The increasing use of peripheral bone densitometry

Osteoporotic fractures are a major cause of excess mortality, morbidity, and expenditure worldwide. There is a strong inverse relation between bone mineral density and the risk of fracture, with a doubling in fracture incidence for each standard deviation reduction in bone mineral density.¹ The World Health Organization has defined osteoporosis as a bone mineral density of more than 2.5 standard deviations (T score <−2.5) below the mean value for young adults.² This definition was made for epidemiological reasons to compare female populations and not as a threshold for intervention. However, the WHO definition has been used increasingly for the diagnosis of osteoporosis in individuals, based on the measurement of bone mineral density at the hip and spine using dual energy x ray absorptiometry. Recently, newer peripheral densitometry devices have been developed, which have the advantage of low cost and portability. It is likely that these will be used increasingly for the diagnosis of osteoporosis and assessment of fracture risk in the community.

The commoner forms of these devices in Europe and North America include heel and forearm dual energy x ray absorptiometry and quantitative ultrasound at the heel. The various devices have similar overall predictive value for estimating fracture risk regardless of the skeletal site measured or technique used, although measurement at any particular site best predicts fracture at that location.³ Ultrasound devices cannot diagnose osteoporosis as they do not measure bone mineral density (on which the WHO definition is based) but measure other factors related to bone strength (broadband ultrasound attenuation, speed of sound, and stiffness). Nevertheless, ultrasound measurements at the heel have been shown in large longitudinal studies to predict future fractures in older subjects, in some cases as accurately as measurement of central bone mineral density.⁴,⁵ It is essential to understand the limitations of these devices because of the potential for misclassification, which may lead to inappropriate reassurance or unnecessary alarm to patients.

For the diagnosis of osteoporosis, it is important to realise that bone mineral density is not the same throughout the skeleton. This “discordance” can be caused by several factors, including differences in bone accretion and loss at various sites, variations in the accuracy of measuring bone mineral density by different techniques, and differences in the normal ranges for young adults between devices. Recent studies have shown significant variation in the prevalence of osteoporosis with measurements at different peripheral and central sites, suggesting potential for misdiagnosis if the WHO criteria are applied at all sites.⁶,⁷

Discordance, and therefore the risk of misclassification, is greater in the early postmenopausal population than in women aged over 65 years. It may therefore be appropriate to measure more than one site in younger women, to reduce the chances of missing a diagnosis of osteoporosis.⁸ In older women, where discordance is less of a problem, the likelihood of missing a diagnosis of osteoporosis when measuring only one skeletal site—such as the wrist, heel, or hip—is reduced. The exception in the elderly is a single measurement of the posterior anterior spine by dual energy x ray absorptiometry, where artefacts such as osteophytes can spuriously increase the value of bone mineral density measurements.

This problem of each person having a different T score depending on the device used needs to be solved urgently. For hip sites, common reference data from the National Health and Nutrition Examination Survey (NHANES III) have been incorporated into the databases of the three major central bone densitometry machines, which has reduced the potential for variation between machines in diagnosing osteoporosis.⁹ For sites other than the hip, however, a common reference database does not exist. With regard to the peripheral devices, one short term solution could be to define “equivalent T score thresholds” for any site or device, which would identify the lowest quintile of the bone mineral density for 60-69 year olds.¹⁰ The WHO definition of osteoporosis was based on identifying people in the lowest quintile of bone mineral density. Specific equivalent device and site thresholds can be obtained by setting the prevalence of the condition at 20% based on 65 year old Caucasian females. A more radical longer term solution being debated is to abandon the WHO definition of osteoporosis in favour of criteria based on absolute values for bone mineral density and risk of fracture for individual sites and devices. Bone mineral density is not the only important determinant of fracture risk, and the ideal method of predicting fracture should incorporate other important factors such as age, history of low trauma fractures, and propensity to fall. The concepts of “absolute risk” (calculated from the relative risks of the relevant factors) and “remaining lifetime fracture risk” could fulfil this role and may be superior to T scores in fracture prediction. However, few studies of sufficient duration exist to examine how bone mineral density and other factors affect the lifetime risk of fracture, and many assumptions are required to extrapolate from short term risk to lifetime risk.¹¹

The role of peripheral devices in monitoring treatment remains unclear, and few data exist on the usefulness of the peripheral devices in men. The case for universal screening for osteoporosis has not been proved, and both peripheral and central bone densitometry are therefore likely to be restricted to those who have risk factors for osteoporosis.¹² It remains to be shown whether peripheral densitometry techniques, which could easily be applied in primary care, can be used in screening to reduce fracture rates in older people.

Although portable peripheral bone densitometry techniques are now available, the diagnosis of osteoporosis is complicated by the issue of discordance between sites which has emerged as a major problem causing misclassification of individuals. While optimal methods for defining osteoporosis and predicting fracture risks are debated, it is imperative that users understand the limitations of these techniques: in the short term, using the device-specific equivalents of T scores in conjunction with other risk factors such as previous fractures and the propensity of a person to fall seems a sensible approach.