Stress fracture injury in young military men and women

Bone 35 (2004) 806 – 816 www.elsevier.com/locate/bone

Stress fracture injury in young military men and women

$

David W. Armstrong III, John-Paul H. Rue, John H. Wilckens, * and Frank J. Frassica National Naval Medical Center, Bethesda, MD 20889, USA United States Naval Academy, Annapolis, MD 21402, USA Department of Orthopaedic Surgery, Johns Hopkins University/Johns Hopkins Bayview Medical Center, Baltimore, MD 21224, USA Received 19 December 2003; revised 13 May 2004; accepted 14 May 2004 Available online 8 July 2004

Abstract Approximately 5% of all military recruits incur stress fracture injuries during intense physical training, predominately in the lower extremity. We compared young men and women with stress fracture injury (subjects) to a matched group of uninjured volunteers (controls) during a summer training program at the United States Naval Academy to identify possible risk factors for stress fracture injury. The subject group was composed of 13 female and 18 male plebes with training-induced stress fracture injury verified by plain radiographs and/or nuclear bone scan. The control group was composed of 13 female and 18 male plebes who remained without injury during plebe summer training but who were matched with the 31 injured plebes for the Initial Strength Test (1-mi run time, means: women, 7.9 min; men, 6.4 min) and body mass index (means: women, 23.4; men, 23.8). We found that the subjects lost significant body weight (mean, 2.63 F 0.54 kg) between Day 1 and the date of their diagnosis of a stress fracture (mean, Day 35) and that they continued to lose weight until the date of their DEXA scan (mean, Day 49). Among female plebes, there was no evidence of the female athlete triad (eating disorders, menstrual dysfunction, or low bone density). Thigh girth was significantly smaller in female subjects than in female controls and trended to be lower in male subjects than in male controls. Total body bone mineral content was significantly lower in the male subjects than in male controls. Bone mineral density of the distal tibia and femoral neck were not significantly different between the groups. DEXA-derived structural geometric properties were not different between subjects and controls. Because, on average, tibias were significantly longer in male subjects than in male controls, the mean bone strength index in male subjects was significantly lower than that of male controls. We conclude that significant, acute weight loss combined with regular daily physical training among young military recruits may be a significant contributing risk factor for stress fracture injuries in young military men and women. D 2004 Elsevier Inc. All rights reserved. Keywords: Structural geometry; Bone mineral density (BMD); Physical training; Recruits; Weight loss

Introduction Each July, approximately 1200 new recruits enter the U.S. Naval Academy in Annapolis, Maryland, as plebes. Plebe summer is a physically challenging, 2-month indoctrination of midshipmen into military life. Physical challenges include the normal activities associated with military life (e.g., instruction, formations, marching, and drilling)

$ The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of the Navy, Department of Defense, nor the U.S. Government. * Corresponding author. c/o Elaine P. Henze, Medical Editor, Department of Orthopaedic Surgery, Johns Hopkins Bayview Medical Center, 4940 Eastern Avenue, #A672, Baltimore, MD 21224-2780. Fax: +1-410-5502899. E-mail address: [email protected] (J.H. Wilckens).

8756-3282/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2004.05.014

and the formal morning physical education program designed to enhance the physical fitness level of the incoming midshipmen. This physical training program, consisting of early morning runs (approximately 10 mi per week; Fig. 1) and calisthenics (60 – 90 min of stretching, push-ups, sit-ups, pull-ups, sprinting, and agility drills) 5 days a week, initiates changes in metabolic events that contribute to a loss of body weight in some recruits. Intense exercise training may also contribute to muscle fatigue and soreness, dehydration, muscle structural damage, muscle swelling, central nervous system fatigue, and increased use of nutrient stores. Specifically, rigorous training may increase metabolism, increase activation of the hypothalamic – pituitary – adrenal axis, deplete muscle glycogen, and contribute to muscle amino acid loss, hepatic gluconeogenesis, and a negative nitrogen balance [1– 3].

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

Fig. 1. Line graph showing relationship of symptoms, miles run, and stress fractures by days of training. SFx, stress fracture.

The physical demands on plebes result in common medical situations such as heat strain, dehydration, blisters, sprains, and various musculoskeletal overuse injuries. Of the overuse injuries among plebes during their first summer training, stress fractures are a persistent medical problem: It affects 3% of the men and 10% of the women (personal communication, Brigade Medical Officer MKD), resulting in substantial loss of training opportunities and a decrease in physical performance. These injuries, which occur during a critical period of training and indoctrination, impact availability for continued training. Stress fractures, also known as fatigue or march fractures, have remained a well-recognized medical condition since Breithaupt [4] initially described them in 1855. Stress fracture injuries are ubiquitous among military organizations and disproportionately affect women [5]. Within the U.S. military, reported rates for stress fracture among new recruits range from 2% to 12%, contributing to substantial lost training time [6 –9]. Stress fractures in military recruits commonly occur in the tibia, femur, metatarsals, and pelvis, and they present with the insidious onset of localized pain that worsens with activity [5,7,8,10]. A low level of physical fitness at entry into military recruit training may be a contributing factor to training-induced musculoskeletal injury. Shaffer et al. [11] have reported that baseline physical fitness among Marine recruits is poor. Less than 15% of recruits were in excellent physical condition, and fewer than half the recruits ran three times per week or averaged a training distance of more than 2.5 mi. Bending forces have been implicated as the most important factor in the pathogenesis of stress fractures [12]. Most military recruits report the onset of symptoms of stress fractures between Days 10 and 12 of training [6]. It has been hypothesized that as the mechanical forces accumulate, the rate of bone resorption exceeds bone remodeling and repair, and the effective loading area is decreased. Therefore, because force remains constant, stress is increased. This process progresses until the stress exceeds the limit of the bone, and fracture occurs. Thus, stress fractures occur

807

when bone strain increases osteoclastic resorption beyond remodeling by osteoblasts [6]. Most military recruits report onset of symptoms of stress fractures early in the training cycle [6], which may represent structural fatigue produced from repetitive mechanical forces. A contributing scenario is that of a progressive decline in the muscular support of the bone, which may be caused by several factors. These factors may include a smaller muscle volume [7], and smaller muscles may be weaker and unable to provide adequate support for the bone. Muscles that are not adapted to repetitive work and therefore lack endurance may fatigue, be unable to support the long bones of the lower extremity, and fail because of large reductions in muscle glycogen stores [13,14] secondary to repetitive, high-intensity training programs and because of the exercising individual’s failure to achieve adequate energy balance and nutritional support from the diet [13,14]. Investigators have attempted to identify risk factors for stress fracture, including previous aerobic fitness and activity level, body somatotype, height, weight, body mass index (BMI), motivation, city versus country origin, hyperpronation of subtalar joint, decreased muscle strength, amenorrhea, oligomenorrhea, decreased calcium, decreased bone density, tibia torsion, narrow tibias, and increased hip external rotation [6,7,10,15– 22]. However, as indicated by Burr [23] and Jones et al. [24] in their extensive literature reviews and to our knowledge, no study has compared subjects sustaining a stress fracture injury during military recruit training with matched uninjured controls based on gender, BMI, and aerobic physical fitness data that were collected before they entered military recruit training. Therefore, the purpose of our study was to compare young men and women at the United States Naval Academy who sustained lower extremity stress fracture during a military summer training program with a matched group of uninjured recruits to identify factors that may increase the risk of a stress fracture.

Materials and methods There were 1224 midshipman in the class of 2000, of which 203 were female (17%). During the study period (July through August 2000), 40 plebes (23 men, 17 women) were diagnosed with 58 stress fractures. The incidence of stress fractures for the plebe indoctrination was 3.3% overall, 2.3% for male midshipmen, and 8.4% for female midshipmen. Of the 58 stress fractures, 74% (n = 43) were in the tibia, 9% (n = 5) were in the metatarsals, 5% (n = 3) were in the femur, 5% (n = 3) were in the fibula, and 5% (n = 4) were in other sites. The left side was involved in 52% of fractures; the right side, in 48%. Of the 40 plebes, 9 (4 women, 5 men) were not included in the analysis. Of those four women, one had a stress fracture of the pelvic ramus and was excluded; one

808

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

declined, and two did not return to clinic after their diagnosis. Of those five men, one left the Academy, two had experienced at least one stress fracture before arrival at the Academy, one had a questionable fracture of the fibula, and one had an upper extremity fracture. The remaining 31 plebes (18 men, 13 women) with 47 lower-extremity stress fractures confirmed by radiography and scintigraphy formed the subject group. Subsequently, an uninjured plebe who matched the subject in terms of gender, BMI (body weight in kilograms/height in meters squared), and initial strength test (IST, described below) 1-mi run time (Table 1) was recruited near the same time for the control group. Controls remained uninjured for the entire plebe summer. All 31 subjects and 31 matched controls provided voluntary, written informed consent as approved by the Committee for the Protection of Human Subjects at the National Naval Medical Center. Initial BMI values were determined from measurements taken by Navy corpsmen on the plebe’s first day (Day 1) at the Academy. The medical department at the Academy supplied Day 1 height in inches and weight in pounds, which were converted to metric equivalents. The standardized IST was conducted by the Naval Academy’s Physical Education Department during the first week of plebe summer training and before the beginning of the mandatory daily morning physical training program. The IST consists of the time to complete a 1-mi run, the number of push-ups performed in 2 min, and the number of sit-ups performed in 2 min. A history of each plebe’s activity level and sport participation before arrival at the Academy was obtained by interview at the time of the dual energy X-ray absorptiometry (DEXA) scan (see below). Plebes seeking medical attention for activity-related musculoskeletal pain were examined by a member of the medical staff. If the plebe’s medical history suggested a stress fracture, plain film radiographs were obtained and interpreted by a staff radiologist (Table 2). If these radiographs were inconclusive, a radionucleotide bone scan was performed. Ten stress fractures were diagnosed by scintitium bone scan. Standard-of-care treatment consisted of protected weightbearing activity, using crutches if normal walking was

Table 2 Radiographic findings Finding

Pretreatment (diagnosis)

Normal Periosteal reaction Cortical thickening Lucent line Cortical irregularity Sclerosis Endosteal reaction Combined: periosteal reaction/cortical thickening Combined: periosteal reaction/sclerosis

12 20 8 2 1 1 1 1 1

painful, and mandatory alternative non-weight-bearing aerobic exercise, such as swimming. Injured plebes are encouraged to maintain their normal summer training schedule (Table 3) during treatment, including attending class and observing physical training, drilling, and marching when participation is precluded by the injury. At entry into the study, the length of the tibia (medial tibia plateau to medial malleolus) and the maximum circumference of both thighs were measured for each participant. A body weight was also obtained at this time. Tibia length determined the ‘‘distal tibia’’ as the point on the tibia 0.67% of the tibia length measured distal to the medial tibia plateau. We recorded the dates that each subject first reported to medical clinic with symptoms consistent with a stress fracture injury, the date of diagnosis of a stress fracture, and the date the plebe was returned to full duty. Using these dates, we calculated the duration of symptoms for each stress fracture. When the plebe had no pain on palpation at the fracture site and could perform a hop on the affected side without pain, plain radiographs were obtained. If the radiograph showed signs of fracture healing, then the plebe was returned to full duty. All participants (dressed in athletic shorts, T-shirts, and socks) underwent DEXA scans of their nondominant hips (right hip, 7; left hip, 55), distal tibias of both legs, and whole body as described below and in other publications [25,26]. Immediately before the scan, each participant was weighed on a digital electronic scale (Lafayette Instruments, Lafayette, IN) to the nearest 0.1 kg, and height was

Table 1 Subject and control demographics on Day 1a Group/gender

N

Mean age F SE (years)

Dominant hand (R/L)

Mean height F SE (cm)

Mean weight F SE (kg)

Mean BMI F SE

Mean DEXA body fat % F SE

Fracture subjects Women Men

13 18

18.5 F 0.17 18.9 F 0.21

11/2 18/0

163.9 F 1.74 181.6 F 1.97

62.9 F 2.31 78.3 F 2.01

23.0 F 0.63 23.8 F 0.54

20.9 F 1.24 12.2 F 0.90

Controls Women Men

13 18

18.4 F 0.17 19.3 F 0.21

12/1 14/4

166.2 F 1.74 177.2 F 1.97

65.0 F 2.31 76.8 F 2.01

23.5 F 0.63 24.5 F 0.54

21.2 F 1.24 10.0 F 0.81

a

There were no statistically significant differences between subjects and controls within gender.

D.W. Armstrong III et al. / Bone 35 (2004) 806–816 Table 3 Sample plebe summer schedule Time

Activity

06:00 – 7:30 a.m. 07:30 – 8:45 a.m. 08:45 – 11:45 a.m. 11:45 – 12:00 p.m.

Physical training program Room clean-up/morning meal Morning training Administrative activities/squad leader discussion Noon meal Afternoon training Sports period Evening meal Drill period Personal time Counsel time Taps/lights out

12:00 – 12:50 01:00 – 03:00 04:00 – 05:50 06:00 – 06:50 07:00 – 09:00 09:10 – 09:30 09:30 – 09:45 10:00 p.m.

p.m. p.m. p.m. p.m. p.m. p.m. p.m.

measured to the nearest millimeter using an Accustat stadiometer (Genentech, Inc., San Francisco, CA). DEXA scans were obtained using a Norland XR-36 Dual Energy X-ray Absorptiometer (Norland Medical Systems, Inc., Ft. Atkinson, WI) that was calibrated daily using a Norland-supplied lumbar spine phantom. The phantom contained 40.83 g of hydroxyapatite within an area of 54.59 cm2. The area density of the phantom was 0.748 g/ cm2. The coefficient of variation for the phantom was 0.54%, calculated from a moving average of the previous 16 phantom scans. The in vivo coefficient of variation was 2.34% at the femoral neck and 1.84% at the distal tibia. The same investigator (DWA) performed and analyzed all scans. Each participant was positioned on the bed of a Norland XR-36 DEXA as follows. For the nondominant hip, the Norland hip-positioning device was placed on the supine plebe. The greater trochanter of the nondominant hip was located, and a pilot scan sequence was initiated. After the pilot scan, the positioning cursor was in the center of the femoral neck. The hip scan sequence was initiated and run to completion (2 min). Then, with the plebe still supine, the lengths of the right and left distal tibia (from the tibial plateau to the center of the medial malleolus) were measured. A point corresponding to a position 67% of the tibial length distal to the tibial plateau was marked on the skin over the tibia. The plebe’s foot was placed in a footpositioning device and rotated internally to an angle of 17.5j. The tibia was scanned for a distance of 10 mm on each side of the point marked on the skin along the tibial crest. Two minutes were required to complete the scan. For the whole-body scan, the hands and feet of the still-supine plebe were held in place by a sheet. The boom was positioned 1 cm above the top center of the plebe’s head, and the point on the scanner bed was marked. The boom was moved into a position lateral to the spine and between the iliac crest and lowest rib, and marked. The scan sequence was initiated and required 6 min for completion. Using software supplied by Dr. Thomas Beck and algorithms described previously, DEXA image data of the distal tibia were used to derive mediolateral bone widths, cross-

809

sectional areas, and cross-sectional moments of inertia (CSMI) [7,10,20,21]. Section modulus was calculated as the ratio of CSMI to one half the mediolateral bone width [20,21]. The bone strength index, calculated as the ratio of section modulus to bone length, is based on the observation that strength of a bone under bending or torsion is inversely dependent on bone length and directly related to the section modulus [27]. Participants completed the Spielberger Self Evaluation Questionnaire (State-Trait Anxiety Inventory) at the time of the DEXA appointment. This standardized self-report instrument is designed to measure anxiety. Participants are directed to respond to 20 forced-choice items as they feel ‘‘right now’’ for the State portion and 20 forced-choice items as they ‘‘generally’’ feel for the Trait portion. Scores are summed over the 20 responses and an aggregate score, ranging from 20 to 80 for each measure, is generated. Female plebes provided a menstrual history. All women in this study reported having normal and regular menstrual periods in the previous 12 months. Female plebes (11 subjects, 13 controls) also completed Garner’s Eating Disorder Inventory (EDI) at the time of the DEXA appointment [28]. This is a standardized self-report measure consisting of 91 forced-choice, Likert format items organized into 11 subscales measuring 11 psychological traits commonly associated with eating disorders. Eight of the subscales are from the original EDI and three provisional subscales were added to the EDI-2. The EDI-2 does not yield a specific diagnosis of an eating disorder; rather, it is designed to measure psychological traits or symptom clusters presumed to have relevance to understanding and treatment of eating disorders. The psychological profile provided by the EDI-2 is consistent with the understanding of eating disorders. Data analyses were conducted using JMP Software version 5.0 (SAS Institute, Cary, NC). Paired Student’s t tests were used to test for differences between measures acquired on Day 1 and data acquired on the date of the DEXA scan (e.g., change in body weight, BMI, height, weight, etc.). Differences between subjects and controls were determined separately for male and female plebes using analysis of variance (ANOVA). Means for all bilateral (i.e., tibia, thigh) measurements were used because statistical analysis did not reveal significant differences between right and left sides. We report adjusted means (LSMEANS) and their standard errors. Significance was set at P V 0.05.

Results Demographics Within the same gender, there were no differences in Day 1 age, height, weight, and BMI between fracture subjects and controls. Comparing genders, male plebes were nearly 8 months older, 15 cm taller, and 13 kg heavier than female

810

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

plebes, but BMI was not different (23.8 F 0.29) between gender. Twenty of 26 women (77%) and 29 of 36 men (81%) were Caucasian. In the subjects, stress fractures (confirmed by radiograph or bone scan) occurred in the tibia (n = 18) in all 13 women, and in the tibia (n = 13), proximal femur (n = 2), metatarsal (n = 2), and fibula (n = 1) in the men.

There were no significant differences between female subjects and controls for the number of push-ups (45 F 4) or sit-ups (63 F 3) completed during the IST. Male subjects performed significantly fewer push-ups than did male controls (means: 59 F 4 and 72 F 4, respectively). There was no difference between male subjects and controls in terms of sit-ups (mean: 67 F 4).

Preadmission sports participation/activities Diagnosis and DEXA scanning During high school, all participants had played a variety of sports. Seventeen of the 26 women (65%) and 24 of the 36 men (67%) were members of a varsity team in high school. There was no association between varsity sport participation and stress fracture. However, plebes that participated primarily in non-weight-bearing high school sports (e.g., swimming) showed a trend ( P = 0.07; Odds Ratio = 3.2) to stress fracture; plebes who participated primarily in weight-bearing sports (e.g., basketball, soccer) did not (Table 4). Comparing subjects and controls, there were no differences in self-reported preadmission running miles per week (8.7 F 1.0 mi*week 1) or additional exercise training minutes (66.7 F 5.4 min*day 1). Physical strength/readiness Within gender, there were no significant differences for the IST 1-mi run time between subjects and controls and no significant differences in the 1-mi run times between study participants and the entire plebe class. The mean 1-mi run time for the entire class of female plebes was 7.54 F 0.07 min and that for the male plebes was 6.37 F 0.02 min. Table 4 Preadmission sports participation among 62 midshipmen subjects Sport

Number participatinga

Number with stress fracture

Cross-country/track Swimming No sport Baseball/softball Soccer Wrestling Football Lacrosse Tennis Basketball Golf Volleyball Water polo Rugby Hockey Badminton Band Cheerleading Crew Marksmanship (rifle/pistol) Sailing Total

19 14 14 7 7 7 4 4 4 3 2 2 2 2 2 1 1 1 1 1 1 99

8 9 7 2 1 2 1 2 4 1 1 2 2 2 0 1 0 0 1 1 1 48

a

Several midshipmen participated in more than one sport.

Male and female controls underwent DEXA scanning 83 F 5 days after Day 1 at the Academy. Female subjects reported to the medical clinic with symptoms at 14 F 3 days after Day 1, were diagnosed with a stress fracture at 34 F 4 days, and underwent DEXA scanning at 49 F 5 days. Male subjects first reported to the medical clinic with symptoms at 16 F 2 days after Day 1, were diagnosed at 38 F 3 days, and underwent DEXA scanning 51 F 4 days. The mean time from onset of symptoms and return to duty for the fracture subjects was 61 days. Body weight From Day 1 to the date of diagnosis (mean, Day 34), the weight of female subjects declined 1.05 F 0.69 kg and continued to decline until the date of their DEXA scans (mean, Day 49) (Fig. 2). Female subjects showed a mean loss of 2.03 F 0.74 kg (0.47 kg*week 1) during the study period. In contrast, weight loss among female controls was not significant: 0.67 kg at recruitment to the study and 0.77 kg at Day 85 (Fig. 2). From Day 1 to the date of diagnosis (mean, Day 38), the weight of male subjects declined 3.69 F 0.57 kg and continued to decline until the date of their DEXA scans (mean, Day 51) (Fig. 2). Male subjects showed a mean loss of body weight 4.03 F 0.78 kg (0.73 kg*week 1) over the study period. Male controls did not lose significant weight (0.50 kg) by recruitment to the study, but their weight loss was significant by the day of their DEXA scans (mean, Day 85): mean, 1.89 F 0.78 kg (0.17 kg*week 1) during the study period. The difference in weight loss between male subjects and controls at diagnosis was significant but showed only a trend to significance ( P = 0.062) by the date of the DEXA. Overall, fracture subjects lost more than four times as much weight as the controls by the date of diagnosis of fracture (2.37 F 0.48 vs. 0.58 F 0.48 kg, respectively). Male controls continued to lose weight until the date of their DEXA, but the difference in their weight loss was one half that of male subjects (1.42 F 0.56 vs. 3.19 F 0.56 kg, respectively). Female parameters Eating disorders Female subjects (n = 11) showed significantly lower scores on three subscales of the EDI (Drive for Thinness,

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

811

35.9 F 2.2; Trait, 32.3 F 2.3). Between male subjects and controls, there was no significant difference in State scores (means: 39.4 F 2.7 and 34.4 F 2.6, respectively), but the Trait scores were significantly higher (although not abnormally so) in male subjects than in controls (mean: 38.3 F 2.1 and 32.1 F 2.1, respectively). Tibia length

Fig. 2. Line graph showing weight change for participants at time of diagnosis and DEXA scan. SFX, stress fracture.

Bulimia, and Body Dissatisfaction) than female controls (n = 11): 4.5 F 2.1 vs. 11.2 F 2.3, respectively. No female participant scored above the 80th percentile on any EDI subscale and none were receiving treatment for an eating disorder at the time of this study or previously. Women not being treated for an eating disorder who have scores above the 80th percentile on the EDI Drive for Thinness, Bulimia, and Body Dissatisfaction subscales (a combined score of z 29 for women) are considered at risk for developing an eating disorder [28]. The use of the EDI-2, by itself, does not provide the definitive diagnosis of an eating disorder; however, we did not conduct follow-up clinical interviews.

Mean tibia length (36.3 F 2.03 cm) was not different in female subjects and controls, but subjects had significantly smaller thigh girth (means: 47.6 F 1.04 and 51.9 F 1.04 cm, respectively). Male subjects had significantly longer tibias than did controls (means: 41.5 F 0.62 and 39.0 F 0.62 cm, respectively) and a trend ( P = 0.075) toward smaller thigh girth than controls (means: 49.3 F 1.06 and 52.1 F 1.06 cm, respectively). Total body mineral content and bone mineral density Female subjects and controls had no significant differences in total body bone mineral content (TBBMC) or bone mineral density (BMD) in the nondominant hip or distal tibia. Male subjects showed significantly lower total body mineral content, exhibited a trend to lower values in the nondominant hip BMD ( P = 0.053), and did not have significantly different tibia BMD than male controls (Table 5). Structural geometry

Menstrual status Female participants reported achieving menarche in their 12th year (12.7 F 0.25 years). There were no differences between groups for self-reported number of menstrual periods (11.5 F 0.35 cycles*year 1) in the previous 12 months. At the time of their DEXA appointments, we did not collect the date of the plebes’ last menstrual periods. Assuming that the menstrual cycle self-reports were accurate and assuming a normal 28-day menstrual cycle, subjects would have been 21 days into their second cycle on the date of their DEXA scans (49 F 5 days from Day 1) and controls would have been on their third cycle (85 F 5 days from Day 1). Extensive reviews have reported no effects of menstrual cycle on body weight change and muscle strength and endurance [29 – 31]. Seven of the 26 females in this study reported using oral contraceptives before their arrival at the Naval Academy; five of the seven reported beginning their use of oral contraceptives 3 months before entering the Academy. All seven women reported ending the use of oral contraceptives shortly after completing plebe summer training. Anxiety inventory There were no significant differences between female subjects and controls concerning response to the Spielberger State-Trait Anxiety Inventory instruments (means: State,

There were no significant differences between female subjects and controls in terms of distal tibia structural geometry characteristics (Table 6), e.g., mediolateral tibia width, cortical bone area, CSMI, section modulus, or bone strength. Comparing male subjects and controls, there were no significant differences in distal tibia mediolateral bone

Table 5 Bone characteristics and confidence interval data Group/gender and CI

N

Mean TBBMC F SE (g)

Mean femoral neck BMD F SE (g*cm 2)

Mean distal tibia BMD F SE (g*cm 2)

Fracture subjects Women 13 Men 18

2623 F 75 3073 F 78a

0.973 F 0.03 1.076 F 0.035

1.343 F 0.03 1.455 F 0.03

Controls Women Men

13 18

2808 F 75 3447 F 78a

1.037 F 0.03 1.176 F 0.035

1.374 F 0.03 1.531 F 0.03

95% CIb Women Men

165 86

1.029 – 1.065 1.121 – 1.188

1.349 – 1.389 1.501 – 1.566

2628 – 2717 3085 – 3250

CI, confidence interval. a Significant difference between subjects and controls within gender. b 95% CI data for female midshipmen data from Ref. [26]. 95% CI data for male midshipmen data from Ref. [25].

812

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

Table 6 Distal tibia structural geometry from DEXA scans Mean width (mm)

Mean cortical area F SE (mm2)

Mean CSMI F SE (mm4)

Mean section modulus F SE (mm3)

Mean tibia length F SE (cm)

Mean bone strength (modulus/tibia length)

Fracture subjects Women 13 Men 18

20.10 F 0.34 22.54 F 0.33

291.3 F 7.5 355.7 F 9.6

10,021 F 573 15,618 F 785

991 F 40 1378 F 53

36.6 F 0.57 41.5 F 0.62a

27.2 F 1.02 33.2 F 1.37a

Controls Women Men

20.66 F 0.34 22.93 F 0.33

302.0 F 7.5 383.7 F 9.9

10,938 F 573 17,109 F 809

1051 F 40 1482 F 54

36.0 F 0.57 39.0 F 0.62a

29.1 F 1.02 38.1 F 1.41a

Group/gender

N

13 18

CSMI, cross-sectional moment of inertia. a Significant difference between subjects and controls within gender.

width, CSMI, or section modulus (Table 6), but the former exhibited a trend ( P = 0.052) to lower cortical bone area. Because the tibias of male subjects were, on average, 2.5 cm longer than those of male controls, male subjects showed a significantly smaller distal tibia bone strength index than did male controls (Table 6).

Discussion To our knowledge, our study is the first to compare 18year-old military men and women with stress fracture to uninjured controls matched by gender, age, BMI, and preadmission aerobic physical performance in terms of demographic, injury, and outcome data for the purpose of identifying possible risk factors for stress fracture injury. Unlike other military recruits [11], plebes enter the Academy with a high level of physical fitness: 85% of this incoming class (of whom 15% were women) had a varsity letter in at least one high school sport and 50% of the female plebes were recruited to play a division I sport. Our study supports the consistent finding in other reports that the most common site of stress fracture in military recruits is the tibia [32,33]. In contrast to controls, most of our fracture subjects reported participating primarily in a nonimpact sport, e.g., swimming, before arrival at the Naval Academy (Table 4). This finding may indicate that a lack of sufficient previous strain-generating, impact-loading exercise may increase susceptibility to stress fracture injury during subsequent intense physical training periods [20,34 –36]. Milgrom et al. [37] reported that participation in ball sports, such as basketball (n = 3 in our study), for at least 2 years before starting basic training significantly reduced the incidence of stress fractures during basic training. Like other investigators [5,8,9,38], we found that female recruits experienced a higher relative incidence of stress fractures at the Naval Academy than did male recruits. Previous studies examining risk factors have suggested that this higher incidence of stress fractures in young women may be secondary to decreased BMD associated with eating disorders and irregular menses [5,8,38]. However, we found no significant differences between female subjects and

female controls in terms of age at menarche or the number of reported menstrual periods in the previous 12 months. We recognize that some women may experience luteal phase disruption, anovulation, and subclinical menstrual disorders, which may contribute to a hypoestrogenic state and therefore may contribute to loss of bone [39]. However, all the women in this study reported normal cyclic menstrual function. The women in our study showed tibia BMD and TBBMC values that were close to or within the Naval Academy 95% confidence interval for the particular bone site measured [26] (Table 5); there were no indications that any of the women in our study were at risk for an eating disorder, and we found no evidence of the Female Athlete Triad [40] in any of our female subjects. Therefore, findings from our study do not support an association between female plebes with stress fracture and altered menstrual function or eating disorders. In our study, TBBMC was significantly lower in men with stress fractures, but not abnormally low. The mean TBBMC of the fracture subjects was very close to the lower limit contained within the 95% confidence interval for 87 men [25] and 165 women [26] derived from previous reports of normal men and women at the Naval Academy who did not experience a bone injury. The mean BMD for the distal tibia was not significantly lower in male or female subjects than in their respective controls. This finding is supported by Giladi et al. [38], who showed no difference in BMD at a point on the tibia located 8 cm above the ankle mortise. The 95% confidence interval derived from our previous work with midshipmen contains the distal tibia BMD values for our female subjects [26] (Table 5) but not, by a small margin, those for male subjects [25] (Table 5). Beck et al. [7] reported findings similar to ours for tibia BMD of male and female Marine Corps recruits with and without stress fracture. However, they reported that tibia BMD in women (but not in men) with stress fracture was significantly lower than that in their respective uninjured recruits [7]. In our study, the distal tibia measurement was approximately 11.4 cm above the medial malleolus in female participants and 12.4 cm above the medial malleolus in male participants. For male subjects in the current study, this distal tibia site was very close to the level 1 site (13.7 cm) on the

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

tibia reported by Milgrom et al. [35], which corresponds to the narrowest width in the mediolateral plane with the tibia rotated 15j internally. Because we used DEXA methodology in one plane with the foot rotated 17.5j medially, we have reported only mediolateral structural characteristics for our subjects. We report CSMI values in the mediolateral orientation for both male subjects and controls, but our values do not appear to be different from the values reported by Milgrom et al. [35]. Our distal tibia section modulus values are not significantly different nor are they predictive of stress fracture risk. However, when we calculated the bone strength indices (Table 6), we found that the indices were significantly lower in our subjects compared to controls due to the significantly longer tibias of the subjects. Bending strength is proportional to the fourth power of the radius for long tubular bones such as the tibia. Thus, a slight decrease in width of the tibia results in a large decrease in resistance to bending. Because the size of bones throughout the skeleton is proportional, a finding of decreased tibia bone width should correlate to overall smaller bones throughout the skeleton, and, hence, smaller bending strengths in all bones throughout the skeleton [7,12,21,27,35]. Although previous studies indicated that geometric parameters were in large part responsible for bone strength differences [7,21,35,38], we found no significant differences between subjects and controls for section modulus. Therefore, we suggest that the difference in bone strength may be secondary to composition rather than to cross-sectional geometry. To determine the structural geometry of the distal tibias of our subjects, we used the algorithms of Beck et al. [7,21] to derive the mechanical properties of each subject via the DEXA data. DEXA technology is widely used to evaluate the skeleton and its mineral status. Its measurements are precise and accurate, it has low radiation exposures, and it affords short examination time. However, DEXA technology is limited by its planar nature, providing only area density or area content [7,12,20,21]. Although structural geometric characteristics may be derived from DEXA images, these measures are accurate only in the plane of the image, and they only partially describe the geometric properties governing bone strength [7,20,21,35]. The mechanical properties of bones show the same characteristics as man-made loadbearing structures, but the skeleton adapts to stress and strain by altering its shape (Wolff’s Law) and bone mineral content. Bone factors that affect the risk of stress fracture include porosity, mineralization, density, trabecular and cortical architecture, and fatigue microdamage, which reduces the elastic modulus [16,23,41]. Our study comparing subjects with stress fracture to matched controls showed no differences in BMD and structural geometry of the distal tibia. The muscles of the body provide the mechanism for movement of skeletal structures through the joints, supply a substantial degree of protection to the skeleton from forces that can cause skeletal injury, help maintain posture and

813

locomotion, and protect bone from bending under impact loading. Complex neuromuscular reflexes manage forces applied to the axial skeleton from activities such as lifting, throwing, running, and jumping. Muscular fatigue engendered by strenuous exercise and unpreparedness (cognitive fatigue) secondary to physical exhaustion and lack of sleep are likely major contributors to skeletal injury [42]. Stress fractures are a common overuse skeletal injury in young military recruits [4,6 –8,38,43], and there appears to be a correlation between the development of such fractures and the level and pattern of activity [12,42 –44]. In our subjects, stress fracture incidence increased with the cumulative number of miles run during the morning exercise training periods (Fig. 1). In the tibias of susceptible individuals, increased weight-bearing physical activity (e.g., approximately 104 load cycles, or approximately 4 weeks of physical training) is likely to create localized peak strains that can result in a stress fracture secondary to muscular fatigue [7,10,18,23,39,42,45]. Most of the stress fractures in our study were confirmed in the tibia after < 4 weeks of plebe summer physical training. Muscle fatigue is likely a contributor to stress fracture in plebes [7,14,18,21,23,42,45,46]. The combination of significantly smaller thigh girth in both male and female subjects and the significant loss of body weight, indicating a hypocaloric state that would lead to increased muscular fatigue and reduced muscular function during the summer training program, provides evidence of reduced muscular support and protection for the bones of the lower extremity. Other studies have reported that men undergoing a simulated march on a treadmill had significant fatigue in muscles that support the lower leg and an increased strain rate on the tibia [16,19,23,45]. A recent study of men performing a 21K hill walk on lowenergy or high-energy food ration showed that reaction time, balance, and the ability to maintain body temperature was significantly impaired during the low-energy trial. The low-energy trial also resulted in greater fatigue and a significant number of lower-extremity musculoskeletal injuries [47]. In 1993, Yoshikawa et al. [46] reported a similar finding in a dog model. Muscle fatigue and the resulting increased bone strain may contribute to stress fracture injury after daily strenuous exercise. Thus, fatigue in the musculature of the lower leg is consistent with the observed incidence of stress fracture and ankle sprain injury in military recruits undergoing rigorous basic training [7,19,44]. Our finding that thigh girth was significantly (approximately 7.5%) smaller in subjects than in controls supports the work of Beck et al. [7,21], who reported a similar finding in Marine Corps recruits. Thus, smaller thigh girth in subjects than in controls provides evidence that the leg muscles in fracture subjects were less likely to generate enough force to protect bone from unnecessary bending [7,18,19,42,46]. This finding is supported in part by the fact that male subjects performed 25 fewer push-ups than did

814

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

male controls, an indication of lower whole body muscular strength and endurance in the injured male plebes at entry into the Naval Academy. Additionally, an important finding of this study was the significant loss of body mass in subjects compared with their matched controls at the time of the diagnosis of a stress fracture. Male subjects lost 3.69 F 0.57 kg and female subjects lost 1.05 F 0.69 kg between Day 1 and the date of diagnosis. Controls also lost weight, but the loss was not statistically significant. The subjects’ significant weight loss indicates a sustained negative energy balance during the preinjury training period. This hypocaloric state among midshipmen may be the result of several factors. At the Naval Academy, plebes have approximately 20 min to consume a meal. Three meals per day are designed to deliver approximately 5000 kcal/day consisting of 40% carbohydrate, 40% fat, and 20% protein (personal communication, Naval Academy dietician). Fresh fruits and vegetables, salads, and dairy products are freely available. During plebe summer, between-meal snacks are extremely limited because plebes may not remove food from the mess hall and because access to food and beverage vending machines is not permitted. If a plebe is not hungry at mealtime or cannot finish the meal, the calories consumed will likely be inadequate to replace calories expended. The evening meal is served at 6:00 pm, and no access to snacks results in plebes going without food for nearly 13 h between that meal and breakfast at 7:30 am (Table 3). In addition, 90 min of morning training occurs before the plebe has breakfast. As shown by the work of Costill et al. [13], muscle glycogen is depleted over 3 days of intense training when a diet consisting of just 40% carbohydrate is consumed, a diet similar to that served at the Academy. This 40% carbohydrate diet, coupled with the lack of food for more than 12 h before intense physical training, may result in significant glycogen depletion in the leg muscles of plebes within days of beginning the mandatory plebe summer physical training program. The physical, mental, and emotional stress of plebe summer, elevated temperature and humidity, and lack of residence hall air conditioning (heat stress) may also contribute to a plebe’s reduced caloric intake [13,14,48]. A sustained negative energy balance also may negatively affect the muscle’s ability to recover from physical exercise [13] and may result in reduced bone collagen synthesis [41]. Therefore, the leg muscles of the subjects in our study were likely chronically fatigued and unable to provide support to the bones of the lower extremity [7,14,16,18,19,42,45,46]. It also is likely that bone collagen synthesis was impaired, which may have contributed to an increased risk of stress fracture in plebes with acute negative energy balance [41]. A limitation of our study may be that, in contrast to the work of Beck et al. [7,21], we did not perform DEXA scans on the entire plebe complement at the

beginning of summer training. Although obtaining DEXA scans for 1200 recruits would provide much information, such scans would have required more than 600 h to complete, and we do not believe that the data would substantially improve the study. In contrast to the work of Beck et al. [7,21], we studied one entire cohort of recruits during a single training cycle rather than recruiting subjects over a 15-month period from several training cycles. We assumed that there would be no substantive change in bone mass, bone density, and structural geometry in the few weeks from onset of the training period and the time of the DEXA scan. In fact, subjects showed BMD measures that were within the ‘‘normal’’ reference population for the Norland XR36 and close to or within the 95% confidence interval [25,26] for BMD and TBBMC of several hundred uninjured midshipmen previously measured during plebe summer training. Another study limitation is that plebes were not weighed regularly (i.e., weekly) and therefore it was not possible to document precisely the change in body weight during the summer training program. However, our data clearly show that plebes sustaining a stress fracture injury lost significant body weight by the time of their stress fracture diagnosis compared with uninjured matched controls. We believe the strength of our study, and the major difference from other studies of stress fracture injury, lies in the careful matching of injured young men and women to uninjured controls based on gender, age, BMI (height and weight), and aerobic physical performance at the beginning of an intensive summer training period. We conclude that an acute negative energy balance contributes to significant weight loss in some young military recruits. Sustained negative energy balance in young recruits undergoing basic military training may pose a risk of increased muscular fatigue, reduced bone collagen synthesis, and reduced muscular support of the long bones of the lower extremity. Therefore, a sustained negative energy balance during basic training may contribute to stress fracture injuries in some young men and women. The etiology of stress fractures continues to be a multifactorial conundrum. Additional research may confirm our preliminary data, which suggests that some factors that contribute to these injuries may be modified and therefore may lead to a reduction in stress fracture injury in young military recruits.

Acknowledgments The authors gratefully acknowledge Thomas Beck, PhD, Department of Radiology, School of Medicine, The Johns Hopkins University, Baltimore, MD, for providing the software for the analysis of the distal tibia DEXA scans. The Chief, Navy Bureau of Medicine and Surgery, Washington, D.C., Clinical Investigation Program, sponsored this study.

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

References [1] Armstrong RB, Warren GL, Warren JA. Mechanisms of exerciseinduced muscle fibre injury. Sports Med 1991;12(3):184 – 207. [2] Nindl BC, Friedl KE, Frykman PN, Marchitelli LJ, Shippee RL, Patton JF. Physical performance and metabolic recovery among lean, healthy men following a prolonged energy deficit. Int J Sports Med 1997;18(5):317 – 24. [3] Wasserman DH. Regulation of glucose fluxes during exercise in the postabsorptive state. Annu Rev Physiol 1995;57:191 – 218. [4] Breithaupt J. Zur pathologie des menschlichen fuses. Med Ztg Berlin 1855;24:169 – 71, 175 – 77. [5] Myburgh KH, Hutchins J, Fataar AB, Hough SF, Noakes TD. Low bone density is an etiologic factor for stress fractures in athletes. Ann Intern Med 1990;113(10):754 – 9. [6] Scully TJ, Besterman G. Stress fracture—A preventable training injury. Mil Med 1982;147(4):285 – 7. [7] Beck TJ, Ruff CB, Shaffer RA, Betsinger K, Trone DW, Brodine SK. Stress fracture in military recruits: gender differences in muscle and bone susceptibility factors. Bone 2000;27(3):437 – 44. [8] Black JR. Stress fractures of the foot in female soldiers: a two-year survey. Mil Med 1982;147(10):861 – 2. [9] Protzman RR, Griffis CG. Stress fractures in men and women undergoing military training. J Bone Joint Surg 1977;59A(6):825. [10] Martin RB. Determinants of the mechanical properties of bones. J Biomech 1991;24(Suppl 1):79 – 88. [11] Shaffer RA, Brodine SK, Almeida SA, Williams KM, Ronaghy S. Use of simple measures of physical activity to predict stress fractures in young men undergoing a rigorous physical training program. Am J Epidemiol 1999;149(3):236 – 42. [12] Milgrom C, Giladi M, Simkin A, Rand N, Kedem R, Kashtan H, et al. An analysis of the biomechanical mechanism of tibial stress fractures among Israeli infantry recruits A prospective study. Clin Orthop 1988;231:216 – 21. [13] Costill DL, Bowers R, Branam G, Sparks K. Muscle glycogen utilization during prolonged exercise on successive days. J Appl Physiol 1971;31(6):834 – 8. [14] Fink WJ, Costill DL, Van Handel PJ. Leg muscle metabolism during exercise in the heat and cold. Eur J Appl Physiol Occup Physiol 1975;34(3):183 – 90. [15] Bennell KL, Malcolm SA, Brukner PD, Green RM, Hopper JL, Wark JD, et al. A 12-month prospective study of the relationship between stress fractures and bone turnover in athletes. Calcif Tissue Int 1998; 63(1):80 – 5. [16] Gefen A. Biomechanical analysis of fatigue-related foot injury mechanisms in athletes and recruits during intensive marching. Med Biol Eng Comput 2002;40(3):302 – 10. [17] Giladi M, Milgrom C, Stein M, Kashtan H, Margulies J, Chisin R, et al. External rotation of the hip A predictor of risk for stress fractures. Clin Orthop 1987;216:131 – 4. [18] Gollhofer A, Komi PV, Miyashita M, Aura O. Fatigue during stretchshortening cycle exercises: changes in mechanical performance of human skeletal muscle. Int J Sports Med 1987;8(2):71 – 8. [19] Mizrahi J, Verbitsky O, Isakov E. Fatigue-related loading imbalance on the shank in running: a possible factor in stress fractures. Ann Biomed Eng 2000;28(4):463 – 9. [20] Sievanen H, Kannus P, Nieminen V, Heinonen A, Oja P, Vuori I. Estimation of various mechanical characteristics of human bones using dual energy X-ray absorptiometry: methodology and precision. Bone 1996;18(1 Suppl.):17S – 27S. [21] Beck TJ, Ruff CB, Mourtada FA, Shaffer RA, Maxwell-Williams K, Kao GL, et al. Dual-energy X-ray absorptiometry derived structural geometry for stress fracture prediction in male U.S Marine Corps recruits. J Bone Miner Res 1996;11(5):645 – 53. [22] Bennell K, Matheson G, Meeuwisse W, Brukner P. Risk factors for stress fractures. Sports Med 1999;28(2):91 – 122.

815

[23] Burr DB. Bone, exercise, and stress fractures. Exerc Sport Sci Rev 1997;25:171 – 94. [24] Jones BH, Thacker SB, Gilchrist J, Kimsey Jr CD, Sosin DM. Prevention of lower extremity stress fractures in athletes and soldiers: a systematic review. Epidemiol Rev 2002;24(2):228 – 47. [25] Armstrong III DW, Shakir KMM, Drake III AJ. Dual X-ray absorptiometry total body bone mineral content and bone mineral density in 18- to 22-year-old Caucasian men. Bone 2000;27(6):835 – 9. [26] Drake AJK, Armstrong III DW, Shakir KMM. Bone mineral density and total body bone mineral content in 18- to 22-year-old women. Bone 2004;34(6):1037 – 43. [27] Selker F, Carter DR. Scaling of long bone fracture strength with animal mass. J Biomech 1989;22(11 – 12):1175 – 83. [28] Garner DM. Eating Disorder Inventory—2. Professional Manual. Odessa (FL): Psychological Assessment Resources, Inc.; 1991. [29] Janse de Jonge XAK. Effects of the menstrual cycle on exercise performance. Sports Med 2003;33(11):833 – 51. [30] Janse de Jonge XAK, Boot CRL, Thom JM, Ruell PA, Thompson MW. The influence of menstrual cycle phase on skeletal muscle contractile characteristics in humans. J Physiol 2001;530(Pt 1):161 – 6. [31] Pelkman CL, Chow M, Heinbach RA, Rolls BJ. Short-term effects of a progestational contraceptive drug on food intake, resting energy expenditure, and body weight in young women. Am J Clin Nutr 2001;73(1):19 – 26. [32] Matheson GO, Clement DB, McKenzie DC, Taunton JE, Lloyd-Smith DR, Macintyre JG. Stress fractures in athletes. A study of 320 cases. Am J Sports Med 1987;15(1):46 – 58. [33] Brudvig TJS, Gudger TD, Obermeyer L. Stress fractures in 295 trainees: a one-year study of incidence as related to age, sex, and race. Mil Med 1983;148(8):666 – 7. [34] Martin AD, McCulloch RG. Bone dynamics: stress, strain and fracture. J Sports Sci 1987;5(2):155 – 63. [35] Milgrom C, Giladi M, Simkin A, Rand N, Kedem R, Kashtan H, et al. The area moment of inertia of the tibia: a risk factor for stress fractures. J Biomech 1989;22(11 – 12):1243 – 8. [36] Swissa A, Milgrom C, Giladi M, Kashtan H, Stein M, Margulies J, et al. The effect of pretraining sports activity on the incidence of stress fractures among military recruits. A prospective study. Clin Orthop 1989;245:256 – 60. [37] Milgrom C, Simkin A, Eldad A, Nyska M, Finestone A. Using bone’s adaptation ability to lower the incidence of stress fractures. Am J Sports Med 2000;28(2):245 – 51. [38] Giladi M, Milgrom C, Simkin A, Danon Y. Stress fractures. Identifiable risk factors. Am J Sports Med 1991;19(6):647 – 52. [39] Loucks AB. Physical health of the female athlete: observations, effects, and causes of reproductive disorders. Can J Appl Physiol 2001;26(6):S176 – 85. [Suppl.]. [40] Cobb KL, Bachrach LK, Greendale G, Marcus R, Neer RM, Nieves J, et al. Disordered eating, menstrual irregularity, and bone mineral density in female runners. Med Sci Sports Exerc 2003; 35(5):711 – 9. [41] Zanker CL, Swaine IL. Responses of bone turnover markers to repeated endurance running in humans under conditions of energy balance or energy restriction. Eur J Appl Physiol 2000;83(4 – 5):434 – 40. [42] Radin EL. Role of muscles in protecting athletes from injury. Acta Med Scand Suppl 1986;711:143 – 7. [43] Milgrom C, Giladi M, Stein M, Kashtan H, Margulies JY, Chisin R, et al. Stress fractures in military recruits A prospective study showing an unusually high incidence. J Bone Jt Surg 1985;67B(5): 732 – 5. [44] Almeida SA, Williams K, Shaffer RA, Luz JT, Badong KE, Brodine SK. A physical training program to reduce musculoskeletal injuries in U.S. Marine Corps recruits. Technical document 97-2B. San Diego: Naval Health Research Center; 1997. [45] Fyhrie DP, Milgrom C, Hoshaw SJ, Simkin A, Dar S, Drumb D, et al. Effect of fatiguing exercise on longitudinal bone strain as related to stress fracture in humans. Ann Biomed Eng 1998;26(4):660 – 5.

816

D.W. Armstrong III et al. / Bone 35 (2004) 806–816

[46] Yoshikawa T, Mori S, Santiesteban AJ, Sun TC, Hafstad E, Chen J, et al. The effects of muscle fatigue on bone strain. J Exp Biol 1994; 188(3):217 – 33. [47] Ainslie PN, Campbell IT, Frayn KN, Humphreys SM, MacLaren DP, Reilly T. Physiological, metabolic, and performance implications of a

prolonged hill walk: influence of energy intake. J Appl Physiol 2003; 94(3):1075 – 83. [48] Costill DL, Cote R, Miller E, Miller T, Wynder S. Water and electrolyte replacement during repeated days of work in the heat. Aviat Space Environ Med 1975;46(6):795 – 800.