Increased awareness of the problem of osteoporotic fractures together with the recent development of effective treatments has increased demands on physicians to manage patients with the disease. This in turn will demand increasing reliance on, and facilities for, the assessment of skeletal status. Bone mineral measurements provide the cornerstone of patient assessment, since osteoporosis is defined in terms of bone mass. Indeed, bone mineral measurements are used to diagnose the disorder, assess prognosis and monitor the natural history of the treated or untreated disease. The field is complicated by the increasing number of sites that can be measured with an ever increasing number of technologies. No one technology or site subserves optimally all the functions demanded of these assessments. This chapter reviews briefly the use of such measurements in clinical practice.

Techniques and Measurement

There are many techniques utilized to assess bone mass (Table Performance characteristics of various techniques of bone mass measurement at various sites). They variously assess mineral content of regional sites, particularly those sites at risk of osteoporotic fracture such as the wrist, spine and hip, but also the whole skeleton. The two most widely used techniques are single energy absorptiometry and dual energy absorptiometry.

Single energy absorptiometry measures bone mineral at a peripheral (appendicular) site such as the heel or the forearm. Single photon absorptiometry (SPA) utilizes a photon emitting source such as iodine-125; the bone mineral in traversed tissue attenuates the photons, from which the mineral content is calculated. Single energy X-ray absorptiometry (SEXA) has now supplanted SPA as a single energy technique for scanning the wrist. It is more precise and avoids the need for isotopes. Sites such as the spine and hip cannot be measured accurately by SPA or SEXA. Dual energy absorptiometry utilizing photons (DPA) or X-rays (dual energy X-ray absorptiometry) permits bone mineral to be measured at these sites.

The amount of calcium present at a specific site of a scan is termed bone mineral content. When the bone mineral content is divided by the area or volume assessed (the region of interest), a value for bone mineral density is provided. With single and dual energy absorptiometry the bone mineral content is divided by the area assessed (because of the two dimensional scan) and is not, therefore, a true volumetric density but an areal density.

Table Performance characteristics of various techniques of bone mass measurement at various sites. Reproduced with permission from World Health Organization. Assessment of fracture risk audits application to screening for postmenopausal osteoporosis. WHO Technical Report Series 843. Geneva: WHO 1994

Technique Site Percentage cancellous bone Precision error in vivo (%) Accuracy error in vivo (%) Scan time (min)
SPA forearm — distal 5 1-2 2-5 10
forearm — ultradistal 40 1-2 2-5 10
heel 95 1-2 2-5 10
dual energy X-ray absorptiometry lumbar — AP 50 1 5-8 10
lumbar — lateral 90 3 5-10 20
proximal — femur 40 1-2 5-8 10
total body 20 1 1-2 20
quantitative computed tomography spine —



100 2-4 5-10 15
spine — dual energy 100 4-5 3-6 20

SPA, single photon absorptiometry; dual energy X-ray absorptiometry, dual energy X-ray absorptiometry; quantitative computed tomography, quantitative computerized tomography; AP, anteroposterior

Quantitative ultrasound (QUS) methods have been introduced in recent years for the assessment of skeletal status in osteoporosis. The most widely evaluated assessments are broad-band ultrasound attenuation (BUA) and speed of sound (SOS) (or ultrasound velocity) at the heel and finger. The interest in their use lies in their low cost and portability, and in the fact that they do not involve ionizing radiation and may provide some information concerning the structural organization of bone in addition to bone mass.

The performance of QUS techniques has been evaluated in a large number of studies. Current evidence supports the use of QUS techniques for the assessment of fracture risk in elderly women, where the prognostic value for future fracture is nearly as good as that of dual energy X-ray absorptiometry. Additional clinical applications of QUS, specifically the assessment of rates of change in bone mineral for monitoring disease progression or response to treatment, require further investigation.

Computerized tomography (CT) has been applied both to the appendicular skeleton and to the spine. The major advantage of CT in the assessment of cancellous bone density is that the result provides a measure of true volumetric density, rather than an areally adjusted result (as is the case with dual energy X-ray absorptiometry). Cancellous bone is more responsive to many interventions than is cortical bone, so the technique is suitable for monitoring of treatment. It also avoids the influence of degenerative disease that is a particular problem with dual energy X-ray absorptiometry at the spine. Although the technique also gives information on the shape and architecture of bone, the resolution of cancellous structure is not optimal. The major disadvantages are radiation exposure and cost.

Magnetic resonance imaging (MRI) provides no direct information on density but, with the positive background given by all types of bone marrow, it is able to provide exceptional resolution of the internal structure of cancellous bone, down to the connections of individual trabeculae. Limitations arise owing to the need for patients to keep still for long periods of time. At present, MRI investigation of the skeleton remains a research procedure.

Of the many techniques that have been developed to assess bone mass, bone mineral or other related aspects of skeletal mass or structure, the technique that has had the greatest attention in terms of technical development and biological validation is dual energy X-ray absorptiometry, which, for now, is the ‘gold standard. Although dual energy X-ray absorptiometry neglects the very many other techniques that are available, it provides a platform against which the performance characteristics of less well established methodologies can be compared in terms of the uses demanded of these techniques.

Application Of Techniques

There are broadly three uses for bone mass measurements in the context of osteoporosis. They may be used for diagnosis, to assess prognosis and to monitor the natural history of the treated or untreated disorder.

Diagnostic use

In 1994, an expert panel of the World Health Organization (WHO) recommended thresholds of bone mineral density in women to define osteoporosis that have been widely but not universally accepted by the international scientific community and by regulatory agencies. Osteoporosis in postmenopausal Caucasian women is defined as a value for bone mineral density (bone mineral density) more than 2.5 standard deviations below the young average value. Severe osteoporosis (established osteoporosis) uses the same threshold, but in the presence of one or more fragility fractures.

The diagnostic threshold identifies approximately 15-20% of postmenopausal women as having osteoporosis when measurements using dual energy X-ray absorptiometry are made at the spine or the hip. Given an approximately linear loss of bone mineral density with age, and because of the Gaussian distribution of bone mineral density values, the incidence of osteoporosis increases exponentially after the age of 50 years, as is also the case for many osteoporosis-related fractures. When measurements are made at the three sites most vulnerable to fracture (the hip, spine and wrist), about 30% of postmenopausal women would have osteoporosis. This approximates the average lifetime risk for any one of these fractures.

The aim of diagnostic assessment is to stratify individuals within this distribution. The ability of techniques to do this depends upon their accuracy errors, which range from 2 to 10% for absorptiometric techniques and are greatest at the lumbar spine. In addition, osteoarthrosis, vascular calcification and fractures confound measurements at this site (Table Sources of diagnostic inaccuracies in the measurement of bone mineral by dual energy X-ray absorptiometry).

Table Sources of diagnostic inaccuracies in the measurement of bone mineral by dual energy X-ray absorptiometry (dual energy X-ray absorptiometry). Guidelines for diagnosis and management of osteoporosis.

  • Osteomalacia
  • Osteoarthritis (spine but also the hip)
  • Vascular calcification (especially the spine)
  • Overlying metal objects
  • Contrast media (spine)
  • Previous fracture (spine, hip and wrist)
  • Severe scoliosis
  • Vertebral deformities due to osteoarthrosis, Scheuermann’s disease
  • Inadequate reference ranges
  • Inadequate machine calibration

A further source of error relates to biological variability. Bone is not a homogeneous structure, and different sites have variable proportions of cancellous and cortical bone. The problem is compounded by variable rates of bone loss at different sites with advancing age. This represents the ‘biological’ inaccuracy in predicting bone mineral density at one site from measurements made at another site.

These factors compound the problem that individuals deemed to be osteoporotic at one skeletal site may not be found to be osteoporotic at another. Correlations between sites or between technologies at the same site are sufficiently poor (r less than 80% in young healthy individuals and generally less than 50% in patients) to be of very low predictive value. Even within the hip, correlation coefficients between regions are too low to be predictive. It is thus clear that the T-score cannot be used as a diagnostic criterion interchangeably with different techniques and at different sites. Even within the hip, there is variation in the degree with which the T-score changes with age. Indeed, were the T-score to be used with different techniques, the prevalence of osteoporosis and proportion of individuals allocated to any diagnostic category would vary so much as to devalue totally the credibility of the field of osteoporosis.

The same holds true in principle for hypertension, where measurements made at the leg may differ substantially from measurements made at the arm. One solution would be to designate individuals with osteoporosis at the spine but not at the hip as having osteoporosis of the spine, rather than using the term osteoporosis alone. This seems unsatisfactory for a systemic disease, and confuses the field still further in much the same way as hypertension of the leg would do. It appears more appropriate, therefore, to select a standardized site for the purpose of diagnosis, but not necessarily for risk assessment.

The hip as a reference site

The foregoing considerations suggest that a gold standard should be adopted in terms of site and technology for diagnostic purposes. No one technique or site subserves all the demands of densitometry, but if one site is to be chosen for diagnosis and prognostic purposes, the total hip or femoral neck is a strong candidate. Measurements at the hip have the highest predictive value for hip fracture, which has been well established in many prospective studies. Moreover, it is the site of greatest biological relevance, since hip fracture is the dominant complication of osteoporosis in terms of morbidity and cost. Several studies have shown that bone mineral density of the femoral neck best predicts cervical fractures, whereas the trochanteric site best predicts trochanteric fractures, but the total hip best reflects the risk of any hip fracture.

The WHO criteria were established largely with dual energy X-ray absorptiometry in mind, since this was the dominant technology at the time. The available evidence suggests that the diagnostic use of T-scores should be reserved for dual energy X-ray absorptiometry at the hip, a view recently endorsed by the International Osteoporosis Foundation. In the case of other sites and techniques, it may be preferable to express deviations of measurements from normal in units of measurement or units of risk. Examples of the latter include 5- or 10-year fracture probability, or an age-standardized relative risk of hip fracture or any fracture. This enfranchises the use of other techniques and sites for risk assessment. Indeed, where techniques give information on the likelihood of fracture, they can all be used in combination perhaps with other risk factors, to determine further investigation or treatment.

Reference ranges

The choice of a reference range is important for the accurate categorization of patients, as too is the estimate of the variance around the mean value. In choosing a cut-off value of standard deviations (SD), the intention of the WHO group was to make osteoporosis a rarity in healthy women before the menopause. Assuming a Gaussian distribution of bone mineral density, 0.7% of the young adult population would be characterized as having osteoporosis.

The management of the menopause 178

Recently, US reference data for the hip have been generated from the National Health and Nutrition Examination Survey (NHANES) III study, and could serve as a standardization platform. The use of NHANES III reference ranges derived from women aged 20-29 years and applied to the total hip decreases the apparent prevalence of osteoporosis in a reference population in the USA from 49% using the femoral neck and laboratory reference ranges of Hologic to 28%, more in line with the thresholds envisaged by the WHO.

Should different countries or different races utilize their own reference ranges or would a common gold standard be sufficient? Normal ranges for dual energy X-ray absorptiometry are available from many countries where the difference in mean bone mineral density and the standard deviations used are relatively small. The use of reference ranges in Whites in the USA accommodates the higher bone mass and lower fracture risk in Blacks.

Variations in bone mineral density between populations appear to be substantially less than variations in fracture risk. Age- and sex-specific risk of hip fracture differs more than ten-fold, even in Europe. These differences are very much larger than can be accounted for by any differences in bone mineral density between these communities. Indeed, in Asia, hip fracture risk is lower than in Northern Europe or the USA, but bone mineral density is lower. In view of the disparity between population fracture risks and bone mineral density, it is uncertain whether reference ranges should be drawn from local populations. There is a case, therefore, particularly for simplicity, to adopt an international reference range and standard deviations, such as the NHANES material, until further work tempers this view. The same holds true for other diagnostic methods, in that reference ranges should be derived from large population bases appropriate for international use.

Diagnosis in men

The reference ranges utilized for the diagnosis of osteoporosis are suggested for women. Cut-off values for men have variously used values derived from female or male populations. Not surprisingly, the prevalence of osteoporosis is greater using male-specific ranges at the hip. In men, the risk of fracture is substantially lower for a bone mineral measurement within their own reference range, so that a more stringent criterion is appropriate to yield the same risk as in women. The use of the same absolute value of bone mineral density as a cut-off in men as that used in women gives approximately the same absolute risk of vertebral and of hip fracture. For this reason, it may be appropriate from both a scientific and pragmatic view to utilize the same absolute threshold in both men and women, but it is important to recognize that the data on men are scanty, and not all studies show comparable gradients of fracture risk with bone mineral density in men.

It is also important to recognize that the use of T-scores for diagnostic assessment has some limitations, even though the presence of osteoporosis over the age of 50 years is a strong reason for considering treatment. However, the T-score of — 2.5 SD in Swedish women aged 50 carries a 10-year hip fracture probability of 1.8%. The hip fracture probability with the same T-score but at the age of 80 years is 11.3%. These probabilities require adjustment for countries other than Sweden, where the incidence of fracture and mortality rates may differ.

Thus, although a T-score of less than — 2.5 SD is an appropriate diagnostic threshold for osteoporosis, this threshold does not necessarily provide an intervention threshold. This will depend upon other factors such as age, history and the medication to be used.

Prognostic use

The use of bone mass measurements for prognosis also depends upon accuracy. Accuracy in this context, however, is the ability of the measurement to predict fracture. In general, all absorptiometric techniques have high specificity but low sensitivity (i.e. detection rate), which varies with the cut-off chosen to designate high risk. Many prospective studies indicate that the relative risk for fracture increases by a factor of 1.5-3.0 for each standard deviation decrease in bone mineral density. The performance depends on the type of fracture. For example, bone mineral density assessments by dual energy X-ray absorptiometry to predict hip fracture are better where measurements are made at the hip rather than the spine or forearm. An individual with a T-score of -3 SD at the hip would have a 2.6 or greater than 15-fold higher risk of hip fracture than an individual with a T-score of 0 SD. The same T-score at the spine (T=-3 SD) would carry only a four-fold increase in hip fracture risk (1.6). Where the intention is to predict any osteoporotic fracture, the commonly used techniques are comparable. The risk of fracture increases approximately 1.5-fold for each SD decrement in measurement. Thus, an individual with a measurement 3 SD below the average value for age would have a 1.5 or greater than three-fold higher risk than an individual with an average bone mineral density.

The ability of bone mineral density to predict fracture is comparable to the use of blood pressure to predict stroke, and significantly better than the ability of serum cholesterol to predict myocardial infarction. Prognostic accuracy is considerably enhanced by the concurrent use of other risk factors. These include biochemical estimates of resorption and/or formation, historical information such as prior fragility fractures, and factors contributing to risk independent of bone density (such as postural stability). In the immediately postmenopausal population, measurements at any site (hip, spine and wrist) predict any osteoporotic fracture equally well. The choice of site depends, therefore, upon the clinical context in which prognostic evaluation is made. In the elderly the hip is likely to be the most favorable site. The measurement of more than one site by absorptiometric techniques provides little added value.

Enhancing prognostic information

The consideration of other risk factors in conjunction with bone mineral density assessments improves the predictive value of the test. Examples are given in Table 6 of factors that contribute significantly to fracture risk over and above that provided by bone density measurements or age. Thus, the presence of multiple risk factors can be used to enhance a case-finding strategy in osteoporosis, and has been incorporated into practice guidelines in the USA. So, the presence, say, of low ultrasound attenuation or velocity, together with independent risk factors, might qualify individuals for treatment, without the need for bone mineral density assessment at the hip. In other words, it is the probability of fracture that is important rather than the fulfilment of a diagnostic criterion. A caveat is that the risk identified by some risk factors is not amenable to particular treatments, so the relationship between total risk and reversible risk is important. Liability to falls is an appropriate example where the risk of fracture is high, but treatment with agents affecting bone metabolism may have little effect to reduce that risk. Other risk factors, particularly a prior fragility fracture, contribute to a risk that is responsive to interventions.bone metabolism may have little effect to reduce that risk. Other risk factors, particularly a prior fragility fracture, contribute to a risk that is responsive to interventions.

In countries such as the UK and other European nations, a more conservative view is taken in that only patients with osteoporosis are offered treatment. The presence of risk factors provides only the opportunity to direct individuals for assessment by dual energy X-ray absorptiometry. Whereas clinical risk factors (e.g. low body mass index, premature menopause, corticosteroid use) are traditionally used to trigger investigation by dual energy X-ray absorptiometry, the wide availability and proliferation of peripheral densitometry and QUS devices suggest that, where low values are found, these might be used to trigger the more formal assessment with dual energy X-ray absorptiometry at the hip. A middle road between these approaches is a strategy of triage in which individuals at very high risk would qualify for treatment, those at very low risk would not be further evaluated, and only those at intermediate risk would have further assessment by dual energy X-ray absorptiometry at the hip.

Table Examples of relative risks of hip fracture in women with and without adjustment for bone mineral density (bone mineral density).

Risk assessment Relative risk Crude Adjusted*
Hip bone mineral density 1 SD below mean population value 2.6
Non carboxylated osteocalcin above normal 2.0 1.8
Biochemical index of bone resorption (CLX) above premenopausal range 2.2 2.0
Prior fragility fracture after the age of 50 years 1.4 1.3
Body weight below 57.8 kg 1.8 1.4
First degree relative with a history of fragility fractures aged 50 years or more 1.7 1.5
Maternal family history of hip fracture 2.0 1.9
Current cigarette smoking 1.9 1.2
Poor visual acuity (<2/10) 2.0 2.0
Low gait speed (/l SD decrease) 1.4 1.3
Increase in body sway (/l SD) 1.9 1.7

*Adjusted for bone mineral density; SD, standard deviation; CLX, C-telopeptide fragments of collagen I

Assessment and treatment thresholds

Thresholds of risk used to characterize multifactorial diseases are often arbitrarily defined. In the case of osteoporosis, fracture risk increases continuously with decreasing bone mineral density, so there is no biological break-point to distinguish absolutely an individual who will fracture from one who will not. Nevertheless, thresholds are useful in a clinical setting, where they give information on prognosis or treatment. In the case of bone mineral density assessment both types of information are given, but need to be cautiously interpreted.

In the case of diagnostic thresholds, it is relevant to recall that a positive (or negative) test may be spurious. The finding of a low bone mineral density should raise the question of why this is so, and other causes of low bone mineral density (technical, confounding and clinical) should be excluded to fulfil a diagnostic criterion. Also, because the same diagnostic threshold in one country will not identify individuals with the same fracture risk in another country with an identical T-score, it is important not to confuse diagnostic thresholds with treatment thresholds. This was not the intention of the WHO, but the diagnostic criterion is interpreted by many practitioners and health-care agencies as an intervention threshold. But intervention thresholds depend not only on risk, which varies with age, but also on the benefits and costs of interventions.

The notion that intervention thresholds may differ from diagnostic thresholds requires the elucidation of intervention thresholds for osteoporosis in much the same way as for cardiovascular disease. This will demand the conversion of bone mineral density results into absolute fracture probabilities. This clearly illustrates that a diagnostic threshold for osteoporosis (T-score=-2.5 SD) has quite different implications at different ages. At the age of 50 years, the 10-year probability of hip fracture in women is 1.7%, but is >10% at the age of 75 years or more. Thus, intervention thresholds are more logically set by threshold probabilities. If an intervention threshold were set (for example) as a 10-year probability of hip fracture that exceeded 10%, this threshold would be attained in women with osteoporosis at the age of 65 years or more. The same threshold of risk is attained in an average population of women aged 75 years or more. The use of absolute risk preserves the utility of the T-score for diagnosis with DXA at the hip, and enhances the value of all technologies to characterize the populations most at risk from fractures.


In contrast to the diagnostic and prognostic use of bone mineral measurements, the use of bone density measurements to monitor changes in bone mass depends upon precision. The long-term errors of the most precise techniques (dual energy X-ray absorptiometry and SEXA) are in the order of 1-2%, a change of 3-6% being required in a single patient to assess treatment effectiveness. Treatment-induced changes are generally most marked at sites of cancellous bone such as the lumbar spine. In most instances, repeating bone mass measurements at an interval shorter than 1 or 2 years after initiating therapy is not helpful for making decisions about treatment. More frequent measurements may, however, aid compliance, but the optimum interval has not been elucidated.


The clinical uses of bone mineral measurements are in diagnosis, prognosis (risk assessment) and monitoring disease progression. For diagnostic purposes, osteoporosis should be diagnosed using dual energy X-ray absorptiometry at the proximal femur. For men, the same threshold for osteoporosis can be used as that computed for women. The T-score is best reserved for diagnostic use. For risk assessment, all validated techniques can provide useful clinical information, although the use of T-scores for this purpose is inappropriate. Rather, measurement results should be converted to fracture probabilities. Risk assessment by bone mineral density can be enhanced by the knowledge of bone mineral density-independent risk factors, again expressed as absolute risk. More work is required to delineate accurately the level of risk that should be used as an intervention threshold.

Tagged with:  
Share →