Diagnostic Performance and Cost-Effectiveness of Technologies to Measure Bone Mineral Density in Postmenopausal Women
Context and policy issues
Osteoporosis is a progressive disease characterized by low bone mass, microarchitectural bone tissue deterioration and loss of bone strength that can lead to an increased risk of fracture. The World Health Organization’s (WHO) definition of osteoporosis is a T score of ≤−2.5, which is equivalent to a hip bone mineral density (BMD) of greater than or equal to 2.5 standard deviations below the young adult female reference mean. Osteoporosis is common in postmenopausal women, and 80% of the approximately one million Canadians with osteoporosis are women.
Different techniques are available to measure BMD. The gold standard for measuring BMD to diagnose osteoporosis is central dual-energy x-ray absorptiometry of the hip (DXA, DEXA). Quantitative ultrasonography (QUS), quantitative computed tomography (QCT), and radiographic absorptiometry can also be used.
The effectiveness of the technologies to assess BMD to diagnose osteoporosis is examined in this report. Because of the limited availability and the cost implications of central DXA devices, an assessment of the effectiveness of alternative devices is required to determine their usefulness for predicting fracture risk in postmenopausal women.
1. What is the diagnostic performance of the technologies used to measure BMD (i.e., peripheral densitometry, quantitative computed tomography, quantitative ultrasonography, radiography absorptiometry) compared to dual-energy x- ray absorptiometry when used to evaluate the risk of osteoporotic fracture in postmenopausal women?
2. What are the comparative economic considerations relative to the use of the different technologies to measure BMD?
Published literature was identified by cross-searching Biosis, Embase, and Medline databases on the OVID search system. Parallel searches were performed on the Cochrane Library (Issue 2, 2007) databases. Retrieval was limited by human population, English or French language, and publications from 2002 to September 13, 2007. RCTs were limited to publications from 2005 to September 13, 2007. Some manual searches of relevant articles were conducted. Filters were applied to limit the retrieval to systematic reviews, RCTs, and economic studies.
One reviewer selected articles and extracted data. Studies were included that compared at least two methods of measuring BMD in postmenopausal women. Studies were excluded if they were not peer-reviewed or were not full publications.
Two systematic reviews, six observational studies, and three studies on economic considerations were included in this report.
Comparative clinical effectiveness
A Canadian systematic review concluded that predicting fracture risk in postmenopausal women was best done using DXA but that QUS was comparable to DXA for the measurement of BMD. Limited information was available from the other systematic review because of language restrictions. The authors concluded that there was uncertainty regarding the comparisons of DXA, QUS, and QCT. The observational studies varied in the methods used to discriminate fractures. The power of a diagnostic test can be expressed as area under the curve (AUC) computed from the receiver operator characteristic (ROC) curve. An AUC close to 1.0 indicates a perfect test, and an AUC close to 0.5 indicates a test of little value. One study comparing DXA to QUS found an AUC of 0.59 to 0.63 for BMD by DXA and 0.62 for QUS. Two other studies found AUCs of 0.86 to 0.95 for DXA and 0.93 for QUS and 0.60 to 0.66 for DXA and 0.60 for QUS. DXA, QUS, and QCT were compared in one study, which found AUCs of 0.843 to 0.896 for DXA, 0.876 for QCT, and 0.604 to 0.869 for QUS. DXA and QCT were compared in one study, which found AUC to be 0.647 for DXA and 0.870 for QCT. There were no differences in odds ratios (ORs) that were reported in some studies, with one exception in which the OR for BMD determined by using DXA was 4.8, and that determined using QCT (MDCT) was 12.7.
Three studies reported on costs for technologies to measure BMD. A study from Thailand reported the cost-effectiveness ratio in US dollars to be US$88.42 per fracture prevented for DXA and US$146.48 per fracture prevented for QUS. A US study found DXA cost US$703,000 per 1,000 women to prevent 7.8 hip fractures, QUS cost US$632,000 per 1,000 women to prevent 6.7 hip fractures, and sequentially testing by QUS then DXA cost US$442,000 per 1,000 women to prevent 5.7 fractures. A Spanish study found that the total cost per correctly detected case of osteoporosis was €23.85 for DXA and €22.00 for QUS.
Conclusions and implications for policy making
The AUC for DXA ranged from 0.59 to 0.95 for the different bones analyzed in the different studies. QUS AUC ranged from 0.60 to 0.93, and QCT AUC ranged from 0.87 to 0.93 in the studies. There do not seem to be major differences between DXA and QUS in the studies. ORs for discriminating fractures were reported for four of the six studies and ranged from 1.35 to 4.8 for DXA, 1.26 to 4.18 for QUS, and 12.7 to 16 for QCT. There were no major differences between the ORs in the studies, except in the study on MDCT and DXA where the OR was higher for MDCT compared to DXA. The costs for DXA and QUS were similar in two cost studies. It is unknown, however, whether any of these studies could be translatable to a Canadian setting.
QCT seems to be at least as effective as DXA. QCT, however, uses more radiation than DXA. Overall, QUS seems to be comparable to DXA for discriminating fractures in postmenopausal women, although this is based on low quality evidence. Other factors such as the lack of radiation used for QUS and the limited availability of DXA may help to determine which screening test would be most useful. None of these were Canadian studies and more research is needed on the methods to measure BMD in postmenopausal women.