Professional Documents
Culture Documents
Alana D. Cline, PhD, RD, G. Richard Jansen, PhD, and Christopher L. Melby, PhD
Department of Food Science and Human Nutrition, Colorado State University, Ft. Collins
Key words: stress fracture, bone density, physical activity, calcium, exercise
Objective: To identify characteristics and factors associated with increased risk for stress fractures in
military women.
Design: Case-control study to retrospectively examine physical activity, prior calcium intake, and bone
density as predictors of stress fractures.
Setting: A military training installation which incorporates physical training for women.
Subjects: Forty-nine female soldiers with confirmed stress fractures (cases) and 78 female soldiers with no
orthopedic injuries (controls), aged 18 to 33 years.
Measures: Retrospective self-reports of habitual exercise, sports participation, and food intake; current
height, weight, and body mass index (BMI); demographic variables (age, ethnicity, menstrual patterns, smoking
habits); and bone density on radiologically defined stress fractures.
Results: Cases and controls were similar in height, weight, and BMI. Measurements of bone density (g/cm2)
at the trochanter (cases, 0.7760.09; controls, 0.7760.08); femoral neck (cases, 0.9460.10; controls,
0.9460.09); Ward’s triangle (cases, 0.9160.11; controls, 0.9360.11); lumbar spine (cases, 1.2160.12; controls,
1.2460.10); and radius shaft (cases, 0.6760.09; controls, 0.6860.05) were not different between groups.
Calcium intake was not different between groups (cases, 11546751 mg/day; controls, 9446513 mg/day) and did
not correlate with bone density (r50.01 to 20.06 at four sites). Sports participation positively correlated with
bone density in the hip (r50.49). Leisure activity energy expenditure (kcal/day) tended toward association with
lower stress fracture risk as expenditure level increased (p50.06).
Conclusion: Stress fracture in female Army recruits was not correlated with bone density or calcium intake
during adolescence, although a weak relationship to prior physical activity was observed.
Disclaimer: “The opinions or assertations contained herein are the private views of the author(s) and are not to be construed as official or as reflecting the views of the
U.S. Army or the Department of Defense.”
Presented in part at the 1995 Experimental Biology Conference, Atlanta GA.
Address reprint requests to: Alana D. Cline, PhD, RD, Pennington Biomedical Research Center, Louisiana State University, 6400 Perkins Road, Baton Rouge, LA
70808-4124.
Journal of the American College of Nutrition, Vol. 17, No. 2, 128 –135 (1998)
Published by the American College of Nutrition
128
Stress Fractures in Female Army Recruits
of stress fracture injury during basic training than men, with were identified upon medical examination during evaluation
approximately 1 to 3% and 10 to 12% of new male and female and treatment of pain. A small group of women (cases) with
recruits, respectively, being affected [4,6]. Although studies of incomplete questionnaires were included in the study because
female athletes [7,8] and ballerinas [9,10] have provided some valuable data on physical characteristics and bone density were
useful information investigating the relationships of bone den- available. All stress fractures were confirmed by bone scan, and
sity, calcium intake, or amenorrhea to stress fractures, there are patients were referred to the principal investigator for recruit-
no studies which have investigated these relationships in fe- ment into the study. Control subjects were randomly selected
male military recruits. Several military studies have considered from women reporting to the troop medical clinic for non-
the relationship of physical activity to stress fractures [4,5,11]. orthopedic, minor health problems. All subjects gave their
However, there is no current consensus on what factors play the informed consent, and the Committee on Human Research at
greatest role in incidence of stress fracture. Colorado State University and the Brooke Army Medical Cen-
With the rising recruitment of women into the armed forces, ter Clinical Investigation Committee approved the research plan.
data are needed on their response to physical training and on Demographic information collected included age, race, geo-
any physical differences that may limit their performance. At graphic location during adolescence, date of menarche, men-
the same time, the number of women engaged in sports is strual patterns, use of oral contraceptives, and smoking habits;
increasing. As a result, there has been a significant increase in anthropometric data collected were height, weight, and BMI
the number of lower extremity stress fractures in women. In the (kg/m2).
few studies that have compared men and women, it has been
shown that women have a higher incidence of stress fractures
than men while participating in the same physical activities.
Food Frequency Questionnaire
The distribution patterns among civilian runners and joggers To evaluate the nutrient intake of participants, an abbrevi-
are very similar to those stated previously for military trainees ated version of the Health Habits and Diet Questionnaire by
[3,6,12]. Block et al [13] was used, with a slight modification in the
The underlying objective of this study was to identify those instructions to ask participants to recall food habits during
characteristics and factors associated with increased risk for adolescence. The questionnaire began with the following state-
stress fractures in military women. The hypotheses tested were ment: “This section is about your usual eating habits. Thinking
that female soldiers who experience stress fractures (cases) back over your TEENAGE (ages 13 to 18) years, how often did
compared to female soldiers without stress fractures (controls) you usually eat the foods listed on the next page?” An addi-
have: a) decreased bone density, b) a history of less physical tional question asked if participants had limited intake of dairy
activity during high school than those with no stress fractures, products for any reason during the same time period. The
and c) a history of lower dietary calcium consumption. The 60-item questionnaire evaluates 18 major nutrients and in-
correlation between bone density and body mass index (BMI) cludes foods representing approximately 93% of total United
was also investigated. States caloric consumption.
Validation of use of a food frequency questionnaire (FFQ)
to recall diet 4 years in the past has been reported by Willett et
al [14], indicating that a self-administered FFQ can be used to
MATERIALS AND METHODS provide information about previous diet. Reasonable correla-
tions with diet 10 to 15 years previous were obtained by
Subjects interview with a FFQ for a variety of nutrients by Sobell et al
The research design was a case-control study of female [15], and Frazier et al [16] reported reproducibility with use of
Army recruits who had recently completed their 12-week basic a FFQ to measure adolescent diet recall in women aged 40 to
training course. Volunteers were asked to complete evaluation 65, suggesting that recall is reasonably reproducible.
procedures, which included anthropometric data, diet and ex- Nutrient estimates for dietary assessment were provided by
ercise questionnaires, and bone density measurement. A total of computer software and are based on the NHANES II nutrient
141 women (63 cases, 78 controls) volunteered to participate, content database [17]. Questionnaires were designed to be
of which 51 (38 cases, 13 controls) received bone density self-administered; they were returned directly to the study
evaluation, and 127 (49 cases, 78 controls) completed ques- dietitian who conferred individually with each subject to verify
tionnaires. There was a disparity between number of women information before coding for computer analysis.
completing questionnaires and those receiving bone density
evaluation due to high drop-out rates by controls for density
evaluation. Study volunteers were reluctant to miss additional
Exercise Questionnaire
training time for non-medical appointments for density evalu- Exercise and energy expenditure from leisure and sports
ation so withdrew from further participation in the study after activities in the year immediately prior to entry into the Army
completing questionnaires. Women with stress fractures (cases) were evaluated by a modification of the Minnesota Leisure
Activities Questionnaire [18], which is designed to assess lei- using a two-tailed test and a 5% level of significance for each
sure activities during the past year. Use of this modified ques- test.
tionnaire has been validated and reported by Jones et al [19] on Relationships between variables were examined using sim-
numerous studies with Army recruits. Energy expenditure from ple correlation procedures, and stepwise multiple regression
activities during the prior year was calculated using reference analysis was used to identify major predictors of bone density.
formulas and intensity codes [20,21]. Participation in organized The stepwise procedure was used so that independent vari-
or varsity sports during high school, and the individual’s per- able(s) remaining in the model would only be those that were
ception of her physical activity level (prior year) on a scale of statistically significant. Chi square analysis was used to exam-
1 (inactive) to 4 (very active) were also reported. ine differences between stress fracture and control subjects
when examined by race, smoking, menstrual status, oral con-
traceptive use, activity level, calcium, and exercise energy
Bone Density Evaluation
expenditure [24].
Single-photon absorptiometry (LUNAR SP2, Lunar Radia- Case and control groups were initially divided into non-
tion Co., Madison, WI) was used to determine bone mineral blacks and blacks for statistical analysis; Hispanics were in-
content and mineral density of the radius [22]. The site of cluded in the non-black group after analysis revealed no dif-
measurement was one-third of the distance from the ulnar ferences between groups with or without their inclusion. Of the
styloid process to the olecranon on the non-dominant arm, 29 black study participants, only seven received bone density
which contains approximately 95% cortical bone. Measurement evaluation. They did not differ from other participants in bone
was also made on the ulna and radius distal to the site of density, calcium intake, or physical activity. Because of the
juxtaposition of the radius and ulna at the wrist sites, which small sample size, no attempt was made to evaluate the influ-
contain predominantly trabecular bone. The coefficient of vari- ence of race on outcome variables in the study. Therefore, cases
ation (CV) of repeated measurements at these two sites using and controls were considered as single groups without dividing
this technique was approximately 2.1 and 2.3%. by race.
Bone mineral measurements of the lumbar spine and hip,
both predominantly trabecular bone, were made using dual-
photon absorptiometry (LUNAR DP3, Lunar Radiation Co.,
Madison, WI) which permits a quantitative assessment of skel- RESULTS
etal bone mineral by use of a gadolinium-153 radionuclide
source [23]. Precision error (CV) of this technique was approx- Table 1 shows no significant differences in age, height,
imately 2.0% for the spine and 2.7% for the hip. All scans were weight, and BMI between injured (case) and control soldiers.
performed by the same experienced technician over a 1-month Because of the narrow age range for enlistment into the mili-
period; standardized calibration techniques were used for qual- tary, there was little between group heterogeneity in age.
ity control at the start of each day.
%,1 SD2
Cases Controls
Location P Standards1
(n538) (n513)
Case Control
Trochanteric 0.7760.09 0.7760.08 .952 0.8160.12 17 15
Femoral neck 0.9460.10 0.9460.09 .861 0.9960.12 19 15
Ward’s triangle 0.9160.11 0.9360.11 .668 0.9260.14 11 8
Lumbar (L2-L4) 1.2160.12 1.2460.10 .443 1.2760.13 27 15
Radius shaft 0.6760.09 0.6860.05 .552 0.7160.05 48 38
Distal radius 0.4060.06 0.3660.08 .062 0.3660.05 14 31
Mean6SD.
1
Young normal bone density values for women [25].
2
Percentage of individual subjects,1 SD below age-predicted standard.
Physical Activity and Amenorrhea vs. Bone Density square53.7, df51, p50.06) almost reaching the 0.05 level of
statistical significance.
The incidence of amenorrhea (temporary cessation of men-
Thirty-seven percent of all subjects reported involvement in
ses) during basic training was 53% and 42% among injured and
varsity sports while in school, with no differences between
control soldiers, respectively, but the difference did not reach
cases and controls. Correlations were found between bone
statistical significance (Table 3). A large number of women,
mineral density of the trochanter and height (r50.36, p,0.01),
63% of injured subjects and 48% of control subjects, were also
weight (r50.29, p,0.05), and high school organized or varsity
on oral contraceptives prior to entering the Army (chi
sports participation (r50.49, p,0.01).
square52.22, df51, p50.14). Of 23 amenorrheic cases, 17
Stepwise multiple regression analysis was used to examine
individuals were also on oral contraceptives. Of 31 amenor-
possible relationships between independent and dependent
rheic controls, 17 were on oral contraceptives. Average age at
variables. The dependent variables of bone mineral density
menarche was 13 years for both groups. A total of 28 women
of the radius (BMDR), bone mineral density of L2-L4 spine
(24%) reported that they had experienced menstrual irregular-
(BMDL2L4), bone mineral density of the femoral neck
ities (less than one menstrual cycle a month) prior to entering
(BMDFN), bone mineral density of Ward’s triangle (BMDWT),
the service; no correlations were seen with bone density or
and bone mineral density of the trochanter (BMDTR) were com-
stress fracture. The percentage of smokers was not significantly
pared with the independent variables of all nutrients, physical
different: 25% of injured subjects and 37% of control subjects
characteristics, demographic characteristics, and stress fracture
smoked.
occurrence. Weak associations at three sites, negative for
There were no significant differences between groups in
BMDWT with percent fat calories eaten (r50.43, P,0.05), pos-
physical activity energy expenditure prior to active duty: in-
itive BMDTR with sports participation (r50.47, p,0.05), and
jured soldiers averaged 611 kcal/day vs. control soldiers who
negative BMDL2L4 with stress fracture (r50.44, p,0.05) were
averaged 617 kcal/day. Chi square analysis, comparing high
most likely due to chance and are not considered to be informative.
exercise (.400 kcal/day) vs. low exercise (,400 kcal/day)
No independent variables tested were significantly associated with
showed that only 42% of injured subjects were in the high
bone mineral density at any skeletal site.
exercise category for leisure-time activities compared to 59%
of control subjects, with differences between groups (chi
Calcium and Nutrient Intake vs. Bone Density
Table 3. Physical, Athletic, and Menstrual Histories of Reported intakes of protein, carbohydrate, fat, and energy
Study Participants during adolescence were not significantly different between
injured and control subjects (Table 4). The intake level of most
Cases (n549) Controls (n578)
nutrients approximated or exceeded Recommended Dietary
Physical activity (kcal/day)1,2 6116836 6176540
Allowances (RDA) levels [26]. Intakes of vitamins, minerals,
Sports participation2,3 2.661.1 2.761.2
Age at menarche (year)2 13.161.9 13.061.6 and fiber and their percentages of RDA also were not different
Training amenorrhea (%)4 53 42 between groups (Table 4). Regular vitamin supplementation
Oral contraceptives (%)5 63 48 (multivitamin plus iron) was reported by 31% of cases and 27%
Cigarette smoking (%)5 25 37 of controls. Additionally, four subjects (controls) reported tak-
1
Calculated for year prior to basic training. ing supplemental vitamins A, C, and E; six cases and six
2
Mean6SD.
3
controls, supplemental iron; and one subject (control), calcium.
Prior to enlistment: 15no sports, 25individual sports, 35organized team sports
(intramurals), 45varsity sports.
Injured subjects averaged a dietary calcium intake of
4
During basic training. 11546751 mg/day, while control subjects averaged 9436513
5
Begun at least one year prior to enlistment; percentage of subjects. mg/day; with intakes ranging from 143 mg to 3407 mg. Chi
square analysis showed no difference between high (.800 mg) normal women in similar age groups for these variables
and low calcium (,799 mg) intake and stress fracture rates (chi [23,27]. Based on this finding, it appears that this sample of
square50.36, df51, p50.55). High calcium values were des- women enlisting in the Army had physical characteristics sim-
ignated as two-thirds of the RDA or greater to provide com- ilar to a cross-section of the general population of women the
parison with values used in previous studies. same age. Women who enter the military are screened for
Servings of dairy products consumed per week were higher medical problems, so they are likely to be in better health than
in the stress fracture (case) group (14.563.1) than in the control their civilian counterparts in this age group, in general. Al-
group (11.168.4) (p,0.05). Assessment of the relationship of though the mean height of this study group was comparable to
patterns of dairy consumption to bone density at each of the ballerinas with stress fractures [9], the ballerinas weighed less.
measured sites indicated that individuals who had at least three
Height and weight of this study group were comparable to
servings of dairy products per day (i.e., a dairy serving with
values in other studies of military females with stress fractures
each meal) (n510) had no significant difference in bone den-
[3–5]. Jones et al [19], in a study on basic trainees, reported
sity when compared with those who reported eating dairy
similar data for height, weight, and BMI. In the current study,
products less frequently (n526).
chi-square analysis revealed no significant relationship be-
No significant correlations were found in the relationship of
tween race, age, or height and stress fracture. These findings
individual nutrients (calcium, phosphorus, iron, sodium, potas-
sium, and dietary fiber) with bone density measurements. Sta- differ from those of Brudvig et al [1], who reported that race
tistical regressions of bone density measures BMDFN, BMDR, and age were factors related to the development of stress
BMDL2L4, BMDTR, and BMDWT, on calcium intake were fractures. Findings of this study are consistent with those of
not statistically significant (p.0.05), with correlation coeffi- Reinker and Ozburn [3], and Hopson et al [2], who found that
cients ranging from 0.009 to 0.055. these factors were unrelated to stress fracture risk.
Bone density values in the current study are lower than
those of white young women as reported by Mazess and Barden
DISCUSSION [25] (Table 2), but not significantly different. One individual
had a vertebral mineral density below the fracture threshold as
Stress Fractures vs. Bone Density defined by Riggs et al [27], of 0.965 g per square centimeter,
Weight, height, and BMI of this study population were not and 12 (10 cases) were in the range of 21 to 22.5 standard
different from comparable values in other studies of young deviations (SD) for osteopenia. Table 2 indicates percentages
of subjects with BMD more than 1 SD below reference value values reported in a study by Lewis [33] that compared athletes
for each site measured. and non-athletes. This suggests that women who enlist in the
Comparison of bone density values from the current study Army tend to be physically active prior to entry. Because
group with ballerinas who experienced stress fractures [9] exercise participation was equally high in both groups, the
showed there were no significant differences between stress effect of exercise on bone may explain bone density values that
fracture and non-stress fracture groups in both studies. Balle- are similar in both groups.
rinas had slightly lower bone density values than the soldiers, The findings of this study agree with those of previous
which may be a reflection of the body build of those girls most studies [2,4,11] that individuals who are less physically active
likely to participate in ballet. prior to entering the military are more likely to get stress
These results differed from a stress fracture study on ath- fractures. This differs from the finding of Ozburn et al [5], that
letes in South Africa [28], in which athletes with stress frac- physical conditioning is not a factor.
tures had significantly lower bone mineral densities than con- Exercise has a positive effect on bone density [34 –36],
trols; the lower density was in both trabecular and cortical which may be reflected in this study by correlations between
bone. varsity or organized sports participation and bone density, both
Winters et al [29] reported lumbar bone mineral densities requiring some degree of physical training. An exception to this
similar to those in the present study in comparisons of trained is a negative correlation between training and bone mineral
runners and moderately active controls. However, half of the density of the spine and femur in all subjects reported by
runners reported stress fractures, compared to one incidence in Myburgh [28]. Cross-sectional studies demonstrated that exer-
the controls, and they had significantly lower total bone cal- cise history in individuals indicated a greater bone mass, but the
cium per kg of soft lean tissue obtained from DEXA scans. difference between adults who exercised and those who did not
was less than 10% [34,37].
Physical Activity and Amenorrhea vs. Bone Density The only correlates in this study that appear to have a
relationship with bone density or stress fracture are high school
The few studies investigating stress fractures in female sports with bone density in the hip, and a trend toward signif-
athletes reported an association between amenorrhea and stress icance of leisure activity energy expenditure with stress fracture.
fracture [28,9]. The association most frequently discussed is
that of amenorrhea and lower bone density values, which then
appears to result in more stress fractures in amenorrheic ath- Calcium and Nutrient Intake vs. Bone Density
letes. The absence of menses reported by the current study
participants was temporary, and occurred only during their When comparing retrospective adolescent food consump-
basic training for a duration of approximately 2 to 3 months. tion data of current study participants with the report by
The short duration may explain why no relationship was evi- McKoy [38] on adolescents from the South, and with NHANES
dent between bone density values and stress fractures in am- III [39,40], intake of current study participants was higher in
enorrheic soldiers. This finding supports a study by Hetland et some nutrients, but differences were not large. Diet analysis in
al [30], which reported that women who began a running other studies utilized either a 3-day diary or a 7-day diary, with
program experienced irregularities in their menstrual cycles, the exception of the study by Frusztajer et al [9], which used a
but found no statistically significant relation between running food frequency questionnaire. Current study participants were
activity and bone mineral measurements or bone turnover. asked to recall diet habits from 1 to approximately 18 years in
Previous studies on athletes reported amenorrhea that lasted the past, which has been reported as being valid and a reason-
from 1 to 7 years or longer resulting in disturbances of estrogen able way of retrospectively assessing diet [14,15]. Consensus is
and other sex-steroid hormone levels. that correlation between past diet and recall of that diet is
Recently it has been reported that ovulatory disturbances greater than the impact of the current diet on recall. Good recall
which are common during exercise training may be detrimental has been seen in milk products, whole milk, fish, and alcohol,
to bone health, even in women who have apparently normal most likely because of the dominant role they play in consump-
menstrual cycles. Physically active women with shortened lu- tion patterns. Also to be considered is that adolescent diets may
teal phases or anovulatory cycles have shown lower BMD than have associated memories that enable better recall, and that
physically active women with normal cycles [29,31,32]. Be- meal consumption in the family situation may be more rigid
cause almost one-fourth of the subjects in this study had re- than adult self-selection. Although written instructions are pro-
ported recent menstrual irregularities, incidence equally distrib- vided with the questionnaires, it remains important for fol-
uted between the two groups, ovulatory disturbances associated low-up instructions and interviews with respondents to clarify
with bone loss may have been present which were not mea- answers and review questionable responses. This became ap-
sured in this study. parent in the present study as questionnaires were reviewed
Energy expenditure for leisure activity of over 600 kcal/day with volunteers upon completion.
in both groups of the current study is comparable with athlete Energy intake in the current study was, on the average,
20. Peterson MS, Peterson K: “Eat To Compete.” Chicago, IL: Year 38. McKoy H, Kenney MA, Kirby A, Disney G, Ercanli FG, Glover E,
Book Medical Publishers, Inc., 1988. Korslund M, Lewis H, Liebman M, Livant E, Moak S, Stallings
21. Berning JR, Steen SN: “Sports Nutrition for the 90’s.” Gaithers- SF, Wakefield T, Schilling P, Ritchey SJ: Nutrient intakes of
burg, MD: Aspen Publishers, 1991. female adolescents from eight southern states. J Am Diet Assoc
22. Cameron JR, Mazess RB, Sorenson JA: Precision and accuracy of 84:1453–1460, 1984.
bone mineral determination by direct photon absorptiometry. In- 39. National Center for Health Statistics. Advance Data. Energy and
vest Radiol 3:141–150, 1968. Macronutrient Intakes of Persons Ages 2 Months and Over in the
23. Mazess RB, Barden HS, Ettinger M, Johnston C, Dawson-Hughes United States: Third National Health and Nutrition Examination
B, Baran D, Powell M, Notelovitz M: Spine and femur density Survey, Phase 1, 1988–91. No 255, DHHS Publ No (PHS)95-1250,
using dual-photon absorptiometry in U.S. white women. Bone Hyattsville, MD: U.S. Dept of Health and Human Services, 1994.
Miner 2:211–219, 1987. 40. National Center for Health Statistics. Advance Data. Dietary intake
24. Steel RGD, Torrie JH: “Principles and Procedures of Statistics.” of vitamins, minerals, and fiber of persons ages 2 months and over
New York, NY: McGraw-Hill, 1980. in the United States: Third National Health and Nutrition Examina-
25. Mazess RB, Barden HA: Interrelationships among bone densitom- tion Survey, Phase 1, 1988–91. No 258, DHHS Publ No (PHS)95-
etry sites in normal young women. Bone Miner 11:347–356, 1990. 1250, Hyattsville, MD: U.S. Dept of Health and Human Services,
26. National Research Council: “Recommended Dietary Allowances,” 1994.
10th ed. Washington, DC: National Academy Press, 1989. 41. National Center for Health Statistics: Dietary Intake Source Data:
27. Riggs BL, Wahner HW, Dunn WL, Mazess RB, Offord KP, United States, 1976–80. Vital Statistics Series 11, No 231. DHHS
Melton LJ: Differential changes in bone mineral density of the Publ No (PHS)83-1681, Hyattsville, MD: U.S. Dept of Health and
appendicular and axial skeleton with aging. J Clin Invest 67:328– Human Services, 1983.
335, 1981. 42. Kenney MA, McCoy JH, Kirby AL, Carter E, Clark AJ, Disney
28. Myburgh KH, Hutchins J, Fataar AB, Hough SF, Noakes TD: Low GW, Floyd CD, Glover EE, Korslund MK, Lewis H, Liebman M,
bone density is an etiologic factor for stress fractures in athletes. Moak SW, Ritchey SJ, Stallings SF: Nutrients supplied by food
Ann Intern Med 113:754–759, 1990. groups in diets of teenaged girls. J Am Diet Assoc 86:1549–1555,
29. Winters KM, Adams WC, Meredith CN, Van Loan MD, Lasley 1986.
BL: Bone density and cyclic ovarian function in trained runners 43. Nieves JW, Golden AL, Siris E, Kelsey JL, Linsay R: Teenage and
and active controls. Med Sci Sports Exerc 28:776–785, 1996. current calcium intake are related to bone mineral density of the
30. Hetland ML, Haarbo J, Christiansen C, Larsen T: Running induces hip and forearm in women aged 30–39 years. Am J Epidemiol
menstrual disturbances but bone mass is unaffected, except in 141(4):342–351, 1995.
amenorrheic women. Am J Med 95:53–60, 1993. 44. Metz JA, Anderson JJB, Gallagher PN: Intakes of calcium, phos-
31. Prior JC, Vigna YM, Schechter MT, Burgess AE: Spinal bone loss phorus, and protein, and physical-activity level are related to radial
and ovulatory disturbances. N Engl J Med 323:1221–1227, 1990. bone mass in young adult women. Am J Clin Nutr 58:537–542,
32. Drinkwater BL: Exercise and bones: lessons learned from female 1993.
athletes. Am J Sports Med 24:S33–S35, 1996. 45. Ulrich CM, Georgiou C, Snow-Harter CH, Gillis DE: Bone min-
33. Lewis A: The effects of habitual activity patterns on the body eral density in mother-daughter pairs: relations to lifetime milk
composition and maximal aerobic capacity of young women. (The- consumption, and calcium supplements. Am J Clin Nutr 63:72–79,
sis). Minneapolis, MN: University of Minnesota, 1977. 1996.
34. Suominen H: Bone mineral density and long term exercise. an 46. King N, Fridlund KE, Askew EW: Nutrition issues of military
overview of cross-sectional athlete studies. Sports Med 16:316– women. J Am Coll Nutr 12:344–348, 1993.
330, 1993. 47. King N, Arsenault JE, Mutter SH, Champagne C, Murphy TC,
35. Turner JG, Gilchrist NL, Ayling EM, Hassall AJ, Hooke EA, Westphal KA, Askew EW: “Nutritional intake of female soldiers
Sadler WA: Factors affecting bone mineral density in high school during the U.S. Army basic combat training.” Natick, MA: US
girls. NZ Med J 105:95–96, 1992. Army Research Institute of Environmental Medicine, USARIEM
36. Greendale GA, Barrett-Connor E, Edelstein S, Ingles S, Haile R: Technical Report T94-17, 1994.
Lifetime Leisure Exercise and Osteoporosis. Am J Epidemiol
141(10):951–959, 1995.
37. Forwood MR, Burr DB: Physical activity and bone mass: exercise
in futility? Bone Min 21:89–112, 1993. Received May 1997; revision accepted October 1997.