An evaluation of the United Kingdom National Osteoporosis Society position statement on the use of peripheral dual-energy X-ray absorptiometry

Abstract

A recent position statement issued by the UK National Osteoporosis Society recommends a triage approach to the use of peripheral dual-energy X-ray absorptiometry (pDXA) devices. Patients with a forearm T-score greater than –1 or less than –2.5 are regarded as normal or osteoporotic, respectively, while those with a T-score between –1 and –2.5 are sent for further assessment with spine and hip DXA. We have evaluated the NOS pDXA algorithm by comparing it with the alternative strategies of relying on forearm BMD alone, or performing spine and hip DXA in every patient. The evaluation was carried out using a mathematical model, and the predictions were compared with in vivo data obtained in patients referred for investigation by their general practitioner. In the model the population distribution of spine, hip, and forearm BMD was described by a trivariant Gaussian function. Relative risks of fracture were taken from a meta-analysis. The three strategies were compared using receiver operating characteristic (ROC) curves in which the percentage of future fracture cases identified was plotted against the percentage of the whole population found to have osteoporosis. ROC curves plotted for the discrimination of hip, vertebral, and Colles fracture risk and the risk of a fracture at any skeletal site were similar for all three strategies, with the curves for the NOS pDXA algorithm nearly identical to those for spine and hip DXA. For the case of hip fracture, vertebral fracture, or a fracture at any site, forearm BMD was slightly inferior to the NOS algorithm, but the reverse was true for Colles fracture. The small difference between the ROC curves suggests that forearm BMD used alone can reproduce clinical decision-making with the NOS pDXA algorithm provided that a T-score threshold of T=–2.1 is used for the diagnosis of osteoporosis, instead of the conventional figure of T=–2.5. Results from the in vivo study were in good agreement with the predictions of the model, although some differences were observed that were explained by inaccuracies in the forearm reference data. We conclude that use of forearm BMD alone with a modified T-score threshold of –2.1 would save the need for spine and hip DXA scans and identify only slightly fewer fracture cases for treatment.