|Home | About | Journals | Submit | Contact Us | Français|
Little is known about factors that affect bone mass and density of infants and toddlers and the means to assess their bone health owing to challenges in studying this population. The objectives of this study were to describe age, sex, race, growth and human milk feeding effects on bone mineral content (BMC) and areal density (aBMD) of the lumbar spine, and determine precision of BMC and aBMD measurements. We conducted a cross sectional study of 307 healthy participants (63 black), ages 1 to 36 months. BMC and aBMD of the lumbar spine were measured by dual energy x-ray absorptiometry. Duplicate scans were obtained on 76 participants for precision determination. Age specific Z-scores for aBMD, weight and length (BMDZ, WAZ, LAZ) were calculated. Information on human milk feeding duration was ascertained by questionnaire. Between ages 1 and 36 mo, lumbar spine BMC increased about 5-fold and aBMD increased 2-fold (p<0.0001). BMC was greater (5.8%) in males than in females (p=0.001), but there was no difference in aBMD (p=0.37). There was no difference in BMC or aBMD between whites and blacks (p≥0.16). WAZ and LAZ were positively associated with BMDZ (r=0.34 and 0.24, p<0.001). Duration of human milk feeding was negatively associated with BMDZ in infants <12 months of age (r=−0.42, p<0.001). Precision of BMC and aBMD measurements was good, 2.20% and 1.84%, respectively. Dramatic increases in BMC and aBMD of the lumbar spine occur in the first 36 months of life. We provide age-specific values for aBMD of healthy infants and toddlers that can be used to evaluate bone deficits. Future studies are needed to identify the age when sex and race differences in aBMD occur, and how best to account for delayed or accelerated growth in the context of bone health assessment of infants and toddlers.
Research on factors that affect bone mineral content (BMC) and density (BMD) and the assessment of bone health of infants and toddlers has received little attention owing to challenges in studying this population. Dual-energy x-ray absorptiometry (DXA) is the most widely used technique for measuring BMC and areal BMD (aBMD) in children and adults due to its low cost, accessibility, ease of use, and safety. Current generation DXAs have markedly enhanced resolution that improves the accuracy for measuring small, less dense bones and require much shorter scan times than earlier models. These improvements are particularly advantageous for measuring BMC and aBMD in infants and young children whose bones are less dense and who have difficulty remaining motionless. BMC and aBMD measurements are central components of a bone health assessment and are obtained in individuals with medical conditions or use medications that adversely affect bone. Clinical evaluation of BMC and aBMD requires information on the distribution of values of BMC and aBMD in healthy, normally growing children of the same age that can be used as a reference. However, data on BMC and aBMD from current generation densitometers on children below 3 years of age are lacking.
Most prior studies in infants and toddlers have used total body scans that provide measurements of total body BMC and aBMD. Older generation pencil beam densitometers acquired scans in a rectilinear fashion, which allowed the infant to be held still while the x-ray beam traversed another portion of the body. However, the scan path of current generation fan-beam densitometers prevents this practice, and the infant must lie completely still throughout the entire scan. Although the scan time is short, remaining motionless for even 2 minutes without use of restraints or sedation is unrealistic for many children younger than 36 months of age. Scans of a smaller region, such as the lumbar spine, are an alternative for infants and very young children because the scan time is rapid (<30 sec), and the infant can be held still without interfering with the scan. Although there are data for lumbar spine BMC and aBMD for infants and toddlers on older generation pencil beam densitometers (1–3), to our knowledge, there are no published age-specific values for BMC or aBMD of the lumbar spine obtained on current generation fan beam densitometers that can be used to clinically evaluate bone deficits of infants and very young children.
Consideration of sex and race effects on age-related trajectories of BMC and aBMD in healthy children is important in the development of reference data. There are notable sex and race differences in BMC and aBMD after puberty (4–9), but it is not clear at what age these differences appear. In older children, consideration of growth relative to same-aged peers is also critical in the evaluation of BMC and aBMD as children who are short for age have lower BMC and aBMD than children of average height (10). The effect of growth status on BMC and aBMD during the infant and toddler period remains to be determined. This is particularly important owing to the different patterns of growth of human milk fed and formula fed infants (11).
The objectives of this project were to describe age, sex and race-related trends in lumbar spine BMC and aBMD using current generation DXA technology among infants and young children 1 to 36 months of age, and to evaluate the effects of attained growth and human milk feeding on bone measures. Lastly, we aimed to determine the precision of BMC and aBMD measurements.
We recruited healthy infants and children between the ages of 1 to 36 months for this cross sectional study. Subjects were recruited by a variety of mechanisms including advertisements around the medical center and in childcare centers and mailings to families of prior study participants. The upper age limit was 36 months as there are aBMD reference data for current generation Hologic densitometers beginning at this age.
Inclusion criteria were weight, length, and weight-for-length within the 3rd and the 97th percentiles for age based on the CDC 2000 growth reference (www.cdc.gov/growthcharts/clinical_charts.htm), and general good health. Exclusion criteria included past medical history of neuromuscular, bone, gastrointestinal, liver, or renal disease; endocrine or growth abnormalities; current or prior use of medications that may affect bone (e.g. intravenous or oral glucocorticoids, growth hormone); preterm birth (< 37 weeks gestation) or low birth weight (< 2500 g); indwelling hardware; prior fracture; and delay in gross motor skills. Gross motor skills were assessed according to the Denver II sub-scale (12). Children were excluded if they could not perform a skill that 90% of children their age can perform (e.g., sit without support at 6.8 months, walk well at 14.9 months, jump up at 28.8 months).
A lumbar spine DXA scan was obtained on the Hologic Discovery A densitometer. Scans were performed in “fast array” mode. The child lay supine upon the scanning table with the parent or technician stabilizing the child at the legs and arms. We attempted one repeat scan in the event of movement if the infant was cooperative. DXA scans were analyzed using the infant spine software (version 12.7) to generate BMC and aBMD values of the lumbar spine region L1–L4. We obtained a second DXA scan on 76 children, after repositioning, to allow calculation of precision (reproducibility).
Weight and length were measured with the child dressed in lightweight clothing and without shoes. Weight (± 0.01 kg) was measured on an electronic balance, and recumbent length (± 0.1 cm) was measured on a length board. Height (± 0.1 cm) was measured with use of a wall-mounted stadiometer for children > 2 years of age. Weight, length and height measurements were acquired in triplicate, and the mean used to compute Z-scores using the CDC 2000 growth reference. Because recumbent length measurements are typically greater than height measurements in children (13), we added a constant of 0.8 cm to height measurements to calculate a standardized variable of length to use in analyses. The constant 0.8 cm is the difference between the 50th percentiles for length and height for children 2 to 3 years of age in the CDC 2000 growth reference. Information on whether they had consumed human milk, the duration of human milk intake, and whether they were currently consuming human milk, was obtained by questionnaire. The Institutional Review Board at Cincinnati Children’s Hospital approved this study, and parents provided informed consent.
Multiple regression was used to evaluate the influence of age, sex, and race (African ancestry, black) on BMC and aBMD. First, age trends in BMC and aBMD (dependent variables) were fitted with linear and quadratic terms. Terms for sex, sex-by-age interaction, and black race were sequentially evaluated in regression models. P values < 0.05 were considered statistically significant. Because growth in infancy follows a logarithmic scale, and the variance of growth measurements increases with age, these regression analyses were performed on log transformed BMC and aBMD data.
BMC and aBMD reference curves were created relative to age using LMS Chartmaker Pro version 2.3 (14). The LMS method utilizes the power for the Box-Cox transformation, median, SD/M to generate BMC and aBMD curves relative to age (15). The LMS analysis generates age-specific values for the median (M), SD/M (S), and power for the Box-Cox transformation (L), which are used to construct centile curves using Equation 1 as follows:
where Z is the Z-score that corresponds to a given percentile. For an individual DXA measurement (X), the Z-score can be calculated using the age-specific L, M, and S parameters and Equation 2 as follows:
The fit of the centile curves was assessed by visual inspection and Q-Q plots.
We used multiple regression to evaluate the effects of attained growth status (i.e. weight-and length-for-age Z-scores) and human milk feeding on BMD Z-scores. BMD Z-scores were normally distributed so logarithmic transformation was not necessary. Because duration of human milk feeding is inherently constrained by age of the infant, we evaluated the influence of human milk feeing duration on BMD Z-scores within 12-month age strata, and tested the interaction of human milk feeding duration and age on BMD Z-scores.
Precision error was calculated from the duplicate measurements according to the approach of Gluer (16).
Of the 369 children enrolled, we obtained acceptable scans without movement on 83% (307/ 369) of subjects. The percentage of subjects with acceptable DXA scans increased with age: 73% (37/51) of infants 1–6.0 mo, 72% (50/69) of infants 6.1–12.0 mo, 80% (51/64) of infants 12.1–18.0 mo, 85% (59/69) of infants 18.1 to 24.0 mo, 93% (56/60) of infants 24.1–30.0 mo, and 96% (54/56) of infants 30.1–36.0 mo (p = 0.0005). There were no differences in weight-or length-for-age, sex and race between infants with and without an acceptable scan (all p > 0.05). Descriptive characteristics of the 307 subjects for whom we obtained an acceptable DXA scan are given in Table 1.
BMC and aBMD increased with age (p < 0.0001) (Figure 1); between ages 1 and 36 mo, lumbar spine BMC increased approximately 5-fold and aBMD increased 2-fold. In multiple regression models accounting for age, we found that boys had 5.8% greater BMC than girls (adjusted means 6.89 vs. 6.51 g, p = 0.001), but there was no difference in aBMD (adjusted means 0.366 vs. 0.362 mg/cm2 respectively, p = 0.37). Furthermore, there was no age-by-sex interaction (p ≥ 0.13). The sex difference in BMC was related to greater weight (+ 0.64 kg) and length (+ 1.3 cm) in boys compared to girls when accounting for age. There was no significant difference (p = 0.77) in BMC between boys and girls when weight and length were included in the multiple regression models. There were no differences between black and white participants in BMC (adjusted means 6.60 vs. 6.80 g respectively, p = 0.16) or aBMD (adjusted means 0.363 vs. 0.364 mg/cm2 respectively, p = 0.87) when adjusted for age and sex. Mixed race and Asian infants were excluded from these analyses to maximize sensitivity of the black - white comparison.
BMC and aBMD curves corresponding to the median, −2 SD and +2 SD (i.e. 50th, 2.3rd and 97.5th percentiles) from the LMS model according to age and sex are shown in Figure 1. Because there were no differences in BMC and aBMD according to race, groups were combined to maximize sample size. All BMC curves for girls were notably lower than for boys, whereas the median aBMD curves were similar for girls and boys; these findings are consistent with results from multiple regression analyses. However, the −2 SD aBMD curve tended to be lower for girls < 1 y of age, and the +2 SD aBMD curves tended to be lower for girls > 2 y of age. Given that we did not statistically detect differences in aBMD between girls and boys, we present modeled LMS parameters and selected percentiles for aBMD by age for girls and boys combined in Table 2. Combining girls and boys enhances the sample size and makes the ± 2 SD curves more robust.
Weight-for-age and length-for age Z-scores were both positively correlated with BMD Z-scores (r = 0.34 and r = 0.24 respectively, p < 0.0001). When both measures were included in a regression model, only weight-for-age Z-score remained a significant predictor of BMD Z-score (p < 0.0001).
Overall, BMD Z-scores of infants and toddlers who were ever fed human milk (n = 231) were lower compared to those who had not received human milk (n = 74) (−0.05 vs. 0.21 respectively, p = 0.047). This difference was more pronounced when comparing BMD Z-scores of those who were currently being fed human milk (n = 46) to those who were not (n = 259) (−0.48 vs. 0.10 respectively, p = 0.0003), and among all study participants, the duration of human milk feeding was negatively associated with BMD Z-scores (r = −0.19, p = 0.009). Because duration of human milk feeding is inherently limited by infant age, and effects of human milk feeding may “wash out” over time, we examined the effects of the duration of human milk feeding by 12 mo age groups. There was a significant interaction between the duration of human milk feeding and age group for predicting aBMD Z-scores (p < 0.005). The duration of human milk feeding was inversely associated with BMD Z-scores among infants < 12 months of age (r = −0.42, p < 0.0001) (Figure 2), whereas there was a negligible association among participants 12–36 mo. These associations persisted when statistically adjusting for weight-for-age Z-scores in regression models.
The precision estimates for BMC and aBMD measures were 2.2% and 1.8%, respectively, and improved with age. Among infants 1–12 mo of age (median = 4 mo, n = 17) the precision error was 2.9% for BMC and 3.1% for aBMD. The precision errors for infants 12–24 mo (median = 18 mo, n = 27) were 2.8% for BMC and 1.9% for aBMD, and for children ages 25–36 mo (median = 29 mo, n = 32) 1.6% for BMC and 1.3% for aBMD.
Progress in the field of bone densitometry of infants and very young children has been hampered by technical challenges in measuring small bones of low density and by the need for infants to remain completely still during the scan. Indeed, movement related artifact has been cited as one of the main challenges in pediatric bone imaging (17). Utilizing current generation fan beam DXA and recently released infant scan analysis software, we successfully obtained BMC and aBMD measurements of the lumbar spine on infants 1 to 36 months of age. The precision errors were small overall, only 2.2% for BMC and 1.8% for aBMD, and only slightly larger than those for children 6–9 years of age (1.5% and 1.1% for BMC and aBMD, respectively) (18). These data demonstrate the feasibility of obtaining lumbar spine DXA scans on infants and toddlers with good precision. Furthermore, we provide age-specific values for aBMD of the lumbar spine for infants 1 to 36 months of age using current generation technology, which provides clinicians the opportunity to evaluate the bone health of infants and young children. The International Society for Clinical Densitometry (ISCD) recommends measurement of the lumbar spine in older children (19), thus these data allow continuity of follow-up measurements at the spine from infancy through childhood and adolescence.
Dramatic increases in BMC and aBMD occur in the first 36 months of life, and these relative increases are much greater than at any other point in life. As such this may be an age period of high susceptibility for bone mineral deficits. Low BMC and aBMD are associated with increased fracture risk in children and adolescents (20–23), thus it is possible that compromised bone status increases fracture risk in this younger age group as well. The long-term consequences of low BMC and aBMD during the first 36 months of life are unknown. Given the ability of growing bone to model and remodel in response to recent conditions and physical challenges, it is possible that compensations in bone occur once factors that resulted in low BMC and aBMD are no longer present (24).
Prior studies have shown that at all skeletal sites, BMC and aBMD are greater in males than in females after puberty (4–8). Evidence of sex differences is inconclusive in young children. Some studies have shown that there are no sex differences in the first year of life in total body (25, 26) or spine BMC and aBMD (2, 3), whereas one study found that males ages 1 to 18 months had a higher total body BMC than females (27). Another study found that males 0–36 mo had a greater mid radius BMC than females (28). We found that lumbar spine BMC was greater in males than females, owing to their greater weight and length, but there was no difference in mean or median aBMD. Though the 50th percentile aBMD curves were similar, the outer centile curves for boys and girls diverged. It is unclear whether this is an artifact due to limited amount of data at the extremes or a real difference. Studies with larger samples sizes are needed to investigate this more fully.
Greater BMC and aBMD in black compared to white children and adults have been well documented. The age at which these differences become evident is less clear. We did not find race differences in BMC and aBMD of the lumbar spine in infants 1–36 months of age in this study. Similarly, no race differences were found in total body BMC measured by pencil beam DXA, or forearm BMC and aBMD by single photon absorptiometry, in infants 1 to 18 months of age (25–27). However, Li et al. (29) found that BMC at the forearm measured by single photon absorptiometry was higher in black children compared to white children 1–6 years of age. Among children participating in the Bone Mineral Density in Childhood Study, we found that black children had greater BMC and aBMD of the lumbar spine as well as other skeletal sites than non-black children at ages ≥ 5 y (30). Thus, it is unclear at what age race differences in BMC and aBMD appear and whether this varies by skeletal site.
Consistent with studies of older children and adolescents, we found that relative growth status reflected by weight- and length-for-age Z-scores, was significantly associated with BMD Z-scores. The International Society for Clinical Densitometry (ISCD) recommends that interpretation of bone measurements in children and adolescents consider growth and maturation (19). Our findings indicate that this recommendation should be expanded to include infants and young children. Unlike studies of older children, in this study of infants and toddlers, weight-for-age Z-score was more strongly associated with lumbar spine Z-score than was height-for-age Z-score (10). Further research is needed to identify the best way to account for delayed or accelerated growth in the assessment of bone health in infants and toddlers. This is particularly important for infants who were born premature or small for gestational age and whose growth is delayed.
The duration of human milk feeding was associated with lower aBMD among infants < 12 months of age. This effect persisted when statistically accounting for weight, which tends to be lower among infants exclusively receiving human milk compared to those receiving formula (11). Human milk has notably lower content than cow milk based infant formula and cow milk of calcium (333 vs. 1333 and 1133 mg/L, respectively), phosphorus (133 vs. 300 and 866 mg/L), protein (10.5 vs. 13.9 and 32.0 g/L) and vitamin D (0.77 vs. 10.1 and 13.0 μg/L) (NDSR 2011, University of Minnesota, Minneapolis, MN). The lower concentration of these nutrients may have resulted in the lower BMD of infants receiving human milk for longer duration. The fact that there was negligible effect of human milk feeding duration among older infants and toddlers is consistent with a transient effect of calcium intake. In a randomized infant feeding trial, Specker et al (31) found that greater calcium and phosphorus intakes in the first 6 mo of life resulted in greater total body BMC, but differences disappeared by 1 year of age when infants received a higher mineral formula or cow milk. Phosphorus from solid foods also may help compensate for the lower levels in human milk as infants transition to a mixed diet. We did not have information on the duration of exclusive human milk feeding nor information on vitamin D supplement usage or other aspects of dietary intake to investigate these relations more fully in the current study. Our relatively simple measure of human milk exposure could have resulted in measurement error, which would attenuate the association with aBMD in this study. Inability to control for variability in vitamin D supplement use according to breastfeeding duration may have resulted in confounding. Given these limitations, further research is needed to better understand the impact of human milk feeding on bone health.
Although our data provide an important advance in bone density assessment of infants and toddlers using current generation DXA technology, there are some limitations. Most notable is the small sample size, especially of black participants, and limited representation by other ancestral and ethnic groups. Our sample was obtained at only one center and may not be representative of infants in general or potential differences in calibration of densitometers. We performed measurements using the Hologic densitometer; additional studies in this young age group are needed to determine the comparability in results obtained with densitometers made by other manufacturers. Furthermore, only 1 skeletal site was measured, which may not be representative of bone health throughout the skeleton. Larger studies that include a more diverse sample from several geographic locations and ancestral origins, and that measure bone mass and density at multiple skeletal sites, are needed to better characterize the normal accrual of bone mass and density in infants and toddlers. Our sample was of healthy individuals. It is unclear how well the infant spine software will identify bone edges (i.e., the bone map) of infants with chronic conditions or who were premature and whose bone density is likely to be lower than what we measured in this study. The ability to obtain DXA scans on infants without movement remains a challenge: we were able to obtain scans without movement on 83% of infants. It is likely that this percentage can be improved upon if more than two attempts are made to obtain a scan in the event of movement or more time is allotted for the infant to settle. Strategies to help infants remain still during scanning are needed.
In sum, we have demonstrated the feasibility of measuring bone density of the lumbar spine in infants 1 to 36 months of age, and provide age-specific values for aBMD that can be used to evaluate bone deficits. Future studies are needed to identify the age when sex and race differences in bone density occur, and how best to account for delayed or accelerated growth in the context of bone health assessment of infants and toddlers.
Funding: This work was funded by USPHS Grant #UL1 RR026314 from the National Center for Research Resources, NIH.
We would like to thank the families that participated in this study as well as the dedicated efforts of Gemma Uetrecht and Arin Fletcher for subject recruitment and data collection.
Disclosures: All authors state that they have no conflicts of interest.
Authors roles: Study design: HJK, JEH, KY. Study conduct: HJK. Data analysis HJK, BSZ; Data interpretation: HJK, BSZ, KY, JEH. Drafting manuscript: HJK. Revising manuscript content BSZ, KY, JEH. Approving final version of manuscript: HJK, BSZ, KY, JEH. HJK takes responsibility for the integrity of the data analysis.