Infant Growth Influences Proximal Femoral Geometry in Adulthood


  • The authors state that they have no conflicts of interest.


The relationship between early growth and adult femoral geometry has not been studied previously. In 333 adults, we were able to show that infant weight predicts femoral width and cross-sectional moment of inertia but not femoral neck length. These results support the hypothesis that growth in early life leads to persisting differences in proximal femoral geometry.

Introduction: Both femoral geometry and bone mass have been shown independently to predict both hip strength and fracture risk. Whereas growth during intrauterine and early postnatal life has been shown to influence adult bone mass, the relationship between growth in early life and adult femoral geometry has not been described previously.

Materials and Methods: We studied the relationship between growth during early life, adult hip geometry, and proximal femur bone mass in a sample of 333 men and women (60–75 years of age), for whom birth weight and weight at 1 year of age were recorded. Hip geometry was derived using Hip Structure Analysis software from proximal femur DXA scans (Hologic QDR 1000).

Results: There were significant (p < 0.002) relationships between weight at age 1 year and measures of femoral width as well as intertrochanteric (IT) cross-sectional moment of inertia (CSMI), but not with femoral neck length. The relationships with measures of femoral width but not CSMI remained after adjusting for adult body weight and were independent of proximal femoral BMC.

Conclusions: These results support the hypothesis that different patterns of growth in utero and during the first year of life lead to persisting differences in proximal femoral geometry, thereby mediating in part the effects of early growth on risk of hip fracture in adulthood.


Osteoporosis is a major cause of morbidity and mortality through its association with fragility fractures. Of the estimated £1.7 billion cost each year incurred by patients with osteoporotic fractures in England and Wales, 80% are caused by hip fracture.(1) Whereas BMD is well recognized as a risk factor for hip fracture,(2) the distribution of bone mineral within the proximal femur seems to be an independent determinant of proximal femoral strength. Among measures of femoral geometry, hip axis length(3) and femoral width(4) have been shown to determine hip fracture risk. These can be used to derive the cross-sectional moment of inertia (CSMI), a biomechanical measure of the bending strength of bone.(5)

Among determinants of osteoporotic fracture, there is a growing body of evidence to suggest that poor growth during intrauterine and early postnatal life might be associated with an increased likelihood of both low bone mass(6) and fracture(7) in later adulthood. However, observed differences in bone mass attributable to poor intrauterine and infant growth do not seem to account for the documented increase in risk of hip fracture in late adulthood. We therefore used a British cohort, which included data on birth weight, growth in infancy, and adult skeletal status (the Hertfordshire Cohort Study), to test the hypothesis that poor growth during early life influences femoral geometry some seven decades later.


We studied 333 healthy subjects (178 men and 155 women), 60–75 years of age, included in the skeletal component of the Hertfordshire Cohort Study.(8) The subjects were born and still reside in the county of Hertfordshire, UK, and are known to be representative of elderly men and women in the country, with regard to body build, cigarette smoking, and other lifestyle determinants of bone mass. The birth weight and weight at 1 year of age of each individual had been recorded in a ledger by a team of midwives and health visitors who had attended each birth in Hertfordshire and had visited the child's home at intervals during the first year of life. After obtaining written permission from each person's general practitioner, we approached each subject by letter asking whether they would be willing to be contacted by one of our research nurses.

At interview, each subject completed a questionnaire including information on previous medical history, current medications, alcohol intake, cigarette smoking, physical activity, and dietary calcium intake. In addition, measurements were made of height (using a stadiometer) and weight (using calibrated electronic scales). BMD was measured in each subject by DXA at the lumbar spine and femoral neck, using a Hologic QDR 1000 instrument. Bone area, BMC, and BMD were obtained directly from the scans. Measurement precision, expressed as the CV, was 1.55% for lumbar spine BMD, 1.45% for total femur BMD, and 1.83% for femoral neck BMD.

DXA scan images were first converted into bone mass images using the automated Hip Structure Analysis (HAS) program.(5) A single operator manually placed four cursers to define the limits of the femoral neck. The software automatically demarcated the linear axes of the neck and shaft as well as three cross-sectional analysis regions: the narrowest width of the femoral neck; the intertrochanteric region; and the femoral shaft. For each of these regions, cross-sectional measurements were derived from the distribution of bone mass across the bone using the method described by Martin and Burr.(9) Reproducibility of the technique in this study was assessed in 50 replicate scans; the CV was derived for the femoral neck length (5.1%), femoral width (0.83%), and CSMI (1.7%).

Data were double entered and analyzed using STATA v 7.0. Variables with a skewed distribution were normalized by an appropriate transformation where necessary. The relationships between birth weight, weight at 1 year, and adult proximal femoral geometry were explored using linear regression. Partial correlation coefficients after adjustment for body size were tested for statistical significance. Potential confounding variables, for example, smoking, were examined using multiple regression. Early growth was further characterized by calculating sex-specific SD scores for birth weight, weight at 1 year, and infant growth conditional on birth weight.(10) This conditional infant growth SD score was free of the artefactual effects of regression to the mean and could be included in analyses simultaneously with birth weight without multicolinearity problems. Use of a conditional characterization of postnatal growth enabled the independent effects of growth in the first year of life on femoral geometry to be separated from the effects of birth weight.


Table 1 shows the anthropometric characteristics of the 333 subjects. Whereas men and women were of similar age and current BMI, the men were significantly taller (p < 0.001) and heavier (p < 0.001) than the women; as infants, males weighed significantly more than females at 1 year (p < 0.001). At the proximal femur, men had greater BMD than women (p < 0.001). Femoral geometry measurements also differed by sex; men had significantly (p < 0.001) greater femoral neck length and width than women. CSMI at the narrow neck, intertrochanteric (IT), and femoral shaft was also greater in the men (p < 0.001). In both men and women, current height was significantly related to neck length (men, r = 0.28, p = 0.0002; women, r = 0.30, p = 0.0001), IT width (men, r = 0.44, p < 0.0001; women, r = 0.51, p < 0.0001), and IT CSMI (men, r = 0.46, p < 0.0001; women, r = 0.51, p < 0.0001). Current height was also significantly (p < 0.001) associated with width and CSMI at the narrow neck and femoral shaft sites.

Table Table 1.. Descriptive Characteristics and Proximal Femoral BMD and Geometric Measurements of 333 Adults for Whom Weight at Birth and 1 Year of Age Were Recorded
original image

Figure 1 and Table 2 show the relationship between femoral geometry and birth weight, weight at 1 year, and conditional infant growth in men and women. In both sexes, a 1-kg increase in weight at 1 year was associated with an increase in IT width by 0.11 cm (0.25 SD of IT width in men and 0.33 SD in women). These relationships persisted even after adjusting for lifestyle determinants of BMD (exercise, dietary calcium intake, cigarette smoking, and alcohol consumption), as well as current height and weight. Weight at 1 year also predicted IT CSMI in men and women and IT cortical buckling ratio in men (Fig. 1). The association between infant weight and IT cortical buckling ratio in the women was also positive but just failed to attain statistical significance (r = 0.14, p = 0.08). Whereas the relationship between IT CSMI and weight at 1 year did not remain statistically significant after adjusting for adult weight and height, weight at 1 year predicted IT cortical buckling ratio even after adjustment for both current height (r = 0.22, p = 0.008) and weight (r = 0.26, p < 0.01). Similar positive associations were found between weight at 1 year and femoral width, CSMI, and buckling ratio at the narrow neck and femoral shaft. Weaker associations were observed between birth weight and measures of femoral width (Table 2); adult CSMI was not, however, significantly associated with birthweight (Table 2).

Table Table 2.. Relationships Between Birth Weight, Weight at 1 Year, Conditional Infant Growth, and Femoral Geometry
original image
Figure Figure 1.

Relationship between infant weight and femoral neck length, intertrochanteric (IT) width, cross-sectional moment of inertia (CSMI), and cortical buckling ratio.

Weight at 1 year was also significantly (p < 0.01) associated with bone mass (BMC) at the proximal femur. We therefore evaluated the independent relationship between weight at 1 year, adult proximal femoral BMC, and proximal femoral geometry. The data relating each of these determinants to proximal femoral cross-sectional moment of inertia are shown in Fig. 2. Figure 2 shows that each third of weight at 1 year and proximal femoral BMC were independent predictors of adult proximal femoral CSMI (p < 0.001). The relationships between femoral geometry and conditional infant growth were strikingly similar to those between femoral geometry and weight at 1 year (Table 2), showing that postnatal growth was related to IT width and CSMI in men and women and buckling ratio in men, independently of size at birth.

Figure Figure 2.

Relationship between infant weight, proximal femoral bone mass, and intertrochanteric cross-sectional moment of inertia in 333 Hertfordshire men and women.

Eighteen men and 16 women had sustained fractures after age 45 years. Twenty-eight of these 34 fractures were at the wrist, with the remainder at the shoulder, hip, and spine. Measures of proximal femoral geometry did not show significant associations with fracture history, but analyses of relationships with hip fracture were not possible because of limited statistical power.


The results of this study suggest that poor growth in infancy is associated with disproportion of the proximal femur in later adult life and a possible corresponding reduction in the mechanical strength of this region. The selective association between weight in infancy and reduced width of the femoral neck, with relative preservation of femoral neck length, may contribute to the previously observed relationship between poor growth in utero and during infancy, with the later risk of hip fracture.

Proximal femoral geometry is an independent determinant of hip fracture risk.(3,11) Whereas hip axis length is important in determining fracture risk, mechanical models suggest that other measures of femoral geometry, such as femoral width and cross-sectional moment of inertia, are also important contributors to strength. Hip structure analysis permits dissection between these various structural components; previous cadaveric studies have suggested that the mechanical characteristics of the proximal femur are more closely assessed by these measures than by BMD measurement alone.(5)

This study adds to the growing body of evidence for the developmental origins of osteoporosis. Programming is the term used to describe lasting changes in structure and function caused by environmental stimuli acting at critical periods during early development.(12) Several epidemiological studies have shown an association between growth in infancy and both adult bone mass(6) as well as hip fracture.(7) Mother-offspring cohort studies have delineated maternal environmental characteristics, such as nutrition, smoking, and physical activity, which alter the trajectory of mineral accrual in the fetal skeleton.(13) However, most studies to date have evaluated the influence of these early environmental characteristics on BMC rather than on femoral geometry. Our study suggests that an adverse intrauterine environment might influence fracture risk not only through reduced bone mass, but also by an alteration in proximal femoral shape and structure. The underlying mechanism for this specific effect of poor intrauterine and early postnatal growth on femoral shape is not known. Poor growth in early life may alter subperiosteal apposition at the proximal femur either during infancy or by altering the trajectory of growth in subsequent years. The timing of this effect might be determined in future studies by assessment of proximal femoral geometry during childhood and adolescence.

Our study has several limitations. These include the use of the DXA images for assessment of proximal femoral geometry. Although these images were not designed for this purpose, the hip structure analysis software has been well validated and limitations in the system described previously.(14) The degree of anteversion of the femoral neck would also alter estimation of neck length and neck shaft angle; such bias would act conservatively in our study. Furthermore, previous studies have shown that at least 15° of anteversion is required before significantly modifying proximal femoral measurements.(15) We found that the CV for femoral neck length was higher than that for femoral width and CSMI. This is because the former measurement is determined as the distance between the center of the femoral head and the intersection of the neck and shaft axes. The latter is automated, but the former is strongly dependent on the user correctly placing the circle on the head. We were unable to measure muscle strength around the hip joint in this study. Although adult grip strength, a measure of muscle strength at a distant site that is moderately well correlated with quadriceps strength, was significantly associated with femoral neck width, the relationship was no longer apparent after adjustment for adult height. Finally, the Hertfordshire Cohort Study does not include detailed information on gestational age at birth or maternal lifestyle (e.g., cigarette smoking). During the period covered by the cohort, it is likely that almost all babies were term births, and we were able to explore relationships between weight in infancy and proximal femoral geometry among subsets in the lowest and highest 10ths of the sex-specific birthweight distribution. These relationships were constant in these two subsamples, suggesting that they act across the population distribution of birth weight rather than being subject to a threshold by gestation. Although information on maternal smoking was not available, we were also able to analyze the relationships between proximal femoral geometry and smoking during adult life among women. No significant relationships were found between any of the geometric parameters and smoking status in adulthood.

In conclusion, the results of this study show that low weight in infancy is associated with reduced femoral neck width, but with relative preservation of femoral neck length. This disproportion results in reduced CSMI and might account for reduced proximal femoral strength several decades later. The association seems independent of that between poor early growth and adult bone mass; it may explain, in part, the developmental origins of hip fracture.


The authors thank the men and women who participated in the study and the nurses and radiology staff of the Hertfordshire hospitals who administered the bone mineral measurements. Computing support was provided by Vanessa Cox, and the manuscript was prepared by Gill Strange. The study was supported by grants from the Medical Research Council of Great Britain, the Arthritis Research Campaign, the Cohen Trust, and the National Osteoporosis Society.