This study was undertaken to identify factors that influence total body bone area (TBBA), total body bone mineral content (TBBMC), and tibial cortical bone measures in 239 children aged 3–5 years. We obtained information on demographic and anthropometric characteristics and measurements of diet, physical activity, and strength. In multiple regression analysis, TBBA correlated with height (p < 0.001), weight (p < 0.001), percent body fat (p < 0.001), and calcium intake (p = 0.02). TBBMC correlated with TBBA (p < 0.001), age (p = 0.001), and weight (p = 0.02) and inversely correlated with height (p < 0.001) and percent body fat (p < 0.001). Children born preterm had lower TBBMC compared with children born at term (p = 0.02). Both periosteal and endosteal circumferences were correlated with weight (both, p < 0.001) and inversely correlated with age (p = 0.006 and p = 0.003, respectively) and percent body fat (p = 0.002 and p = 0.005 respectively). Endosteal circumference was greater and cortical bone area was lower in children born preterm compared with those born at term (both, p = 0.04). Findings of higher TBBA and lower TBBMC in children with high percent body fat indicate undermineralization of bone and suggest that obesity in preschool children may have detrimental effects on total body bone mass accretion. A smaller tibial periosteal circumference and thus cross-sectional area in children with the same weight but higher percent body fat also would lead to a biomechanical disadvantage in these children. Findings of low TBBMC and cortical bone area among children born preterm need to be confirmed in other populations. We speculate that differences in these measurements between children born preterm and at term may be caused by differences in activity.
THE AMOUNT of bone gain early in life is a major factor influencing an individual's risk of developing osteoporosis. Reduced bone mineral density (BMD) during childhood is a significant predictor of childhood fracture risk.(1, 2) Because of the immediate and long-term consequences of low bone mass, it is important to identify which factors influence bone gain so that preventive strategies that lead to improved bone health can be implemented. Pediatric studies primarily have been conducted in children ≥8 years of age.3-15) Results of these studies indicate that anthropometric characteristics, as well as childhood diet and physical activity, are important predictors of bone mineral content (BMC) or BMD in children.
A difficulty with interpreting pediatric bone data obtained using dual-energy X-ray absorptiometry (DXA) stems from the fact that BMD is expressed as an areal density or BMC per projected bone area (BA). The calculation of BMD (BMC/BA in g/cm2) incorrectly assumes that BMC and BA are directly proportional.(16) A variety of mathematical methods have been suggested to adjust areal BMD to obtain volumetric BMD measurements.16-18) Molgaard and coworkers recently suggested using a three-step process for investigating bone mass in children. This process includes determining height for age, BA for height, and BMC for BA.(19) These three steps are thought to correspond to the three different causes of reduced bone mass: short bones, narrow bones, and light bones.
This study was undertaken to identify factors associated with total body BA (TBBA) and total body BMC (TBBMC) in children aged 3-5 years using the approach of Molgaard and coworkers.(19) We also investigated factors influencing periosteal and endosteal circumferences and cortical BA of the distal tibia, as measured by peripheral quantitative computed tomography (pQCT). We used baseline data from the South Dakota Children's Health Study, an ongoing randomized trial designed to determine whether calcium intake modifies the bone response to physical activity in young children. As part of the baseline examination, we collected detailed information on demographic and anthropometric characteristics. Measurements also were made of dietary intake, physical activity, gross motor development, and strength. In addition to identifying factors associated with bone measurements, the following specific hypotheses also were tested: (1) TBBMC and TBBA are greater in children with higher levels of physical activity and (2) TBBMC and TBBA are greater in children with higher intakes of calcium. Few studies have reported on the relationship between BMC and diet or activity in a large group of preschool children.
MATERIALS AND METHODS
We obtained baseline data on 239 children aged 3-5 years who were randomized in the ongoing trial. Eligibility for the study required that the children be enrolled in one of the 11 participating child-care centers in eastern South Dakota. The presence of a chronic illness that may affect bone mineralization (i.e., cerebral palsy, intestinal malabsorption, and juvenile rheumatoid arthritis) was reason for ineligibility. Because of confidentiality issues, we were not able to determine the population characteristics of children not enrolled in the study. The Human Subjects Committee at South Dakota State University approved the study and written informed consent was obtained from the parent.
Parents completed questionnaires that provided information on gender, race, history of prematurity, and type of feeding during the first month of life. Baseline examinations were conducted in a mobile unit and consisted of anthropometric measurements, TBBA, and TBBMC measurements using DXA (Hologic 4500A; Hologic, Inc., Waltham, MA, USA) and periosteal and endosteal circumference and cortical BA of the 20% distal tibia site of the left leg using pQCT (Norland/Stratec XCT2000; Norland Corp., Fort Atkinson, WI, USA). The pediatric whole body software was used for analysis of TBBA, TBBMC, and body composition as recommended by the manufacturer for children up to 11-13 years of age. Percent body fat was the total fat mass measured by DXA divided by the total mass. The difference between the adult and pediatric whole body software is a lower bone detection threshold used in the pediatric analysis, which identifies and measures bone of a lower density than the standard adult algorithm. The percent body fat results are about 2.7-2.9% higher than those obtained using the adult software. Percent body fat results obtained using the standard adult and pediatric whole body analyses are highly correlated and show a strong linear relationship with low SEs of the estimates (QDR4500 Pediatric Option User's Guide; Hologic, Inc.). The pQCT settings used for this study are described elsewhere.(20) pQCT scans with movement occurred frequently and a pediatric tibia restraint was developed by Norland/Stratec to reduce movement. A total of 110 pQCT scans were not included in this analysis because of significant movement before the development of the leg restraint. We also omitted one DXA scan from the analysis that showed significant movement. The CV for TBBMC supplied by the manufacturer and our CV using a phantom are <1%. Our CVs for pQCT in children aged 3-5 years are 3.6% and 5.4% for periosteal circumference and cortical BA, respectively.
Parents and child-care providers completed 3-day food records (2 weekdays and 1 weekend day) and study personnel reviewed the records for completeness. A registered dietitian supervised the nutrient intake analysis that was completed using the Nutritionist V database (First Data Bank, The Hearst Corp., San Bruno, CA, USA). Calcium intake was expressed as both milligrams per day and milligrams per kilocalorie. Quartiles of calcium intake (milligrams per day) also were determined.
Each child wore an activity sensor (Actiwatch motion sensor; Mini Mitter, Sun River, OR, USA) for a 2-day period (48 h). The sensor contains an omnidirectional sensor capable of detecting acceleration in two planes. This type of sensor integrates the degree and speed of motion and produces an electrical current that varies in magnitude and is sensitive to 0.01g (0.098 m/s2). An increased degree of speed and motion produces an increase in voltage, and the sensor stores this information as activity counts per minute. Study personnel placed a belt on which the sensor was secured on the waist of each child, with the base of the instrument positioned against the lumbar spine. Other investigators have found that placement on the torso provides a better measure of reliability compared with placement on the wrist and/or ankle using a similar type of movement-sensing device.(21) We downloaded and reviewed sensor data to determine whether the child wore the sensor for the entire 2-day period. We determined that 25 2-day readings were invalid and we obtained only 1 day of valid data in 12 children. Through 6 h of direct observation of activity we previously found that sensor counts between 500/minute and 999/minute represent moderate activity and counts >1000/minute represent vigorous activity (B. Specker, unpublished data, 2000). We expressed sensor data as percent of time in moderate or vigorous activity (counts > 500/minute), percent of time in vigorous activity (counts > 1000/minute), or quartiles of average total counts per day. The validity of the Actiwatch motion sensor in measuring activity in this population has been reported elsewhere.(22)
We used the Peabody Developmental Motor Scales (PDMS; Pro-Ed, Inc., Austin, TX, USA) to assess gross motor development. Normalized T scores are presented with a score of 50 representing the age-adjusted average score. Study personnel followed standardized procedures to obtain high jump and long jump distances and we used a Martin vigorimeter (Elmed, Inc., Addison, IL, USA) to obtain grip strength measurements using the smallest ball available. The pressure applied to the ball of the vigorimeter is measured in bars. The lowest measurable pressure is 0.06 bars, and in several cases the children were able to squeeze the ball without any measurable change in pressure. In those situations, we assumed the grip strength to be 0.03 bar, which is one-half the pressure between zero and the lowest measurable pressure. We made all strength measurements in triplicate and recorded the largest observed measurement. Strength measurements were added after the study began, and we obtained baseline measurements on approximately one-half of the children in the study.
We entered data onto an Access Database and performed data analysis using JMP statistical software (SAS Institute, Cary, NC, USA) after testing the outcome variables for normality. Initially, we determined the univariate relationships between the bone measurements and anthropometric measurements (height, weight, and percent total body fat), demographic characteristics (age, gender, and race) and general information (season, history of prematurity, type of feeding during the first month of life, and PDMS). To identify significant predictors of TBBA and TBBMC, we performed a forward-backward stepwise regression analysis including those variables that were significant at p ≤ 0.05 after including height or TBBA in the models predicting TBBA or TBBMC, respectively. Once we obtained the final models, we added our measures of calcium intakes (mg/day, mg/kcal, or quartiles of intake), activity (percent of time in moderate or vigorous activity, percent of time in vigorous activity, average daily counts, and quartiles of daily counts), and strength (high and long jump distances and grip strength) singly to determine whether they explained a significant amount of the remaining error. We did not include these variables in the original stepwise analysis because of missing data. The data are presented as means ± SD unless otherwise stated.
Table 1 summarizes the population characteristics by gender. TBBMC, weight, height, percent body fat, caloric intake (kilocalories per day), percent of time in moderate or vigorous activity, percent of time in vigorous activity, average daily sensor counts, PDMS gross motor index, long jump distance, and grip strength were greater in male children compared with female children (Table 1). Fourteen percent of the mothers reported their child's birth to be preterm with a range of 2-9 weeks of prematurity. Forty percent of the children received only human milk during the first month of life, 24% received human milk plus formula, and 36% received only formula. Average daily sensor counts (×10,000) of less than 21.3, 21.3-26.6, and 26.6-33.3, and greater than 33.3 defined the quartiles of activity from lowest to highest. Percent of time in moderate or vigorous activity ranged from 2% to 26% and percent of time in vigorous activity ranged from <1% to 12%. Quartiles of calcium intake were defined from lowest to highest as less than 709 mg/day, 709-875 mg/day, and 876-1036 mg/day and more than 1036 mg/day.
Table Table 1.. Characteristics of the Study Population by Gender (Mean ± SD [n])
TBBA was associated univariately with age (r = 0.65; p < 0.001), height (r = 0.88; p < 0.001), weight (r = 0.92; p < 0.001), average calcium intake per day (r = 0.18; p < 0.01), high jump distance (r = 0.28; p < 0.01), long jump distance (r = 0.39; p < 0.001), and grip strength (r = 0.48; p < 0.001). Figure 1 illustrates the relationship between TBBA and height. In a multiple regression analysis, TBBA was correlated with height, weight, percent body fat, and calcium intake per kilocalorie (Table 2). TBBA by quartile of calcium intake was 938 ± 4.4 cm2, 939 ± 4.3 cm2, 948 ± 4.3 cm2, and 952 ± 4.4 cm2 (least square means ± SEM). TBBA was not associated with any of the activity measures.
Table Table 2.. Multiple Regression Models Predicting TBBA, TBBMC, Periosteal and Endosteal Circumferences, and Cortical BA of the Distal Tibia
TBBMC differed univariately by gender (Table 1), was higher in children born at term compared with preterm (582 ± 87g vs. 551 ± 90 g; p = 0.057), and correlated with age (r = 0.93; p < 0.001), height (r = 0.84; p < 0.001), weight (r = 0.85; p < 0.001), gross motor index (r = 0.30; p < 0.05), calcium intake per day (r = 0.20; p < 0.01), high jump distance (r = 0.28; p < 0.01), long jump distance (r = 0.41; p < 0.01), and grip strength (r = 0.51; p < 0.001). Figure 2 illustrates the relationship between TBBMC and TBBA. In a multiple regression analysis, TBBMC was correlated directly with TBBA, age, and weight and inversely correlated with height and percent body fat (Table 2). Children born at term had higher TBBMC compared with children born preterm (least square means [±SEM] of 579 ± 1.9 g vs. 567 ± 4.7 g, respectively). TBBMC was not associated with any of the measures of calcium intake or physical activity, and the interaction between quartiles of calcium intake and quartiles of activity counts was not statistically significant (p = 0.09).
Age, weight, percent body fat, and long jump distance were significant predictors of tibial periosteal circumference in a multiple regression analysis (Table 2). Heavier children and children who could jump far had greater periosteal circumferences than did lighter children and children who could not jump as far. Children of similar weight and age who had greater percent body fat had a smaller circumference than did children with less percent fat. Endosteal circumference was associated with age, weight, and percent body fat in a similar manner as with periosteal circumference (Table 2). In addition, children with history of preterm birth had a mean endosteal circumference of 41.0 ± 1.1 mm compared with 38.5 ± 0.4 mm in children born at term (p = 0.04). The larger endosteal circumference and similar periosteal circumference resulted in a smaller cortical BA in children born preterm compared with those born at term (50.5 ± 2.2 mm2 vs. 55.2 ± 0.8 mm2, respectively; Table 2). Differences between children born preterm and at term are summarized in Table 3. There were no differences in anthropometric measurements or calcium intake, but children born preterm were found to have significantly lower measures of physical activity than children born at term. Although these measures of physical activity were not independent predictors of the bone measurements, when they were included in the analysis, history of preterm birth was no longer a significant predictor of endosteal circumference or cortical BA.
Table Table 3.. Characteristics of Children with History of Preterm Birth (Mean ± SEM)
Several investigators have suggested using BMC rather than BMD to determine whether bone mineralization is normal in children.(18, 23) BMD measurements from DXA are expressed as areal densities or BMC per projected BA. Although BMC is approximately proportional to BA and compensates in part for skeletal size, this normalization does not fully correct for bone volume. Thus, when expressed as areal BMD, bone density may be overestimated in larger bones.(16, 24) Molgaard and coworkers recently recommended the use of a three-step process to evaluate bone mineralization during childhood: (1) height for age, (2) BA for height, and (3) BMC for BA.(23) These three steps are suggested to correspond to three causes of reduced bone mass in children: short bones, narrow bones, and light bones. We used this approach in our current analyses to determine anthropometric, demographic, and environmental factors that influence TBBA independent of height and TBBMC independent of TBBA. We also investigated factors influencing periosteal and endosteal circumferences and cortical BA of the distal tibia.
We found an association between TBBA and current calcium intake when expressed as milligrams per kilocalories of intake, an indicator of the consumption of calcium-rich foods. Although most previous pediatric calcium supplementation trials did not find an effect of calcium supplementation on BA,25-28) two did report increased BA in the intervention groups.(29, 30) Although most of these trials used supplements as the calcium source, the majority of dietary calcium in our population is likely to be from dairy sources. One of the studies that found an increase in BA with increased calcium exposure was the study by Bonjour and coworkers in which girls were randomized to receive foods fortified with calcium from milk extract rather than calcium supplements.(30) Cadogan and coworkers found higher serum insulin-like growth factor 1 (IGF-1) concentrations among girls randomized to receive 1 pint of milk per day for 18 months compared with girls not receiving milk.(31) Growth hormone and IGF-1 are involved in bone remodeling and promote bone growth.(32) Thus, it is possible that the association we observed between TBBA and calcium intake is a reflection of the effects of other nutrients or growth factors present in dairy products or results from dairy product consumption rather than from calcium intake alone.
Traditionally, body weight has been identified as a significant predictor of BMC and BMD. The reason often provided for the relationship between either BMC or BMD and body weight is that this represents an osteogenic bone response from increased bone loading among individuals with greater body weight. In this analysis, we investigated the relationships of TBBA and TBBMC with percent body fat in addition to overall body weight. We found positive relationships between both TBBA and TBBMC and body weight, indicating a greater skeletal size with greater overall weight. We also found a positive relationship between TBBA and percent body fat but an inverse relationship between TBBMC and percent body fat. These findings indicate that children with higher percent body fat will have larger TBBA but less bone mass compared with children of similar weight who have lower percent body fat. Therefore, these children will have undermineralized bones. By definition, percent body fat is inversely correlated with percent lean mass. Therefore, it is not possible to determine whether it is a high percent body fat or low percent lean mass that is responsible for the findings of a higher TBBA and lower TBBMC. These results may help explain the recent finding by Goulding and coworkers that high body weight is a significant risk factor for fractures in girls aged 3-15 years.(1) Weiler and coworkers also recently reported a negative impact of percent body fat on TBBMC in girls aged 10-19 years.(33) These findings have significant public health implications and underscore the potential adverse effects of high percent body fat or the potential benefits of high percent lean mass during childhood on current fracture risk and attainment of maximal peak bone mass.
Osteopenia and rickets during the first year of life occur most often in preterm infants and is caused by interruption of mineral accretion that occurs during the last trimester.34-36) Preterm infants have lower growth percentiles and BMC than term infants.(37) One study found that human milk-fed preterm infants failed to achieve catch-up bone mineralization until their second birthday.(38) We found a 2.1% lower TBBMC in children born preterm compared with children born at term. We also found a larger endosteal circumference, with a similar periosteal circumference, resulting in a lower cortical BA in children born preterm versus those born at term. Weeks preterm ranged from 2 to 9 weeks, with a median of 3 weeks. Therefore, these infants were not likely to be very low birth weight infants. The only measured differences between children born preterm and at term were lower levels of physical activity among children born preterm. When the activity measures were included in the statistical analyses, history of preterm birth was no longer a significant predictor of endosteal circumference or cortical BA. However, differences in TBBMC by history of preterm birth were still significant when activity measures were included in the analyses. These findings of an effect of preterm birth on TBBMC and bone size at 3-5 years of age in a group of infants whose births were only mildly preterm needs to be confirmed. If mild prematurity has a long-term impact on bone size and mass, possible interventions for this group of infants may be warranted.
Few studies have investigated factors influencing bone geometric measurements in preschool children. There was an inverse relationship between both periosteal and endosteal circumferences and age after controlling for body weight. However, there was a positive relationship between cortical BA and age and we speculate that the endosteal circumference is increasing at a slower rate than the periosteal circumference at this age. Metacarpal bone diameter increases little between the ages of 3 and 5 years, whereas cortical thickness increases significantly.(24) This appositional phase on the endosteal surface of cortical bone in children aged 3-5 years was illustrated elegantly in an article by Parfitt.(39) The greater cortical BA in the older children would lead to greater bone strength.
Studies in immature rats have shown a significant increase in the periosteal circumference of the tibia as well as an increase in cortical BA in response to jumping.(40) Cortical BA of the distal radius also increases with strength training in postmenopausal women.(41) Our finding of an association between periosteal circumference and long jump distance is consistent with these reports, although it is not clear which comes first. Our findings of lower periosteal and endosteal circumferences and lower cortical BA in children with high percent body fat compared with children of similar weight with low percent body fat may be a result of differences in physical activity. Children with a high percent body fat have lower lean mass compared with children of similar weight and it is possible that children with high percent body fat are less physically active, although we did not observe a relationship between percent body fat and activity measures in this study. A smaller periosteal circumference and, thus, cross-sectional area in children with the same weight but higher percent body fat also would lead to a biomechanical disadvantage in these children. The smaller tibial cross-sectional area but larger TBBA among children with greater percent body fat suggests that larger bone size is likely at other (nonmeasured) bone sites.
There are reports of increased BMD with increased physical activity levels in children(13) and jumping trials in older children also show an increase in BMD.(42, 43) We did not find a relationship between TBBMC or TBBA and activity levels in this study possibly because of relatively low levels of activity at this age. The children in this study were younger than children previously studied and Slemenda and coworkers found that the younger children (aged 5-7 years) had lower levels of activity than the older children (>7 years). Preschool children in this study spent on average <5% of their time in vigorous activity with a total of approximately 13% of their time in moderate and vigorous activity. These findings are consistent with other reports of activity levels in preschool children.44-46)
This study includes baseline data from an ongoing randomized intervention trial designed specifically to investigate the interaction between physical activity and calcium intake on bone mass accretion over a 1-year period. We did not observe a significant interaction between quartiles of calcium intake and quartiles of baseline activity counts in the current analyses. It is possible that calcium intake does not modify the bone response to activity within the ranges of activity levels or calcium intakes that normally are observed in this population of preschool children. Whether such an interaction occurs at higher levels of activity is not yet known.
In summary, we found that TBBA was greater in children with a greater calcium intake per kilocalorie, a surrogate for greater consumption of calcium-rich foods. A lower TBBMC and cortical BA was observed in children with higher percent body fat compared with children of similar weight with lower percent body fat. We also found a lower TBBMC and smaller cortical area among preschool children who were born preterm compared with those born at term. It is possible that some of these bone differences can be attributed to lower levels of physical activity in children born preterm compared with those born at term. However, these findings need to be confirmed in other populations.
This research was supported by National Institutes of Health (NIH) grant R01-AR45310.