A simple tool to identify Asian women at increased risk of osteoporosis

Abstract

Patients with low bone mineral density (BMD) have a high risk of future fractures, and should be actively considered for treatment to reduce their risk. However, BMD measurements are not widely available in some communities, because of cost and lack of equipment. Simple questionnaires have been designed to help target high-risk women for BMD measurements, thereby avoiding the cost of measuring women at low risk. However, such tools have previously focused on evaluation of non-Asian women. We collected information about numerous risk factors from postmenopausal Asian women in eight countries in Asia using questionnaires, and evaluated the ability of these risk factors to identify women with osteoporosis as defined by femoral neck BMD T-scores ≤-2.5. Multiple variable regression analysis and item reduction yielded a final tool based on only age and body weight. This risk index had a sensitivity of 91% and specificity of 45%, with an area under the curve of 0.79. Previously published risk indices based on larger numbers of variables performed similarly well in this Asian population. Large differences in risk were identified using our index to create three categories: 61% of the high-risk women had osteoporosis, compared with only 15% and 3% of the intermediate- and low-risk women, respectively. The low-risk group represented 40% of all women, for whom BMD measurements are probably not needed unless important risk factors, such as prior nonviolent fracture or corticosteroid use, are present. An existing population-based sample of postmenopausal Japanese women was used to validate our index. In this sample of Japanese women the sensitivity was 98% and specificity was 29%; the low-risk category, for whom BMD is probably unnecessary, represented 25% of all women. We conclude that our index performed well for classifying the risk of osteoporosis among postmenopausal Asian women and applying it would result in more prudent use of BMD technology.