; Pollock et al., 1977~. With the exception of volume of training, other risk factors for injury associated with physical training have not been clearly established. Physical fitness and body composition are suspected to affect the risks of injury during physical activity for civilians and military personnel (Bensel, 1976; Cowan et al., 1988; Jones, 1983; Koplan et al., 1985; Macera et al., 1989a), but the exact nature of that relationship has not been clearly estab- lished. Another possible risk factor of importance to the military that may be associated with both fitness and body composition is gender. During basic training the incidence of injury for women has consistently been high- er than that for men (Bensel and Kish, 1983; Kowal, 1980), but civilian studies have not identified gender as a risk factor (Koplan et al., 1982; Macera et al., 1989a,b). TWO ARMY STUDIES OF BODY COMPOSITION, FITNESS, AND INJURY Rational decisions regarding Army policy on fitness, fatness, and training are best made when based on data from military populations. As a founda- tion for decision making, this paper will examine data from two epidemio- logic studies of male and female Army trainees that were conducted by the U.S. Army Research Institute of Environmental Medicine and that provide further insight into the following areas: · the relationship of percent OF and BMI with physical fitness and their relative importance as predictors of physical fitness in male and fe- male Army trainees; · the degree of association of percent BE and BMI with risks of training related injuries in men and women; · the degree of association between physical fitness and risks of injury in men and women; · the degree of association between past physical activity or training and current risks of injury; · the relative importance of different parameters of body composition, physical fitness, and physical training (activity) on risks of injury using a multivariate model; and · the implications of the above determinations for screening, selecting, and training military personnel.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY Methods 145 The two studies described below were prospective follow-up studies of initial entry trainees. Both were conducted at Fort Jackson, South Carolina. One was completed in 1984 and the other in 1988, and both followed train- ees through the full course of the 8-week basic training cycle. Subjects In 1984 potential volunteers were all trainees coming to the Fort Jack- son reception station on one weekend. Ninety-nine percent volunteered to participate. In 1988 volunteers were solicited from all women being pro- cessed at the Fort Jackson reception station during 1 month. Male volun- teers were recruited from men destined to be assigned to companies in the same battalions as the female volunteers during the same 1-month period. The volunteer rate from the second group in 1988 was 92 percent in this group. The 1984 data are from a population of 310 trainees (124 men and 186 women). The 1988 data are from three training battalions, a total of 2,245 trainees (1,349 men and 896 women). Because not all trainees in either study were able to take all portions of the testing due to scheduling con- flicts or assignment to other duties, the number of subjects Was not identical in all portions of the analysis. Also, roughly 5 percent of men and 7 percent of women trainees were lost from follow-up due to discharge from the Army or transfer to another unit. Both studies were conducted in two phases: a prescreening phase, which consisted of a series of body composi- tion and physical fitness measures along with a questionnaire, and a follow- up phase, which included a medical records review. Both studies used a similar series of prescreening measures including height, weight, percent body fat (BF), body mass index (BMI), a health and fitness questionnaire, and Army physical fitness test results. Prescreening measures were made on all individuals over a period of 1 or 2 days, with the exception of physical fitness tests. Prior to screening, trainees were informed of the nature of the study. Those who volunteered signed a con- sent form, immediately after which they were screened and given the ques- tionnaire. Follow-up consisted of medical records review and documenta- tion of training. Prescreening Measures At Fort Jackson in 1984, BE was estimated from four skinfold measure- ments using the equations of Durnin and Womersley (1974~.~ For the 1988
46 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK study, circumference measures were used to estimate percent BE (separate sites and equations for men and women as specified in Army Regulation 600-9; U.S. Army, 19861. BMI for both men and women was calculated as weight divided by height squared. Physical fitness was assessed with the Army physical fitness test, which was taken within the first 3 days of the onset of basic training. Measures taken were 1-mile run or 2-mile run times and the number of sit-ups and push-ups performed in a 2-minute time period. Entire units (companies) ran either a 1- or a 2-mile "diagnostic fitness" test. At Fort Jackson in 1984, past physical activity and sports participation were assessed by a questionnaire delivered to groups of 50 or more recruits. Each question was read aloud by trained personnel. The primary question to assess physical activity level prior to entering the Army was: How active are you compared to others of your age and sex? Subjects were asked to rate their activity on a 4-point scale from inactive to very active. A similar question was validated by Washburn et al. (1987~. Total kcals of energy expended in leisure time recreational and sports activities per week were estimated from questionnaire data. Study partici- pants were asked to check activities they had done in the last year on a list of common activities. For each activity checked they were asked to list how many days per week on average they performed the activity and how many minutes per performance. The average number of performances per week was multiplied by the average number of minutes per performance a subject reported doing an activity in the last 6 months. The number of kcals per week was attained by multiplying minutes per week by an estimate of the average number of kcals expended in a specified activity per minute. All estimates were then summed for each individual. The question was modeled after the Minnesota Leisure Time Physical Activity questionnaire (Taylor, 1978~. The 1984 questionnaire also queried trainees about their prior athletic status (nonathlete = 1, recreational athlete = 2, nonschool team or intramu- ral athlete = 3, and varsity athlete = 4~. The usual energy intensity (kcals expended per minute) of the trainees' leisure time and sports activity was also estimated by the investigators and rated on a four point scale (1 = sedentary, 2 = low, 3 = moderate, and 4 = high). A more extensive ques- tionnaire was delivered at Fort Jackson in 1988; the analyzed results are not yet available. Medical Follow-up Medical follow-up was achieved by a periodic 100 percent medical record review of every chart of every study participant. In 1984 a single records review was conducted during the last week of training. In 1988
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 147 records were reviewed every 2 to 3 weeks. An injury was defined as a sick- call visit to a troop medical clinic for a musculoskeletal complaint that received an injury diagnosis by a medical caretaker, usually a physician's assistant or a physician. Physical Training Physical training was assessed by scrutinizing company training sched- ules and verbal reports from company cadre in 1984. In 1988 daily training logs were also used to document training. Physical training for companies of men or women trainees for a specific year was similar. In 1984 trainees ran and performed calisthenics 4 to 6 days per week. Men generally started running 1 mile and progressively increased the distance of runs by about 0.5 mile per run each week up to 3 miles per run. On occasion they may have run 4 or 5 miles at a time near the end of the training cycle. Women began running distances of 0.5 mile per run and progressed in distance 0.5 mile per week up to 3 miles per run. At Fort Jackson in 1988, trainees ran only 3 times per week; otherwise routine physical training was fairly similar to that in 1984. Both years, each company of trainees was required to complete a 5- to 10-mile road march while carrying a light load (20 to 25 pounds [lbsi) in the middle of the training cycle and another 8- to 12-mile march at the end of the cycle with a heavier load (40 to 45 lbs). Every company also conducted training and ran a time trial on an obstacle course and a confidence course. Analysis Pearson product-moment correlation coefficients were calculated to de- scribe the relationship between continuous variables such as percent BF, BMI, and physical fitness measurements (that is, run times and numbers of sit-ups and push-ups). Also, to determine whether endurance performance (run times) of trainees at different points along the spectrum of percent BF and BMI was different, trainees were divided into quintiles (five roughly equal-sized groups) on the basis of BF measures and weight-height ratios (BMI). The mean run times of men and women in successive groups by percent BF or BMI were compared to each other for significance using a one-way analysis of variance (ANOVA). For significant ANOVAs, signif- icant between-group differences were identified using a least significant difference post hoc test. A stepwise multiple regression model was developed to predict mile run times for men and women from other physical measurements and ques- tionnaire data at Fort Jackson in 1984. Changes in R2 values from the stepwise regression output were interpreted as indicators of the amount of
148 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK variance in endurance performance explained by successive predictors step- ping into the equations. The point estimate of the significance of B coeffi- cients for each successive predictor variable are reported as p values in the text. The significance of F scores are also reported for the successive steps in the models for both men and women. Risks of injury were calculated as the cumulative incidence (percent) of trainees incurring one or more training-related injuries during the 8-week basic training cycle. Relative risks (percent injured in contrast group divid- ed by percent injured in referent group) were used to compare the incidence of injury in groups possessing different supposed risk factors or exposed to different levels of a risk factor. Significance of contrasted risks was tested with simple chi-square tests or partitioned chi-squares. Mantel-Haenszel chi-squares for linear trends were used when a trend was suspected on inspection of the data. To compare the risks of individuals exhibiting different levels of a continuous risk factor such as percent BE or mile run time, subjects were divided into successive quartiles (four roughly equal-sized groups) or quin- tiles (five nearly equal-sized groups) of risk based on their measured value of the variable of interest. For the 1984 data, trainees were placed into quartiles of risk based on continuous measured variables because the sam- ple size lacked power to demonstrate differences between smaller-sized groups. Trainees in the 1988 study were divided into quintiles to obtain a clearer picture of trends and because the sample size was large enough to support more divisions without sacrificing power. In both studies for all potential risk factors examined, a referent level of risk was chosen, and each other level was compared to it. Referent levels were usually the lowest level of risk observed or the level believed to have the lowest risk based on other knowledge. Relative risks were calcu- lated for each contrast. For the 1984 data, 90 percent confidence intervals are reported in the tables below because this was a hypothesis-generating study, and we did not want to fail to recognize a possible significant associ- ation due to lack of power secondary to a small sample size. For 1988 data, both 90 and 95 percent confidence intervals are reported in the tables be- low. Point estimates of significance (p values) are reported in the text when appropriate. To control for the influence of body composition and physical fitness on the risks of injury for women compared to men, Mantel-Haenszel chi- squares stratified on percent BF and mile run times, respectively, were performed. Finally, a stepwise logistic regression was also performed to determine the most important factors contributing to the risk of injury in a model where the effects of multiple factors were controlled for simulta neously.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY Results 149 The mean physical characteristics and physical fitness test results for men and women trainees in 1984 and 1988 are listed in Table 9-1. Compar- isons of the descriptive characteristics and fitness of men in 1984 and 1988 indicate that they were very similar in age, height, weight, percent BE, and BMI, but in 1988 they appeared to be slightly less fit; the same was true for women. Comparing men and women, the men were taller, heavier, and had higher BMIs than women in both 1984 and 1988, while women had higher percentages of BF. In both years, men ran faster and performed more push TABLE 9-1 Mean Descriptive Characteristics and Physical Fitness Test Results of Men and Women Army Trainees at Fort Jackson, South Carolina, in 1984 and 1988 Men Women Variable n Mean (SD) n Mean (SD) 1984 Age (years) 124 20.2 (2.7) 186 21.2 (3.6)* Height (cm) 123 175.2 (6.6) 186 163.3 (6.6) Weight (kg) 124 73.6 (10.9) 186 58.7 (5.8) Body mass index (weight/height2) 123 24.3 (3.1 ) 186 22.4 (2.0)* Body fat (%) 124 16.9 (4~9) 186 25.2 (9.4)* 1-Mile run (minutes) 79 7.2 (1.0) 140 9.7 (1.4)* Sit-ups (no.) 98 54.5 (13.8) 163 39.7 (11.9)* Push-ups (no.) 97 31.0 (9.3) 138 12.4 (9~9)* 1988 Age (years) 1,056 20.1 (3.3) 921 20.2 (3.5)* Height (cm) 1,053 175.2 (7.1) 895 162.0 (6.5) Weight (kg) 1,053 75.7 (12.2) 895 58.3 (6.5) Body mass index (weight~eight2) 1,053 24.6 (3.6) 895 22.2 (2.0) Body fat (%) 1,053 16.1 (5.8) 895 26.8 (3.0) 1-Mile run (minutes) 756 7.6 (0~9) 541 10.3 (1.8)* 2-Mile run (minutes) 593 16.4 (2.2) 355 20.3 (2.3)* Sit-ups (no.) 1,357 44.3 (12.2) 902 33.9 (13.8)* Push-ups (no.) 1,357 30.5 (12.9) 792 10.3 (7~3) *Difference between men and women significant at p < .05.
150 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK ups and sit-ups than women. Cutoff points for quartiles and quintiles of percent BF, BMI, and run times are listed in Table 9-2 for 1984 and Table 9-3 for 1988. Correlation of percent BF and Body Mass Index The correlation between percent BF by skinfolds and BMI among men trainees in 1984 was .81 (p < .000), and the correlation between BF by circumferences and BMI in 1988 was .84 (p < .000~. For women trainees in 1984, the correlation between body fat by skinfolds and BMI was .64, while in 1988 the correlation of body fat by circumferences and BMI was .86. TABLE 9-2 Body Composition and Fitness Variable Medians, Quartile Cutoff Points, and Ranges for Men and Women Army Trainees at Fort Jackson, South Carolina, 1984 * Variable Median Quartile Cutoff Point Range Men Percent body fat 16.6 Q1 13.1 7-29 Q3 20.6 Body mass index (kg/m2) 23.7 Q1 22.1 19-31 Q3 26.5 1-Mile run (minutes) 7.0 Q1 6.4 5.9-11.5 Q3 7.7 Sit-ups 52 Q 1 46.8 16-99 Q3 64.0 Push-ups 31 Q 1 26.5 4-53 Q3 36.0 Women Percent body fat 25.1 Q122.4 14-37 Q3.4 Body mass index (kg/m2) 22.5 Q121.1 18-27 Q323.6 1-Mile run (minutes) 9.8 Q19.0 6.0-16.3 Q310.4 Sit-ups 51 Q 130.0 6-66 Q346.0 Push-ups 11 Q15.0 Q317.0 Median = Q2.
151 TABLE 9-3 Body Composition and Fitness Variable Medians, Quintile Cutoff Points, and Ranges for Men and Women Army Trainees at Fort Jackson, South Carolina, 1988 Variable Median Quintile Cutoff Point Range Men Percent body fat 15.4 Q 1 10.98 2.13-36.12 Q2 14.00 Q3 17.38 Q4 21.50 Body mass index (kg/m2) 24.3 Q1 21.38 17.22-34.32 Q2 23.34 Q3 25.14 Q4 28.07 1-Mile run (minutes) 7.5 Q1 6.83 5.5-10.9 Q2 7.27 Q3 7.73 Q4 8.38 2-Mile run (minutes) 16.4 Q1 14.60 11.4-26.0 Q2 15.67 Q3 16.56 Q4 17.83 Sit-ups 45 Q 1 34 2-85 Q2 41 Q3 47 Q4 54 Push-ups 29 Q 1 19 1-87 Q2 26 Q3 32 Q4 40 Women Percent body fat 27.00 Q1 23.50 15.8~2.6 Q2 25.86 Q3 27.90 Q4 30.10 Body mass index (kg/m2) 22.4 Q1 20.27 16.36-27.20 Q2 21.79 Q3 22.97 Q4 24.11 1-Mile run (minutes) 10.0 Q1 8.94 5.6-19.3 Q2 9.72 Q3 10.41 Q4 11.50 continued on next page
152 TABLE 9-3 Continued BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK Variable Median Quintile Cutoff Point Range 2-Mile run (minutes) 20.4 Q1 18.50 13.9-29.8 Q2 19.71 Q3 20.81 Q4 21.98 Sit-ups 34 Q 1 23 1 -92 Q2 31 Q3 37 Q4 45 Push-ups 9 Q 1 4 1-52 Q2 7 Q3 11 Q4 16 Past Activity Level and Body Composition On entry to the Army in 1984, the least active men were also the fattest. For men trainees, a trend was observed of decreasing average percentage BF (assessed by skinfolds) with increasing self-reported activity level prior to entry. Percent BF decreased from 20.7 percent for the least active group to 19.5 percent for the average group, to 16.2 percent for the next, and 15.5 percent for the most active. The extreme groups (inactive versus very active) were significantly different (p < .05~. For women in 1984, however, there was no apparent association between activity levels prior to entry to the service and percent BF. The percent BF of women varied from 24.3 percent BF for the least active group, to 26.8 percent for the average groups, to 24.7 percent for the next group, to 23.9 percent for the most active group. Body Composition and Physical Performance Positive correlations between percent BF and mile run times and be- tween BMI and mile run times were observed among both men (Table 9-4) and women (Table 9-5) in 1984 and 1988. Results indicate that for both men and women, as percent BF and BMI increase, run times become slow- er. The magnitude of the correlations between body composition measures and endurance performance and their degree of significance was greater for men than women in both years. For men, roughly 7 to 28 percent of the variance in mile run times can be explained by percent BF, while only 1 to 3 percent of the variance among women's times can be explained on this basis.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY TABLE 9-4 Correlations of Percent Body Fat (BF) and Body Mass Index (BMI) with Entry Level Physical Fitness, and Correlations of Run Times With Sit-ups and Push-ups, Men Army Trainees, Fort Jackson, South Carolina Correlation r p n 1984 Percent BF with: 1-Mile run .27 .009 77 Sit-ups -. 1 7 .05 1 97 Push-ups -.1 1 .057 96 BMI with: 1 -Mile run . 18 .069 76 Sit-ups -.17 .336 96 Push-ups -.02 .407 95 1-Mile run time with: Sit-ups -.47 .000 77 Push-ups -.23 .000 76 1 988 Percent BF with: 1-Mile run .53 .000 525 2-Mile run .36 .000 376 Sit-ups -. 1 8 .000 907 Push-ups -.29 .000 912 BMI with: 1-Mile run .42 .000 525 2-Mile run .36 .000 376 Sit-ups -.19 .000 907 Push-ups -.04 .291 912 1-Mile run time with: Sit-ups -.4 1 .000 756 Push-ups -.29 .000 751 2-Mile run time with: Sit-ups -.3 1 .000 589 Push-ups -.30 .000 591 153 Negative correlations were noted between percent BF and numbers of sit-ups and push-ups. These data indicate that successively "fatter" men and women trainees on average perform fewer sit-ups and push-ups. A1- though the correlations between body fat and sit-ups and push-ups were significant for both men and women in both years, the magnitude of corre- lations were generally lower than for percent BF and run times. Although
154 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPlK TABLE 9-5 Correlations of Percent Body Fat (BF) and Body Mass Index (BMI) with Entry Level Physical Fitness, and Correlations of Run Times With Sit-ups and Push-ups, Women Army Trainees, Fort Jackson, South Carolina Correlation r p n 1984 Percent BF with: 1 -Mile run .12 .075 135 Sit-ups -.14 .035 158 Push-ups -.02 .410 133 BMI with: 1-Mile run .00 .478 135 Sit-ups -.02 .393 158 Push-ups -.10 .116 133 1-Mile run time with: Sit-ups -.24 .002 133 Push-ups -.01 .445 109 1988 Percent BF with: 1 -Mile run.16 .004 339 2-Mile run.12 .022 342 Sit-ups-.11 .000 605 Push-ups-.18 .003 686 BMI with: 1 -Mile run.13 .017 339 2-Mile run.09 .079 342 Sit-ups-.03 .405 686 Push-ups-.08 .041 606 1-Mile run time with: Sit-ups-.23 .000 536 Push-ups-.26 .000 467 2-Mile run time with: Sit-ups-.22 .000 353 Push-ups-.21 .000 314 the direction of correlations of BMI with sit-ups and push-ups was also negative, their magnitude was small (less than r = 0.21. Tables 9-6 and 9-7 show mean run times for men and women for differ- ent quintiles of BF and BMI. Mean run times increased significantly with successively higher levels of BF above the third quintile for men. The leanest men ran 1 mile in an average 7.1 minutes compared to 8.4 minutes
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 155 TABLE 9-6 Mean Run Times by Quintile of Percent Body Fat (BF) and Body Mass Index (BMI) for Men Army Trainees, Fort Jackson, South Carolina, 1988 Variable Mean SD Significantly different from n Q 1 Q2 Q3 Q4 I-Mile Run BF quintile Qllean 7.1 0.74 87 Q2 7.2 0.71 117 Q3 7.4 0.75 111 * Q4 7.8 0.84 108 * * * Q5 fat 8.4 0.93 102 * * * * 2-Mile Run BF quintile Q1 lean 15.8 1.74 100 Q2 15.6 1.85 70 Q3 15.8 1.51 71 Q4 17.1 2.16 72 * * * Q5 fat 17.9 2.53 63 * * ~* 1-Mile Run BMI quintile Q1 low 7.2 0.80 106 Q2 7.3 0.76 84 Q3 7.4 0.80 112 Q4 7.6 0.81 121 * * * Q5 high 8.4 0.99 102 * * * * 2-Mile Run BMI quintile Q1 low 15.6 1.81 78 Q2 15.8 1.60 94 Q3 16.3 2.03 74 * Q4 16.7 2.30 65 * * Q5 high 17.8 2.38 65 * * * * Significantly different at p < .05 by least significant difference. for the fattest quintile (p < .05). Among women, only the highest percent BF (fifth quintile) was significantly different from the others. Women with the lowest percentages of BF ran the mile in a mean time of 10.4 minutes compared to 11.2 minutes for the fattest (p < .05~. The relationship and trends in time for the 2-mile run versus quintile of percent BF for men and women, respectively, are similar to those seen for the 1 mile. Patterns of relationship of quintile of BMI and mean run times for men
156 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK TABLE 9-7 Mean Run Times by Quintile of Percent Body Fat (BF) and Body Mass Index (BMI) for Women Army Trainees, Fort Jackson, South Carolina, 1988 Significantly different from Q1 Q2 Q3 Q4 Variable Mean SD n 1-Mile Run BF quintile Q1 lean 10.4 1.84 73 Q2 10.2 1.60 74 Q3 10.5 1.81 59 Q4 10.6 1.91 63 QS fat 11.2 2.17 70 * * * * 2-Mile Run BF quintile Q1 lean 20.0 2.11 66 Q2 19.8 2.18 68 Q3 20.3 2.48 76 Q4 20.4 2.32 70 Q5 fat 20.9 2.09 62 * * I-Mile Run BMI quintile Q1 low 10.3 1.81 70 Q2 10.4 1.88 76 Q3 10.5 1.89 66 Q4 10.4 1.76 66 O5 high 11.2 2.06 61 * * * * 2-Mile Run BMI quintile Q1 low 20.3 2.40 62 Q2 19.7 2.12 60 Q3 20.0 2.42 74 Q4 20.6 2.47 74 * Q5 high 20.8 2.06 73 * * *Significantly different at p < .05 by least significant difference. and women were virtually identical to those for BF. Men and women in the lowest quintiles of BMI ran significantly faster than those in the highest quintiles. Prediction of Endurance Performance Because of the importance of aerobic fitness to the Army's mission, an exploratory model was devised to predict endurance performance (mile run
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 157 times) using data from the pilot study at Fort Jackson in 1984. Potential predictor variables for both men and women were: Age (years) Height (HT, cm) · Body fat (BF, %) Sit-ups (SU, number) Push-ups (PU, number) Total calories (TCAL, total kcals/week) · Activity level (ACT: 1 = inactive, 4 = very active) · Athletic status (ATHS: 1 = nonathlete, 4 = varsity) · Intensity (INT: 1 = sedentary, 4 = high) Predictor variables entered the model in stepwise fashion in order of impor- tance. The final predictive regression equation for men trainees using the above variables was: Mile time (minutes) =-.019(SU) + .055(BF)-.227(ACT) - .142(lN~ + 8.47. Sit-ups (p = .004) entered the equation first, explaining 20 percent of the variance in mile times. Percent BF ~ = .009) entered the equation next, which explains an additional 11 percent of the variance in times. Following percent BF, activity level (p = .04) stepped in, contributing another 4 per- cent to the explanatory power of the equation. The last predictor to enter the model was the intensity of the trainee's past recreational and sports activities (p = .13), accounting for another 2 percent of the variance in run times for men. The final multiple regression model explained 37 percent of the variance in endurance performance of men trainees as measured by run times (p < .000). Multiple regression coefficients (R) increased with each step from .45 to .56 to .60 to .62. All steps were significant at p < .009. For women the endurance performance predictive equation was: Mile time (minutes) =-.39(ACT)- .038(HT)- .018(SU) + .045(BF) - .22(ATH) + 16.9. The trainees' self-assessed activity level (p = .004) entered the model first, and explained 15 percent of the variance in mile times for women. Height stepped into the equation next, contributing an additional 9 percent to the explanatory ability of the model. Number of sit-ups (p = .07) followed height into the model, which boosted the explained variance 4 percent more. Percent BF ~ = .08) entered the model for women fourth, contributing 3 percent to the explained variance in run times. The last variable to enter the model was athletic status (p = .13), which accounted for 2 percent of the explained variance. The multiple regression model explained 33 percent of the apparent variance in the run times of women trainees (p < .000). The
158 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK multiple regression coefficients (R) for successive steps in the model increased from .39 to .49 to .53 to .56 to .57. All steps were significant at p < .0009. Injuries At Fort Jackson in 1984 over the course of the 8-week basic training cycle, significantly more women suffered training-related injuries than men: 50.5 percent of women compared to 27.4 percent of men, with a risk ratio of 1.84 (p < 0.000~. In 1988 during the 8 weeks of basic training, 43.5 percent of women were injured, and only 27.2 percent of men experienced an injury, a risk ratio of 1.61 (p < 0.000~. In 1984, over 90 percent of all musculoskeletal complaints for both men and women were due to lower extremity injuries, and in 1988 about 85 percent of the injuries of men and women were lower extremity training-related injuries, such as stress frac- tures, patellofemoral syndrome, achilles tendonitis, and ankle sprains. Body Composition and Injury Tables 9-8 and 9-9 display the risks of injury for men by quartile (1984) and quintile (1988) of percent BF. For men in both 1984 and 1988, a higher incidence of injury was evident among the fattest quartiles and quintiles of trainees. In 1988 the fattest three quintiles of men were at significantly greater risk than the leanest two, 25.3 versus 20.7 (risk ratio = 1.2, p = .05~. Tables 9-10 and 9-11 show the risks of injury by quartile (1984) and quin- tile (1988) of percent BF for women. There were no significant differences in risk to women by percent BF in 1984. In 1988, contrary to what was observed for men, the incidence of injury for the leanest two quintiles of women was greater than the third and fourth quintiles: 42.4 percent com- pared to 33.8 percent (risk ratio of 1.3, p = .05~. The risk of injury by quartiles or quintiles of BMI is shown for men in Tables 9-8 and 9-9 and for women in Tables 9-10 and 9-11. The relation- ship between BMI for men and women in 1984 appeared as if it might be bimodal, with both the lowest and highest quartiles at greater risk of injury than the middle groups. For men in 1984 the risk of injury for the lowest quartile of BMI was 35.5 percent versus 18.0 percent for the middle two quartiles, a risk ratio of 2.00 (p = .111. For the highest quartile versus the middle two, the risk ratio was 38.7 percent to 18.0 percent (risk ratio = 2.15, p = .031. For women in 1984 the pattern of association of BMI with risk of injury was similar to that of men. The risk for the lowest quartile of women was 55.6 percent compared to 38.3 percent for the third quartile (risk ratio = 1.45, p = .09~. The risk of the highest quartile of women trainees was 63.0 percent versus 38.3 percent for the third quartile, with a risk ratio of 1.65 (p = .02~.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 159 TABLE 9-8 Risk of Musculoskeletal Injury by Quartile of Percent Body Fat, Body Mass Index, and Mile Run Time for Men Army Trainees, Fort Jackson, South Carolina, 1984 Relative Risk Confidence Variable Risk (%) (versus baseline) n Interval (90%) Percent body fat Q1 lean 27.3 1.29 33 (0.62-2.65) Q2 26.7 1.26 30 (0.60-2.64) Q3 21.2* 1.00 33 Q4 fat 35.7 1.68 28 (0.84-3.36) Total 123 Body mass index Q1 low 35.5 2.06 31 (0.94~.48) Q2 18.8 1.10 32 (0.44-2.68) Q3 17.2* 1.00 29 Q4 high 38.7 2.25 31 (1.04~.83)t Total 124 Run time Q1 fast 14.3 1.43 21 (0.35-5.86) Q2 10.0* 1.00 20 - Q3 26.3 2.63 19 (0.74-9.30) Q4 slow 42.1 4.21 19 (1.28-13.83)t Total 79 *Referent level (denominator for risk ratio). tp < .1. This bimodal relationship of BMI with injury risk was not clearly evi- dent at Fort Jackson in 1988. Although the extremes of the distribution of BMI did tend to have a higher incidence of injury than middle quintiles (see Table 9-9 for men and Table 9-11 for women), no significant difference in risk between quintiles of BMI was identified among men trainees. For women in 1988 the extreme quintiles of BMI were at the greatest apparent risk, but only the lowest two quintiles were in significantly greater jeopar- dy, 42.5 percent risk, compared to the fourth quintile, which had the lowest risk: 30.7 percent (risk ratio = 1.3, p = .011. Physical Fitness and Injury The relationship between physical fitness and injury is more pronounced and more directional than that for body composition and injury. A signifi- cant association between low aerobic fitness (endurance) as measured by mile run times and elevated risk of injury for both men (p for trend = .02)
60 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK TABLE 9-9 Risk of Musculoskeletal Injury by Quintile of Percent Body Fat, Body Mass Index, and Run Time for Men Army Trainees, Fort Jackson, South Carolina, 1988 Relative Risk Confidence Variable Risk (%) (versus baseline) n Interval (90~0/95%) Percent body fat Q1 lean 22.4 1.18210 (0.86-1.62/0.81-1.72) Q2 19.0* 1.00211 - Q3 25.6 1.35211 (1.00-1.82/0.94-1.93)t Q4 23.2 1.23211 (0.90-1.67/0.85-1.77) Q5 fat 27.6 1.46210 (1.09-1.96/1.03-2.07)t Total 1,053 Body mass index Qllow 23.3 1.12210 (0.83-1.51/0.78-1.60) Q2 25.6 1.23211 (0.92-1.64/0.87-1.74) Q3 20.9* 1.00211 Q4 23.2 1.11211 (0.82-1.51/0.78-1.60) QS high 24.8 1.19210 (0.88-1.60/0.83-1.69) Total 1,053 Run time Q1 fast 23.5 1.03 277 (0.80-1.33/0.76-1.40) Q2 22.8* 1.00 268 - Q3 29.2 1.28 267 (1.01-1.63/0.96-1.71)t Q4 26.7 1.17 270 (0.91-1.50/0.87-1.57) QS slow 34.1 1.50 267 (1.19-1.88/1.14-1.97)§ Total 1,349 *Referent level (denominator for risk ratio). tp < .05. up ~ .1. §p ~ .01. and women (p for trend = .03) was seen in 1984 (see Table 9-8 for men and Table 9-10 for women). The slowest two quartiles of men had a higher risk of injury than the fastest two: 34 percent versus 12 percent, a risk ratio of 2.8 (p = .03~. For women trainees, a similarly significant association was observed between mile times and risk of injury. The slowest two quartiles of women had a higher risk of injury, 59 percent, versus 35 percent for the fastest two quartiles (risk ratio = 1.7, p = .01~. At Fort Jackson in 1988, a trend similar to those observed in 1984 was noted between run times and risk of injury among men trainees (see Table 9-9 for men and Table 9-11 for women). The slowest three quintiles of men trainees had a combined average incidence of injury of 30 percent compared to 23.1 percent for the fastest two, a risk ratio of 1.3 (p = .005~. A signifi
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY TABLE 9-10 Risk of Musculoskeletal Injury by Quartile of Percent Body Fat, Body Mass Index, and Mile Run Time for Women Army Trainees, Fort Jackson, South Carolina, 1984 161 Relative Risk Confidence Variable Risk (%) (versus baseline) n Interval (90%) Percent body fat Q1 lean 41.3 0.78 46 (0.54-1.12) Q2 61.7 1.16 46 (0.86-1.56) Q3 53.2* 1.00 47 Q4 fat 45.7 0.86 46 (0.61-1.21) Total 186 Body mass index Q1 low 55.6 1.45 45 (1.00-2.11)t Q2 45.8 1.20 48 (0.80-1.78) Q3 38.3* 1.00 47 Q4 high 63.0 1.64 46 (1.15-2.35)t Total 186 Run time Q1 fast 36.3 1.08 36 (0.64-1.84) Q2 33.3* 1.00 36 Q3 57.1 1.71 35 (1.09-2.71)t Q4 slow 60.6 1.82 33 (1.16-2.86)t Total 140 * Referent level (denominator for risk ratio). tp < .1. cant linear trend between slower run time and higher risk of injury was also identified (p =.003~. The association between run times and injury risk was not so distinct for women in 1988. The risk ratio of the slowest two quintiles was contrasted with the fastest three, yielding a risk ratio of 1.2 (p = .02~. When the 1988 run time data for men trainees and lower extremity injuries only was examined for associations, an even more pronounced trend of increasing injury risk with slower run time was observed (Mantel-Haens- zel chi-square for trend, p = .0006~. Risks for men descended from 30.3 percent for the slowest quintile to 23.7 percent, to 24.7 percent, to 18.3 percent, and slightly up to 19.3 percent. There was also a stronger relation- ship between quintiles of run time and risk of injury for women trainees in 1988 when only lower extremity injuries were analyzed (Mantel-Haenszel chi-square for trend, p = .081. Risks for women for successive quintiles of run time for slowest to fastest went from 38.5 percent to 48.6 percent, then down to 37.9 percent for two quintiles and down to 34.1 percent.
162 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK TABLE 9-11 Risk of Musculoskeletal Injury by Quintile of Percent Body Fat, Body Mass Index, and Run Time for Women Army Trainees, Fort Jackson, South Carolina, 1988 Relative Risk Confidence Variable Risk (%) (versus baseline) n Interval (90%/95%) Percent body fat Q1 lean 44.8 1.32 181 (1.06-1.64/1.02-1.71)t Q2 39.8 1.16 176 (0.92-1.48/0.88-1.52) Q3 33.9* 1.00 183 - Q4 34.1 1.01 176 (0.79-1.28/0.75-1.34) Q5 fat 39.0 1.15 177 (0.91-1.45/0.88-1.51) Total 893 Body mass index Q1 low 43.6 1.42 179 (1.13-1.78/1.08-1.86)t Q2 41.3 1.35 179 (1.07-1.70/1.02-1.78)t Q3 38.0 1.24 179 (0.97-1.57/0.93-1.65) Q4 30.7* 1.00 179 Q5 high 37.4 1.22 179 (0.96-1.55/0.91-1.63) Total 895 Run time Q1 fast 41.7 1.08 180 (0.87-1.33/0.84-1.39) Q2 41.6 1.06 178 (0.86-1.31/0.83-1.37) Ok Q3 38.7- 1.00 181 Q4 53.6 1.39 179 (1.15-1.68/1.11-1.74)t Q5 slow 43.3 1.12 178 (0.91-1.38/0.87-1.43) Total 896 * Referent level (denominator for risk ratio). tp < .05. tp < .01. Physical Activity and injury in 1984 a stepwise trend of decreasing risk with increasing activity level was evident for men (Table 9-12~. Risks decreased from 43 percent for the least active group to 17 percent for the most active (p for trend = .06~. Comparing the risks of the average and inactive groups with the active and very active groups, the risk ratio is 1.6 (36.4 percent/22.5 per- cent, p = .091. For women there did not appear to be an association between physical activity and risk of training-related injuries (Table 9-121. Gender, Physical Fitness and Risk of injury The crude relative risk of injury for women compared to men at Fort Jackson in 1984 was 1.8 (50.5 percent/27.2 percent, p < .000~. When risks
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY TABLE 9-12 Association of Self-Assessed Activity Level of Men and Women Trainees Prior to Entering the Army with Risk of Injury, Fort Jackson, South Carolina, 1984 163 Relative Risk Confidence Activity Level Risk (%) (versus baseline) n Interval (90%) Men Very active 17.2* 1.00 29 Active 25.5 1.48 51 (0.68-3.21) Average 35.1 2.04 37 (0.95~37) Not very active 42.9 2.49 7 (0.93-6.63) Total 124 Women Very active 48.5* 1.00 33 Active 52.2 1.08 69 (0.84-1.31) Average 48.4 1.00 64 (0.56-0.96) Not very active 55.0 1.14 20 (0.57-1.21) Total 186 *Referent level (denominator for risk ratio). for women versus men were stratified by level of fitness (mile run times) so that women were compared to men of the same degree of fitness, there was no difference in risk between genders, and the overall risk ratio was .98 (Mantel-Haenszel chi-square = 0.00, p = 1.00; Table 9-13~. When risks were stratified on percent BF, the risk ratio remained unchanged at 1.8, which indicates that BF did not affect the risk of injury. Stratification on several other factors, including age, race, sit-ups, and push-ups, did not affect the magnitude of the risk ratio. Two logistic regression models were also created to determine the im- portance of various risk factors for injury. The variables included in the first model were: gender, age, race, athletic status, self-assessed activity level, height, weight, percent BF, push-ups, and sit-ups. In this regression, without fitness/run time included, the only factor that stepped into the mod- el was gender, with an estimated odds ratio for women versus men of 2.5 (p = .005~. The second model created was identical except that mile run time was included as a variable. In this second model, gender did not approach the significance required to enter the model, but mile run time did, with an estimated odds ratio for slow versus fast of 3.5 (p = .0011. Percent BF did not approach the required Fs for entry into either model, nor did any other variable. Again when a measure of physical fitness was a candi- date for entry into the stepwise model, gender differences disappeared, and
64 BRUCE H. JONES, MAITHEW W. BOVEE, AND JOSEPH J. KNAPIK TABLE 9-13 Risk of Injury for Women Versus Men Army Trainees by Tertiles* of Mile Run Time, Fort Jackson, South Carolina, 1984 Risk of Injury (%)t Run Time Confidence Interval Fertile Women Men Risk Ratio (95%) 2 3 20.0% 17.5% 1.1 (~3~5) (2tlO) (1 1/63) 37.3% 46.7% 0.8 (.4-1.5) (22/59) (7/15) 57.7% 0.0% (47/71) (0/1) NOTE: Mantel-Haenszel summary risk ratio = .98 (.4 - 2.3); Mantel-Haenszel chi-square = 0.00, p = 1.00.) *Tertiles were: T1 = S.9 - 7.9 minutes; T2 = 7.9 - 9.7 minutes; T3 = > 9.7 minutes. "Percent risk = injured/(injured + not injured). endurance as measured by run times was the best predictor of training- related injuries. Discussion These two studies at Fort Jackson provided a unique opportunity to prospectively examine the relationships among body composition, physical fitness, and injury in men and women. The assemblage of basic trainees at an Army reception station for several days prior to the onset of basic train- ing permitted the collection of baseline data from direct physical measure- ments and questionnaires. Access to medical records of this young, healthy population provided an opportunity that would be rare outside the military. Also, the records represent all health care received, because basic trainees do not have access to any other health care system. A final unique aspect of this study was that, unlike most epidemiologic studies of this nature on civilian sports and exercise populations, all individuals in the study were engaged in similar types and amounts of physical training and other daily . . . activities. Many of the results of this study, such as the correlation between in- creasing percent BE and decreasing endurance performance, were similar to those reported by previous investigators. Other findings, such as the associ- ation between lower levels of physical fitness and higher risks of injury, were unique. These singular findings may be explained by characteristics of this study design that were different from previous studies of this nature.
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 165 Results of this study have important implications for the military and phys- ically active civilian populations. Correlation Between Body Composition and Physical Fitness Because an underlying assumption of Army policy is that fatter soldiers are less fit, it was deemed important to examine that premise. Others have found significant correlations between measures of BF and fitness. Vogel and Friedl (chapter 6) found a significant correlation (r = -.48) between percent BF and maximum oxygen uptake for men. In another recent study, Jette et al. (1990) reported correlations between BMI and estimated maxi- mum oxygen uptake of -.41 for men and -.54 for women. Cureton et al. (1979) found negative correlations between percent BF and run times of men and women of-.30 and-.22, respectively. The findings presented here were parallel to those; significant positive correlations were observed between percent BF and 1- or 2-mile run times of .27 to .53 for men, but only .12 to .16 for women, which indicates that fatter men and women run slower. In this study, correlations between BMI and run times were also signif- icant and positive but of lower magnitude than for percent BF. This lower correlation probably occurred because BMI is only a surrogate measure of percent BF, and it is the inert fat tissue that detracts from weight-bearing endurance performance. BMI accounted for only 65 to 70 percent of the variance in percent BF among men trainees and between 40 and 70 percent of the variance for women trainees. Negative correlations between BMI and number of sit-ups performed in 1-minute intervals have been reported by Jette et al. (1990~: r = -.24 and r = -.15 for men and women, respectively. In this study, negative correla- tions were found between percent BF and number of sit-ups in 2 minutes of -.17 to -.29 for men and -.12 to -.14 for women. Jette et al. (1990) also observed negative correlations between BMI and push-ups with r = -.22 for both men and women. Correlations in this study between percent BF and push-ups for men ranged from -.17 to -.29, and those for women ranged from -.02 to-.18. Correlations between BMI and sit-ups and push-ups were lower than for percent BF and these calisthenics. In general, the correlations between measures of BF and either push- ups or sit-ups were lower than for those with weight-bearing endurance performance such as running. These lower correlations with BF are attrib- uted to the fact that individuals must lift only a portion of their body weight against gravity to perform push-ups and sit-ups, but they must lift their entire body weight, including the fat, to run. To further understand the degree and significance of changes in run times as level of BF increases, these changes were analyzed for successive quintiles of BF in 1988. Significant differences in run times were found
66 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPlK between successive quintiles of percent BF among men, but only between the extreme quintiles of BF, the fattest and the lower ones, among women. This finding suggests that in this study, percent BF is not as good a discrim- inator of fitness for women as for men. Vogel and Friedl (chapter 6) found significant decreases in run time between quartiles of active-duty men and women soldiers with successively higher percentages of BF. As an aside, the relative run times of women may not be as strongly affected by increases in percent BF because the relative range of fatness for women is less than for men. The range of fatness for women is from 16 to 34 percent, a relative difference of 2.3 between extremes, while for men the range is 2 to 30 percent, a 15-fold difference (Friedl et al., 1989~. For this reason, percent BF provides less discriminating power for women. The consistency and significance of the correlation between measures of BF and endurance performance are important to the military because current regulations and policy assume such a relationship. Also, the stron- ger correlations between measures of percent BF and physical fitness (that is, run times, sit-ups, and push-ups) than between BMI and fitness have important implications for the military. Stronger correlations with percent BF suggest that using BF standards rather than BMI or height-weight tables as criteria for enlistment and retention would provide a better indicator of recruit and soldier fitness not to mention a better measure of body composition. Predicting Endurance Because of the universal requirement for soldiers to march and carry loads, models were developed to predict the endurance performance of men and women. A multiple regression model was used to determine the rela- tive importance of multiple factors suspected of contributing to weight- bearing endurance as measured by 1-mile run times. For both men and women, the same 10 potential predictors of physical performance were can- didate variables for the models: age, height, weight, percent BF, sit-ups, push-ups, total leisure-time, kcals per week, self-assessed activity level, level of sports participation, and aerobic intensity of usual leisure-time ac- tivities. Four variables contributed to the final model for men, and five variables contributed to the final model for women. In the men's endurance prediction model, percent BF stepped into the equation second-explaining 11 percent of the variance in run times-be- hind sit-ups, which explained 19 percent of the variance. For the women's model, percent BF stepped into the equation fourth behind self-assessed activity, height, and sit-ups-explaining only 3 percent of the variance in the run times of women trainees. Activity level was important in both
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 167 models and explained more of the variance in endurance (15 percent) among women than any other variable. Results of this modeling suggest several things. First the regression models coupled with the lower correlations between percent BF and run times for women reported above suggest that percent BF is not as good an indicator of fitness for women as it is for men. Second, the models for predicting endurance performance suggest that in addition to percent BF, other simple measures such as sit-ups-during the selection process might contribute significantly to the Army's ability to recruit fit soldiers. Al- though it might be difficult to use questions on self-assessed activity like those in this study in the context of recruiting soldiers, it is clear that past activity is an important factor in the prediction of fitness. Risks of Injury Previous studies have reported the incidence of musculoskeletal com- plaints ranging from 42 to 54 percent for women Army trainees and 23 to 26 percent for men (Bensel and Kish, 1983; Jones, 1983; Kowal, 19804. The cumulative incidence of injuries among trainees in this study was 51 percent for women and 27 percent for men, and the data here suggest that risks of injury have been relatively stable over almost a decade. Association Between Body Composition and Risk of Injury hew studies have examined the association of percent BF and BMI with the risk of training-related injuries, and no studies have systematically looked at the relationship of BF and weight-bearing training injuries. A few stud- ies of runners have examined the relationship of BMI to injuries (Blair et al., 1987; Macera et al., 1989b; Marti et al., 19881. No association between BMI and injury was reported for men or women runners in a study by Macera et al. (1989b), while Blair et al. (1987) reported a slight but signifi- cant positive correlation (r = .1) between BMI and risk of injury among runners. More consistent with the findings here is Marti et al.'s (1988) report of a bimodal distribution of injuries among men runners, in which the groups with the highest and lowest BMIs in a population of distance runners suffered the highest incidence of injuries. Macera et al. (1989a) in a pro- spective study of exercising adults reported that a high BMI at baseline was a risk factor for men but not for women. During this study in 1984 it was felt that the relationships between percent BF or BMI and risk of injury both might be bimodal. The hypothe- sis was that men and women of more "average" BF, those in the middle groups, would be at lower risk of injury than those at the high and low
68 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK extremes. For this reason, the middle quartiles and quintiles of body fat and BMI were chosen as the referent level for contrasting risks. It now appears, at least among this sample of men and women Army trainees that the patterns of risk are different for men and women. The univariate analysis suggests that the men trainees with the highest percent- ages of BF are at greatest risk of injury. Certainly this was true in 1988 when the men trainees in the highest quintile of BF were at 1.5 times greater risk than the lower ones. In contrast, it appears that women with the lowest percentages of BF are at greater risk than those of average percent BF as seen in 1988, when the women with the lowest body fats were at 1.3 times greater risk than those women in the middle. With BMI, the distribution of risk of injury appears to be bimodal. However, the only significant association for men occurred in 1984 when the trainees with the highest BMI were at 2.1 times greater risk of injury than those of average BMI. For women, the only significant associations occurred in 1988. At that time, women with both the highest and lowest BMI were at 1.5 and 1.6 times greater risk, respectively, than the more "average" referent group. Assuming that this observation is correct-that the fattest and highest BMI men and the leanest and lowest BMI women represent the tails of the distribution of BF at greatest risk of injury then a plausible explanation for these findings is necessary. It may be that the men trainees with the highest BF were at greater risk than their peers because they were carrying so much extra weight as fat fat that would not only contribute to greater fatigue at any given level of weight-bearing performance, but also would impose an additional stress on the musculoskeletal system. Paradoxically, it may be that the least fat women trainees were at greater risk for the converse reason: too little lean body mass. Perhaps women with low percentages of BP who are still relatively fat compared to men may not have enough lean body mass to support their total body weight without undue stress. In any case, distinct and consistent patterns of relationship between percent BF or BMI and risk of injury are not evident. Some of this lack of correspondence at least for percent BF may be attributable to different tech- niques of measurement used in 1984 and 1988: skinfolds versus circumfer- ences, respectively. Also, the apparently different pattern of association between BF and injury for men and women in this study may hypothetically be due to the fact that the Army height-weight selection standards artificial- ly truncate the distribution of percent BF among women trainees. The height-weight standards effectively exclude 30 percent of eligible women but only 5 percent of men (Friedlet al., 1989~. Regardless of what accounts for the differences between men and women, the current upper limits of height-weight are not effectively excluding the women at greatest risk of . . 1nJury.
BODY COMPOSITION, PlIYS1CAL FITNESS, AND INJURY Association of Physical Fitness with Risk of injury 169 The association between physical fitness and risk of injury in this study is more consistent for both men and women than the association with BE. In fact, as the stratified and logistic regression analyses suggest, endurance or weight-bearing fitness was the factor most strongly associated with risk of injury. Men and women in this study with the least endurance that is, the slowest run times-were at greatest risk of injury. The slowest men were at 1.4 to 2.8 times greater risk than their slower counterparts, and the slower women were at 1.3 to 1.8 times greater risk. Other authors have not reported such a relationship between fitness and injury. In fact, most report an increase in risk of training injuries for the most fit individuals (Blair et al., 1987; Macera et al., 1989b; Marti et al., 1988~. Blair et al. (1987) and Marti et al. (1988) both reported a positive association between high levels of fitness and high risks of injury on univariate analysis that disappeared when the amount of training (miles run) was ac- counted for in a multivariate analysis. This result suggests that in these studies the relationship between fitness and injury was confounded by the association of fitness with greater amounts of training. Studies by others on the relationship of physical fitness to injury pri- marily investigated runners of different fitness levels who ran for different numbers of miles at various intensities (Blair et al., 1987; Koplan et al., 1982; Macera et al., 1989b; Marti et al., 19884. In this study, men and women within companies (150 to 250 trainees) and to some extent across companies-ran, marched, and exercised similar amounts and at similar intensities, intensities that were dictated by the group and Army policy rather than individual predilections. Thus this study provided controls for confounding due to varied volume and intensity of training, which other studies have not. It is not surprising that a measure of weight-bearing fitness is associat- ed with injury among Army trainees. The single most common physical stress during basic training results from weight-bearing physical training, a stress that is secondary to running, drill and ceremony, marching to and from training sites, and road marching with loads. Even when not training, weight-bearing musculoskeletal stress is unavoidable. Walking is usually the only mode of transportation to and from the mess hall and other sites during available leisure time. The more aerobically fit trainees are under less physiological stress at any given activity level and may also have more prior exposure to musculoskeletal stress. thus decreasing their risk of injury. Whatever the underlying reason, the data here suggest that a measure of endurance fitness might provide additional information to assist in identify- ing injury-prone Army volunteers.
170 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK Association of Physical Activity and Risk of Injury It is well known that higher volumes (amounts) of training are associat- ed with higher risks of injuries among runners (Koplan et al., 1985; Powell et al., 19861. But data from this study demonstrate that risks of injuries among men trainees at Fort Jackson in 1984 decreased in a stepwise fashion as self-reported levels of prior physical activity increased, and sedentary men trainees were more than twice as likely to suffer training injuries. This finding is similar to that from a study of marine recruits (Gardner et al., 1988) in which a highly significant trend was observed of decreasing inci- dence of stress fractures with increasing self-reported activity levels. These data suggest that for men recruits higher prior physical activity levels may protect against current injury when they are engaged in a uniform training program, and are performing the same amounts of exercise as individuals with less prior exposure to the stress of vigorous physical activity. Other studies have looked at runners all of whom ran different distances, in which case the "dose" or volume of running was the primary risk factor (Blair et al., 1987; Koplan et al., 1982; Macera et al., 1989b; Marti et al., 1988~. Gender, Physical Fitness, and Risk of injury In the studies reported here, women were injured significantly more often than men, between 1.6 and 1.8 times more often. This finding is in agreement with those of previous Army studies of basic trainees (Bensel and Kish, 1983; Kowal, 1980) but is not consistent with civilian studies (Koplan et al., 1982; Macera et al., 1989a,b). The primary risks during Army basic training are lower extremity injuries associated with weight- bearing activities such as running and marching. Also, the pattern and distribution of these injuries is similar to that for civilian runners and jog- gers (Jones, 1983~. Despite these apparent similarities of trainee activity and injuries to those of civilian runners and joggers, civilian studies have not found women to be at higher risk (Koplan et al., 1982; Macera et al., 1989b, Powell et al., 1986~. Powell et al. (1986) concluded that "gender per se does not appear to be an important risk factor for injuries." Macera and her colleagues (1989a,b) have shown in both civilian runners and in exercising adults that gender is not a risk factor for injury. Also, Koplan et al. (1982) found no differences in risk of running injuries between men and women. Selection bias could account for these contradictory findings between military and civilian studies. Koplan et al. (1985) indicated that civilian studies of physical activity, fitness, and related injury suffer from selection bias. These studies (Koplan et al., 1982; Macera et al., 1989b; Powell et al., 1986) are biased in that the populations studied include only individuals
BODY COMPOSITION, PHYSICAL FITNESS, AND INJURY 171 who were fit enough to tolerate routine vigorous training and had not quit due to injury or for other reasons. If women were actually at greater risk of injury, what would be expected is that fewer women would be represented in the populations studied since women on average are less physically fit than men (Table 9-1, Vogel et al., 1986~. In fact this is what is found. In all the cited studies of runners (Koplan et al., 1982; Macera et al., 1989b) and exercise participants (Macera et al., 1989a), the number of women in the population examined were only 16 to 23 percent that of men. If the hypothesis that only men and women who are fit enough to survive training remain in the population of routine exercisers is true, then we might expect that the injury rates among men and women of the same high fitness levels would be similar. The results of the study of Army trainees in 1984 support such a conclusion. Although the crude risks of injury were higher for women, when risks of injury were stratified on run times (physical fitness), differences in risk between women and men disap- peared, and the risk ratio approached 1. Also, with the logistic regression model, gender remained the predominant and only significant risk factor for injury with an odds ratio of 2.5 (p < .0005) until mile run time was entered as a potential predictor, whereupon gender ceased to even approach signifi- cance as a risk factor. Run time (aerobic fitness) replaced gender as the sole and best Dredictor of injury (odds ratio = 3.5, p ~ .0001~. ., , , - O ~ ~ ~ . . The possible implications of this finding are important for the Army, the other military services, and possibly civilian exercise enthusiasts and medical practitioners for two reasons. First, it suggests that low levels of aerobic fitness or some related factor are a primary risk factor for muscu- loskeletal injuries associated with military and possibly other vigorous weight- bearing training activities such as running. Second, it indicates that gender per se is not the major risk factor that a crude analysis of military training injury data might imply, and that low physical fitness may be the underly- ing predisposing factor. Conclusion In general, men enter the Army with lower percentages of BF than women and are able to perform more sit-ups and push-ups and run faster than women. They also suffer fewer injuries than women because of their relatively higher levels of endurance and possibly other associated factors. Data from this study suggest that measures of percent BF are not as good at predicting physical fitness for women as they are for men. For both men and women, physical fitness as measured by even simple techniques such as sit-ups in combination with BF is a better predictor of other types of fitness such as mile run times or other forms of weight-bearing endurance than percent BF alone. Furthermore, higher percentages of BF for men are
72 BRUCE H. JONES, MATTHEW W. BOVEE, AND JOSEPH J. KNAPIK associated with increased risk of injury, but for women they are not. Phys- ical fitness is a better predictor of injury than BE or gender. Therefore, if physical fitness and freedom from injury are important to the Army, it would make sense to at least include some simple measure of fitness in the screening process for prospective enlisters. Several important conclusions can be made from this study and also the process of analysis. A few multivariate analyses can be much more infor- mative than numerous univariate analyses. Although univariate analyses are the foundation of multivariate approaches such as those used here, they are not a substitute for more complex models. More information generated from larger populations and yielding more powerful multivariate models is needed. With these models, the Army should be able to predict injuries and also such factors as career success or discharge. This information could provide the Army with a rational foundation from which to select and retain men and women who are most likely to possess the combination of fitness, fatness, and freedom from injury that is desired for military readiness. REFERENCES AR 600-9. 1986. See U.S. Department of the Army. 1986. AR 40-501. 1987. See U.S. Department of the Army. 1987. AR 350-15. 1989. See U.S. Department of the Army. 1989 Bensel, C. K. 1976. The effects of tropical and leather combat boots on lower extremity disorders among Marine recruits. Technical Report No. 76t49/CEMEL. U.S. Army Natick Research and Development Command, Natick, Mass. Bensel, C. K., and R. N. Kish. 1983. Lower extremity disorders among men and women in Army basic training and the effects of two types of boots. Technical Report No. TR-83/ 026. U.S. Army Natick Research and Development Laboratories, Natick, Mass. Blair, S. N., W. K. Harold, and N. N. Goodyear. 1987. Rates and risks for running and exercise injuries: Studies in three populations. Res. Q. 58:221-228. Buskirk, E., and H. L. Taylor. 1957. Maximal oxygen intake and its relation to body composi- tion, with special reference to chronic physical activity and obesity. J. Appl. Physiol. 11 :72-78. Cowan, D., B. Jones, P. Tomlinson, J. Robinson, D. Polly, P. Frykman, and K. Reynolds. 1988. The epidemiology of physical training injuries in U.S. Army infantry trainees: Methodology, population and risk factors. Technical Report No. T4-89. U.S. Army Research Institute of Environmental Medicine, Natick, Mass. Cureton, K. J., L. D. Hensley, and A. Tiburzi. 1979. Body fatness and performance differences between men and women. Res. Q. 50:333-340. Durnin, J.V.G.A. and J. Wormersley. 1974. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged 16 to 72 years. Br. J. Nutr. 32:77-79. Friedl, K. E., J. A. Vogel, M. W. Bovee, and B. H. Jones. 1989. Assessment of body weight standards in male and female Army recruits. Technical Report No. 1 15-90. U.S. Army Research Institute of Environmental Medicine, Natick, Mass. Gardner, L. I., J. E. Dziados, B. H. Jones, J. R. Brundage, J. M. Harris, R. Sullivan, P. Gill.
BODY COMPOSlTlON, PHYSICAL FITNESS, AND INJURY 173 1988. Prevention of lower extremity stress fractures: A controlled trial of a shock absor- bent insole. Am. J. Public Health 78(12):1563-1567. Jette, M., W. Lewis, and K. Sidney. 1990. Fitness, performance and anthropometric character- istics of 19,185 Canadian Forces personnel classified according to body mass index. Milit. Med. 155:120-126. Jones, B. H. 1983. Overuse injuries of the lower extremities associated with marching, jog- ging, and running: A review. Milit. Med. 148:783-787. Jones, B. H., R. Manikowski, J. A. Harris, J. E. Dziados, S. Norton, T. Ewart, J. A. Vogel. 1988. Incidence of and risk factors for injury and illness among male and female Army basic trainees. Report number: T 19/88. U.S. Army Research Institute of Environmental Medicine, Natick, Mass. Koplan, J. P., K. E. Powell, R. K. Sikes, and R. W. Shirley. 1982. An epidemiologic study of the benefits and risks of running. J. Am. Med. Assoc. 248:3118-3121. Koplan, J. P., D. S. Siscovick, and G. M. Goldbaum. 1985. The risks of exercise: A public health view of injuries and hazards. Public Health Rep. 100(2):189-195. Kowal, D. M. 1980. Nature and causes of injuries to women resulting from an endurance training program. Am. J. Sports Med. 8(4):265-269. Macera, C. A., K. L. Jackson, G. W. Hagenmaier, J. J. Kronenfeld, H. W. Kohl, and S. N. Blair. 1989a. Age, physical activity, physical fitness, body composition, and incidence of orthopedic problems. Res. Q. 60:225-233. Macera, C. A, R. R. Pate, K. E. Powell, K. L. Jackson, G. W. Hagenmaier, J. J. Krmenfeld, H. W. Kohl, and S. N. Blair. 1989b. Predicting lower-extremity injuries among habitual runners. Arch. Intern. Med. 149:2565-2568. Marti, B., J. P. Vader, C. E. Minder, and T. Abelin. 1988. On the epidemiology of running in- juries. The 1984 Bern Grand-Prix study. Am. J. Sports Med. 16:285-294. Miller, A. T., Jr., and C. S. Blyth. 1955. Influence of body type and body fat content on the metabolic cost of work. J. Appl. Physiol. 8:139-141. Pollock, M. L., L. R. Gettman, C. A. Milesis, M. D. Bah, L. Durstine, and R. B. Johnson. 1977. Effects of frequency and duration of training on attrition and incidence of injury. Med. and Sci. Sports 9:31-36. Powell, K. E., H. W. Kohl, C. J. Caspersen, and S. N. Blair. 1986. An epidemiological per- spective on the causes of running injuries. Phys. Sports Med. 14:100-114. Reister, F. A. 1975. Medical Statistics in World War II. Office of the Surgeon General, Depart- ment of the Army, Washington, D.C. Taylor, H. L., D. R. Jacobs, B. Schucker, J. Knudsen, and A. S. Leon. 1978. A questionnaire for the assessment of leisure time physical activities. J. Chronic Dis. 31:741-755. Tomlinson, J. P. W. M. Lednar, and J. D. Jackson. 1987. Risk of injury in soldiers. Milit. Med. 152:60-64. U.S. Department of the Army. 1985. Army Field Manual 21-20: Physical Fitness Training. Headquarters, Washington, D.C. U.S. Department of the Army. 1986. Army Regulation 600-9. "The Army Weight Control Program." September 1. Washington, D.C. U.S. Department of the Army. 1987. Army Regulation 40-501. "Standards of Medical Fitness." December 1. Washington, D.C. U.S. Department of the Army. 1989. Army Regulation 350-15. "The Army Physical Fitness Program." Washington, D.C. Vogel, J. A., J. F. Patton, R. P. Mello, W. L. Daniels. 1986. An analysis of aerobic capacity in a large United States population. J. Appl. Physiol. 60:494-500. Washburn, R. A., L. L. Adams, and G. T. Haile. 1987. Physical activity assessment for epide- miologic research: The utility of two simplified approaches. Prev. Med. 16:636-646.