点击显示 收起
1 From MRC Human Nutrition Research, Cambridge, United Kingdom (SJS, AP, and SCJ), and the Department of Paediatric Epidemiology and Biostatistics, Institute of Child Health, London (TJC).
2 Presented in part at the 45th Annual Meeting of the American College of Sports Medicine, Orlando, FL, 3–6 June 1998. 3 Supported by the Medical Research Council, United Kingdom, and an award from the Mead Johnson Research Fund. The supplement and placebo tablets were donated by Shire Pharmaceuticals (United Kingdom) and Nycomed Pharma (Norway). 4 Address reprint requests to A Prentice, MRC Human Nutrition Research, Elsie Widdowson Laboratory, Fulbourn Road, Cambridge CB1 9NL, United Kingdom. E-mail: ann.prentice{at}mrc-hnr.cam.ac.uk.
ABSTRACT
Background: Osteoporosis may be prevented or delayed by maximizing peak bone mass through diet modification and physical activity during adolescence.
Objective: We studied whether increases in calcium intake and physical activity effectively increase the bone mineral status of adolescent girls aged 16–18 y.
Design: We conducted a 15.5-mo study of calcium supplementation (1000 mg Ca/d as carbonate) in 144 adolescent girls aged 17.3 ± 0.3 y (
Conclusions: Calcium supplementation and exercise enhanced bone mineral status in adolescent girls. Whether this is a lasting benefit, leading to the optimization of peak bone mass and a reduction in fracture risk, needs to be determined.
Key Words: Adolescent girls • bone mineral content • calcium supplementation • dietary calcium • exercise intervention • osteoporosis
INTRODUCTION
Osteoporosis—the loss of bone and degeneration of skeletal architecture with an increased risk of fragility fractures—is a major worldwide health problem, particularly for elderly women. Bone mineral status in later life is a function of the maximum bone mineral mass attained in young adult life and of subsequent age-related bone loss (1). Increasing peak bone mass in young persons could reduce the incidence of later osteoporotic fractures. Modification of diet and physical activity during childhood and adolescence may be an effective strategy for maximizing peak bone mass (2–7).
Many supplementation studies have suggested a positive effect of increased calcium intake on bone mineral status in young persons, but to date all have been conducted in children and younger adolescents (8–14). To investigate the effects toward the end of linear growth, adolescent girls aged 16–18 y were recruited to participate in a calcium and exercise intervention study. This is an age that is targeted for health messages and in whom significant bone mineral accretion still occurs, whereas interindividual variation associated with the timing of puberty is minimized. The aim of the intervention was to determine whether increases in calcium intake and physical activity enhance bone mineral status in adolescent girls.
SUBJECTS AND METHODS
Subjects
One hundred forty-four female students were recruited from the 2 main sixth-form colleges in Cambridge, United Kingdom. The subjects were aged 17.3 ± 0.3 y (
Subjects were randomly assigned, double-blind, to receive a calcium supplement (group S; n = 65) or matching placebo (group P; n = 66) for 15.5 mo. Randomization was stratified by college in permuted blocks of 4. A staff member not involved in data collection performed the randomization and held the code. The calcium supplement was chewable, orange-flavored calcium carbonate containing 500 mg Ca/tablet (Calcichew-500; distributed by Shire Pharmaceuticals, United Kingdom, and manufactured by Nycomed Pharma, Oslo); the placebo was indistinguishable in shape, taste, and texture from the supplement. Subjects were asked to consume 2 tablets/d (total supplement: 1000 mg Ca/d), one midmorning and one late in the afternoon, and to avoid taking them at meals to minimize possible adverse effects on the absorption of other minerals, especially iron.
For each subject, supplementation started immediately after the baseline measurements were made and finished after all outcome measurements were completed. Jars of tablets, each containing enough tablets to last 25 wk, were issued to each subject in 3 batches during the intervention period. Compliance in both the S and P groups was monitored by using a check diary completed by each subject and by a count-back of tablets remaining in the jar at the end of each 25-wk period. The tablets were well accepted, and no side effects were reported. Tablet compliance, defined by the number of tablets consumed during the supplementation period relative to the number of tablets allocated, was 70 ± 27% (
The subjects were randomly allocated by the supplement-code holder to 1 of 2 exercise groups, stratified by supplement group to ensure equal numbers in the S and P groups: group E (n = 75) was invited and encouraged to attend three 45-min exercise classes a week; group N (n = 56) was not invited to these sessions. The ratio of group E to group N subjects was 2:1 to allow for dropouts from group E. A typical exercise session consisted of a 7–10-min warm-up, a 30-min workout, and a 5–8-min warm-down and stretch section and was conducted to music. One of the principal investigators (SJS), who was both a fully qualified and an experienced fitness instructor, conducted all the exercise classes. The workout component (30 min of the 45-min class) of the exercise-to-music aerobics class was performed at moderate-to-vigorous intensity and consisted of weight-bearing, moderate-to-high impact movements. All sessions were constructed in a similar fashion, enabling exercise intervention to be quantified as the number of sessions attended.
Classes were scheduled during lunch breaks and immediately after school to facilitate regular attendance. Eight sessions were offered per week at different times to provide flexibility. The exercise classes started in January 1996, after all baseline measurements had been completed, to enable all subjects in group E to start together. The classes continued until all outcome measurements on the last subject had been completed. For logistic reasons, the classes were organized during term-time and cancelled during examination times, resulting in a total of 24 wk when classes were run. Attendance at the sessions was monitored by using a register. The exercise classes were well received, but attendance was relatively low (
Measurements
Baseline measurements were completed between October and December 1995, and outcome measurements (after the intervention) were completed between January and March 1997. The mean (± SD) time interval between the baseline (before intervention) and outcome measurements was 465 ± 21 d (15.5 ± 0.7 mo).
Bone mineral content (BMC; in g) and bone area (BA; in cm2) of the whole body, lumbar spine (L1–L4), nondominant forearm (total, ultradistal, and distal third radius), and left hip (total, femoral neck, greater trochanter, and intertrochanter) were measured by dual-energy X-ray absorptiometry (DXA) with the use of a Hologic QDR 1000/W (Hologic Inc, Waltham, MA). Software version 5.61 was used for the whole body (enhanced analysis) and forearm; the performance mode was used for the spine and hip measurements (software version 4.47P). The outcome scan was analyzed with reference to the individual subject’s baseline image with the use of the DXA compare facility. Quality assurance and long-term instrument stability were assessed by scanning the Hologic spine phantom at the beginning of each measurement day. Over the 18 mo of the study, the CV for both BMC and BA was < 0.4%, with no significant drift over time.
Height (cm) and weight (kg) were measured on the same day as was bone mineral. Height was measured to the nearest 0.1 cm with a wall-mounted stadiometer. Weight was measured to the nearest 0.1 kg with an electronic digital scale (Sauter weighing scales; Todd Scales Ltd, Norwich, United Kingdom). Customary dietary calcium intake was assessed in 2 ways. First, before and after the intervention, the structured food-frequency questionnaire Calquest (M Nelson, King’s College, London) was used to assess calcium intake in the week before the measurements were made and over the previous year. Calquest is a validated dietary instrument commonly used in the United Kingdom, and is based on the consumption frequency of the main sources of calcium in the British diet (milk, dairy products, and calcium-fortified flour products). Second, a 7-d food diary based on household measures was completed partway through the study period. Calcium ingested from dietary supplements or antacids was included in all assessments. The data obtained from these dietary measurements gave similar information and behaved in a similar fashion in the regression analyses. For simplicity, only the data obtained with the food-frequency questionnaire in the week before baseline are presented. Total time (h/wk) spent playing sports or exercising was calculated from a physical-activity questionnaire adapted from the Allied Dunbar Fitness Survey United Kingdom (1992).
Statistical analysis
Statistical analyses were carried out with analysis of variance, analysis of covariance, multiple linear regression, and logistic regression by using linear model software in DATA DESK 6.1.1 (Data Description Inc, Ithaca, NY). To minimize problems due to multiple testing, Scheffé’s post hoc test was used, where appropriate. Summary statistics are presented as means ± SDs and differences between groups or across time as means ± SEs. BMC, BA, weight, height, calcium intake, and time spent at exercise and sports were transformed to natural logarithms to facilitate the exploration of power relations between continuous variables and of proportional effects of discrete variables (15). The regression coefficient for a discrete variable, when the dependent variable is in natural logarithms, once multiplied by 100, corresponds closely to the percentage effect: (difference/mean) x 100 (15, 16). All percentage differences were derived in this manner.
The percentage changes in bone variables between baseline and outcome were analyzed by hierarchical analysis of variance and analysis of covariance models followed by Scheffé post hoc tests with the following independent variables: time point, subject number, and intervention group—nested by subject number.
Multiple linear regression models were constructed to examine the separate effects of the supplementation and exercise interventions on the change in bone mineral status over the study period by including either a supplement or exercise group term in the analyses. An interaction between supplement and exercise was also tested for. These analyses were initially performed on an intention-to-treat basis by including all subjects irrespective of their compliance. The effect of compliance was then tested in several ways: 1) by including an interaction term between intervention group and either tablet compliance (expressed as % of tablets consumed) or attendance at the exercise sessions (expressed as % of expected attendance) and 2) by repeating the analyses after restricting the data to those subjects with better tablet compliance in the S and P groups (tablet compliance > 75%: n = 32 in group S and 30 in group P) or exercise attendance (attendance at > 50% of the exercise sessions: n = 20 in group E and 56 in group N). These cutoffs were arbitrary but provided a reasonable compromise between high compliance and sample numbers. Other cutoffs were also investigated; however, because they showed a similar overall pattern, they are not presented here. No significant differences in any variable were found at baseline between subjects with high or low tablet compliance or between those with high or low exercise attendance, except that there were fewer current smokers in the group with high tablet compliance than in the group with low tablet compliance (6% compared with 20%; P = 0.03).
The effect of the calcium and exercise interventions on bone mineral expressed as size-adjusted BMC (SA-BMC, ie, BMC adjusted for BA, body weight, and height) was examined to determine the effect on the skeleton independently of bone and body size (15). The separate effects of the interventions on the 2 components of bone mineral status—absolute mineral mass (BMC) and skeletal size (BA)—were also evaluated. For variables measured at both baseline and outcome, the change () between each and the means of the baseline and outcome values were used in the models to minimize interdependence (17). Baseline values were always included in the analyses, regardless of significance, to minimize effects of regression toward the mean.
Conventionally, BMC is adjusted for bone size by calculating areal bone mineral density (BMD = BMC/BA; in g/cm2). This is only a partial correction for BA at many skeletal sites (15). The differences observed in this study with BMD were similar to those obtained with SA-BMC but had greater variance and lower significance because of the incomplete adjustment for BA. For clarity, BMD results were not included, and data are presented after full size-adjustment.
The full regression model for SA-BMC was as follows: y = ln BMC and x = ln BMC (baseline), ln BA, mean ln BA, ln weight, mean ln weight, ln height, mean ln height, treatment group (group S/group P or group E/group N = 1/0). The coefficient for the treatment group, after multiplying by 100, can be interpreted either as the percentage difference between the treatment and placebo groups at outcome after adjustment for the initial value or as the difference between the treatment groups in the percentage change from baseline to outcome after adjustment for the initial value; these are equivalent interpretations. A consistent approach to variable selection was adopted throughout by using the simultaneous analysis of covariance modeling procedure (Linear Model Type 3, DATA DESK 6.1.1). First, all variables of interest were included in each initial model, followed by backward elimination of nonsignificant factors (P > 0.05), the least significant being removed first to produce a final parsimonious model.
RESULTS
The characteristics of the 4 study groups at baseline were not significantly different (Tables 1 and 2). Calcium intake varied over a wide range (230–2139 mg/d), and the distribution of time spent at exercise and sports was positively skewed (median: 2.1 h/wk; range: 0–16 h/wk).
View this table:
TABLE 1 . Characteristics of the subjects at baseline1
View this table:
TABLE 2 . Bone absorptiometric measures at baseline1
The changes in BMC, BA, and SA-BMC for all subjects over the study period at the whole body, spine, and radius are shown in Table 3 and at the hip are shown in Table 4. Significant increases in BMC and BA were observed at the whole body, lumbar spine, and forearm, whereas significant decreases were observed at the total hip and intertrochanter. These changes were associated with significant increases in SA-BMC at the whole body, spine, and forearm, and a significant decrease at the trochanter. In addition, there were significant increases in body weight and height (Table 3) but no significant changes in calcium intake and time spent at exercise and sports, other than the changes associated with the interventions (data not shown).
View this table:
TABLE 3 . Effect of time and interventions on anthropometric measures and bone variables at the whole body, spine, and radius1
View this table:
TABLE 4 . Effect of time and interventions on bone variables at the hip1
The effect of the calcium and exercise interventions on BMC, BA, and SA-BMC are shown in Tables 3, 4 and 5. No evidence of an interaction between supplement effect and exercise intervention was found at any skeletal site, as indicated by a P value > 0.05 for a supplement x exercise group term in multiple regression models or a time point x supplement x exercise group term in hierarchical analysis of covariance models.
View this table:
TABLE 5 . Effect of the interventions on bone measures in subjects with good compliance1
The calcium intervention resulted in greater bone mineral status at the spine, ultradistal radius, and hip. The effect of the calcium intervention on SA-BMC was apparent in the intention-to-treat analysis at the spine, ultradistal radius, and hip (femoral neck and greater trochanter) (Tables 3 and 4). The significance of these results was stronger, and the magnitude of the effect was generally greater when restricted to subjects with higher compliance (Table 5). A positive interaction between the treatment group and tablet compliance was noted at the total hip, trochanter, and intertrochanter, which suggested a dose effect on the bone response to calcium supplementation at these sites but not at other sites [coefficient ± SE: total hip, 0.034 ± 0.03 (P = 0.03); trochanter, 0.074 ± 0.025 (P = 0.003); and intertrochanter, 0.036 ± 0.019 (P = 0.05). For those with compliance > 75%, the largest increases were at the total hip (2.7%) and greater trochanter (4.8%). Because of the overall pattern of bone changes (Tables 3 and 4), the calcium supplement increased bone mineral accretion at the whole body, spine, and forearm and ameliorated the bone mineral loss at the hip. This is illustrated in Figure 1, in which the changes in SA-BMC in those with high compliance are shown separately for the S and P groups.
FIGURE 1. . Mean (± SE) longitudinal changes (outcome - baseline) in size-adjusted bone mineral content (BMC) in subjects in the calcium-supplemented (; n = 32) and placebo (; n = 30) groups with high compliance (consumption of > 75% of the tablets provided), obtained with the use of heirarchical, nested analysis of covariance models. Significance of the difference between outcome and baseline in each intervention group: *P 0.05, **P 0.01, ***P 0.001. Significance of the difference between the supplement and placebo groups in change over time after adjustment for baseline (Table 5) is given at the left-hand side of the figure: #P 0.05, ##P 0.01, ###P 0.001.
The exercise intervention had no significant effect on SA-BMC in the intention-to-treat analysis (Tables 3 and 4), but SA-BMC increased significantly at the total hip and trochanter in subjects who attended > 50% of the target number of exercise sessions (Table 5). A similar pattern of longitudinal changes in SA-BMC was observed between the exercise and nonexercise groups with that observed between the supplement and placebo groups; the intervention resulted in significant increases in both groups at the whole body and forearm and a reduction in the decrease at the trochanter.
To test whether the interventions had effects on bone growth in addition to bone mineralization, the patterns of change were assessed separately for BMC and BA without size adjustment and are shown in Table 5 for those with the better compliance. At the whole body, spine, and ultradistal radius, no indication that the increases in unadjusted BMC were accompanied by increases in BA was evident. However, in the various regions of the hip, the magnitude of the differences in BMC between the supplemented and placebo groups were greater before than after size adjustment, and significant differences in BA were observed. This finding suggests that, at the hip, the calcium supplement may have influenced bone mineral mass partly through effects on bone size. Because the overall pattern of change over time was a decrease in BMC and BA at the hip (Table 4), the effect of the supplement was to reduce the magnitude of these decreases compared with the placebo. No such effects at the hip were detected as a result of the exercise intervention (Table 5).
No significant effect of the supplement on change in weight or height was observed during the study period. Significant interactions between the bone response to calcium supplementation and body size were noted in the different regions of hip but not at the total hip or at the other sites. The supplement effect was greater at the femoral neck in the subjects with the smallest weight gain or greatest weight loss and at the trochanter in those who had the greatest mean height and at the intertrochanter in those with the greatest weight gain [supplement compliance > 75%, coefficient ± SE for interaction term: femoral neck, -0.27 ± 0.13 (P = 0.038); trochanter, 0.82 ± 0.20 (P = 0.0002); intertrochanter, 0.26 ± 0.11 (P = 0.021)]. No significant effects of body size on the response to the exercise intervention were detected.
No evidence in the regression models of an interaction between habitual dietary calcium intake and response to the calcium supplement was found, as indicated by a P value > 0.05 for a calcium intake x supplement group term when added to multiple regression models. To further evaluate the effects of calcium supplementation in relation to habitual calcium intake, the subjects were classified into subgroups according to their median calcium intake of 833 mg/d (data not shown). No significant differences were noted between the subgroups in response to supplementation, and no consistent trends in the sign of the coefficients were observed. Similarly, no significant interactions were noted between the response to the exercise intervention and baseline physical activity level.
DISCUSSION
We tested the hypothesis that an increase in calcium intake and physical activity would enhance bone mineral acquisition in adolescent girls toward the end of linear growth. The 15.5-mo calcium intervention resulted in higher bone mineral status in 16–18-y-old teenage girls at all skeletal sites investigated. The observed increases in bone mineral are biologically significant; for example, the 5% higher SA-BMC at the trochanter in the calcium-supplemented subjects, in whom the population CV (SD/mean) was 11%, represents nearly half an SD, which, if maintained into old age, would reduce the relative risk of hip fractures in this group by 30%.
This study illustrates that bone mineral accretion is still occurring at the whole body, spine, and forearm in British adolescent girls aged 16–18 y, a finding that is consistent with the findings of other longitudinal studies (17). Conversely, we observed a decline in bone mineral at the hip, which was noted in one other study in girls of similar age (18). Whether this decline illustrates that peak bone mass has already been attained at the hip by 16 y of age is uncertain. The observed decreases in scanned BA in the hip regions may suggest a reorientation of the hip with increasing age, a redistribution of mineral within the hip, or alterations in bone-edge detection by DXA. Superimposing the effect of the calcium supplement on this process resulted in a greater increase in bone mineral accretion at the whole body, spine, and forearm in the calcium group, with a reduction in the apparent bone mineral loss at the hip.
Our finding that the consumption of calcium supplements by teenage girls enhances bone mineral status is consistent with studies in children and younger adolescents (8–14). In contrast, this study was in an age group who, on average, were 4 y postmenarche and in whom linear growth had largely ceased but peak bone mass had yet to be attained for most skeletal regions, with the possible exception of the hip.
Neither the exercise intervention nor the customary calcium intake significantly altered the effects of calcium supplementation. However, the relations between supplement effect and tablet compliance at the hip, and the fact that the magnitude and strength of the bone responses were greater in those with higher compliance, suggest that there may have been a dose response. This finding would agree with the hypothesis that increases in bone mineral status as a result of calcium supplementation are the result of decreases in the bone remodeling space that are proportionate to the change in calcium intake achieved (19).
The exercise intervention did not affect bone mineral status in the intention-to-treat analysis, but SA-BMC was significantly higher at the total hip and greater trochanter in subjects who attended > 50% of the exercise classes. The magnitudes of these effects were more modest than were those observed with the calcium intervention. Two other exercise-intervention studies were conducted in female adolescents (20, 21): one showed an effect of exercise (21); the other did not (20).
The exercise regimen chosen was appropriate for the age group in the current study and took into account their academic time constraints; however, despite the enthusiasm of the subjects, attendance was poor, which might explain why the exercise intervention had only a small effect. This supports the possibility that, unless the exercise regimen imposes severe changes on the range of motion or the magnitude of loading, substantial changes in bone mineral status do not occur. Two studies in military recruits (18–21 y) showed that 14 wk of strenuous physical training can load the bones sufficiently to increase BMD in the lower limbs (22, 23). A longer intervention (43 wk) with good compliance in prepubertal girls led to large increases in BMD (21). One study in older women (20–35 y) combined an exercise and calcium intervention, but only the exercise intervention increased BMD (23). Several studies have shown that certain types of exercise can improve bone mineral status; however, despite our subjects being highly motivated, their attendance was poor. This limits the power of the study to define an effect of exercise on bone mineral status or an interaction with the effect of calcium. It also illustrates that, at a public health level, when encouraging young people to exercise more as a means of preventing osteoporosis, consideration needs to be given to effective promotion and implementation.
In summary, the effect of the calcium intervention shows the magnitude of effect likely to be produced by a population-based intervention. However, the effect was similar regardless of habitual calcium intake, which varied by > 10-fold, which makes it difficult to judge the optimal customary calcium intake for this age group. It was previously hypothesized that the effect of calcium supplementation reduces the rate of bone remodeling, leading to modest increases in bone mass by decreasing the remodeling space (24–27). This concept is supported by the findings of 2 studies that included measurements of biochemical indexes (8, 14). Follow-up of these studies has provided evidence that the effect of supplementation disappears once the supplement is withdrawn (28–30). This suggests that the increase in calcium intake may have no lasting benefit unless the supplementation is continued indefinitely. Our subjects would need to be followed after the withdrawal of supplementation to investigate whether the effects of calcium supplementation would persist or whether they represented a temporary bone remodeling phenomenon.
The findings of the current study add weight to existing evidence that lifestyle interventions enhance bone mineral status in children and adolescents. Whether this is a lasting benefit, leading to optimization of peak bone mass and accompanied by a reduced risk of osteoporosis, has yet to be determined.
ACKNOWLEDGMENTS
We are grateful to the principals, staff, students, and families of Hills Road Sixth Form College and Long Road Sixth Form College, Cambridge, United Kingdom, for their enthusiasm and commitment. We also thank the following staff of MRC Human Nutrition Research for their invaluable contributions: S Levitt (data entry), MA Laskey (DXA scan quality control), and S Austin (code holder).
SJS, AP, and TJC designed the study, SJS and SCJ carried out the fieldwork and collected the data, SJS analyzed and interpreted the data and wrote the first draft of the paper, and AP and TJC assisted with the analysis and interpretation of the data and the writing of the paper. AP, SJS, and TJC are employees of the Medical Research Council, United Kingdom, and SJS held an MRC PhD Studentship. There were no financial or personal relationships with the other sponsors of the research.
REFERENCES