Elsevier

Journal of Clinical Densitometry

Volume 16, Issue 4, October–December 2013, Pages 472-481
Journal of Clinical Densitometry

2013 Position Development Conference on Bone Densitometry
2013 International Society for Clinical Densitometry Position Development Conference: Task Force on Normative Databases

https://doi.org/10.1016/j.jocd.2013.08.001Get rights and content

Abstract

Following the standard protocol for development of Official Positions for the International Society for Clinical Densitometry, the Expert Panel heard the report and recommendations from the Task Force on Normative Databases; using the RAND methodology, agreement was reached on the following statements:

1. Manufacturers should continue to use their own databases for the lumbar spine as the reference standard for T-scores.

2. Manufacturers should continue to use National Health and Nutrition Examination Survey III data as the reference standard for femoral neck and total hip T-scores.

3. If local reference data are available, they should be used to calculate only Z-scores but not T-scores.

4. A uniform Caucasian (non-race adjusted) female reference database should be used to calculate T-scores for men of all ethnic groups.

Introduction

Bone mineral density (BMD) testing was developed in the 1970s, initially with neutron activation analysis (whole body) 1, 2, 3, followed by single-photon absorptiometry (forearm) 4, 5. Finally, dual-photon absorptiometry 6, 7, 8—subsequently dual-energy X-ray absorptiometry (DXA) (9)—allowed for specific measurements of central sites (spine and hip) that are frequently fractured in patients with osteoporosis. Over time, the source of radiation has changed (from gadolinium to X-ray photon generation) and image acquisition has evolved (from pencil beam to fan beam). Over the past 3 decades, physical variables have also changed (people are taller and heavier, less likely to smoke, with different levels of physical activity, and different intakes of calcium and caffeine) (10).

Initially, there was some uncertainty as to how to express results from DXA. The main output of DXA is areal BMD (bone mineral content [BMC] divided by the 2-dimensional area), which has units of grams of bone mineral per square centimeter of bone. However, with different mean BMD values for different anatomic regions (spine > hip > forearm) and different manufacturers having different calibration for BMC and BMD (for the spine and total hip, GE Lunar > Hologic ~ Norland and for the femoral neck, GE Lunar > Norland > Hologic), the raw data could not easily be put into context; the primary use for actual BMD values (in grams per square centimeter) has been and remains to compare serial results over time (same machine and same anatomic region). For clinical use, the challenge was to come up with a way of handling these multiple permutations (and trying to compare results with different technologies, such as quantitative computed tomography and quantitative ultrasonometry). Using Z-scores, an accepted statistical concept (comparing an individual with age-matched controls and expressing the difference as a standard deviation [SD] score) was not satisfactory because BMD declines with age (a Z-score of 0 at age 30 yr is good, but a Z-score of 0 at age 80 yr is not so good—the prevalence of osteoporosis would not change with age). The solution was to use a young adult Z-score (subsequently named the “T-score”—an interesting side story) (11), which compares an individual's BMD with the mean value for young normal individuals and expresses the difference as a SD score. The 2001 ISCD Official Position was to use a non-race-adjusted database regardless of the patient's race/ethnicity and use young female norms for women and male norms for men (12).

The formula for calculating the T-score is deceptively simple:patient’sBMDyoungnormalmean(YNM)youngnormalstandarddeviation(YNSD)

Clearly, the numbers used for YNM and YNSD are of critical importance. At the same BMD, individual's T-score will be better if the YNM is low or the YNSD is broad. Initially, manufacturers developed their own normative data 13, 14, 15. Again, BMD cannot be compared between manufacturers because of differences in calibration, but T-scores should be similar regardless of equipment, assuming the “normal” populations are similar. In a seminal study, Faulkner et al (16) compared T-scores of women measured on both Lunar (now GE) and Hologic (Fig. 1). T-scores for the spine were similar for both manufacturers, but T-scores for the femoral neck were consistently lower for Hologic compared with Lunar by about 0.9 T-score units. The explanation was that (corrected for calibration differences) not only was the Hologic YNM was higher than that of Lunar but also was the Hologic YNSD was less broad.

Between 1988 and 1994, the National Health and Nutrition Examination Survey (NHANES) III collected data on a variety of health measures from a representative sample of the US population in 2 phases 17, 18, 19, 20. Femoral neck BMD was measured in 382 men and 409 women aged 20-29 yr using Hologic pencil beam densitometers (QDR-1000) (NHANES III did not measure spine BMD). Of note, the femoral neck mean and SD values were slightly different in the 2 phases of NHANES III (for women, 0.849 ± 0.111 g/cm2 in phase 1 and 0.858 ± 0.120 in phase 2 and for men, 0.930 ± 0.136 in phase 1 and 0.934 ± 0.137 in phase 2) (side note—in an Australian study, from a large data set, different definitions of “normal” led to different means and SDs (21)).

The NHANES III YNM was lower, and the YNSD was broader than the Hologic manufacturer's database, resulting in femoral neck T-scores approx 0.7 T-score units better using NHANES III normative data instead of Hologic's own hip data. Not only did this difference result in “overdiagnosing” osteoporosis in individual subjects when the manufacturer's database was used 22, 23, but 2 large trials recruiting subjects based on a T-score of −2.0 and below with the Hologic database wound up with subjects with T-scores as high as −1.5 if norms from NHANES III had been used 24, 25.

In 1997, Hologic changed their hip database to that of NHANES III (for the femoral neck and trochanter, Hologic used the values from the first phase of NHANES III; for the total hip, they used values from the second phase) (K. Wilson, personal communication, 2013). This brought Lunar and Hologic T-scores into general agreement for both spine and hip. Eventually, although GE Lunar's hip norms were similar to Hologic, GE Lunar moved to NHANES III normative data as well around 2005 26, 27.

In addition to differences in BMD, there were other differences between NHANES III and later NHANES. Comparing results from 2000 to 2004 with 1998–1994, 25-hydroxyvitamin D levels were 2–4 ng/mL lower in men (but not in women), due in part to changes in assay methodology but also differences in BMI, milk intake, and sun protection (28). Comparing 2005–2008 with 1988–1994, there were also differences in body size, use of bone active medications, self-reported health status, and calcium and caffeine consumption (10).

Normative data for the spine in use today are those originally developed by the manufacturers 25–30 yr ago. The problem of T-score discrepancies between manufacturers seemed to be limited to the hip, so there has not been a strong argument for changing spine normative data. However, when a later round of NHANES measures were done (2005–2008), both spine and hip BMD were obtained (285 men and 262 women aged 20–29 yr), this time with Hologic fan beam equipment (Hologic 4500A and Discovery; Hologic, Waltham, MA). For the first time, there were normative data for both spine and hip from a well-characterized sample of young normal individuals. In the final stages of revising the manuscript for presentation to the ISCD Expert Panel (February 2013), an additional 2 yr of data from NHANES became available (A.C. Looker, personal communication, February 2013). NHANES “2005–2010” provides results from a larger sample size and “regresses to the mean” of NHANES III.

Previously, the ISCD recommended using a young normal Caucasian database to calculate T-scores, regardless of race (12) but using a male database for men and a female database for women (12). It was emphasized in the 2001 positions that the recommendation not to adjust T-score derivation based on race should apply only to the US population. After the 2005 ISCD Position Development Conference (PDC), it was clarified that application of the recommendation to use a uniform Caucasian (non-race adjusted) normative database for all ethnic groups may vary according to local requirements (29).

It is important to consider the role of T-score. In the past, it was a surrogate for risk. However, many or most fractures occur in people with T-scores better than −2.5 30, 31. Today, the T-score serves as an arbitrary determinant of osteoporosis or low bone mass for epidemiologic studies and in some cases for reimbursement of health care costs, but the T-score has been at least partly supplanted by FRAX and other fracture risk calculators, tools that address the need to incorporate additional clinical information and perhaps other quantitative measures (bone turnover markers, trabecular bone structure, etc.) as well, to aid in intervention decisions. As the field moves to absolute fracture risk, T-scores (and the normative data used to calculate them) will become less important.

With the newer NHANES data for the hip, YNMs are higher and YNSDs are less broad compared with NHANES III; this is more apparent with NHANES 2005–2008 than with NHANES 2005–2010. The evidence for an increase in hip BMD was strongest among non-Hispanic white women, with little evidence for parallel changes among non-Hispanic white men or other ethnicities (10). These differences could be due in part to methodology (fan beam for the newer data vs pencil beam for NHANES III hip and the original manufacturers' spine) but could also be because of changes in the population (compared with NHANES III, the more recent young normal individuals are heavier [3.8 kg for men and 6.8 kg for women]). A large population-based study from Canada found that major osteoporotic fracture rates declined substantially and linearly from 1996 to 2006 and that this was explained by improvements in BMD rather than greater rates of obesity or osteoporosis treatment (32). In the spine, YNM for men from the later NHANES is lower and for women slightly higher compared with the manufacturer's data; YNSDs are essentially no different.

Table 1 shows the YNM and YNSD values for women, Table 2 shows data for men, and Table 3 shows the BMD at which a T-score of −2.5 would be reached.

Several formulas have been offered to convert BMD from grams per square centimeter to milligrams per square centimeter (to eliminate the decimal) and correct for differences in calibration (standardized BMD [sBMD]). These formulas do not completely correct for differences in calibration. These formulas were developed in the mid-1990s for the existing pencil beam technology (and for L2–L4, not L1–L4); slightly different results are seen using formulas developed for current generation fan beam technology (33). For purposes of comparing BMD from different manufacturers' equipment, we prepared tables showing YNM and YNSD in sBMD. Tables 4 and 5 show NHANES III, NHANES 2005–2008, NHANES 2005–2010, and current manufacturer normative data as sBMD; Table 6 shows the sBMD threshold at which the T-score equals −2.5.

The Task Force was asked to address 5 issues, initially formulated as questions. Because the research and development process requires a statement to be discussed and approved, the questions were reformulated as statements. To preserve the atmosphere of the Task Force deliberations and what was presented to the Expert Panel, both the questions assigned to the Task Force and the statements reformulated by the Task Force were included in the materials presented to the Expert Panel. Because the Expert Panel decided to keep the status quo (continue to use the manufacturers' databases for spine and NHANES III for hip), it did not address the question, in the event that a change was recommended, whether should it be to NHANES 2005–2010 or 2005–2008.

Section snippets

Methodology

The Task Force members met briefly face-to-face, had 1 or more conference calls, and extensive email correspondence. Several straw votes were taken on the issues, and the questions were transformed to positive or negative statements based on the feelings of the Task Force members (for or against). The Task Force chair drafted the manuscript that was circulated and revised several times. Material was sought from NHANES and bone densitometry manufacturers. Numbers in the tables were prepared by

Acknowledgments

Thanks are owed to Anne Looker (National Health and Nutrition Examination Survey), Tom Webb (GE Lunar), Kevin Wilson and Tom Kelly (Hologic), and Tom Sanchez (Cooper Surgical/Norland) for material provided and review of tables of normative data and standardized bone mineral density calculations.

References (74)

  • N. Vallarta-Ast et al.

    Densitometric diagnosis of osteoporosis in men

    J Clin Densitom

    (2002)
  • C.H. Chesnut et al.

    Quantitation of bone mass in osteoporosis: recent advances

    Calcif Tissue Res

    (1976)
  • S.H. Cohn et al.

    Measurement of total-body calcium, sodium, chlorine, nitrogen and phosphorus in man by neutron activation analysis

    J Nucl Med

    (1971)
  • J.R. Cameron et al.

    Precision and accuracy of bone mineral determination by direct photon absorptiometry

    Invest Radiol

    (1968)
  • R.B. Mazess et al.

    Accuracy of bone mineral measurement

    Science

    (1964)
  • R.B. Mazess

    Measurement of the skeleton by noninvasive methods

    Calcif Tissue Int

    (1979)
  • R.B. Mazess et al.

    Dual-photon absorptiometry

    Phys Med Biol

    (1983)
  • J.T. Stalp et al.

    Determination of bone density by coherent-Compton scattering

    Med Phys

    (1980)
  • A.C. Looker et al.

    Changes in femur neck bone density in US adults between 1988-1994 and 2005-2008: demographic patterns and possible determinants

    Osteoporos Int

    (2012)
  • N.B. Watts

    T scores and osteoporosis

    Menopausal Med

    (2002)
  • N.C. Binkley et al.

    What are the criteria by which a densitometric diagnosis of osteoporosis can be made in males and non-Caucasians?

    J Clin Densitom

    (2002)
  • CooperSurgical. Bone densitometry. Available at: http://www.coopersurgical.com/ourproducts/boneden/Pages/default.aspx....
  • T.L. Kelly

    Bone mineral density reference databases for American men and women (abstract)

    J Bone Miner Res

    (1990)
  • R.B. Mazess et al.

    Bone density of the spine and femur in adult white females

    Calcif Tissue Int

    (1999)
  • K.G. Faulkner et al.

    Discrepancies in normative data between Lunar and Hologic DXA systems

    Osteoporos Int

    (1996)
  • A.C. Looker et al.

    Prevalence of low femoral bone density in older U.S. women from NHANES III

    J Bone Miner Res

    (1995)
  • A.C. Looker et al.

    Prevalence of low femoral bone density in older U.S. adults from NHANES III

    J Bone Miner Res

    (1997)
  • A.C. Looker et al.

    Updated data on proximal femur bone mineral levels of US adults

    Osteoporos Int

    (1998)
  • A.C. Looker et al.

    Proximal femur bone mineral levels of US adults

    Osteoporos Int

    (1995)
  • L.J. Melton

    How many women have osteoporosis now?

    J Bone Miner Res

    (1995)
  • L.J. Melton et al.

    Perspective. How many women have osteoporosis?

    J Bone Miner Res

    (1992)
  • S.R. Cummings et al.

    Effect of alendronate on risk of fracture in women with low bone density but without vertebral fractures—results from the fracture intervention trial

    JAMA

    (1998)
  • N. Binkley et al.

    Recalculation of the NHANES database SD improves T-score agreement and reduces osteoporosis prevalence

    J Bone Miner Res

    (2005)
  • W.D. Leslie et al.

    The impact of hip subregion reference data on osteoporosis diagnosis

    Osteoporos Int

    (2005)
  • E.S. Siris et al.

    Bone mineral density thresholds for pharmacological intervention to prevent fractures

    Arch Intern Med

    (2004)
  • Leslie WD, Lix LM, Yogendran MS, et al. 2013 Osteoporosis treatment does not explain decreasing temporal trends in...
  • B. Fan et al.

    Does standardized BMD still remove differences between Hologic and GE-Lunar state-of-the art DXA systems?

    Osteoporos Int

    (2010)
  • Cited by (67)

    • Generation of a Reference Dataset to Permit the Calculation of T-scores at the Distal Femur and Proximal Tibia in Persons with Spinal Cord Injury

      2022, Journal of Clinical Densitometry
      Citation Excerpt :

      While the introduction of T-score cutoff criteria has improved assessment of fracture risk considerably compared to aBMD, several studies have demonstrated that the use of T-score values have not been completely successful in reaching agreement in T-scores among the DXA densitometer manufacturers (13,14) and a reference dataset (15). Furthermore, while the use of a female Caucasian reference dataset to calculate T-scores and to diagnose osteoporosis in men is standard practice because men and women fracture at similar aBMD values (10), this practice is remains somewhat controversial in the AB population (16,17) with no evidence to support the application of this concept in persons with traumatic SCI. Premenopausal woman and men under the age of 50 yr are recommended to be compared to an age-matched aBMD reference dataset for the calculation of the Z-score.

    • Management of Hip Fractures at an Academic Center: Challenges and Opportunities

      2020, Journal of Clinical Densitometry
      Citation Excerpt :

      This is in concordance with findings of our previous study (12), and possibly explained by the lack of osteoporosis assessment in men, the high frequency of osteoporosis risk factors, such as smoking (41%), and low rates of Ca/vitamin D supplementation (<11%), despite a high prevalence of vitamin D deficiency in our population (18–20). Femoral neck T-scores were comparable between sexes, consistent with growing evidence that men and women fracture at similar BMD, and again confirms the International Osteoporosis Foundation and the International Society for Clinical Densitometry recommendations of deriving T-scores using the National Health and Nutrition Examination Survey III women database (21). In Lebanon, osteoporosis awareness increased when the densitometry became available in 1995 (12).

    View all citing articles on Scopus
    View full text