• osteoporosis;
  • hip fracture;
  • BMD;
  • hip strength analysis;
  • DXA


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

In this prospective, case-control study, femoral neck diameter, cross-sectional moment of inertia, or section modulus was an independent predictor of hip fracture risk after adjustment for BMD. However, the contribution of each of these indices to hip fracture prediction was modest in the presence of BMD.

Introduction: The relative contribution of measures of hip strength to hip fracture prediction is unclear. This study was designed to characterize the association between hip strength indices and hip fracture risk in relation to BMD in elderly men and women.

Materials and Methods: Seventy-one women and 25 men ≥60 years of age, who sustained a hip fracture during the study period of 1989–2003, were selected from the prospective, population-based Dubbo Osteoporosis Epidemiology Study. These fracture cases were randomly matched for age and sex in a 1:2 ratio with nonfracture individuals. BMD at the femoral neck was measured before the fracture event by DXA (Lunar DPX-L). Hip strength indices, including femoral neck diameter (FND), cross-sectional moment of inertia (CSMI), and section modulus (Z), were estimated by reanalysis of the image files using hip strength analysis software.

Results: In women, after adjustment for BMD, increased risk of hip fracture was associated with smaller FND (OR, 1.6; 95% CI, 1.0, 2.7), lower CSMI (OR, 1.8; 95% CI, 1.0, 3.2), or Z (OR, 1.6; 95% CI, 1.1, 5.1). In men, none of these hip strength indices were significant predictors of fracture risk. However, using the results in women as a prior distribution, it was estimated that the BMD-adjusted OR for FND (OR, 1.5; 95% CI, 1.0, 2.3), CSMI (OR, 1.6; 95% CI, 1.0, 2.5), or Z (OR, 2.3; 95% CI, 1.4, 3.9) was each significantly associated with hip fracture risk in men. In the logistic regression model, BMD alone accounted for 32% and 16% of the variance of fracture liability in women and men, respectively. The addition of FND, CSMI, or Z to the model increased the respective variance proportion to 34% and 19%.

Conclusions: These data suggest that smaller FND and lower CSMI or Z is an independent risk factor for hip fracture in both women and men. However, the contribution of these measures to hip fracture prediction over and above BMD is likely modest.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

HIP FRACTURE AMONG the elderly population represents a major public health problem because it is associated with increased mortality, morbidity, and health care costs. (1–3) Low BMD assessed by DXA is one of the major risk factors for hip fracture. (4) Although a BMD measurement can provide information on fracture risk, it cannot accurately identify individuals who will or will not actually sustain a fracture. (5)

In recent years, it has been increasingly recognized that the probability of fracture is a function of bone strength and nonskeletal factors. Bone strength does not only depend on the amount of bone mineral in the bone; it also depends on certain structural characteristics of the skeleton. At the proximal femur, in vitro studies have shown an association between bone strength and both BMD and several geometrical measurements of the femoral neck, such as increased femoral neck diameter and biomechanical indices of bending resistance. (6–8) Logically, measurements of bone structure may improve the prediction of hip fracture when used in conjunction with BMD. (9, 10)

Several in vivo studies have shown that variation in geometrical measurements, such as increased hip axis length(9–13) or femoral neck-shaft axis angle, (11, 13, 14) was associated with increased hip fracture risk. Additionally, both a bigger(11, 13–16) and a smaller(17–20) femoral neck diameter have been reported to be a risk factor for hip fracture. However, the evidence is thus far predominantly based on retrospective case-control studies in women, (10–17, 19, 21, 22) and the possibility of posttraumatic bone loss must be considered in the evaluation of geometrical measurements and their relation to BMD.

In addition to hip axis length and femoral neck diameter, it is possible to estimate hip strength indices such as the bending resistance of the femoral neck from data generated by DXA. (23, 24) This hip strength analysis has so far mostly been used to evaluate age-related changes in bone structure, (23, 25, 26) and to characterize differences between genders(25, 27–29) and races. (30–33) However, few studies have evaluated the association between these hip strength indices and hip fracture risk. (10, 18, 19) Although previous studies have suggested that these measurements might improve the prediction of hip fracture, it remains unclear if hip strength indices are BMD-independent predictors.

This study was therefore designed to characterize the association between DXA-based hip strength indices and hip fracture risk and to determine whether hip strength indices could improve hip fracture risk prediction over and above the contribution caused by BMD alone.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

Study population

This study was part of a larger study, the Dubbo Osteoporosis Epidemiology Study (DOES), a prospective, population-based epidemiological study evaluating risk factors for and outcomes of fractures. (34, 35) In 1989, all women and men ≥60 years of age living in Dubbo, a semiurban city of ∼32,000 people 400 km northwest of Sydney, Australia, were invited to participate in the study. From 1989 to 1993, a total of 1902 white males and females from an initial population of 1960 males and 2161 females were recruited into the study. The study was approved by the St Vincent's Campus Research Ethics Committee and informed written consent was obtained from each participant.

Selection of study sample

Between 1989 (baseline) and 2003, 73 women and 25 men sustained a hip fracture after a fall from no higher than the standing position. The closest DXA measurement before the fracture event in each of the subjects was identified. Using the age of the subjects with hip fracture at this visit, these subjects were randomly age- and gender-matched in a 1:2 ratio with controls who had not sustained any fracture during the study period. Because two of the DXA scans were not possible to retrieve from their storage discs, the final study sample consisted of 71 women and 25 men with hip fractures and 142 female and 50 male age-matched controls. Based on the primary aim of the study, the sample size was determined before the study to detect an OR of 2.0 of the association between each hip strength indices and hip fracture at the significance level of 5% and power of 80%.

Data collection

All subjects were interviewed by a nurse coordinator at initial and subsequent visits, which took place at approximately every second year. Each subject was administered a standardized questionnaire to solicit information on general health, anthropometric data, and fracture history. BMD (g/cm2) was measured at the femoral neck by DXA using a Lunar DPX-L densitometer (Lunar Corp., Madison, WI, USA) at each visit. The right hip was scanned in all cases unless there had been a hip replacement, in which case the left hip was scanned.

To identify all hip fractures that occurred in the study population after the baseline measurements, all radiography reports from the two area radiology services that service the entire Dubbo were reviewed. (34, 35)

Hip strength analysis

All image files were reanalysed by one of the authors, who was blinded for fracture status, using hip strength analysis software provided by Lunar Instruments Corp. (Madison, WI, USA). The X-ray absorption data of the proximal femur are, with this software, extracted from the output image data file, and the amount of bone mineral and its distribution within the femoral neck are calculated. First, the operator has to manually define the center of the femoral head and place the femoral neck axis as accurately as possible along the femoral neck. The femoral neck region of interest is placed in the proximal part of the femoral neck, and finally, the femoral shaft axis is defined centrally along the shaft. The software will iteratively assess all cross-sections in the femoral neck region of interest and identify the plane with the least cross-sectional moment of inertia (CSMI, cm4).

Because the estimate of CSMI is dependent on the orientation of the neck axis, the neck axis was placed in the orientation that accounted for the minimum CSMI. CSMI is an estimate of the ability of the femoral neck to withstand bending forces and was calculated using the mass distribution of the absorption curve. (23) The CSMI estimated with DXA has been found to be highly correlated with the CSMI measured directly on cadaver specimens (r2 = 0.96). (23) The automatic identification of the weakest cross-section of the femoral neck is the central part of the hip strength analysis software, and this cross-section level is used for the subsequent calculations of section modulus (Z, cm3), cross-sectional area (CSA, cm2), and compressive stress (Cstress, N/mm2).

The section modulus is also an estimate of the ability of the femoral neck to withstand bending forces and was calculated as CSMI divided by the distance from the center of the mass to the superior neck margin. CSA, a measure of the resistance of the bone to axial forces, represents the area of mineral packed together in the defined cross-section of the femoral neck and is, in principle, proportional to BMC. Cstress represents the compressive stress on the superior surface of the femoral neck during a fall on the greater trochanter. The fall force, which is dependent on body weight and height, generates both a compressive force and a bending moment on the femoral neck. Cstress was calculated by combining the ratio of compressive force to CSA and bending moment to CSMI. (23)

The femoral neck diameter (FND, cm) is the average diameter of the femoral neck within femoral neck region of interest and is also calculated by the hip strength analysis software. The femoral neck axis length (FNAL, cm) was defined as the linear distance from the apex of the femoral head to the lateral aspect of the femur along the femoral neck axis defined in the hip strength analysis and measured using the DXA ruler option.


The reproducibility of the hip strength analysis was determined by five repeated scans in six young healthy subjects after repositioning of the subject. The SE and the CV of BMD and hip strength indices were calculated. On average, the SE of hip strength measurements was between 1.6% and 3.7% of their mean, except for the composite parameter Cstress, whose CV was 11.2% (Table 1).

Table Table 1.. Reproducibility of the Hip Strength Analysis Determined by Five Repeated Scans in Six Young Healthy Subjects After Repositioning of the Subjects*
Thumbnail image of

Statistical analysis

From a statistical viewpoint, the specific aims of this study were (1) to estimate the strength of association between hip strength indices and hip fracture risk and (2) to assess whether the association was independent of BMD. If there was an independent relationship, how much of fracture risk variance was accounted for by the hip strength indices?

To address the first aim, the conditional logistic regression model was used to estimate the OR of fracture for each SD lower in the measurement of a hip strength index. To address the second question, each hip strength index was modeled as a predictor of fracture risk simultaneously with BMD in the conditional logistic regression model. Because hip strength indices are known to correlate with each other, it was also of interest to find a set of reasonably independent predictors that may yield a maximal discrimination between fracture and nonfracture. The backward elimination algorithm was used to search for this set of predictors.

To quantify the relative importance of each and combined risk factors, the Cox and Snell's pseudo r2 and the likelihood ratio index (LRI) were used. Both statistics are expressed in relation to the likelihood function, (36) and despite the contention of the usefulness of these statistics, they may be interpreted as an estimate of the proportion of risk variance accounted for by the risk factors in a logistic regression model. All analyses were performed using the SAS statistical analysis system (SAS Institute, Cary, NC, USA).

It was recognized that, because the study sample for men was modest, with a small number of fracture cases, the association between fracture risk and hip strength measurements may not attain the statistical significance threshold of p < 0.05. To provide a meaningful result for men, the Bayesian approach was used to estimate the association. In this approach, the question was as follows: given the observed association in women, what is the posterior association in men? Therefore, the source of prior information was derived from the female sample, and the likelihood was the observed result in men. With these two sets of information, it was possible to estimate the “likely” true relationship between hip strength indices and hip fracture in men. The mathematical details for this approach can be found in standard Bayesian textbooks. (37)


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

Characteristics of study subjects

All measurements, except Cstress, were normally distributed with constant variances. Women with hip fractures were both significantly shorter and lighter than their age-matched controls (Table 2).

Table Table 2.. Characteristics of Women and Men With Hip Fractures and Their Respective Age-Matched Controls
Thumbnail image of

In both genders, BMD, CSMI, Z, and CSA were all significantly lower and Cstress significantly higher in those with hip fracture compared with those without a fracture. FND was significant smaller in women with hip fractures compared with their age-matched controls (p < 0.001), but the difference was not statistically significant in men (p = 0.18). Furthermore, in both genders, BMD, Z, and CSA were the measurements with the highest standardized difference between those with and without hip fracture, and the effect size was on average stronger in women than in men (Table 2).

Correlation among variables

Correlation among all analyzed variables was on average similar in both genders. Advancing age was associated with lower BMD (r = −0.20) and bigger FND (r = 0.17; Table 3). BMD was positively correlated with CSMI (r = 0.61), Z (r = 0.73), and CSA (r = 0.89) but was negatively correlated with Cstress (r = −0.43). BMD was also positively correlated with FND (r = 0.37). However, when fracture cases and controls were analyzed separately, this correlation was only observed in women and men with hip fracture (r = 0.27 and r = 0.48, respectively; Fig. 1).

Table Table 3.. Correlation Between All Included Variables, Both Genders Combined*
Thumbnail image of
thumbnail image

Figure FIG. 1.. Relation between BMD and femoral neck diameter and section modulus at the femoral neck in elderly women and men. •, subjects with hip fracture; {hollow circle}, subjects without any fracture. The linear regression lines are solid in the fracture group and dotted in the control group.20

Download figure to PowerPoint

Among the hip strength measurements, CSMI, Z, and CSA were highly correlated (r = 0.88-0.97). FND was positively associated with all these variables (r = 0.71-0.90), whereas Cstress was negatively associated, albeit to a lesser degree (−0.41 to −0.51).

Association between hip strength measures and hip fracture

Each SD lowering in BMD was associated with a 7.5 (95% CI, 3.6, 15.4) OR of hip fracture in women and 2.5 (1.4, 4.4) in men. In univariate unadjusted analyses, FND, CSMI, Z, CSA, and Cstress were each associated with hip fracture risk in women. For men, the associations, except for FND, were also significant, albeit with generally lower ORs (Table 4). These results remained unchanged after adjustment for weight, as well as after adjustment for height, with the exception of FNAL, whose association with hip fracture risk was statistically significant after taking account the effect of height in women (Table 4).

Table Table 4.. Unadjusted, Weight-Adjusted, and BMD-Adjusted ORs for Hip Fracture Risk
Thumbnail image of

In women, after adjustment for BMD, smaller FND, lower CSMI, or Z was each significantly associated with hip fracture risk. In men, none of the hip strength indices were significant risk factors after adjustment for BMD (Table 4).

In the multivariate model where all factors were considered simultaneously, the only two significant independent factors in women were BMD (OR, 4.6; 95% CI, 2.0, 10.3) and Z (OR, 2.3; 95% CI, 1.1, 5.1).

The BMD-adjusted ORs for hip fracture risk in women with DXA measurements and hip fracture at the ipsilateral side were as follows: FND, 1.7 (0.6, 4.5); CSMI, 2.3 (0.7, 8.1); Z, 3.3 (0.6, 18.4). In women with DXA measurements and hip fracture at the contralateral side, the ORs were as follows: FND, 1.7 (0.9, 3.2); CSMI, 2.0 (0.9, 4.4); Z, 2.2 (0.8, 5.9).

Contribution of independent risk factors

In the conditional univariate logistic regression model, the variation in BMD accounted for 32% of the variation in fracture risk among women. When each hip strength index was considered as a single factor in the regression model, it was estimated that the proportion of variance of fracture risk explained by FND was 7%, CSMI was 22%, and Z was 29%. However, when both BMD and each of the hip strength indices were analyzed in a multivariate regression model, BMD and FND, CSMI, or Z accounted for 34% of the variance of fracture (i.e., a 2% improvement in predictive variance over and above that of BMD). In men, the variation in BMD or Z accounted for 16% of the variation in fracture risk, and the combination of BMD and Z improved the predictive variance by 3%.

Bayesian analysis

Using the results from the conditional logistic model in women as a prior distribution, it was estimated that the BMD-adjusted OR for FND (1.5 {1.0, 2.3}), CSMI (1.6 {1.0, 2.5}), or Z (2.3 {1.4, 3.9}) were also each significantly associated with hip fracture risk in men.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

One of the most frequently mentioned conjectures in fracture assessment is that measures of bone strength can contribute to the prediction of fracture risk. The results of this study partially supports that conjecture. In a sample of elderly women and men, smaller FND and lower CSMI or Z was each a significant determinant of hip fracture risk, independent of BMD. However, the fracture risk attributed to each of these hip strength indices was on average lower than the contribution of BMD, and when combined with BMD, each of these indices provided a modest improvement to the BMD-alone predictive value.

The association between low BMD and increased hip fracture risk is well documented(5) with each SD decrement in BMD being associated with a 2- to 3-fold increase in fracture risk. In this study, the magnitude of association between BMD and hip fracture risk was considerably higher than previously reported in other case-control studies(10, 38) and also stronger than previously reported from the entire cohort of DOES. (35, 39) Thus, it is likely that the strength of association between BMD and hip fracture risk in this study was overestimated, and accordingly, probably also the effect size of the hip strength measurements, because these were derived from the same BMD scan. However, this would not necessarily affect the magnitude of effect of the hip strength measurements relative to BMD, which, for both Z and CSA, was of a similar magnitude.

All of the hip strength indices examined in this study, except FNAL, were a significant predictor of hip fracture risk. However, only the effect of FND, CSMI, or Z was independent of BMD. These three indices were all highly correlated with each other, and although Z was the single strongest predictor of them, all contributed to a similar extent to the fracture prediction over and above BMD alone. However, the magnitude of this contribution was modest, and their clinical relevance is questionable. Although the distribution of mineral within a cross-section of the femoral neck seems to be an important determinant of bone strength, the additional information obtained by the hip strength analysis may not be sufficient to improve the discrimination between hip fractures and controls beyond BMD. Accordingly, the usefulness of the hip strength analysis in addition to the BMD measurement for fracture prediction among elderly individuals may be overestimated.

BMD, assessed by a 2D image technique such as DXA, is an estimate of the average amount of mineral per unit area in a section of bone facing the detector. (40) The hip strength indices, derived from the hip strength analysis software, are estimated by use of the same mass distribution curve. (9, 23, 24) Therefore, the hip strength indices, whether it is CSA, CSMI, Z, or Cstress, are all, to various extents, dependent of the amount of mineral present in the femoral neck, and accordingly, none of the measurements is totally independent of BMD. However, FND is independent of the amount of mineral, although it is theoretically possible that a low mineral content might influence the edge detection of the bone, and thus the measure of FND. On the other hand, BMD, as well as CSMI, Z, and Cstress, are all, to some extent, also dependent on FND. Therefore, the magnitude of correlation between these measurements can determine the degree of improvement (or contribution) to the predictive value when used in combination.

The finding that a smaller neck diameter was a risk factor for hip fracture is consistent with previous reported associations in both women(18–20, 41) and men. (16) However, more studies, albeit retrospective case-control studies, have shown that a bigger neck diameter was associated with an increased risk of hip fracture in women(11, 13–16) as well as in men. (14, 15) Notably, our contradictory finding is supported by an in vitro study showing that a bigger neck diameter was associated with increased, and not decreased, mechanical strength of the proximal femur. (6)

The finding that FNAL was not a significant predictor of hip fracture risk is also consistent with earlier findings. (11, 15, 42) However, in women, this study suggested that longer FNAL was associated with increased risk of hip fracture after adjusting for height. This finding is consistent with previous finding that longer hip axis length (which is highly correlated with FNAL) was a risk factor for hip fracture. (9–13) In addition to the linear distance of FNAL, the measurement of hip axis length also includes the distance between the apex of the femoral head and the inner rim of the pelvis. However, because the bone scan in this study did not include the inner pelvic rim portion, hip axis length could not be determined.

Age-related periosteal apposition has previously been reported to occur in both the long bones(43, 44) and the femoral neck. (26, 28) The periosteal apposition compensates in part for the decreased bone strength caused by the age-related bone loss from the endosteal surface and may be a compensatory mechanism to preserve bone strength as BMD decreases with advancing age. (43, 44) In this study, advancing age was associated with an increased FND and decreased BMD. Moreover, women with hip fractures had, compared with controls, a smaller FND in relation to BMD. The same tendency was observed in men. This supports the conjecture that a deficit in periosteal apposition could be an underlying mechanism in the pathogenesis of hip fracture. (17) However, this hypothesis must be tested in a study with longitudinal measurements of BMD and FND.

An important question is whether the strength of the association between hip strength indices and hip fracture risk might be stronger if measured at the same hip that was going to be fractured compared with the contralateral hip. An analysis was carried out to address the question, and the results suggested that the strength of association of the hip strength indices and hip fracture risk in women with hip fracture and DXA measurement at the same hip was virtually identical to those with hip fracture and DXA measurement at the contralateral hip separately.

This study has some advantages, in which study subjects were matched for the most important risk factor, namely age. Unlike other traditional case-control studies, this study followed subjects for up to 14 years, making it possible to measure the hip strength indices within 2 years before the fracture event. In addition, given the possibility of post-traumatic bone loss, a longitudinal study design as in this study is preferable when evaluating risk factors in relation to BMD.

However, some potential limitations should also be considered in interpreting these results. Women with hip fracture had lower weight than their age-matched controls, which could lead to inaccuracies in BMD measurements by DXA, possibly characterized by an underestimation of BMD and other bone parameters in women with hip fractures. (45) The sample size of this study, particularly in men, was modest, which could limit the statistical power to detect a modest association. Furthermore, it is possible that hip strength indices, such as CSMI, estimated by a 3D measurement technique (compared with the 2D measurement technique used in this study), may improve the contribution to hip fracture risk when used in conjunction with BMD assessed by DXA. Another possible limitation of the study was that potentially confounders, such as fall-related factors and soft tissue padding around the hips, was not measured in this study.

In summary, these data indicate that certain DXA-based measures of hip strength, such as FND, CSMI, or Z, were BMD-independent predictors of hip fracture risk in women and men. However, in the presence of BMD, the liability to fracture attributed to each of these hip strength indices was modest. Although the clinical use of these measures in an elderly population seems limited, the contribution to long-term prediction of hip fracture in a younger population remains to be evaluated.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References

The authors thank Janet Watters and Donna Reeves for assistance with data collection and measurements of bone densitometry, the staff of Dubbo Base Hospital for invaluable help, and Natasa Ivankovic for help in managing the database. HGA was supported by grants from the Tegger Foundation, Wiberg Foundations, Swedish Orthopaedic Society, and Lund University Foundation. This work was supported by the National Heath and Medical Research Council of Australia and untied educational grants from GE-Lunar, Merck Australia, Eli Lilly International, and Aventis Australia.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  8. References
  • 1
    Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA 1999 Mortality after all major types of osteoporotic fracture in men and women: An observational study. Lancet 353:878882.
  • 2
    Cooper C, Atkinson EJ, Jacobsen SJ, O'Fallon WM, Melton LJ III 1993 Population-based study of survival after osteoporotic fractures. Am J Epidemiol 137:10011005.
  • 3
    Ismail AA, O'Neill TW, Cooper C, Finn JD, Bhalla AK, Cannata JB, Delmas P, Falch JA, Felsch B, Hoszowski K, Johnell O, Diaz-Lopez JB, Lopez Vaz A, Marchand F, Raspe H, Reid DM, Todd C, Weber K, Woolf A, Reeve J, Silman AJ 1998 Mortality associated with vertebral deformity in men and women: Results from the European Prospective Osteoporosis Study (EPOS). Osteoporos Int 8:291297.
  • 4
    Cummings SR, Bates D, Black DM 2002 Clinical use of bone densitometry: Scientific review. JAMA 288:18891897.
  • 5
    Marshall D, Johnell O, Wedel H 1996 Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 312:12541259.
  • 6
    Cheng XG, Lowet G, Boonen S, Nicholson PH, Brys P, Nijs J, Dequeker J 1997 Assessment of the strength of proximal femur in vitro: Relationship to femoral bone mineral density and femoral geometry. Bone 20:213218.
  • 7
    Augat P, Reeb H, Claes LE 1996 Prediction of fracture load at different skeletal sites by geometric properties of the cortical shell. J Bone Miner Res 11:13561363.
  • 8
    Bouxsein ML, Courtney AC, Hayes WC 1995 Ultrasound and densitometry of the calcaneus correlate with the failure loads of cadaveric femurs. Calcif Tissue Int 56:99103.
  • 9
    Faulkner KG, Cummings SR, Black D, Palermo L, Gluer CC, Genant HK 1993 Simple measurement of femoral geometry predicts hip fracture: The study of osteoporotic fractures. J Bone Miner Res 8:12111217.
  • 10
    Peacock M, Turner CH, Liu G, Manatunga AK, Timmerman L, Johnston CC Jr 1995 Better discrimination of hip fracture using bone density, geometry and architecture. Osteoporos Int 5:167173.
  • 11
    Boonen S, Koutri R, Dequeker J, Aerssens J, Lowet G, Nijs J, Verbeke G, Lesaffre E, Geusens P 1995 Measurement of femoral geometry in type I and type II osteoporosis: Differences in hip axis length consistent with heterogeneity in the pathogenesis of osteoporotic fractures. J Bone Miner Res 10:19081912.
  • 12
    Bergot C, Bousson V, Meunier A, Laval-Jeantet M, Laredo JD 2002 Hip fracture risk and proximal femur geometry from DXA scans. Osteoporos Int 13:542550.
  • 13
    Gnudi S, Ripamonti C, Gualtieri G, Malavolta N 1999 Geometry of proximal femur in the prediction of hip fracture in osteoporotic women. Br J Radiol 72:729733.
  • 14
    Alonso CG, Curiel MD, Carranza FH, Cano RP, Perez AD 2000 Femoral Bone Mineral Density, Neck-Shaft Angle and Mean Femoral Neck Width as Predictors of Hip Fracture in Men and Women. Multicenter Project for Research in Osteoporosis. Osteoporos Int 11:714720.
  • 15
    Karlsson KM, Sernbo I, Obrant KJ, Redlund-Johnell I, Johnell O 1996 Femoral neck geometry and radiographic signs of osteoporosis as predictors of hip fracture. Bone 18:327330.
  • 16
    Duan Y, Beck TJ, Wang XF, Seeman E 2003 Structural and biomechanical basis of sexual dimorphism in femoral neck fragility has its origins in growth and aging. J Bone Miner Res 18:17661774.
  • 17
    Seeman E, Duan Y, Fong C, Edmonds J 2001 Fracture site-specific deficits in bone size and volumetric density in men with spine or hip fractures. J Bone Miner Res 16:120127.
  • 18
    Duboeuf F, Hans D, Schott AM, Kotzki PO, Favier F, Marcelli C, Meunier PJ, Delmas PD 1997 Different morphometric and densitometric parameters predict cervical and trochanteric hip fracture: The EPIDOS Study. J Bone Miner Res 12:18951902.
  • 19
    Crabtree NJ, Kroger H, Martin A, Pols HA, Lorenc R, Nijs J, Stepan JJ, Falch JA, Miazgowski T, Grazio S, Raptou P, Adams J, Collings A, Khaw KT, Rushton N, Lunt M, Dixon AK, Reeve J 2002 Improving risk assessment: Hip geometry, bone mineral distribution and bone strength in hip fracture cases and controls. The EPOS study. European Prospective Osteoporosis Study. Osteoporos Int 13:4854.
  • 20
    Karlamangla AS, Barrett-Connor E, Young J, Greendale GA 2004 Hip fracture risk assessment using composite indices of femoral neck strength: The Rancho Bernardo study. Osteoporos Int 15:6270.
  • 21
    Michelotti J, Clark J 1999 Femoral neck length and hip fracture risk. J Bone Miner Res 14:17141720.
  • 22
    Gnudi S, Ripamonti C, Lisi L, Fini M, Giardino R, Giavaresi G 2002 Proximal femur geometry to detect and distinguish femoral neck fractures from trochanteric fractures in postmenopausal women. Osteoporos Int 13:6973.
  • 23
    Yoshikawa T, Turner CH, Peacock M, Slemenda CW, Weaver CM, Teegarden D, Markwardt P, Burr DB 1994 Geometric structure of the femoral neck measured using dual-energy x-ray absorptiometry. J Bone Miner Res 9:10531064.
  • 24
    Beck TJ, Ruff CB, Warden KE, Scott WW Jr, Rao GU 1990 Predicting femoral neck strength from bone mineral data. A structural approach. Invest Radiol 25:618.
  • 25
    Beck TJ, Ruff CB, Scott WW Jr, Plato CC, Tobin JD, Quan CA 1992 Sex differences in geometry of the femoral neck with aging: A structural analysis of bone mineral data. Calcif Tissue Int 50:2429.
  • 26
    Beck TJ, Looker AC, Ruff CB, Sievanen H, Wahner HW 2000 Structural trends in the aging femoral neck and proximal shaft: Analysis of the Third National Health and Nutrition Examination Survey dual-energy X-ray absorptiometry data. J Bone Miner Res 15:22972304.
  • 27
    Crabtree N, Lunt M, Holt G, Kroger H, Burger H, Grazio S, Khaw KT, Lorenc RS, Nijs J, Stepan J, Falch JA, Miazgowski T, Raptou P, Pols HA, Dequeker J, Havelka S, Hoszowski K, Jajic I, Czekalski S, Lyritis G, Silman AJ, Reeve J 2000 Hip geometry, bone mineral distribution, and bone strength in European men and women: The EPOS study. Bone 27:151159.
  • 28
    Kaptoge S, Dalzell N, Loveridge N, Beck TJ, Khaw KT, Reeve J 2003 Effects of gender, anthropometric variables, and aging on the evolution of hip strength in men and women aged over 65. Bone 32:561570.
  • 29
    Looker AC, Beck TJ, Orwoll ES 2001 Does body size account for gender differences in femur bone density and geometry. J Bone Miner Res 16:12911299.
  • 30
    Nakamura T, Turner CH, Yoshikawa T, Slemenda CW, Peacock M, Burr DB, Mizuno Y, Orimo H, Ouchi Y, Johnston CC Jr 1994 Do variations in hip geometry explain differences in hip fracture risk between Japanese and white Americans. J Bone Miner Res 9:10711076.
  • 31
    Nelson DA, Barondess DA, Hendrix SL, Beck TJ 2000 Cross-sectional geometry, bone strength, and bone mass in the proximal femur in black and white postmenopausal women. J Bone Miner Res 15:19921997.
  • 32
    Nelson DA, Pettifor JM, Barondess DA, Cody DD, Uusi-Rasi K, Beck TJ 2004 Comparison of cross-sectional geometry of the proximal femur in white and black women from Detroit and Johannesburg. J Bone Miner Res 19:560565.
  • 33
    Yan L, Crabtree NJ, Reeve J, Zhou B, Dequeker J, Nijs J, Falch JA, Prentice A 2004 Does hip strength analysis explain the lower incidence of hip fracture in the People's Republic of China? Bone 34:584588.
  • 34
    Jones G, Nguyen T, Sambrook PN, Kelly PJ, Gilbert C, Eisman JA 1994 Symptomatic fracture incidence in elderly men and women: The Dubbo Osteoporosis Epidemiology Study (DOES). Osteoporos Int 4:277282.
  • 35
    Nguyen T, Sambrook P, Kelly P, Jones G, Lord S, Freund J, Eisman J 1993 Prediction of osteoporotic fractures by postural instability and bone density. BMJ 307:11111115.
  • 36
    Hosmer D, Lemeshow S 2000 Applied Logistic Regression, 2nd ed, Wiley, New York.
  • 37
    Congdon P 2001 Wiley Series in Probability and Statistics, John Wiley & Sons, Chichester, UK.
  • 38
    Greenspan SL, Myers ER, Maitland LA, Resnick NM, Hayes WC 1994 Fall severity and bone mineral density as risk factors for hip fracture in ambulatory elderly. JAMA 271:128133.
  • 39
    Center JR, Nguyen TV, Pocock NA, Eisman JA 2004 Volumetric bone density at the femoral neck as a common measure of hip fracture risk for men and women. J Clin Endocrinol Metab 89:27762782.
  • 40
    Nilsson BE, Johnell O, Petersson C 1990 In vivo bone-mineral measurement. How and why-a review. Acta Orthop Scand 61:275281.
  • 41
    Reeve J 2004 Measurement issues with bone apposition. J Bone Miner Res 19:689692.
  • 42
    Center JR, Nguyen TV, Pocock NA, Noakes KA, Kelly PJ, Eisman JA, Sambrook PN 1998 Femoral neck axis length, height loss and risk of hip fracture in males and females. Osteoporos Int 8:7581.
  • 43
    Smith R, Walker R 1964 Femoral expansion in aging women: Implications for osteoporosis and fractures. Science 217:945948.
  • 44
    Ahlborg HG, Johnell O, Turner CH, Rannevik G, Karlsson MK 2003 Bone loss and bone size after menopause. N Engl J Med 349:327334.
  • 45
    Bolotin HH, Sievanen H 2001 Inaccuracies inherent in dual-energy X-ray absorptiometry in vivo bone mineral density can seriously mislead diagnostic/prognostic interpretations of patient-specific bone fragility. J Bone Miner Res 16:799805.