SEARCH

SEARCH BY CITATION

Keywords:

  • osteoporosis;
  • fractures;
  • absolute risk;
  • FRAX;
  • dual energy X-ray absorptiometry

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The World Health Organization (WHO) fracture risk assessment system (FRAX) allows for calibration from country-specific fracture data. The objective of this study was to evaluate the method for imputation of osteoporotic fracture rates from hip fractures alone. A total of 38,784 women aged 47.5 years or older at the time of baseline femoral neck bone mineral density (BMD) measurement were identified in a database containing all clinical dual energy X-ray absorptiometry (DXA) results for the Province of Manitoba, Canada. Health service records were assessed for the presence of nontrauma osteoporotic fracture codes after BMD testing (431 hip, 787 forearm, 336 clinical vertebral, and 431 humerus fractures). Ten-year hip and osteoporotic fracture rates were estimated by the Kaplan-Meier method. The population was stratified by age (50 to 90 years, 5-year width strata) and again by femoral neck T-scores (−4.0 to 0.0, 0.5 SD width strata). Within each stratum, the ratio of hip to osteoporotic fractures was calculated and compared with the predicted ratio from FRAX. Increasing age was associated with greater predicted hip-to-osteoporotic ratios (youngest 0.07 versua oldest 0.41) and observed ratios (youngest 0.10 versus oldest 0.48). Lower T-scores were associated with greater predicted (highest 0.04 versus lowest 0.71) and observed ratios (highest 0.06 versus lowest 0.44). There was a strong positive correlation between predicted and observed ratios (Spearman r = 0.90–0.97, p < .001). For 14 of the 18 strata, the predicted ratio was within the observed 95% confidence interval (CI). Since collection of population-based hip fracture data is considerably easier than collection of non–hip fracture data, this study supports the current emphasis on using hip fractures as the preferred site for FRAX model calibration. © 2010 American Society for Bone and Mineral Research


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Worldwide, approximately 9 million new osteoporotic fractures occur each year,1 with the global burden of osteoporosis projected to increase markedly over the next few decades as the number of elderly individuals increases.2 The ability to accurately determine fracture risk is critical in identifying cost-effective thresholds for intervention.3, 4

Recently, there has been a shift from risk assessment based on T-score categories to absolute fracture risk based on 10-year absolute fracture risk.4 The World Health Organization (WHO) created a fracture risk assessment tool (FRAX) that can be used to derive individualized estimates of 10-year fracture risk.5 Since osteoporotic fracture rates can vary greatly between countries, the FRAX system should be calibrated to the target population.6 Creation of a country-specific FRAX model minimally requires age- and gender-specific annual hip fracture and all-cause mortality data. These data are used to estimate 10-year risk of hip fracture based on age, gender, femoral neck bone density, and clinical risk factors. The gradients of risk for femoral neck bone density to predict fracture and for clinical risk factors to predict fracture have been derived from pooling multiple large cohort studies and are assumed to apply equally to all countries. The FRAX tool also reports 10-year risk of osteoporotic fracture where this represents a composite of hip, clinical spine, forearm, and proximal humerus fractures. Although calibration of osteoporotic fracture rates ideally is achieved by collecting country-specific incidence data stratified by sex and age as for hip fractures, it is well recognized that it is much more difficult to obtain accurate data on non–hip fractures because these usually do not result in hospitalization or surgery. Therefore, the FRAX tool can use imputed non–hip fracture rates based on the ratio of hip to non–hip fractures observed in the Swedish population from Malmo, Sweden.7, 8 These predictions were derived from 1 to 2 years of X-ray reports and did not directly assess 10-year fracture outcomes.

Examination of the Web-based FRAX tool (www.shef.ac.uk/FRAX) indicates that the ratio of predicted 19-year risk of hip to all osteoporotic fractures is very sensitive to age and bone density. This reflects the rarity of hip fractures prior to age 60. The ratio increases exponentially thereafter, whereas other osteoporotic fractures (particularly the wrist) show an age-related increase that precedes hip fractures by many years.9, 10 Conversely, there is a much steeper gradient of risk for femoral neck bone density to predict hip fractures than to predict non–hip osteoporotic fractures.11 The FRAX risk estimates reflect these relationships, with an increase in the hip-to-osteoporotic-fracture ratio as a function of older age and as a function of lower femoral neck T-score. The proportion of projected osteoporotic fractures that affect the hip can vary from as low as 3% (age 50 with the femoral neck T-score 0.0) to as high as 76% (age 50 with the femoral neck T-score −4.0).

The validity of the methodology used to impute non–hip osteoporotic fracture rates and derive estimates of 10-year osteoporotic fracture risk based on the preceding assumptions was assessed in the Manitoba Bone Mineral Density Program database. Using administrative data, hip and osteoporotic fracture rates were assessed up to 10 years following bone density measurement. The observed proportion of hip fractures relative to all osteoporotic fractures then was compared with the FRAX projections.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Study population

A historical cohort was drawn from the Manitoba Bone Density Program and its regionally based clinical database, both of which have been described in detail previously.12 The population comprised women over age 47.5 years who underwent baseline femoral neck bone mineral density (BMD) testing between January 1990 and March 2007 using one of the program's primary dual energy X-ray absorptiometry (DXA) instruments (DPX or Prodigy, GE Lunar, Madison WI, USA). Age was calculated from the date of DXA testing using information on date of birth obtained from the population registry. For women with more than one eligible femoral neck measurement, only the first record for that individual was included. Men and younger women were not included in the study because of small numbers and greater potential for referral bias. The study was approved by the Research Ethics Board for the University of Manitoba and the Health Information Privacy Committee of Manitoba Health.

Bone density measurements

All clinical bone densitometry in the Province of Manitoba, Canada, is performed within a single program structure that maintains uniform testing indications, requisition process, and reporting. The program's database includes all DXA test results since the first instrument was installed in 1990, with a database that is over 99% complete.13 Canada has a publicly funded health care system, and BMD testing, along with other essential health services, are available to all Manitoba residents without charge. Prior to 2000, DXA measurements were performed with a pencil-beam instrument (Lunar DPX, Lunar Corp., Madison WI, USA), and after that date, a fan-beam instrument was used (Lunar Prodigy, GE Healthcare). Instruments were cross-calibrated in vivo with 59 volunteers, and no clinically significant differences were identified (femoral neck T-score differences < 0.1). Therefore, all analyses are based on the unadjusted numerical results provided by the instrument. Femoral neck T-scores (number of SDs above or below young-adult mean BMD) and Z-scores (number of SDs above or below age-matched mean BMD) were based on the Third National Health and Nutrition Examination Survey (NHANES III) reference data (Prodigy Version 8.8). All equipment and technologist performance are subject to a program-wide quality assurance program. Densitometers underwent daily assessment with an anthropomorphic spine phantom and showed stable long-term performance [coefficient of variation (CV) = standard deviation/mean < 0.5%] with in vivo precision for femoral neck BMD of 1.9% to 2.4%.14

Fracture outcomes

Manitoba Health maintains computerized databases of physician billing claims and hospital separations for all residents of the province eligible to receive health services. Each health system contact includes information on a patient's demographics, date of service, type of service and/or procedure, and diagnoses coded using the International Classification of Disease, Vol. 9: Clinical Modification (ICD-9-CM). After April 1, 2004, hospitals reported under the ICD-10-CA system, and therefore, fracture diagnosis codes were converted into the equivalent ICD-9-CM codes. Anonymous linkage of these databases to the BMD database was possible via a unique scrambled health identification number, thereby allowing for the creation of a longitudinal record of health services.

Each subject's longitudinal health service record was assessed from the date of bone density measurement to March 31, 2007 for the presence of noncraniofacial ICD-9-CM fracture codes using previously described definitions.15 Fractures were classified as incident if they occurred after the bone density test. Specific fracture sites of interest were the hip (ICD-9-CM 820, 821), spine (ICD-9-CM 805), forearm (ICD-9-CM 813), and humerus (ICD-9-CM 812) because they are the basis for the 10-year absolute fracture risk estimates published by Kanis and colleagues.8 We excluded fractures associated with ICD-9-CM trauma codes (ICD-9-CM E800–E879 and E890–E999). In addition, we required that hip fractures and forearm fractures be accompanied by a site-specific fracture reduction, fixation, or casting code in order to enhance the diagnostic and temporal specificity for an acute fracture. Hip, spine, forearm, and proximal humerus fractures defined in this way were collectively designated as “osteoporotic” fractures. Our use of administrative health data to define fractures in this way shows that BMD measurements predict fractures in our clinical cohort as well as has been reported in large meta-analyses.15

Statistical analysis

All statistical analyses were performed with Statistica (Version 6.1, StatSoft, Inc., Tulsa, OK, USA). A p value of less than 0.05 was taken to indicate a statistically significant result. We stratified our cohort into subgroups using 5-year age strata (±2.5) from 50 to 90 years. We then restratified the cohort into subgroups using 0.5 SD T-score strata (±0.25) from −4.0 to 0.0 (or greater). Ten-year hip and osteoporotic fracture rates were estimated for each age and T-score subgroup using a direct method (up to 10 years of continuous observation) from Kaplan-Meier curves. Observations were censored for relocation out of the province but not for death, in accordance with the WHO fracture risk system.4 The ratio of observed 10-year hip fracture risk to observed 10-year osteoporotic fracture risk was calculated for each subgroup. Ninety-five percent CIs were generated using Monte Carlo simulation (1000 replications). Predicted ratios for hip to osteoporotic fractures were obtained as the ratio of 10-year hip fracture risk and 10-year osteoporotic fracture risk from the FRAX Web site (www.shef.ac.uk/FRAX) using the mean age and T-score for that stratum (Swedish FRAX tool, height 165 cm, weight 70 kg, no clinical risk factors). The correlation between predicted and observed ratios was assessed using the Spearman rank-order correlation coefficient.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Study population

The final study population consisted of 38,784 women with a mean age of 64.9 years (SD 10.2). The baseline characteristics of the participants are summarized in Table 1. Mean femoral neck T-score was −1.46 (SD 0.98), which is in the osteopenic range. Mean Z-score was 0.07 (SD 1.02), indicating close agreement with manufacturer reference data. During 175,680 person-years of follow-up (mean 4.5 years), 2182 (5.6%) of the women died, and 1125 (2.9%) were lost owing to relocation. Incident osteoporotic fracture codes occurring after the BMD measurement were identified in 1840 (4.7%) women. This included 431 hip fractures, 787 forearm fractures, 336 clinical vertebral fractures, and 431 humerus fractures (some women had more than one fracture type). Overall, 23.4% of the women with osteoporotic fractures sustained a hip fracture.

Table 1. Characteristics of the Study Population (n = 38,784)
  1. Data are mean ± SD unless otherwise specified.

Age, years64.9 ± 10.2
White ethnicity, n (%)38,139 (98.5)
Femur neck BMD 
 T-score−1.46 ± 0.98
 Z-score+0.07 ± 1.02
Osteoporotic (T-score ≤ −2.5), n (%)5340 (13.8)
Final status: 
 Alive, n (%)35,477 (91.5%)
 Dead, n (%)2182 (5.6%)
 Lost (moved), n (%)1125 (2.9%)
Years of follow-up, total (mean, range)175,680 (4.5, 0.1–10.0)

Table 2 shows that mean age within each 5-year age subgroup was very close to the age that was being approximated (within 0.4 years). As mean age for the subgroup increased, there was a progressive decrease in mean femoral neck T-score (from −0.91 at age 50 to −2.49 at age 90). For strata defined by T-score, mean values within each 0.5 SD subgroup were within 0.06 SD of the nominal value. As femoral neck T-score for the stratum decreased, there was a progressive increase in mean age (from 59.2 years from T-score 0.0 and above to 75.5 years for T-score −4.0).

Table 2. Baseline Variables, Number of Incident Hip and Osteoporotic Fractures, and Hip-to-Osteporotic-Fracture Ratios According to Age Stratum and Femoral Neck T-Score Stratum
Age stratumNMean ageMean T-scoreNumber (%) with hip fracturesNumber (%) with any osteoporotic fracturesObserved ratio of hip to osteoporotic fracturesPredicted ratio of hip to osteoporotic fracturesa
50472750.3−0.916 (0.1)116 (1.8)0.05 (0.02–0.11)0.07
55655155.0−1.0816 (0.3)192 (3.3)0.08 (0.05–0.13)0.11
60624459.9−1.2627 (0.6)219 (4.6)0.12 (0.09–0.18)0.14
65582065.0−1.4442 (1.3)246 (7.8)0.17 (0.13–0.22)0.17
70556069.9−1.6567 (1.4)303 (6.3)0.22 (0.18–0.27)0.25
75483174.9−1.85111 (1.8)327 (5.2)0.34 (0.29–0.39)0.34
80313579.8−2.0185 (1.5)267 (4.8)0.32 (0.27–0.38)0.39
85147384.6−2.2754 (3.7)124 (8.4)0.44 (0.36–0.53)0.40
9044389.9−2.4923 (5.2)46 (10.4)0.50 (0.37–0.67)0.41
T-score stratumNMean ageMean T-scoreNumber (%) with hip fracturesNumber (%) with any osteoporotic fracturesObserved ratio of hip to osteoporotic fracturesPredicted ratio of hip to osteoporotic fracturesa
  • a

    From FRAX Web site (Swedish model) assuming female, height 165 cm, weight 70 kg, and no additional clinical risk factors.

−4.019975.5−4.0320 (10.1)38 (19.1)0.53 (0.39–0.71)0.71
−3.567775.6−3.4456 (8.3)127 (18.8)0.44 (0.36–0.54)0.58
−3.0230072.8−2.96106 (4.6)256 (11.1)0.41 (0.36–0.48)0.50
−2.5493870.1−2.4895 (1.9)378 (7.7)0.25 (0.21–0.30)0.36
−2.0743466.9−1.9984 (1.1)411 (5.5)0.20 (0.17–0.25)0.25
−1.5795264.1−1.5048 (0.6)309 (3.9)0.16 (0.12–0.20)0.17
−1.0655662.1−1.0112 (0.2)172 (2.6)0.07 (0.04–0.12)0.11
−0.5442860.4−0.527 (0.2)93 (2.1)0.08 (0.04–0.15)0.06
0.0 or greater430059.20.333 (0.1)56 (1.3)0.05 (0.02–0.16)0.04

Fractures according to age

Table 2 shows that older age was associated with a stepwise increase in the proportion of the subgroup with incident hip fractures (from 0.1% to 5.2%) and osteoporotic fractures (from 1.8% to 10.4%). In the youngest age stratum, the predicted ratio to osteoporotic fractures was 0.07, and the observed ratio based on numbers of patients with fractures was 0.05 (95% CI 0.02–0.11). In the oldest stratum, the predicted ratio was 0.41 and the observed ratio was 0.50 (95% CI 0.37–0.67). All the predicted values were contained within the observed 95% CIs.

Similar results were obtained when the observed ratio for 10-year risk of hip fracture to osteoporotic fracture was assessed (Fig. 1). Once again, the observed and predicted ratios increased with older age. Of the nine age strata, the predicted value was contained within the observed 95% CI in eight cases. For the age 85 stratum, the observed ratio (0.63, 95% CI 0.45–0.81) exceeded the predicted ratio (0.40). At age 50, the predicted ratio was 0.07 and the observed ratio was 0.10 (95% CI 0–0.21). At age 90, the predicted ratio was 0.41 and the observed ratio was 0.48 (95% CI 0.27–0.69). The correlation between predicted and observed hip-to-osteoporotic-fracture ratios was high (Spearman r = 0.97, p < .001).

thumbnail image

Figure 1. Ratio of hip to osteoporotic fractures by age category (5-year width) and femoral neck T-score category (0.5 SD width) as predicted from the FRAX Web site and observed clinically. Error bars are 95% CIs.

Download figure to PowerPoint

Fractures according to femoral neck bone density

Table 2 shows that lower femoral neck T-scores were associated with a stepwise increase in the proportion of the subgroup with incident hip fractures (from 0.1% to 10.1%) and osteoporotic fractures (from 1.3% in the stratum with lowest risk to 19.2%). At a femoral neck T-score of 0.0 or greater, the predicted ratio of hip to osteoporotic fractures was 0.04. Based on the numbers of women with fractures, the observed ratio was 0.07 (95% CI 0.02–0.016). In the lowest bone density stratum with T-score −4.0, the predicted ratio was 0.71 and the observed ratio was 0.53 (95% CI 0.39–0.71). In the nine strata, the predicted value was contained within the observed 95% CI in six cases, and in the remaining three cases, the predicted ratio exceeded the observed ratio.

Figure 1 indicates the ratios of hip fractures to osteoporotic fractures based on predicted and observed 10-year fracture risk. Observed ratios increased with lower T-scores down to −3.0 and then appeared to plateau. Although there was general agreement between the predicted and observed ratios, among three of the nine strata, the observed ratio was significantly less than the predicted ratio. A strong correlation again was seen between predicted and observed hip-to-osteoporotic-fracture ratios (Spearman r = 0.90, p < .001).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We directly assessed the proportion of osteoporotic fractures that affect the hip in a large clinical cohort of women referred for BMD testing with DXA. By comparing the ratio of hip to all osteoporotic fractures within strata defined by age and femoral neck T-score, it is possible to see how these factors affect the observed ratios. As expected, older age and lower T-scores were associated with a greater proportion of osteoporotic fractures that involve the hip. More important, by comparing observed ratios with those predicted under the WHO fracture risk assessment tool (FRAX), it is possible to directly evaluate the procedure used by the WHO for imputation of non–hip fractures from hip fractures for purposes of model calibration. The level of agreement generally is considered to be quite good across a broad range of ages and T-scores. The only significant discrepancy related to a lower than predicted proportion of hip fractures in women with extremely low femoral neck T-scores (−3.5 to −4.0). In these women, hip fractures were predicted to comprise the majority of the osteoporotic fractures, whereas the observed values were slightly less than half. Therefore, the WHO system would slightly underestimate the 10-year risk for osteoporotic fractures. This is unlikely to be a significant problem in clinical practice, however, because women with such low femoral neck T-scores would be considered high risk for fracture under any system.

The procedure for calibrating a FRAX tool involves collection of population-specific mortality and fracture incidence data stratified by sex and age. Under ideal circumstances, detailed fracture information would be collected for all sites considered in this system, which includes clinical spine, distal forearm, proximal humerus, and hip. Correction of hip fracture data is relatively easy because these fractures result in hospitalization and a surgical procedure. They are therefore amenable to identification from a variety of administrative data sources on large populations. The same is not true for the other fracture sites, which usually do not result in hospitalization or surgery. Although non–hip fracture data can be identified in cohort studies, these may not be representative of the overall population and may be underpowered for detailed analysis of fracture patterns. The method developed by the WHO for imputation of non–hip fracture rates was based on detailed review of health care interactions and X-ray reports for a single population in Malmo, Sweden.7, 8 Generalizability of these ratios to other population makes it possible to calibrate a FRAX tool based on hip fracture data alone.

Strengths to our study include the large size of the cohort. Although based on a clinical population, it accurately depicts how these measurements perform in routine practice because our database captures virtually all bone densitometry for the Province of Manitoba. The provincial data repository also allows for highly complete tracking of reported fractures and is not dependent on patient recall. Some limitations are also acknowledged. Predicted hip-to-osteoporotic-fracture ratios assumed no additional clinical risk factors but would be slightly different in the presence of additional clinical risk factors because these show slightly different effects on hip fractures and non–hip fractures. Fracture ascertainment from administrative data for is less reliable than direct radiographic review, particularly for vertebral fractures, because the majority are not diagnosed clinically.16 Nonetheless, hip fractures are reported accurately, and our previous analyses have shown good detection of non–hip fractures (including clinical vertebral fractures).15 Finally, our study population did not include men or a significant proportion of nonwhite women; therefore, results may not be applicable to these populations.

In summary, the WHO method for imputation of 10-year osteoporotic fracture rates from hip fractures alone is broadly consistent with clinical observational data. Since collection of population-based hip fracture data is considerably easier than collection of non–hip fracture data, this supports the current emphasis on using hip fractures as the preferred site for FRAX model calibration, although wherever available non–hip fracture data should be analyzed to confirm the imputation and calibration for osteoporotic fracture rates.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

WDL has received honoraria for lectures from Merck Frosst Canada; research support from Merck Frosst Canada; and unrestricted educational and research grants from the Alliance for Better Bone Health, Sanofi-Aventis and Procter & Gamble Pharmaceuticals Canada, Inc., Novartis Pharmaceuticals Canada, Inc., Amgen Pharmaceuticals, Inc., and Genzyme Canada, Ltd. LML has received unrestricted research grants from Amgen Pharmaceuticals, Inc. The results and conclusions of this study are those of the authors, and no official endorsement by Manitoba Health is intended or should be inferred.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We are indebted to Manitoba Health for providing data (HIPC File No. 2007/2008-35). This article has been reviewed and approved by the members of the Manitoba Bone Density Program Committee.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  • 1
    Johnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with osteoporotic fractures. Osteoporos Int. 2006; 17: 17261733.
  • 2
    Melton LJ III. Epidemiology worldwide. Endocrinol Metab Clin North Am. 2003; 32: 113.
  • 3
    Kanis JA, Oden A, Johnell O, Jonsson B, De Laet C, Dawson A. The burden of osteoporotic fractures: a method for setting intervention thresholds. Osteoporos Int. 2001; 12: 417427.
  • 4
    Kanis JA, Borgstrom F, De Laet C, et al. Assessment of fracture risk. Osteoporos Int. 2005; 16: 581589.
  • 5
    Kanis JA, Johnell O, Oden A, Johansson H, McCloskey E. FRAX and the assessment of fracture probability in men and women from the UK. Osteoporos Int. 2008; 19: 385397.
  • 6
    Kanis JA, Johnell O, De Laet C, Jonsson B, Oden A, Ogelsby AK. International variations in hip fracture probabilities: implications for risk assessment. J Bone Miner Res. 2002; 17: 12371244.
  • 7
    Kanis JA, Johnell O, Oden A, et al. Long-term risk of osteoporotic fracture in Malmo. Osteoporos Int. 2000; 11: 669674.
  • 8
    Kanis JA, Johnell O, Oden A, Dawson A, De Laet C, Jonsson B. Ten year probabilities of osteoporotic fractures according to BMD and diagnostic thresholds. Osteoporos Int. 2001; 12: 989995.
  • 9
    Singer BR, McLauchlan GJ, Robinson CM, Christie J. Epidemiology of fractures in 15,000 adults: the influence of age and gender. J Bone Joint Surg Br. 1998; 80: 243248.
  • 10
    van Staa TP, Dennison EM, Leufkens HG, Cooper C. Epidemiology of fractures in England and Wales. Bone. 2001; 29: 517522.
  • 11
    Johnell O, Kanis JA, Oden A, et al. Predictive value of BMD for hip and other fractures. J Bone Miner Res. 2005; 20: 11851194.
  • 12
    Leslie WD, Metge C. Establishing a regional bone density program: lessons from the Manitoba experience. J Clin Densitom. 2003; 6: 275282.
  • 13
    Leslie WD, Caetano PA, MacWilliam LR, Finlayson GS. Construction and validation of a population-based bone densitometry database. J Clin Densitom. 2005; 8: 2530.
  • 14
    Leslie WD, Ward LM. Bone density monitoring with the total hip site: time for a re-evaluation? J Clin Densitom. 2004; 7: 269274.
  • 15
    Leslie WD, Tsang JF, Caetano PA, Lix LM. Effectiveness of bone density measurement for predicting osteoporotic fractures in clinical practice. J Clin Endocrinol Metab. 2007; 92: 7781.
  • 16
    Nevitt MC, Ettinger B, Black DM, et al. The association of radiographically detected vertebral fractures with back pain and function: a prospective study. Ann Intern Med. 1998; 128: 793800.