Peripheral or central densitometry: does it matter which technique we use?

Abstract

Over the past decade, bone density scans have assumed an essential role in the diagnosis of osteoporosis. Although dual X-ray absorptiometry (DXA) scans of the central skeleton remain widely used, a variety of different types of equipment for measuring peripheral sites is now available. However, the poor correlation between different types of measurement and a lack of consensus on how results from peripheral sites should be interpreted have proved a barrier to the more widespread use of these devices. These issues prompt the following questions: Which technique best identifies patients at risk of fracture? What approaches to scan interpretation ensure the closest agreement among different methods? Does it matter if different patients are selected for treatment on the basis of different techniques? The relative risk (RR)of fracture derived from prospective studies is a key parameter for comparing the clinical value of different techniques. Recent reports confirm the advantages of hip bone mineral density compared with peripheral measurements for predicting hip fracture risk, although for fractures at other sites the differences are inconclusive. Using receiver operating characteristic curves, we show that the guidelines adopted for scan interpretation are of crucial importance for ensuring that the information provided is used effectively. The closest agreement among different techniques is achieved by setting thresholds for peripheral devices that target either the same percentage of the population or the same percentage of future fracture cases as femur DXA. Different methods select different groups of individuals from the total pool of patients who will later sustain a fracture, with the most successful technique being the one with the largest RR value. The emphasis placed by many studies on validating new techniques by studying their correlation with DXA may lead to the clinical value of peripheral devices being underestimated when the key datum is the RR value inferred from prospective fracture studies.