Performance of risk indices for identifying low bone density in postmenopausal women

Objective: To examine the ability of 4 published osteoporosis risk indices to identify women with low bone density. Subjects and Methods: Subjects included postmenopausal women 45 years and older consecutively recruited from US clinics, women from general practice centers in the Netherlands (age range, 50-80 years), women in the Rotterdam Study (the Netherlands) 55 years and older, and women aged 55 to 81 years old, screened for a clinical trial of alendronate. Bone mineral density (BMD) was measured at the femoral neck or lumbar spine; T scores represent the number of SDs below the mean for young healthy women. One risk index was calculated from age and weight; the other risk indices included up to 4 additional variables obtained by questionnaire. We calculated the sensitivity and specificity for identifying women with BMD T scores of -2.5 or less or -2.0 or less in the US clinic sample and created 3 risk categories, using each of the 4 indices. Results: Data were available for 1102 women from the US clinic sample, 3374 women in the Rotterdam Study, 23,833 women screened for a clinical trial of alendronate, and 4204 women from general practice centers in the Netherlands. Specificity for identifying BMD T scores of -2.5 or less ranged from 37% to 58% (depending on risk index) when sensitivity was approximately 90%. The prevalence of osteoporosis (defined, as T scores less than or equal to-2.5) differed widely. across the 3 risk categories, ranging from 2% to 4% for the low-risk category to 47% to 61% for the high-risk category in the US clinic sample. For spine BMD in the US clinic sample, the prevalence of T scores of -2.5 or less ranged from 7% (low risk) to 38% (high risk). The large differences in prevalence across risk categories were consistent across the other 3 samples of postmenopausal women in the United States and the Netherlands for all 4 risk indices. Conclusions: We recommend measuring BMD in women who are classified as having an increased risk of osteoporosis by using any of these risk indices because all 4 indices appear to predict low bone mass equally well. The Osteoporosis Self-assessment Tool index is easiest to calculate and therefore may be most useful in clinical practice.