SEARCH

SEARCH BY CITATION

Keywords:

  • osteoporosis;
  • falls;
  • bone quality

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

In an age- and sex-stratified population sample (n = 700), we estimated fall-related loads and bone strength indices at the UDR and FN. These load/strength ratios more closely simulated patterns of wrist and hip fractures occurring in the same population than did measurement of vBMD.

Introduction: Areal BMD measurements, although associated with fracture risk, incompletely explain patterns of fragility fractures. Moreover, population-based assessments relating applied loads and whole bone strength to fracture patterns have not been made.

Materials and Methods: Using QCT, we assessed volumetric BMD (vBMD), cross-sectional geometry, and axial (EA) and flexural (EI) rigidities (indices of bone's resistance to compressive and bending loads, respectively) at the ultradistal radius (UDR) and femoral neck (FN) and estimated the loads applied to the wrist and hip during a fall. We used fall load (FL)/bone strength ratios to estimate fracture risk.

Results: vBMD in young adults was similar between sexes. Decreases in vBMD over life were also similar (30% and 28%) at UDR but were somewhat greater (46% and 34%) at FN in women versus men, respectively. In young adults, FL/strength ratios at UDR were 32–51% lower (better) in men than in women and increased (worsened) over life less in men (+4% to +22%) than in women (+20% to +33%). In young adults, FL/strength ratios at FN were only marginally better in men than in women but worsened less over life in men (+22% to +36%) than in women (+40% to +62%).

Conclusions: The 6:1 female preponderance and the virtual immunity of men for age-related increases in wrist fractures are largely explained by the more favorable FL/strength ratios at UDR in young adult men (because of larger bone size and more favorable geometry) versus women and to maintaining this advantage over life. The 2-fold lower incidence of hip fractures in men versus women is largely explained by age-related increases (worsening) of FL/bone strength ratios that are only one-half of the increases in women. The moderate increases in these ratios with aging are insufficient to explain the >4-fold increase in hip fracture incidence after age 75 in both sexes, suggesting contributions of other factors, especially the well-documented increased frequency of injurious falls among the elderly.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

EPIDEMIOLOGIC STUDIES HAVE established a strong association between BMD, as evaluated by DXA, and fracture incidence, (1, 2) and DXA measurements are widely used clinically to assess fracture risk in individual patients. However, DXA assesses areal BMD (aBMD) rather than volumetric BMD (vBMD), overestimates the true vBMD of larger bones, (3, 4) cannot separate trabecular from cortical bone, and fails to assess bone geometry or structure. (5) QCT can address each of these limitations, although neither BMD measurements by 2-D DXA(6) or by 3-D QCT(7) track well with the age- and sex-specific patterns of fragility fractures.

Fractures occur when the load applied to a bone exceeds its ability to resist that load. Thus, in principle, the ratio of applied load to whole bone strength should predict fracture risk more efficiently than BMD measurements alone. (8) To assess this ratio, the activity or event associated with a fracture must be identified, and the applied load and bone strength for that event estimated. Previous studies have found that forward and sideways falls are the most common events associated with wrist and hip fractures, respectively. (9–12) The ability of a bone to resist fracture at a given loading configuration depends not only on its mass but also on the spatial distribution of bone tissue and the intrinsic properties of bone material. (13–16) Attempts have been made to estimate certain biomechanical indices of bone strength from DXA measurements, (17, 18) but this approach involves many assumptions with respect to bone shape and the distribution of bone tissue and fails to predict fracture risk better than aBMD measurements by hip DXA. (19)

Fractures of the wrist and hip are two of the three most common and important fractures associated with osteoporosis. However, they have strikingly different age- and sex-specific incidence patterns. (20) Thus, in a population sample, we sought to determine if the estimated ratio of the applied load in falls to whole bone strength at the ultradistal radius (UDR) and femoral neck (FN) was superior to measurement of vBMD in explaining the pattern of wrist and hip fragility fractures occurring in the same population from which the study sample was obtained.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

Study subjects

We studied an age-stratified, random sample of the population of Rochester, MN, as described in detail previously. (7) The sample spanned the age range from 21 to 97 years and included 375 women and 325 men. Ninety-four postmenopausal women were receiving estrogen therapy and six postmenopausal women and three men were receiving bisphosphonate or raloxifene therapy for osteopenia. Because analysis with and without the inclusion of these two groups of subjects gave similar results, all were included.

Bone densitometry

As previously described, (7) we used single-energy central QCT (Light Speed QX-I scanner; GE Medical Systems, Wakesha, WI, USA) and high-resolution pQCT (Densiscan 1000; Scanco Medical AG, Basserdorf, Switzerland) to assess vBMD and bone cross-sectional geometry at the FN and UDR, respectively. The UDR scanning site has been shown to represent the site of Colles' fracture of the wrist. (21) For the FN, we acquired a contiguous data set at the midportion of the femoral neck and analyzed a single reformatted oblique section orthogonal to the femoral neck axis. (22) The reconstructed slice thickness was 2.5 mm, and the in-plane pixel dimension was 0.74 × 0.74 mm. A calibration standard, scanned with the patient, was used to convert CT numbers (Hounsfield units) directly to equivalent vBMD (g/cm3). (22, 23) For the UDR, pQCT measurements were made between 7 and 20 mm proximal to the joint space between the radius and carpal bones (six contiguous slices, 1.5 mm thick, in-plane pixel size = 0.35 mm).

Geometric variables

From the QCT data, we determined the total cross-sectional area (CSA), moment-of-inertia (I), and section modulus (SM). CSA is an index of bone size, I represents the distribution of bone about any arbitrary axis, and SM is the I about an axis divided by the radius of the bone along that axis. The sum of bending moments about any two orthogonal axes is the polar moment or torsional I. For the FN, we assumed that the femur would bend principally about an antero-posterior oriented axis at the midfemoral neck during a sideways fall and, thus, computed the I about this axis (Iap). For the UDR, the manufacturer's software computes the polar I (Ip).

Bone strength indices

In addition to measurements of bone geometry and vBMD, we used the 3-D QCT data to compute indices of whole bone strength using an approach based on engineering structural analysis. (24) Briefly, assuming that bone tissue fails at a constant strain, (25) the failure load of a whole bone is proportional to the structural rigidity at its weakest cross-section. (26, 27) Structural rigidity measurements combine the intrinsic mechanical behavior of the bone material (i.e., elastic modulus {E} in megapascals {MPa}) with the relevant cross-sectional geometric properties (i.e., CSA in cm2 for compression and tension and I in cm4 for bending or torsion). Each bone voxel is assigned a unique value for E based on its vBMD. (28)

Axial rigidity (EA), in Newtons, reflects the resistance of bone to tensile or compressive loading and was computed as the product of the E of each bone voxel times its CSA, summing over all bone voxels in the cross-section as follows:

  • equation image

where Ei is the compressive E of the individual bone pixel, and Ai is the area of that bone pixel perpendicular to the primary loading direction.

Flexural or torsional rigidity (EI), in Nmm2, reflects the resistance of bone to bending or torsion, respectively. To compute the flexural rigidity at the FN, we assumed that, during a sideways fall, bending would occur principally about an anterior-posterior axis at the midfemoral neck. Thus, EIap was computed as follows:

  • equation image

where, in this case, x is the anterior-posterior axis through the midfemoral neck, N is the total number of pixels, Ei is the elastic modulus of the individual bone pixel, w and h are the width and height of the bone pixel, and d is the distance from the bending axis. For flexural rigidity at the UDR, the Densiscan device computes a polar I for the cortex and trabecular regions separately. We combined these values with mean E values in a “two-compartment” model computation of torsional rigidity (EIp). Although expressed in the same units, torsional and bending indices of structural strength cannot be directly compared.

Estimation of applied loads and load/strength ratios

For these estimates, we used the loading conditions for the most common type of trauma associated with each fracture—a forward fall for wrist fractures and a sideways fall for hip fractures. (9–12) For the hip, the applied load was estimated from data on the kinematics of sideways falls with impact at the hip. (12, 29, 30) The load applied to the wrist during a forwards fall was estimated from data predicting impact forces on the upper extremity during falls on the outstretched hand. (31) For each type of fall, individual height and weight data were used to estimate subject-specific loads that would be applied to bone as a result of a fall. We assessed the fall load/bone strength ratio at the UDR and FN scanning sites as estimates of fracture risk using both compressive rigidity (EA) and flexural rigidity (EI) as indices of bone strength. Because the fall load is in the numerator and the bone strength index is in the denominator, higher values for the ratio indicate increasing fracture risk.

Pattern of fragility fractures

The indices of bone strength and load/strength ratios were compared with updated incidence rates for cervical hip and distal forearm fractures occurring in the Rochester population. Fractures were ascertained using the data resources of the Rochester Epidemiology Project, which incorporates inpatient and outpatient data from the local providers of medical care. (32) Fractures from severe trauma or pathological fractures were not included. Incidence rates were estimated for adults assuming the entire Rochester population age 35 years and over to be at risk. Age- and sex-specific incidence rates are shown here for FN (cervical) fractures in 1985–1992 updated through 2001 (excluding intertrochanteric hip fractures) and fractures of the distal forearm in 1985–1994. (33, 34)

Statistical analysis

The relationships between variables relating to bone strength indices, geometric properties, and fall force with age and sex were studied using Pearson correlation and least squares regression, where the natural spline of age was included in the regression model. Each model was compared with a linear relationship, and the simplest model was used for analysis. Changes in variables between 20 and 90 years of age were based on predicted values from these models. Differences in changes over life between men and women were tested using an age-sex interaction term in a regression model. An adjustment for the effect of differences in bone size on geometric properties and bone strength indices was made by dividing the original values by height; the need for this adjustment was determined by analyzing the correlation of the variables with height among subjects 20–29 years of age. The Student's t-test was used to compare the mean values for young adult males versus females. A lowess smoother, (35) essentially a type of moving average, was used to explore the data in Figs. 1, 2, 3, and 4.

thumbnail image

Figure FIG. 1.. Composite of bone variables that affect risk of fracture at the ultradistal radius plotted against age. The panels consist of (A) vBMD (B) cross-sectional area/Ht, (C) moment-of-inertia/Ht, (D) compressive rigidity (EA/Ht), (E) flexural rigidity (EIp/Ht), and (F) applied loads in falls/Ht. Changes with age are given by natural splines shown by solid line in men and by broken line in women. All variables are normalized to the mean value for young women, 20–29 years of age, which is given the arbitrary value of 1.0.20

Download figure to PowerPoint

thumbnail image

Figure FIG. 2.. Fall load (FL)/bone strength ratios at ultradistal radius vs. age, normalized to mean values in females, 20–29 years of age, which is given an arbitrary value of 1.0. (Left) FL/EA. (Right) FL/EIp. Changes with age are given by natural splines shown by solid line in men and by broken line in women.20

Download figure to PowerPoint

thumbnail image

Figure FIG. 3.. Composite of bone variables that affect risk of fracture at the femoral neck plotted against age. The panels consist of (A) vBMD, (B) cross-sectional area/Ht, (C) moment-of-inertia/Ht, (D) compressive rigidity (EA/Ht), (E) flexural rigidity (EIap/Ht), and (F) applied loads in falls/Ht. Changes with age are given by natural splines shown by solid line in men and by broken line in women. All variables are normalized to the mean value for young women, 20–29 years of age, which is given the arbitrary value of 1.0.20

Download figure to PowerPoint

thumbnail image

Figure FIG. 4.. Fall load (FL)/bone strength ratios at femoral neck vs. age, normalized to mean values in females, 20–29 years of age, which is given an arbitrary value of 1.0. (Left) FL/EA. (Right) FL/EIap. Changes with age are given by natural splines shown by solid line in men and by broken line in women.20

Download figure to PowerPoint

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

Skeletal variables at ultradistal radius

Table 1 provides both unadjusted and height-adjusted variables relating to vBMD, geometric properties, bone strength indices, fall-related loads, and fall load/bone strength ratios at the UDR. It also provides the mean and SD for the absolute values for variables in young adult women and men (20-29 years of age) and proportional change in these variables over life. Figure 1 shows age changes in women and men at the UDR scanning site for vBMD, CSA/Ht, Ip/Ht, EA/Ht, EIp/Ht, and fall loads, normalized to the mean for women, ages 20–29, which was arbitrarily set at 1.0. Figure 2 shows analogous changes for fall load/bone strength ratios at the UDR. In young adults, values for total vBMD were similar in men and women at all ages, although initial values were 10% higher in men. There were age-related decreases in absolute values for total vBMD of about 30% in both sexes. However, geometric properties, including CSA/Ht and I/Ht, were much higher in young men than in young women. As a result, bone strength indices were also substantially higher in young men (EA/Ht was 41% and EIp/Ht was 97% higher, respectively). Over life, CSA/Ht increased minimally and more in men than in women, Ip/Ht increased 2-fold more in men than women, and SM/Ht decreased slightly in women but was unchanged in men. With aging, bone strength indices decreased in both sexes but decreased more in women than in men. Estimated loads applied to the wrist during a fall were only 5% higher in men, despite their greater body size, and these values remained relatively constant over life. In young adults, FL/EA was 32% lower (better) and FL/EIp was 51% lower (better) in men versus women. Over life, the load/strength ratios worsened in both sexes, but worsened more in women (FL/EA by 33% and FL/EIp by 22%) than in men (FL/EA by 21% and FL/EIp by 4%).

Table Table 1.. Variables Related to Bone Strength and Fracture Risk at the Ultradistal Radius, Adjusted and Unadjusted for Height
Thumbnail image of

Skeletal variables at FN

Table 2 and Figs. 3 and 4 provide analogous information for variables at the FN. In young adults, total vBMD was slightly higher in women than in men but decreased with aging somewhat more in women (−46%) than in men (−34%). However, young adult values for geometric properties were much higher in men (33-83%) than in women. With aging, CSA/Ht increased by a similar extent (by 14–21%) in both sexes, whereas Iap/Ht and SM/Ht were unchanged in women but declined somewhat in men. Bone strength indices were also much higher in young adulthood in men than in women (EA/Ht was 14% higher and EIap/Ht was 75% higher, respectively) and both declined moderately with aging, somewhat more in women than in men. The predicted load on the hip caused by a sideways fall was 40% higher in men than women as a consequence of their greater height and body mass and decreased only slightly in both sexes with aging. Load/strength ratios for axial and flexural rigidities in young adults were not clearly better in men as they were at the UDR: men had 11% higher (worse) FL/EA than women but a 26% lower (better) FL/EIap than women. The load/strength ratios increased (worsened) over life in both sexes, but increases were 2-fold greater in women (FL/EA by 62% and FL/EIap by 40%) than in men (FL/EA by 36% and FL/EIap by 22%).

Table Table 2.. Variables Related to Bone Strength and Fracture Risk at the Femoral Neck, Adjusted and Unadjusted for Height
Thumbnail image of

Community fracture patterns

The age- and sex-specific pattern for incidence of fragility fractures of the wrist (distal forearm or Colles' fracture) and FN are shown in Fig. 5. The pattern of these two types of fragility fractures differed strikingly from each other. In adult women, the incidence of wrist fractures began to increase over basal levels at about age 45, rose sharply by 6-fold until about age 60 years, and increased only gradually over the remainder of life. In contrast to findings in women, the incidence of wrist fractures in men did not increase with age and remained at low basal levels throughout life. Consequently, older women had an excess of wrist fractures over men of about 6-fold. Hip fractures increased as a single exponential function in both sexes, although, at any given age, the incidence was 2-fold greater in women than in men. The increase in hip fracture incidence began to rise rapidly in both sexes around age 70 and increased by >8-fold over the remainder of life in both sexes. Thus, wrist fractures in women and hip fractures in both sexes exhibit a steep increase in incidence with age, but for hip fractures, this occurs ∼25 years later than for wrist fractures.

thumbnail image

Figure FIG. 5.. Age- and sex-specific patterns of wrist and cervical hip fracture incidence in Rochester, MN, men and women. Fractures caused by severe trauma and pathological fractures have been excluded. These data have been updated from a previous publication. (20)20

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

Although our primary objective was to relate fall load/bone strength ratios to patterns of wrist and hip fractures, we also addressed two other unresolved questions. First, what is the net effect of the various age-related changes in BMD, geometry, and structure on estimates of bone strength? We showed previously(7) that aging results in both unfavorable changes, such as decreases in trabecular and cortical vBMD and CSA/Ht that weaken the skeleton, and in favorable changes, such as increased CSA and outward displacement of the cortex that strengthen it. In this study, we found that, although the bone strength indices, EA/Ht and EI/Ht, decrease with aging, the proportional decrease was less than for total vBMD. Thus, the adverse effect of age-related bone loss on bone strength is attenuated by age-related compensatory changes in bone structure.

Second, does their larger skeletal size endow men with a greater resistance to loads that produce fractures compared with women? Because of sex differences in pubertal growth patterns, men reach adulthood with larger skeletons than women do, (36) and studies on bone breaking strength in vitro have clearly established that larger bones are stronger, even when adjusted for differences in BMD. (13) However, men's larger skeletons also will increase the force applied to bone in falls. In this study, we find that when the greater bone strength indices associated with larger bone size were adjusted for the corresponding increase in fall load, the larger skeletal size remained an advantage at the UDR, whereas this advantage was largely lost at the FN.

Our fall load/bone strength analyses help to explain some of the striking differences between the age- and sex-specific patterns of wrist (Colles') and hip fractures and are particularly helpful in understanding the highly idiosyncratic pattern of wrist fractures. Differences between sexes in vBMD at the UDR cannot explain the large predominance of women and the virtual immunity of men for age-related wrist fractures because the vBMD values in young adults and the moderate decline in these values with aging are similar between sexes. Nor can the high incidence of wrist fractures in women over men be explained by greater trauma. Although there are few data on the sex-specific incidence of falls, postmenopausal women have been reported to have, at most, only 30% more falls than men of the same age. (37) This represents but a small fraction of the 600% higher incidence of wrist fractures in women versus men. However, because of their larger bone size and more favorable geometry, young adult men have fall load/bone strength ratios at UDR that predict a much lower risk of fracture than in young adult women. Moreover, these ratios (and, thus, the estimated fracture risk) change little with aging in men, whereas they worsen markedly with aging in women. In fact, values for the load/strength ratios in elderly men are lower (better) than values in young adult women.

Neither differences between sexes in midlife values for vBMD nor the more favorable geometric properties of bone in young adult men explain the sharp midlife rise in wrist fracture incidence in women. However, menopausal-induced increases in bone turnover lead to perforative resorption of trabecular plates and loss of trabeculae. (38) This disruption of trabecular microarchitecture weakens the skeleton out of proportion to the decrease in BMD and could be a major contributor to the midlife increase in wrist fracture incidence. (39) After menopause, bone turnover increases rapidly to a level that is 80% above the premenopausal mean but changes little thereafter. (40, 41) Interestingly, in women, this pattern of change in bone turnover is very similar to their incidence pattern for wrist fracture, with rapid increases in the early postmenopausal period followed by a relatively sustained plateau. Until recently, quantification of changes in trabecular microarchitecture required iliac biopsy, but high-resolution imaging technology has been recently developed for noninvasive assessment at peripheral sites, including the site of Colles' fracture. (42–45) Availability of such data will likely improve estimates of the failure load and fracture risk at the distal radius. (15)

We found that fall load/bone strength analyses were also helpful in understanding mechanisms of the observed pattern of hip fractures. The female predominance can be only partially explained by sex differences in total vBMD. In fact, vBMD in young adulthood was slightly higher in women than in men, and decreases with age were only slightly more in women. In contrast, the fall load/bone strength ratios track more closely with the sex difference in incidence of hip fractures. Although young adult men had much more favorable load/strength ratios at UDR than young adult women did, sex differences in these ratios were not as pronounced at the FN. However, the changes in the load/strength ratios with aging were 2-fold higher (worse) in women than in men, which is consistent with their 2-fold higher incidence of hip fractures.

Another major characteristic of hip fractures is the large increase in incidence after age 75 in both sexes. The moderate worsening in load/strength ratios in the elderly of either sex, however, can explain only a small part of the increased incidence of hip fractures with age. This suggests that there are additional causal factors that we did not assess. We speculate that the most important of these is the well-documented increased incidence of falls, augmented by impaired protective neuromuscular reflexes in the elderly. (46, 47) When bone strength decreases with aging to below a critical level, each fall carries a statistical risk of fracture, and the more frequent the falls, the greater the cumulative risk. In addition, a variable, but significant, proportion of elderly subjects may develop vitamin D deficiency. In some instances, this may induce secondary hyperparathyroidism(48) or histologic osteomalacia, (48, 49) either of which would lead to impaired bone strength and hip fractures. Finally, because the very old of both sexes have a more shuffling gait, they tend to fall sideways or backward, which predisposes them to hip fracture; in contrast, younger postmenopausal women have a more brisk gait and tend to fall forward, which predisposes them to wrist fracture. (10, 50) These age-specific gait differences may also contribute to the predilection for women to sustain wrist fracture in middle life and to sustain hip fractures in late life.

Our study has several limitations. First, because it is cross-sectional, the data may have been affected by secular increases in bone size that occurred during the nearly 80-year age span of our population cohort. However, we attempted to offset this by adjusting variables for height when appropriate. Second, although the bone strength indices that we used, axial and flexural rigidities, predict the failure of engineering structures, empiric relationships between them and whole bone strength at the hip and wrist have not been reported. This limited our ability to compare bone strength directly in terms of a predicted failure load (in Newtons) to the estimated loads applied to the bone (in Newtons) and to derive an absolute, rather than a relative, factor-of-risk for fracture. (8) Third, our estimates of loads applied to the wrist and hip were projected for only a single fall configuration for each fracture. Clearly a host of additional factors influence the load applied to bone during actual fall events, including the specific characteristics of the fall, the thickness of soft tissue overlying the hip, protective responses that slow the fall, and the impact surface itself.

In conclusion, we found that fall load/bone strength ratios more closely simulate the age- and sex-specific incidence patterns of wrist and hip fractures than do measurements of vBMD made in the same study or measurements of aBMD made in previous studies. (6) This was particularly true for the differences in fracture incidence between sexes. Prospective observational or nested case-control studies should now be made to define the efficacy of this approach in fracture prediction, and these results should be compared with those obtained with the more complex and expensive method of finite element analysis. (16) Relating fall loads to bone strength indices will likely play an increasingly important role in the clinical assessment of risk for fragility fractures in the near future.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References

The authors thank Margaret Holets for making the pQCT measurements, Lisa McDaniel and Louise McCready for assistance in recruitment and management of the study subjects, James M Peterson for assistance with data management and file storage, and Sara J Achenbach for statistical assistance. Ronald A Karwoski and Mahlon C Stacy of the Biomedical Imaging Center provided valuable assistance in analysis of the spiral QCT scans. This work was supported in part by NIH Grants AR-027065 and RR00585.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. References