- Email: [email protected]
Bone mineral status in adolescent girls: effects of eating disorders and exercise
JOURNAL OF ADOLESCENT HEALTH 2000;26:322–329
ORIGINAL ARTICLE
Bone Mineral Status in Adolescent Girls: Effects of Eating Disorders and Exercise BETTY R. CARRUTH, Ph.D., AND JEAN D. SKINNER, Ph.D.
Purpose: To compare whole-body, lumbar, total spine, and pelvis bone mineral density (BMD), body mass index (BMI), body composition, energy expenditure for physical activity, and dietary intake of adolescents, aged 16 –22 years. Methods: Three study groups included 25 girls with histories of eating disorders, 25 girls with no histories of eating disorders who exercised < 7 hours/week, and 15 girls with no history of eating disorders who exercised > 7 hours/week. Bone mineral density was measured by dual-energy x-ray absorptiometry (DEXA), body composition by bioelectric impedance and DEXA, energy expenditure by Personal Activity Computer, nutrient intake by 4-day dietary recalls/records, and BMI by measures of height/weight. General linear models, LSM ⴞ SEM, Student’s t-tests, and correlation analyses were used to determine group differences. Results: No significant differences in whole-body, spinal, and pelvis BMD were found among the three groups. Mean body fat (percent) was significantly higher (p ⴝ .0001) for the group with histories of eating disorders than other groups. Dietary intakes of adolescents with histories of eating disorders were significantly lower for energy (p ⴝ .0001), fat (p ⴝ .0001), calcium (p ⴝ .0007), vitamin D (p ⴝ .0180), and zinc (p ⴝ .0057) than those without eating disorder histories who exercised < 7 hours/week. Conclusion: Except for body fat (percent), measures of BMD, energy expenditure, and BMI were not significantly different among groups. Our data suggest that with full recovery from eating disorders, teenage girls can achieve normal bone mass and body composition. © Society for Adolescent Medicine, 2000
From the Department of Nutrition, University of Tennessee-Knoxville, Knoxville, Tennessee. Address correspondence to: Betty R. Carruth, Ph.D., Department of Nutrition, University of Tennessee, 1215 Cumberland Avenue, Knoxville, TN 37996-1900. Manuscript accepted September 15, 1999. 1054-139X/00/$–see front matter PII S1054-139X(99)00089-0
KEY WORDS: Bone mineral density Calcium Dual X-ray absorptiometry Eating disorders Exercise Vitamin D
The advent of low-radiation technology enables investigators to determine maximum bone mineral mass acquired during childhood (1) and to study determinants thereof in both normal and diseased states (2). In a cross-sectional study, 83% of adult bone mass existed in premenarcheal 12-year-old females (3). In a study of 2 to 20-year-old females, maximum mean values for whole-body bone mineral content (WBBMC) were found in the 16-year-old female group. A significant difference (p ⬍ .01) in WBBMC was noted between Tanner stages 1–2 and 2–3 but not for 3– 4 and 4 –5 (4). Tanner stage 2–3 in females occurs on the average around 11–12 years of age (5). Body size, eumenorrhea, and body composition influence attainment of bone mineral mass during adolescence. Late puberty and amenorrhea are risk factors for reduced bone mineral density (BMD). For 45 healthy prepubertal and pubertal girls, pubertal development and body size were important determinants of bone acquisition (6). In a cross-sectional study of 295 females, girls who had earlier menarche had higher BMD (7). Ninety-nine percent of the changes in wholebody mineral were related to increases in bone size. Body fat as a determinant of bone mineral status is not clearly understood. Although regular menses is positively associated with increased bone mineral mass, there is no specific level of fat mass at which menses occurs (8).
© Society for Adolescent Medicine, 2000 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
The relationship between extended periods of amenorrhea (estrogen deficiencies), low body weight, and bone mineral mass are well documented in adult females with anorexia nervosa (9,10). However, most published studies about females with eating disorders combine data for adolescents and adults and present group results (10 –13). Treatment outcome literature summarizes findings in the same manner (14). No studies in the literature that assessed bone mineral status of adolescent patients whose primary health care provider judged them to be rehabilitated from their eating disorder and who were not currently in a therapeutic regimen were found. In general, adolescents in published studies represent a clinic population characterized by amenorrhea, low body weight, food restriction, and/or purging behaviors (9,15). The duration of illness may not be treated as a covariant (11). The control group may represent a wide range of ages and/or of varying stature (16). Calcium intake; dietary supplement usage; and other lifestyle factors, such as smoking and alcohol, are not consistently reported (6,17). Body composition data are infrequently given for adolescent subjects. This study was planned to control for some factors that may differentially affect bone mineral density and body composition, (i.e., eating disorders, body size, lean body mass, exercise, and diet). The study objectives were to compare whole-body, spinal, and pelvis BMD (g/cm2), body composition (% fat and % lean body mass), body mass index (BMI), estimates of energy expenditure for physical activity, and dietary intakes of energy, macronutrients, calcium, vitamin D, magnesium, and zinc in three groups of eumenorrheic females aged 16 –22 years. We hypothesized that adolescent females with a history of eating disorders would have significantly lower whole-body, spinal, and pelvis BMD (g/cm2), BMI, and percent body fat (% BF) than the two comparison groups. The amount of exercise, as measured in this study, was hypothesized to be significantly higher for those with a history of eating disorders than the two other groups. Mean energy (kilocalories) and selected nutrient intakes of adolescents with a history of eating disorders were expected to be significantly lower than those of the two comparison groups.
Methods Subjects White, eumenorrheic, adolescent females with histories of eating disorders (n ⫽ 25) were referred from primary physicians and therapists. The health care
323
providers’ diagnosis and judgment about recovery status were the basis for acceptance into the study. The group with histories of eating disorders included those initially diagnosed with anorexia nervosa (AN) (n ⫽ 15), bulimia nervosa (BN) (n ⫽ 5), and anorexia– bulimia nervosa (AN/BN) (n ⫽ 5). Individuals were excluded from participating in the study if they were in therapy, received medication, or were amenorrheic (ⱕ 9 periods/year) or had missed more than three consecutive periods in the last year. The two comparison groups included: (1) 25 adolescents with no history of eating disorders who exercised ⱕ 7 hours/week, and (2) 15 adolescents who exercised ⬎ 7 hours/week. To address the issues of stature and age differences, the 25 adolescents with no history of eating disorders were matched for age, weight, and height with the 25 adolescents with histories of eating disorders. To address the influence of exercise apart from a history of eating disorders, two levels of exercise (ⱕ 7 hours/week and ⬎ 7 hours/week) were used in the comparison groups with no history of eating disorders. The comparison groups were recruited from schools, girls’ clubs, and gymnasium sports counselors and by posters. In addition to the criterion of eumenorrhea, other exclusion criteria included body weight (⬍ 25th or ⬎ 90th percentile), hormone/ endocrine disorders that negatively influence bone health, smoking, drinking ⬎ 200 g ethanol/week, receiving hormonal medications to regulate menses (except for oral contraceptive agents), birth weights ⬍ 2500 g, and ⬍ 36 weeks’ gestational age at birth. Adolescents aged ⱕ 18 years and their parents signed separate informed consent forms, which described research objectives, an assurance of confidentiality, and the potential risk/benefits to participants. Females aged 19 –22 signed informed consent forms. The protocol was approved by the committee for use of human subjects in research at the participating academic and medical institutions. Data Collection At the initial interview, one of two trained interviewers used a questionnaire that provided information about chronologic age, menstrual history, socioeconomic status, foods consumed in the last 24 hours, and the adolescents’ self-reported heights/weights. To prepare for the next interview, adolescents were instructed about keeping a 2-day food record (including a weekend day) and the use of a CALTRAC
324
CARRUTH AND SKINNER
Personal Activity Computer (Hemokenetics, Inc., Madison, Wisconsin). CALTRAC is the size of a small pocket computer that attaches to a belt or waistline of the garment and records the quantity of body movements. Estimates of energy expenditure (in kilocalories) were customized for weight, height, age, and gender. Participants wore the CALTRAC over 4 continuous days, including at least one weekend day. Interviewers were available to participants during the week and weekends to enhance cooperation and accuracy of recordings. The second interview occurred in a nuclear medicine clinic in an urban tertiary medical center. To determine nonpregnancy status prior to bone mineral assessments, a venipuncture blood sample (20 mL) was taken by a registered nurse and human chorionic gonadotropin level assayed. Individual blood profiles for dehydroepisterone sulfate, androstenedione, luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, and testosterone were analyzed by the medical center’s laboratory. Bone Mineral Density was measured with dualenergy x-ray absorptiometry (DEXA) (Model QDR Hologic 2000, Bedford, MA). Body composition was measured with RJL bioelectric impedance analyzer (BIA Model 101, RAL Systems, Detroit, Michigan). To compare bioelectric impedance analysis (BIA) with another measure of body composition, DEXA assessments were done on the 25 females with histories of eating disorders, 16 of the comparison group who exercised ⱕ 7 hours/ week, and the 15 girls who exercised ⬎ 7 hours/ week. Weights and heights were measured on a balance-beam scale and stadiometer. A registered dietitian checked the 2-day food records for completeness and took a 24-hour recall. Recordings for CALTRAC were checked and interviewers probed for events that could affect CALTRAC results. Analyses Frequencies and means for demographic and other group characteristics were determined using SAS (System for Statistical Analysis, Cary, North Carolina, 1988). Least Squares Means (LSM ⫾ SEM) were calculated and Student’s t-tests used for computing statistically significant differences between groups for percent body fat, percent lean body mass, BMI, height, weight, and CALTRAC. To predict factors that significantly contributed to BMD, general linear models (GLM) were developed for those with histories of eating disorders and the 25 girls who were matched for age, height, weight (exercised ⱕ 7
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
hours/week). For the eating-disorder group only, Pearson correlation coefficients for whole-body, spinal, and pelvis BMD (g/cm2) and the years since diagnosis of the eating disorder were computed and a two-tail test of significance performed. The degree of agreement between the adolescents’ self-reported weight/height and clinic measurements was determined by Bland-Altman estimates (18). Estimates of body composition (percent fat and lean body mass) were derived from information provided by the manufacturer of BIA, published studies (19,20), and DEXA results. Two 24-hour recalls and 2-day food records were coded using the Nutritionist IV software (Version 3.5.2 for Windows, N-Square Computing, San Bruno, California). Mean nutrient intakes for protein, calcium, magnesium, zinc, vitamin D, and energy (in kilocalories) were compared with the Recommended Dietary Allowances (RDAs) (21) and 1998 Dietary Reference Intakes (22). To assess mean differences in dietary intake by those with histories of eating disorders and the comparison groups, GLM procedures and MANOVA were performed with nutrients as the dependent variables.
Results Table 1 describes whole-body, lumbar spine, total spine, and pelvis BMD. These skeletal sites contain trabecular bone that may be negatively affected by eating disorders (23,24). No significant group differences for whole-body BMD or any skeletal sites were found or differences by diagnoses of AN, BN, or AN/BN for those with histories of disorders. With CALTRAC and years since onset of menses in a GLM, no significant differences among groups were found for any of the bone mineral results. The Pearson product-moment correlation coefficients computed between number of years since diagnosis of an eating disorder and BMD results were not significant. With CALTRAC and age at onset of menses in a GLM, no significant group differences in whole-body, spinal, or pelvis BMD were found (Table 1). For the total group with histories of eating disorders (n ⫽ 25), mean years since diagnosis was 3.1 years (range 1–5 years). Body composition results from BIA among the three groups were significantly different for % BF (Table 2). Group 3, who exercised ⬎ 7 hours/week had significantly lower % BF than those with a history of eating disorders (Group 1, p ⫽ .0001) and comparison Group 2 (p ⫽ .03), who exercised ⱕ 7
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
325
Table 1. Bone Mineral Density of Adolescent Females (Aged 16 –22 Years)
Group Group 1§ (n ⫽ 25) Group 2£ (n ⫽ 25) Group 3† (n ⫽ 15)
Lumbar Spine Pelvis Total Spine BMD* (g/cm2) (g/cm2) (g/cm2) (g/cm2) —————————————————————LSM ⫾ SEM ————————————————————— (range) 1.07 ⫾ .01 (0.91–1.24) 1.07 ⫾ .01 (0.94 –1.21) 1.10 ⫾ .07 (1.0 –1.23)
1.02 ⫾ .02 (0.79 –1.29) 1.04 ⫾ .03 (0.70 –1.24) 1.06 ⫾ .03 (0.83–1.31)
1.11 ⫾ .02 (0.86 –1.3) 1.12 ⫾ .02 (0.89 –1.27) 1.17 ⫾ .02 (1.03–1.33)
0.77 ⫾ .01 (0.67– 0.97) 0.79 ⫾ .02 (0.61– 0.97) 0.80 ⫾ .01 (0.73– 0.91)
* General linear model included exercise and gynecologic age. No significant differences among groups in whole-body bone mineral density (BMD) or other skeletal sites among the three groups. § Group 1—History of eating disorders. £ Group 2—No history of eating disorders, matched individually with Group 1 by age, height, and weight; exercised ⱕ 7 hours/week. † Group 3—No history of eating disorders; exercised ⬎ 7 hours/week. LSM ⫽ least square means; SEM ⫽ standard error of mean.
hours/week. Adolescents with histories of eating disorders (Group 1) had significantly higher % BF (p ⫽ .0001) than Groups 2 and 3. No significant differences were noted among the three groups for mean height, weight, or BMI. At the initiation of the study, BIA was selected for body composition because it was inexpensive and noninvasive, and published data for adolescent females were available. After the study began, assessing % BF using DEXA became available. Body composition using DEXA differed from results based on BIA (Table 2). DEXA results for % BF were as follows: Group 1 ⫽ 28.7 ⫾ 5.4, Group 2 ⫽ 29.5 ⫾ 6.4, and Group 3 ⫽ 24.4 ⫾ 5.7. Body fat (percent) was not significantly different between those with histories of eating disorders (Group 1) and Group 2. Group 3 (exercise ⬎ 7 hours/week) had significantly lower % BF than Group 1 (p ⫽ .03) and Group 2 (p ⫽ .03). A further analysis showed
that mean % BF results were consistent when only 16 of those with histories of eating disorders were matched with 16 in Group 2 and those adolescents ⬎ 19 years were excluded. The degree of agreement between adolescents’ self-reported body weight and clinic measurements showed that on the average those with a history of eating disorders and Group 2 (no history of eating disorders, exercised ⱕ 7 hours/week) underreported their weight by 2.2 kg. Hormone profiles were within normal ranges for pubertal adolescent females. Based on CALTRAC results, mean energy expenditure for physical movement was 1976 kcal/day (1522–2498 kcal/day) for those with a history of eating disorders, compared with 1991 kcal/day (1681–2366 kcal/day) for Group 2 (exercise ⱕ 7 hours/week) and 1994 ⫾ 332 (1548 –2574) for Group 3 (exercise ⬎ 7 hours/week). CALTRAC results were not significantly different by group.
Table 2. Body Composition of Adolescent Females (Aged 16 –22 Years) Group Group 1§ (n ⫽ 25) Group 2£ (n ⫽ 25) Group 3† (n ⫽ 15)
Body Fat (%)* Lean Body Mass (%) ———————LSM ⫾ SEM ——————— 32.2 ⫾ .01‡ (26 – 40) 26.8 ⫾ .01 (11– 41) 23.5 ⫾ .01Ł (16 –32)
67.8 ⫾ .01 (60 –74) 73.2 ⫾ .01 (59 – 89) 76.5 ⫾ .01 (68 – 84)
Wt (kg) BMI (wt/ht2) Ht (cm2) ——————————Mean ⫾ SD —————————— 163.1 ⫾ 6.5僐 (154.4 ⫾ 180.7) 164.2 ⫾ 5.1 (154.4 –172.8) 161.6 ⫾ 5.2 (153.3–171.5)
57.5 ⫾ 8.7 (43.8 –78.5) 57.4 ⫾ 6.7 (45.8 –72.6) 56.0 ⫾ 7.4 (43.4 –70.9)
21.6 ⫾ 2.6 (17.0 ⫾ 28.1) 21.4 ⫾ 2.7 (17.5 ⫾ 28.1) 21.4 ⫾ 2.5 (17.0 ⫾ 27.2)
* Body composition determined by bioelectric impedance analyzer. Group 1—History of eating disorders. £ Group 2—No history of eating disorders and matched individually for age, height, and weight to Group 1; exercised ⱕ 7 hours/week. † Group 3—No history of eating disorders; exercised ⬎ 7 hours/week. ‡ Group 1 (history of eating disorders) had significantly higher percent body fat than Group 2 (p ⫽ .0001) and Group 3 (p ⫽ .0001). Ł Group 3 (exercised ⬎ 7 hours/week) had significantly lower percent body fat than Group 1 (p ⫽ .0001) and Group 2 (p ⫽ .03). 僐 No significant difference among groups for height, weight, and BMI. BMI ⫽ body mass index; LSM ⫽ least square means; SD ⫽ standard deviation; SEM ⫽ standard error of mean. §
326
CARRUTH AND SKINNER
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
Table 3. Energy and Nutrient Intakes of Adolescent Females (16 –22 years)
Nutrient Energy (Kcal) Carbohydrate (g) Protein (g)
RDA DRI* 2200
44
Fat (g) Calcium (mg)
1300
Magnesium (mg)
1000 360
Zinc (mg) Vitamin D (g)
310 12 5
Group 1§ (n ⫽ 25) (% RDA) [DRI]
Group 2£ (n ⫽ 25) (% RDA) [DRI]
Group 3† (n ⫽ 15) (% RDA) [DRI]
1502 ⫾ 306‡ (68) 242 ⫾ 61Ł 52 ⫾ 17 (118) 38 ⫾ 24僐 617 ⫾ 315 [47] [62] 171 ⫾ 64 [47] [55] 6⫾3 (50) 2⫾2 [40]
2062 ⫾ 457 (94) 291 ⫾ 75 74 ⫾ 22 (168) 69 ⫾ 22 977 ⫾ 366 [75] [98] 242 ⫾ 83 [67] [78] 8⫾3 (67) 4⫾2 [80]
1765 ⫾ 444 (80) 270 ⫾ 60 66 ⫾ 21 (150) 49 ⫾ 26 756 ⫾ 311 [58] [76] 213 ⫾ 62 [59] [69] 7⫾3 (58) 5⫾9 [100]
* 1989 Recommended Dietary Allowances (RDA) for females (aged 15–18). Differences in RDAs for ages 19 –24: protein increases 2 g. 1998 Dietary Reference Intakes (DRI), Adequate Intakes, for ages 14 –18 and 19 –30. § Group 1: History of eating disorders. £ Group 2: No history of eating disorders and matched for height, weight, and age with Group 1; exercised ⱕ 7 hours/week. † Group 3: No history of eating disorders; exercised ⬎ 7 hours/week. ‡ ⫻ ⫾ SD intake based on two 24-hour recalls and 2-day food records. Ł,僐 No RDA for carbohydrate and fat.
As shown in Table 3, none of the three groups had mean intakes of energy that met estimated energy allowances (15–24 years of age) or any of the nutrients, except vitamin D for Group 3. Calcium, magnesium, zinc, and vitamin D intakes of those with histories of eating disorders ranged from 40% to 62% of the recommended levels. Although Group 2 (exercised ⱕ 7 hours/week) had energy mean intake of 2062 kcal/day, essential vitamins and minerals needed for bone health were less than 100% of the RDA. Dietary recommendations consider factors such as age, gender, and moderate amounts of exercise. Although Groups 1 and 2 were matched for body size and age, those with a history of eating disorders (Group 1) had significantly lower energy and selected nutrient intakes: kcal, F ⫽ 21.06, p ⫽ .0001; fat, F ⫽ 19.53, p ⫽ .0001; calcium, F ⫽ 13.06, p ⫽ .0007; vitamin D, F ⫽ 6.00, p ⫽ .0180; and zinc, F ⫽ 8.39, p ⫽ .0057. In all groups protein intakes exceeded Recommended Dietary Allowances.
Discussion Our findings do not support the hypotheses that adolescents with a history of eating disorders had significantly lower whole-body, spinal, or pelvis
BMD (gm/c2), BMI, or % BF than those who were matched for age, height, and weight (exercise ⱕ 7 hours/week) or those who exercised ⬎ 7 hours/ week. The level of exercise and age since onset of menses in a GLM showed no significant differences in BMD measurements among the three groups. Whole-body BMD results in this study were similar to published data for healthy pubertal girls as follows: females aged 16 –20, range by age, mean ⫾ SD ⫽ 1.075 ⫾ .079 –1.093 ⫾ .093 g/cm2; school girl volunteers aged 11–17.9, mean ranges by age, 0.699 – 1.256 g/cm2 (1). Our findings may be partially explained by our research methodology. Subjects in the eating-disorder group were judged by their health care provider as rehabilitated. The adolescents were eumenorrheic, and menstrual histories did not reveal extended durations of amenorrhea that have been associated with reduced BMD of the lumbar and/or total spine. Using dual-photon absorptiometry, Bachrach et al (25) studied 15 adolescents with AN prospectively (12–16 months) and found that whole-body BMD increased with weight; lumbar spine BMD did not increase. The researchers proposed an independent effect of estrogen on bone mass in AN patients. In our study, LH and FSH values were within normative values for pubertal females with regular menses.
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
Iketani et al (11) suggested that return of menses and weight gain were necessary to normalize BMD in AN patients. The reported influences of BMI and age on BMD were addressed in this study by matching age, height, and weight of those without a history of eating disorders and a moderate activity level to those adolescents with a history of eating disorders. Mean BMI was not significantly different between groups and was not a factor associated with group differences in BMD. In a study of children (5–20 years), chronologic age, Tanner stage, weight, height, and BMI were significantly associated with BMD (26). Comparing those with a history of eating disorders with two groups of healthy adolescents addressed age-related physical development and body mass changes that influence bone mineral acquisition during adolescence (6,27–29). The age range of adolescents in this study targeted the period of adolescence when peak bone mass is acquired. The reported differential effects of AN, BN, and AN/BN on whole-body BMD were nonsignificant in this study. In a study by Iketani et al (11) of females (ⱕ 18 years), mean BMD of those with BN (mean ⫽ 0.90 g/cm2) and AN/BN (mean ⫾ SD ⫽ 1.05 ⫾ 0.16 g/cm2) were not significantly different from the mean BMD of controls (mean ⫾ SD ⫽ 0.94 ⫾ 0.05 g/cm2). In our study, inclusion of healthy groups who exercised ⱕ 7 hours/week and ⬎ 7 hours/week provided some measure of physical activity levels over time. CALTRAC results represented one point in time measurement of energy expenditure for whole-body movement over 4 continuous days. Two measures of physical activity were used because results of published studies about BMD and physical activity during adolescence are inconsistent. For example, in an 8-month controlled study of healthy females (mean age ⫽ 19.9 years), lumbar spine bone mineral values increased with running or resistant exercise (30). In Finnish youth (6 –19 years), BMD was significantly correlated with physical activity (33) whereas Lonzer et al (26) found no association in children 6 –20 years old. In six AN patients (17–30 years), more energy was expended in physical activity than controls despite low body weight and reduced basal metabolic rate (BMR) (17). The duration of amenorrhea and exercise (hours/week) significantly predicted spinal bone density in adult females (17). In our study, level of physical activity (ⱕ 7 or ⬎ 7 hours/week) was not associated with differences in BMD among the three groups. No comparable pub-
327
lished studies were found that assessed the influence of physical activity on BMD in adolescents who were judged as rehabilitated from their eating disorders and were not participating in a clinical treatment program. The % BF determined by either BIA or DEXA in this study was not consistent with published findings for patients with eating disorders (17), and the two methods gave different results. Ellis (32) compared BIA, DEXA, and total body electrical conductivity to estimate total body fat and % BF in children and adolescents; results were highly dependent on the method used. Support for the “the use of BMI as a fatness measure in groups of children and adolescents” as suggested by Pietrobelli et al (33) would not be appropriate for adolescents in this study because body composition differed and BMI did not. In our study, the group who exercised ⬎ 7 hours/week had significantly higher-percent lean body mass than the group with a history of eating disorders (p ⫽ .0001) and those who exercised ⱕ 7 hours/week (p ⫽ .03). However, whole-body, spinal, and pelvis BMD were not significantly different among the three groups. As hypothesized, adolescents with a history of eating disorders had significantly lower energy, fat, calcium, vitamin D, and zinc intakes than the other two groups and less than the amounts currently recommended to achieve peak bone mass during adolescence. Dietary intakes in this study were consistent with the literature about female adolescents’ diets (34,35). Avoidance of foods containing fat, including milk products and lean muscle meats, contributed to the low calcium, vitamin D, and zinc intakes of adolescents in this study. They consumed dairy products, low fat cheeses, and enriched grain products that provided protein (118 –168% of RDA) but were not sources of vitamin D. Consumption of carbonated drinks and unsweetened fruit drinks also contributed to the inadequacy of the diets. Wyshak and Frisch (36) found that cola beverage consumption had a strong association with bone fractures in 8- to 17-year-old girls, but a high intake of dietary calcium was protective against fractures. For adolescents who avoid dairy products because of perceived fat content, vitamin and mineral supplementation may be a prudent option to improve calcium and vitamin D intakes. A small sample (n) may reduce the power of a study. The calculated power of this study was Ɑ .99, using one-way ANOVA with one-repeated-measures factor. Several studies of adolescents (aged ⱕ 19) with eating disorders have reported on a small number of subjects (9,11,13,16,37) but did not publish
328
CARRUTH AND SKINNER
power calculations. The paucity of empirical data about rehabilitated adolescents (27) and/or adult patients who had eating disorders (38) limits a comparison of our results other than with a clinic population of adult women with eating disorders or with an age cohort of adolescents in the general population. The eumenorrheic status of all groups for at least the last 18 months and the relative short time period from age of diagnosis (1–5 years) to time of bone mineral assessments likely influenced BMD results, compared with BMD of adult females whose menstrual history involved years of amenorrhea. We conclude that except for body fat (percent), measures of BMD, energy expenditure (CALTRAC), and BMI were not significantly different among adolescents with a history of eating disorders and the two comparison groups. BMD values were within ranges of published data for pubertal, white adolescents of cohort ages. With full recovery from eating disorders, the findings suggest that teenage girls may achieve normal bone mass and body composition. Sources of support: Tennessee Agricultural Experiment Station 37901-1071, University of Tennessee, Knoxville, TN 37996.
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
10. Salisbury JJ, Mitchell JE. Bone mineral density and anorexia nervosa in women. Am J Psychiatry 1991;148:768 –74. 11. Iketani T, Kiriike N, Nakanishi S, Nakasuji T. Effects of weight gain and resumption of menses on reduced bone density in patients with anorexia nervosa. Biol Psychiatry 1995;37:521–7. 12. Frusztajer NT, Dhuper S, Warren MP, et al. Nutrition and the incidence of stress fractures in ballet dancers. Am J Clin Nutr 1990;51:779 – 83. 13. Casper RC, Schoeller DA, Kushner R, et al. Total daily energy expenditure and activity level in anorexia nervosa. Am J Clin Nutr 1991;53:1143–50. 14. Schoemaker C. Does early intervention improve the prognosis in anorexia nervosa? A systematic review of the treatmentoutcome literature. Int J Eat Disord 1997;21:1–15. 15. Carmichael KA, Carmichael DH. Bone metabolism and osteopenia in eating disorders. Medicine 1995;74:254 – 67. 16. Bachrach LK, Guido D, Katzman D, et al. Decreased bone density in adolescent girls with anorexia nervosa. Pediatrics 1990;86:440 – 47. 17. van Marken Lichtenbelt WD, Heidendal GAK, Westerterp KR. Energy expenditure and physical activity in relation to bone mineral density in women with anorexia nervosa. Eur J Clin Nutr 1997;51:826 –30. 18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10. 19. Houtkooper LB, Lohman TG, Going SB, Hall MC. Validity of bioelectrical impedance for body composition assessment in children. J Applied Physiol 1989;66:814 –21. 20. Segal KR, Loan MV, Fitzgerald PI, et al. Lean body mass estimation by bioelectrical impedance analysis: a four-site cross-validation study. Am J Clin Nutr 1988;47:7–14.
References 1. Hannan WJ, Cowen SJ, Wrate RM, Barton J. Improved prediction of bone mineral content and density. Arch Dis Child 1995;72:147–9. 2. Weaver CM, Martin BR, Plawecki KL, et al. Differences in calcium metabolism between adolescent and adult females. Am J Clin Nutr 1995;61:577– 81. 3. Lloyd T, Rollings N, Andon MB, et al. Determinants of bone density in young women. Relationships among pubertal development, total body bone mass, and total body bone density in premenarchal females. J Clin Endocrinol Metab 1992;75: 383–7.
21. Food and Nutrition Board. Recommended Dietary Allowances, 10th ed. Washington, DC: National Academy of Sciences; 1989. 22. Food and Nutrition Board. Dietary Reference Intakes: Recommended Levels for Individual Intake. Washington, DC: National Academy of Sciences; 1998. 23. Newton JR, Freeman CP, Hannan WJ, Cowen S. Osteoporosis and normal weight bulimia nervosa—Which patients are at risk? J Psychosom Res 1993;3:239 – 47. 24. Andersen AE, Woodward PJ, LaFrance N. Bone mineral density of eating disorder subgroups. Int J Eat Disord 1995; 18:335– 42.
4. Zanchetta JR, Plotkin H, Alvarez Filgueira ML. Bone mass in children: normative values for the 2–20-year-old population. Bone 1995;16(Suppl):393S–9S.
25. Bachrach LK, Katzman DK, Litt IF, et al. Recovery from osteopenia in adolescent girls with anorexia nervosa. J Clin Endocrinol Metab 1991;72:602– 6.
5. Tanner JM. Issues and advances in adolescent growth and development. J Adolesc Health Care 1987;8:470 – 8.
26. Lonzer MD, Imrie R, Rogers D, et al. Effects of heredity, age, weight, puberty, activity, and calcium intake on bone mineral density in children. Clin Pediatr 1996;185–9.
6. Katzman DK, Bachrach LK, Carter DR, Marcus R. Clinical and anthropometric correlates of bone mineral acquisition in healthy adolescent girls. J Clin Endocrinol Metab 1991;73: 1332–9. 7. Boot AM, deRidder MAJ, Pols HAP, et al. Bone mineral density in children and adolescents: relation to puberty, calcium intake, and physical activity. J Clin Endocrinol Metab 1997;82:57– 62. 8. Hergenroeder AC. Bone mineralization, hypothalamic amenorrhea, and sex steroid therapy in female adolescents and young adults. J Pediatr 1995;120:683–9. 9. Biller BMK, Saxe V, Herzog DB, et al. Mechanisms of osteoporosis in adult and adolescent women with anorexia nervosa. J Clin Endocrinol Metab 1989;68:548 –54.
27. Glastre C, Braillon P, David L, et al. Measurement of bone mineral content of the lumbar spine by dual energy x-ray absorptiometry in normal children: correlations with growth parameters. J Clin Endocrinol Metab 1990;70:1330 –3. 28. Uusi-Rasi K, Haapasalo H, Kannus P, et al. Determinants of bone mineralization in 8 to 20 year old Finnish females. Eur J Clin Nutr 1997;51:54 –9. 29. Teegarden D, Proulx WR, Martin BR, et al. Peak bone mass in young women. J Bone Miner Res 1995;10:711–5. 30. Snow-Harter C, Bouxsein ML, Lewis BT, et al. Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial. J Bone Miner Res 1992;7:761–9.
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
31. Kro¨ger H, Kotaniemi A, Vainio P, Alhava E. Bone densitometry of the spine and femur in children by dual-energy x-ray absorptiometry. Bone Miner 1992;17:75– 85. 32. Ellis KJ. Measuring body fatness in children and young adults: comparison of bioelectric impedance analysis, total body electrical conductivity, and dual-energy X-ray absorptiometry. Intl J Obesity 1996;20:866 –73. 33. Pietrobelli A, Faith M, Allison DB, et al. Body mass index as a measure of adiposity among children and adolescents: A validation study. J Pediatr 1998;132:204 –10. 34. Devaney BL, Gordon AR, Burghardt JA. Dietary intake of students. Am J Clin Nutr 1995;61(Suppl):205S–12S. 35. French SA, Perry CL, Leon GR, Fulkerson MA. Food preferences, eating patterns, and physical activity among adoles-
36.
37.
38. 39. 40.
329
cents: correlates of eating disorders symptoms. J Adolesc Health 1994;15:286 –94. Wyshak G, Frisch RE. Carbonated beverages, dietary calcium, the dietary calcium/phosphorus ratio, and bone fractures in girls and boys. J Adolesc Health 1994;15:210 –15. Abrams SA, Silber TJ, Esteban NV, et al. Mineral balance and bone turnover in adolescents with anorexia nervosa. J Pediatr 1993;123:326 –31. Treasure JL, Treasure GFM. Reversible bone loss in anorexia nervosa. Br Med J 1987;295:474 –75. Hoffman A, ed. Adolescent Medicine. Menlo Park, California: Addison-Wesley, 1983:435. Lexi-Comp, ed. Laboratory Manual. Stow: Lexi-Comp, Inc., 1989:70, 398.
ORIGINAL ARTICLE
Bone Mineral Status in Adolescent Girls: Effects of Eating Disorders and Exercise BETTY R. CARRUTH, Ph.D., AND JEAN D. SKINNER, Ph.D.
Purpose: To compare whole-body, lumbar, total spine, and pelvis bone mineral density (BMD), body mass index (BMI), body composition, energy expenditure for physical activity, and dietary intake of adolescents, aged 16 –22 years. Methods: Three study groups included 25 girls with histories of eating disorders, 25 girls with no histories of eating disorders who exercised < 7 hours/week, and 15 girls with no history of eating disorders who exercised > 7 hours/week. Bone mineral density was measured by dual-energy x-ray absorptiometry (DEXA), body composition by bioelectric impedance and DEXA, energy expenditure by Personal Activity Computer, nutrient intake by 4-day dietary recalls/records, and BMI by measures of height/weight. General linear models, LSM ⴞ SEM, Student’s t-tests, and correlation analyses were used to determine group differences. Results: No significant differences in whole-body, spinal, and pelvis BMD were found among the three groups. Mean body fat (percent) was significantly higher (p ⴝ .0001) for the group with histories of eating disorders than other groups. Dietary intakes of adolescents with histories of eating disorders were significantly lower for energy (p ⴝ .0001), fat (p ⴝ .0001), calcium (p ⴝ .0007), vitamin D (p ⴝ .0180), and zinc (p ⴝ .0057) than those without eating disorder histories who exercised < 7 hours/week. Conclusion: Except for body fat (percent), measures of BMD, energy expenditure, and BMI were not significantly different among groups. Our data suggest that with full recovery from eating disorders, teenage girls can achieve normal bone mass and body composition. © Society for Adolescent Medicine, 2000
From the Department of Nutrition, University of Tennessee-Knoxville, Knoxville, Tennessee. Address correspondence to: Betty R. Carruth, Ph.D., Department of Nutrition, University of Tennessee, 1215 Cumberland Avenue, Knoxville, TN 37996-1900. Manuscript accepted September 15, 1999. 1054-139X/00/$–see front matter PII S1054-139X(99)00089-0
KEY WORDS: Bone mineral density Calcium Dual X-ray absorptiometry Eating disorders Exercise Vitamin D
The advent of low-radiation technology enables investigators to determine maximum bone mineral mass acquired during childhood (1) and to study determinants thereof in both normal and diseased states (2). In a cross-sectional study, 83% of adult bone mass existed in premenarcheal 12-year-old females (3). In a study of 2 to 20-year-old females, maximum mean values for whole-body bone mineral content (WBBMC) were found in the 16-year-old female group. A significant difference (p ⬍ .01) in WBBMC was noted between Tanner stages 1–2 and 2–3 but not for 3– 4 and 4 –5 (4). Tanner stage 2–3 in females occurs on the average around 11–12 years of age (5). Body size, eumenorrhea, and body composition influence attainment of bone mineral mass during adolescence. Late puberty and amenorrhea are risk factors for reduced bone mineral density (BMD). For 45 healthy prepubertal and pubertal girls, pubertal development and body size were important determinants of bone acquisition (6). In a cross-sectional study of 295 females, girls who had earlier menarche had higher BMD (7). Ninety-nine percent of the changes in wholebody mineral were related to increases in bone size. Body fat as a determinant of bone mineral status is not clearly understood. Although regular menses is positively associated with increased bone mineral mass, there is no specific level of fat mass at which menses occurs (8).
© Society for Adolescent Medicine, 2000 Published by Elsevier Science Inc., 655 Avenue of the Americas, New York, NY 10010
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
The relationship between extended periods of amenorrhea (estrogen deficiencies), low body weight, and bone mineral mass are well documented in adult females with anorexia nervosa (9,10). However, most published studies about females with eating disorders combine data for adolescents and adults and present group results (10 –13). Treatment outcome literature summarizes findings in the same manner (14). No studies in the literature that assessed bone mineral status of adolescent patients whose primary health care provider judged them to be rehabilitated from their eating disorder and who were not currently in a therapeutic regimen were found. In general, adolescents in published studies represent a clinic population characterized by amenorrhea, low body weight, food restriction, and/or purging behaviors (9,15). The duration of illness may not be treated as a covariant (11). The control group may represent a wide range of ages and/or of varying stature (16). Calcium intake; dietary supplement usage; and other lifestyle factors, such as smoking and alcohol, are not consistently reported (6,17). Body composition data are infrequently given for adolescent subjects. This study was planned to control for some factors that may differentially affect bone mineral density and body composition, (i.e., eating disorders, body size, lean body mass, exercise, and diet). The study objectives were to compare whole-body, spinal, and pelvis BMD (g/cm2), body composition (% fat and % lean body mass), body mass index (BMI), estimates of energy expenditure for physical activity, and dietary intakes of energy, macronutrients, calcium, vitamin D, magnesium, and zinc in three groups of eumenorrheic females aged 16 –22 years. We hypothesized that adolescent females with a history of eating disorders would have significantly lower whole-body, spinal, and pelvis BMD (g/cm2), BMI, and percent body fat (% BF) than the two comparison groups. The amount of exercise, as measured in this study, was hypothesized to be significantly higher for those with a history of eating disorders than the two other groups. Mean energy (kilocalories) and selected nutrient intakes of adolescents with a history of eating disorders were expected to be significantly lower than those of the two comparison groups.
Methods Subjects White, eumenorrheic, adolescent females with histories of eating disorders (n ⫽ 25) were referred from primary physicians and therapists. The health care
323
providers’ diagnosis and judgment about recovery status were the basis for acceptance into the study. The group with histories of eating disorders included those initially diagnosed with anorexia nervosa (AN) (n ⫽ 15), bulimia nervosa (BN) (n ⫽ 5), and anorexia– bulimia nervosa (AN/BN) (n ⫽ 5). Individuals were excluded from participating in the study if they were in therapy, received medication, or were amenorrheic (ⱕ 9 periods/year) or had missed more than three consecutive periods in the last year. The two comparison groups included: (1) 25 adolescents with no history of eating disorders who exercised ⱕ 7 hours/week, and (2) 15 adolescents who exercised ⬎ 7 hours/week. To address the issues of stature and age differences, the 25 adolescents with no history of eating disorders were matched for age, weight, and height with the 25 adolescents with histories of eating disorders. To address the influence of exercise apart from a history of eating disorders, two levels of exercise (ⱕ 7 hours/week and ⬎ 7 hours/week) were used in the comparison groups with no history of eating disorders. The comparison groups were recruited from schools, girls’ clubs, and gymnasium sports counselors and by posters. In addition to the criterion of eumenorrhea, other exclusion criteria included body weight (⬍ 25th or ⬎ 90th percentile), hormone/ endocrine disorders that negatively influence bone health, smoking, drinking ⬎ 200 g ethanol/week, receiving hormonal medications to regulate menses (except for oral contraceptive agents), birth weights ⬍ 2500 g, and ⬍ 36 weeks’ gestational age at birth. Adolescents aged ⱕ 18 years and their parents signed separate informed consent forms, which described research objectives, an assurance of confidentiality, and the potential risk/benefits to participants. Females aged 19 –22 signed informed consent forms. The protocol was approved by the committee for use of human subjects in research at the participating academic and medical institutions. Data Collection At the initial interview, one of two trained interviewers used a questionnaire that provided information about chronologic age, menstrual history, socioeconomic status, foods consumed in the last 24 hours, and the adolescents’ self-reported heights/weights. To prepare for the next interview, adolescents were instructed about keeping a 2-day food record (including a weekend day) and the use of a CALTRAC
324
CARRUTH AND SKINNER
Personal Activity Computer (Hemokenetics, Inc., Madison, Wisconsin). CALTRAC is the size of a small pocket computer that attaches to a belt or waistline of the garment and records the quantity of body movements. Estimates of energy expenditure (in kilocalories) were customized for weight, height, age, and gender. Participants wore the CALTRAC over 4 continuous days, including at least one weekend day. Interviewers were available to participants during the week and weekends to enhance cooperation and accuracy of recordings. The second interview occurred in a nuclear medicine clinic in an urban tertiary medical center. To determine nonpregnancy status prior to bone mineral assessments, a venipuncture blood sample (20 mL) was taken by a registered nurse and human chorionic gonadotropin level assayed. Individual blood profiles for dehydroepisterone sulfate, androstenedione, luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, and testosterone were analyzed by the medical center’s laboratory. Bone Mineral Density was measured with dualenergy x-ray absorptiometry (DEXA) (Model QDR Hologic 2000, Bedford, MA). Body composition was measured with RJL bioelectric impedance analyzer (BIA Model 101, RAL Systems, Detroit, Michigan). To compare bioelectric impedance analysis (BIA) with another measure of body composition, DEXA assessments were done on the 25 females with histories of eating disorders, 16 of the comparison group who exercised ⱕ 7 hours/ week, and the 15 girls who exercised ⬎ 7 hours/ week. Weights and heights were measured on a balance-beam scale and stadiometer. A registered dietitian checked the 2-day food records for completeness and took a 24-hour recall. Recordings for CALTRAC were checked and interviewers probed for events that could affect CALTRAC results. Analyses Frequencies and means for demographic and other group characteristics were determined using SAS (System for Statistical Analysis, Cary, North Carolina, 1988). Least Squares Means (LSM ⫾ SEM) were calculated and Student’s t-tests used for computing statistically significant differences between groups for percent body fat, percent lean body mass, BMI, height, weight, and CALTRAC. To predict factors that significantly contributed to BMD, general linear models (GLM) were developed for those with histories of eating disorders and the 25 girls who were matched for age, height, weight (exercised ⱕ 7
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
hours/week). For the eating-disorder group only, Pearson correlation coefficients for whole-body, spinal, and pelvis BMD (g/cm2) and the years since diagnosis of the eating disorder were computed and a two-tail test of significance performed. The degree of agreement between the adolescents’ self-reported weight/height and clinic measurements was determined by Bland-Altman estimates (18). Estimates of body composition (percent fat and lean body mass) were derived from information provided by the manufacturer of BIA, published studies (19,20), and DEXA results. Two 24-hour recalls and 2-day food records were coded using the Nutritionist IV software (Version 3.5.2 for Windows, N-Square Computing, San Bruno, California). Mean nutrient intakes for protein, calcium, magnesium, zinc, vitamin D, and energy (in kilocalories) were compared with the Recommended Dietary Allowances (RDAs) (21) and 1998 Dietary Reference Intakes (22). To assess mean differences in dietary intake by those with histories of eating disorders and the comparison groups, GLM procedures and MANOVA were performed with nutrients as the dependent variables.
Results Table 1 describes whole-body, lumbar spine, total spine, and pelvis BMD. These skeletal sites contain trabecular bone that may be negatively affected by eating disorders (23,24). No significant group differences for whole-body BMD or any skeletal sites were found or differences by diagnoses of AN, BN, or AN/BN for those with histories of disorders. With CALTRAC and years since onset of menses in a GLM, no significant differences among groups were found for any of the bone mineral results. The Pearson product-moment correlation coefficients computed between number of years since diagnosis of an eating disorder and BMD results were not significant. With CALTRAC and age at onset of menses in a GLM, no significant group differences in whole-body, spinal, or pelvis BMD were found (Table 1). For the total group with histories of eating disorders (n ⫽ 25), mean years since diagnosis was 3.1 years (range 1–5 years). Body composition results from BIA among the three groups were significantly different for % BF (Table 2). Group 3, who exercised ⬎ 7 hours/week had significantly lower % BF than those with a history of eating disorders (Group 1, p ⫽ .0001) and comparison Group 2 (p ⫽ .03), who exercised ⱕ 7
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
325
Table 1. Bone Mineral Density of Adolescent Females (Aged 16 –22 Years)
Group Group 1§ (n ⫽ 25) Group 2£ (n ⫽ 25) Group 3† (n ⫽ 15)
Lumbar Spine Pelvis Total Spine BMD* (g/cm2) (g/cm2) (g/cm2) (g/cm2) —————————————————————LSM ⫾ SEM ————————————————————— (range) 1.07 ⫾ .01 (0.91–1.24) 1.07 ⫾ .01 (0.94 –1.21) 1.10 ⫾ .07 (1.0 –1.23)
1.02 ⫾ .02 (0.79 –1.29) 1.04 ⫾ .03 (0.70 –1.24) 1.06 ⫾ .03 (0.83–1.31)
1.11 ⫾ .02 (0.86 –1.3) 1.12 ⫾ .02 (0.89 –1.27) 1.17 ⫾ .02 (1.03–1.33)
0.77 ⫾ .01 (0.67– 0.97) 0.79 ⫾ .02 (0.61– 0.97) 0.80 ⫾ .01 (0.73– 0.91)
* General linear model included exercise and gynecologic age. No significant differences among groups in whole-body bone mineral density (BMD) or other skeletal sites among the three groups. § Group 1—History of eating disorders. £ Group 2—No history of eating disorders, matched individually with Group 1 by age, height, and weight; exercised ⱕ 7 hours/week. † Group 3—No history of eating disorders; exercised ⬎ 7 hours/week. LSM ⫽ least square means; SEM ⫽ standard error of mean.
hours/week. Adolescents with histories of eating disorders (Group 1) had significantly higher % BF (p ⫽ .0001) than Groups 2 and 3. No significant differences were noted among the three groups for mean height, weight, or BMI. At the initiation of the study, BIA was selected for body composition because it was inexpensive and noninvasive, and published data for adolescent females were available. After the study began, assessing % BF using DEXA became available. Body composition using DEXA differed from results based on BIA (Table 2). DEXA results for % BF were as follows: Group 1 ⫽ 28.7 ⫾ 5.4, Group 2 ⫽ 29.5 ⫾ 6.4, and Group 3 ⫽ 24.4 ⫾ 5.7. Body fat (percent) was not significantly different between those with histories of eating disorders (Group 1) and Group 2. Group 3 (exercise ⬎ 7 hours/week) had significantly lower % BF than Group 1 (p ⫽ .03) and Group 2 (p ⫽ .03). A further analysis showed
that mean % BF results were consistent when only 16 of those with histories of eating disorders were matched with 16 in Group 2 and those adolescents ⬎ 19 years were excluded. The degree of agreement between adolescents’ self-reported body weight and clinic measurements showed that on the average those with a history of eating disorders and Group 2 (no history of eating disorders, exercised ⱕ 7 hours/week) underreported their weight by 2.2 kg. Hormone profiles were within normal ranges for pubertal adolescent females. Based on CALTRAC results, mean energy expenditure for physical movement was 1976 kcal/day (1522–2498 kcal/day) for those with a history of eating disorders, compared with 1991 kcal/day (1681–2366 kcal/day) for Group 2 (exercise ⱕ 7 hours/week) and 1994 ⫾ 332 (1548 –2574) for Group 3 (exercise ⬎ 7 hours/week). CALTRAC results were not significantly different by group.
Table 2. Body Composition of Adolescent Females (Aged 16 –22 Years) Group Group 1§ (n ⫽ 25) Group 2£ (n ⫽ 25) Group 3† (n ⫽ 15)
Body Fat (%)* Lean Body Mass (%) ———————LSM ⫾ SEM ——————— 32.2 ⫾ .01‡ (26 – 40) 26.8 ⫾ .01 (11– 41) 23.5 ⫾ .01Ł (16 –32)
67.8 ⫾ .01 (60 –74) 73.2 ⫾ .01 (59 – 89) 76.5 ⫾ .01 (68 – 84)
Wt (kg) BMI (wt/ht2) Ht (cm2) ——————————Mean ⫾ SD —————————— 163.1 ⫾ 6.5僐 (154.4 ⫾ 180.7) 164.2 ⫾ 5.1 (154.4 –172.8) 161.6 ⫾ 5.2 (153.3–171.5)
57.5 ⫾ 8.7 (43.8 –78.5) 57.4 ⫾ 6.7 (45.8 –72.6) 56.0 ⫾ 7.4 (43.4 –70.9)
21.6 ⫾ 2.6 (17.0 ⫾ 28.1) 21.4 ⫾ 2.7 (17.5 ⫾ 28.1) 21.4 ⫾ 2.5 (17.0 ⫾ 27.2)
* Body composition determined by bioelectric impedance analyzer. Group 1—History of eating disorders. £ Group 2—No history of eating disorders and matched individually for age, height, and weight to Group 1; exercised ⱕ 7 hours/week. † Group 3—No history of eating disorders; exercised ⬎ 7 hours/week. ‡ Group 1 (history of eating disorders) had significantly higher percent body fat than Group 2 (p ⫽ .0001) and Group 3 (p ⫽ .0001). Ł Group 3 (exercised ⬎ 7 hours/week) had significantly lower percent body fat than Group 1 (p ⫽ .0001) and Group 2 (p ⫽ .03). 僐 No significant difference among groups for height, weight, and BMI. BMI ⫽ body mass index; LSM ⫽ least square means; SD ⫽ standard deviation; SEM ⫽ standard error of mean. §
326
CARRUTH AND SKINNER
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
Table 3. Energy and Nutrient Intakes of Adolescent Females (16 –22 years)
Nutrient Energy (Kcal) Carbohydrate (g) Protein (g)
RDA DRI* 2200
44
Fat (g) Calcium (mg)
1300
Magnesium (mg)
1000 360
Zinc (mg) Vitamin D (g)
310 12 5
Group 1§ (n ⫽ 25) (% RDA) [DRI]
Group 2£ (n ⫽ 25) (% RDA) [DRI]
Group 3† (n ⫽ 15) (% RDA) [DRI]
1502 ⫾ 306‡ (68) 242 ⫾ 61Ł 52 ⫾ 17 (118) 38 ⫾ 24僐 617 ⫾ 315 [47] [62] 171 ⫾ 64 [47] [55] 6⫾3 (50) 2⫾2 [40]
2062 ⫾ 457 (94) 291 ⫾ 75 74 ⫾ 22 (168) 69 ⫾ 22 977 ⫾ 366 [75] [98] 242 ⫾ 83 [67] [78] 8⫾3 (67) 4⫾2 [80]
1765 ⫾ 444 (80) 270 ⫾ 60 66 ⫾ 21 (150) 49 ⫾ 26 756 ⫾ 311 [58] [76] 213 ⫾ 62 [59] [69] 7⫾3 (58) 5⫾9 [100]
* 1989 Recommended Dietary Allowances (RDA) for females (aged 15–18). Differences in RDAs for ages 19 –24: protein increases 2 g. 1998 Dietary Reference Intakes (DRI), Adequate Intakes, for ages 14 –18 and 19 –30. § Group 1: History of eating disorders. £ Group 2: No history of eating disorders and matched for height, weight, and age with Group 1; exercised ⱕ 7 hours/week. † Group 3: No history of eating disorders; exercised ⬎ 7 hours/week. ‡ ⫻ ⫾ SD intake based on two 24-hour recalls and 2-day food records. Ł,僐 No RDA for carbohydrate and fat.
As shown in Table 3, none of the three groups had mean intakes of energy that met estimated energy allowances (15–24 years of age) or any of the nutrients, except vitamin D for Group 3. Calcium, magnesium, zinc, and vitamin D intakes of those with histories of eating disorders ranged from 40% to 62% of the recommended levels. Although Group 2 (exercised ⱕ 7 hours/week) had energy mean intake of 2062 kcal/day, essential vitamins and minerals needed for bone health were less than 100% of the RDA. Dietary recommendations consider factors such as age, gender, and moderate amounts of exercise. Although Groups 1 and 2 were matched for body size and age, those with a history of eating disorders (Group 1) had significantly lower energy and selected nutrient intakes: kcal, F ⫽ 21.06, p ⫽ .0001; fat, F ⫽ 19.53, p ⫽ .0001; calcium, F ⫽ 13.06, p ⫽ .0007; vitamin D, F ⫽ 6.00, p ⫽ .0180; and zinc, F ⫽ 8.39, p ⫽ .0057. In all groups protein intakes exceeded Recommended Dietary Allowances.
Discussion Our findings do not support the hypotheses that adolescents with a history of eating disorders had significantly lower whole-body, spinal, or pelvis
BMD (gm/c2), BMI, or % BF than those who were matched for age, height, and weight (exercise ⱕ 7 hours/week) or those who exercised ⬎ 7 hours/ week. The level of exercise and age since onset of menses in a GLM showed no significant differences in BMD measurements among the three groups. Whole-body BMD results in this study were similar to published data for healthy pubertal girls as follows: females aged 16 –20, range by age, mean ⫾ SD ⫽ 1.075 ⫾ .079 –1.093 ⫾ .093 g/cm2; school girl volunteers aged 11–17.9, mean ranges by age, 0.699 – 1.256 g/cm2 (1). Our findings may be partially explained by our research methodology. Subjects in the eating-disorder group were judged by their health care provider as rehabilitated. The adolescents were eumenorrheic, and menstrual histories did not reveal extended durations of amenorrhea that have been associated with reduced BMD of the lumbar and/or total spine. Using dual-photon absorptiometry, Bachrach et al (25) studied 15 adolescents with AN prospectively (12–16 months) and found that whole-body BMD increased with weight; lumbar spine BMD did not increase. The researchers proposed an independent effect of estrogen on bone mass in AN patients. In our study, LH and FSH values were within normative values for pubertal females with regular menses.
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
Iketani et al (11) suggested that return of menses and weight gain were necessary to normalize BMD in AN patients. The reported influences of BMI and age on BMD were addressed in this study by matching age, height, and weight of those without a history of eating disorders and a moderate activity level to those adolescents with a history of eating disorders. Mean BMI was not significantly different between groups and was not a factor associated with group differences in BMD. In a study of children (5–20 years), chronologic age, Tanner stage, weight, height, and BMI were significantly associated with BMD (26). Comparing those with a history of eating disorders with two groups of healthy adolescents addressed age-related physical development and body mass changes that influence bone mineral acquisition during adolescence (6,27–29). The age range of adolescents in this study targeted the period of adolescence when peak bone mass is acquired. The reported differential effects of AN, BN, and AN/BN on whole-body BMD were nonsignificant in this study. In a study by Iketani et al (11) of females (ⱕ 18 years), mean BMD of those with BN (mean ⫽ 0.90 g/cm2) and AN/BN (mean ⫾ SD ⫽ 1.05 ⫾ 0.16 g/cm2) were not significantly different from the mean BMD of controls (mean ⫾ SD ⫽ 0.94 ⫾ 0.05 g/cm2). In our study, inclusion of healthy groups who exercised ⱕ 7 hours/week and ⬎ 7 hours/week provided some measure of physical activity levels over time. CALTRAC results represented one point in time measurement of energy expenditure for whole-body movement over 4 continuous days. Two measures of physical activity were used because results of published studies about BMD and physical activity during adolescence are inconsistent. For example, in an 8-month controlled study of healthy females (mean age ⫽ 19.9 years), lumbar spine bone mineral values increased with running or resistant exercise (30). In Finnish youth (6 –19 years), BMD was significantly correlated with physical activity (33) whereas Lonzer et al (26) found no association in children 6 –20 years old. In six AN patients (17–30 years), more energy was expended in physical activity than controls despite low body weight and reduced basal metabolic rate (BMR) (17). The duration of amenorrhea and exercise (hours/week) significantly predicted spinal bone density in adult females (17). In our study, level of physical activity (ⱕ 7 or ⬎ 7 hours/week) was not associated with differences in BMD among the three groups. No comparable pub-
327
lished studies were found that assessed the influence of physical activity on BMD in adolescents who were judged as rehabilitated from their eating disorders and were not participating in a clinical treatment program. The % BF determined by either BIA or DEXA in this study was not consistent with published findings for patients with eating disorders (17), and the two methods gave different results. Ellis (32) compared BIA, DEXA, and total body electrical conductivity to estimate total body fat and % BF in children and adolescents; results were highly dependent on the method used. Support for the “the use of BMI as a fatness measure in groups of children and adolescents” as suggested by Pietrobelli et al (33) would not be appropriate for adolescents in this study because body composition differed and BMI did not. In our study, the group who exercised ⬎ 7 hours/week had significantly higher-percent lean body mass than the group with a history of eating disorders (p ⫽ .0001) and those who exercised ⱕ 7 hours/week (p ⫽ .03). However, whole-body, spinal, and pelvis BMD were not significantly different among the three groups. As hypothesized, adolescents with a history of eating disorders had significantly lower energy, fat, calcium, vitamin D, and zinc intakes than the other two groups and less than the amounts currently recommended to achieve peak bone mass during adolescence. Dietary intakes in this study were consistent with the literature about female adolescents’ diets (34,35). Avoidance of foods containing fat, including milk products and lean muscle meats, contributed to the low calcium, vitamin D, and zinc intakes of adolescents in this study. They consumed dairy products, low fat cheeses, and enriched grain products that provided protein (118 –168% of RDA) but were not sources of vitamin D. Consumption of carbonated drinks and unsweetened fruit drinks also contributed to the inadequacy of the diets. Wyshak and Frisch (36) found that cola beverage consumption had a strong association with bone fractures in 8- to 17-year-old girls, but a high intake of dietary calcium was protective against fractures. For adolescents who avoid dairy products because of perceived fat content, vitamin and mineral supplementation may be a prudent option to improve calcium and vitamin D intakes. A small sample (n) may reduce the power of a study. The calculated power of this study was Ɑ .99, using one-way ANOVA with one-repeated-measures factor. Several studies of adolescents (aged ⱕ 19) with eating disorders have reported on a small number of subjects (9,11,13,16,37) but did not publish
328
CARRUTH AND SKINNER
power calculations. The paucity of empirical data about rehabilitated adolescents (27) and/or adult patients who had eating disorders (38) limits a comparison of our results other than with a clinic population of adult women with eating disorders or with an age cohort of adolescents in the general population. The eumenorrheic status of all groups for at least the last 18 months and the relative short time period from age of diagnosis (1–5 years) to time of bone mineral assessments likely influenced BMD results, compared with BMD of adult females whose menstrual history involved years of amenorrhea. We conclude that except for body fat (percent), measures of BMD, energy expenditure (CALTRAC), and BMI were not significantly different among adolescents with a history of eating disorders and the two comparison groups. BMD values were within ranges of published data for pubertal, white adolescents of cohort ages. With full recovery from eating disorders, the findings suggest that teenage girls may achieve normal bone mass and body composition. Sources of support: Tennessee Agricultural Experiment Station 37901-1071, University of Tennessee, Knoxville, TN 37996.
JOURNAL OF ADOLESCENT HEALTH Vol. 26, No. 5
10. Salisbury JJ, Mitchell JE. Bone mineral density and anorexia nervosa in women. Am J Psychiatry 1991;148:768 –74. 11. Iketani T, Kiriike N, Nakanishi S, Nakasuji T. Effects of weight gain and resumption of menses on reduced bone density in patients with anorexia nervosa. Biol Psychiatry 1995;37:521–7. 12. Frusztajer NT, Dhuper S, Warren MP, et al. Nutrition and the incidence of stress fractures in ballet dancers. Am J Clin Nutr 1990;51:779 – 83. 13. Casper RC, Schoeller DA, Kushner R, et al. Total daily energy expenditure and activity level in anorexia nervosa. Am J Clin Nutr 1991;53:1143–50. 14. Schoemaker C. Does early intervention improve the prognosis in anorexia nervosa? A systematic review of the treatmentoutcome literature. Int J Eat Disord 1997;21:1–15. 15. Carmichael KA, Carmichael DH. Bone metabolism and osteopenia in eating disorders. Medicine 1995;74:254 – 67. 16. Bachrach LK, Guido D, Katzman D, et al. Decreased bone density in adolescent girls with anorexia nervosa. Pediatrics 1990;86:440 – 47. 17. van Marken Lichtenbelt WD, Heidendal GAK, Westerterp KR. Energy expenditure and physical activity in relation to bone mineral density in women with anorexia nervosa. Eur J Clin Nutr 1997;51:826 –30. 18. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10. 19. Houtkooper LB, Lohman TG, Going SB, Hall MC. Validity of bioelectrical impedance for body composition assessment in children. J Applied Physiol 1989;66:814 –21. 20. Segal KR, Loan MV, Fitzgerald PI, et al. Lean body mass estimation by bioelectrical impedance analysis: a four-site cross-validation study. Am J Clin Nutr 1988;47:7–14.
References 1. Hannan WJ, Cowen SJ, Wrate RM, Barton J. Improved prediction of bone mineral content and density. Arch Dis Child 1995;72:147–9. 2. Weaver CM, Martin BR, Plawecki KL, et al. Differences in calcium metabolism between adolescent and adult females. Am J Clin Nutr 1995;61:577– 81. 3. Lloyd T, Rollings N, Andon MB, et al. Determinants of bone density in young women. Relationships among pubertal development, total body bone mass, and total body bone density in premenarchal females. J Clin Endocrinol Metab 1992;75: 383–7.
21. Food and Nutrition Board. Recommended Dietary Allowances, 10th ed. Washington, DC: National Academy of Sciences; 1989. 22. Food and Nutrition Board. Dietary Reference Intakes: Recommended Levels for Individual Intake. Washington, DC: National Academy of Sciences; 1998. 23. Newton JR, Freeman CP, Hannan WJ, Cowen S. Osteoporosis and normal weight bulimia nervosa—Which patients are at risk? J Psychosom Res 1993;3:239 – 47. 24. Andersen AE, Woodward PJ, LaFrance N. Bone mineral density of eating disorder subgroups. Int J Eat Disord 1995; 18:335– 42.
4. Zanchetta JR, Plotkin H, Alvarez Filgueira ML. Bone mass in children: normative values for the 2–20-year-old population. Bone 1995;16(Suppl):393S–9S.
25. Bachrach LK, Katzman DK, Litt IF, et al. Recovery from osteopenia in adolescent girls with anorexia nervosa. J Clin Endocrinol Metab 1991;72:602– 6.
5. Tanner JM. Issues and advances in adolescent growth and development. J Adolesc Health Care 1987;8:470 – 8.
26. Lonzer MD, Imrie R, Rogers D, et al. Effects of heredity, age, weight, puberty, activity, and calcium intake on bone mineral density in children. Clin Pediatr 1996;185–9.
6. Katzman DK, Bachrach LK, Carter DR, Marcus R. Clinical and anthropometric correlates of bone mineral acquisition in healthy adolescent girls. J Clin Endocrinol Metab 1991;73: 1332–9. 7. Boot AM, deRidder MAJ, Pols HAP, et al. Bone mineral density in children and adolescents: relation to puberty, calcium intake, and physical activity. J Clin Endocrinol Metab 1997;82:57– 62. 8. Hergenroeder AC. Bone mineralization, hypothalamic amenorrhea, and sex steroid therapy in female adolescents and young adults. J Pediatr 1995;120:683–9. 9. Biller BMK, Saxe V, Herzog DB, et al. Mechanisms of osteoporosis in adult and adolescent women with anorexia nervosa. J Clin Endocrinol Metab 1989;68:548 –54.
27. Glastre C, Braillon P, David L, et al. Measurement of bone mineral content of the lumbar spine by dual energy x-ray absorptiometry in normal children: correlations with growth parameters. J Clin Endocrinol Metab 1990;70:1330 –3. 28. Uusi-Rasi K, Haapasalo H, Kannus P, et al. Determinants of bone mineralization in 8 to 20 year old Finnish females. Eur J Clin Nutr 1997;51:54 –9. 29. Teegarden D, Proulx WR, Martin BR, et al. Peak bone mass in young women. J Bone Miner Res 1995;10:711–5. 30. Snow-Harter C, Bouxsein ML, Lewis BT, et al. Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial. J Bone Miner Res 1992;7:761–9.
May 2000
EATING DISORDERS, EXERCISE, AND BONE MINERAL STATUS
31. Kro¨ger H, Kotaniemi A, Vainio P, Alhava E. Bone densitometry of the spine and femur in children by dual-energy x-ray absorptiometry. Bone Miner 1992;17:75– 85. 32. Ellis KJ. Measuring body fatness in children and young adults: comparison of bioelectric impedance analysis, total body electrical conductivity, and dual-energy X-ray absorptiometry. Intl J Obesity 1996;20:866 –73. 33. Pietrobelli A, Faith M, Allison DB, et al. Body mass index as a measure of adiposity among children and adolescents: A validation study. J Pediatr 1998;132:204 –10. 34. Devaney BL, Gordon AR, Burghardt JA. Dietary intake of students. Am J Clin Nutr 1995;61(Suppl):205S–12S. 35. French SA, Perry CL, Leon GR, Fulkerson MA. Food preferences, eating patterns, and physical activity among adoles-
36.
37.
38. 39. 40.
329
cents: correlates of eating disorders symptoms. J Adolesc Health 1994;15:286 –94. Wyshak G, Frisch RE. Carbonated beverages, dietary calcium, the dietary calcium/phosphorus ratio, and bone fractures in girls and boys. J Adolesc Health 1994;15:210 –15. Abrams SA, Silber TJ, Esteban NV, et al. Mineral balance and bone turnover in adolescents with anorexia nervosa. J Pediatr 1993;123:326 –31. Treasure JL, Treasure GFM. Reversible bone loss in anorexia nervosa. Br Med J 1987;295:474 –75. Hoffman A, ed. Adolescent Medicine. Menlo Park, California: Addison-Wesley, 1983:435. Lexi-Comp, ed. Laboratory Manual. Stow: Lexi-Comp, Inc., 1989:70, 398.