Comparison of fracture risk prediction by the US Preventive Services Task Force strategy and two alternative strategies in women 50-64 years old in the Women's Health Initiative

Abstract

Context: The United States Preventive Services Task Force (USPSTF) recommends osteoporosis screening for women younger than 65 years whose 10-year predicted risk of major osteoporotic fracture (MOF) is at least 9.3% using the Fracture Risk Assessment Tool. In postmenopausal women age 50-64 years old, it is uncertain how the USPSTF screening strategy compares with the Osteoporosis Self-Assessment Tool and the Simple Calculated Osteoporosis Risk Estimate (SCORE) in discriminating women who will and will not experience MOF.

Objective: This study aimed to assess the sensitivity, specificity, and area under the receiver operating characteristic curve of the three strategies for discrimination of incident MOF over 10 years of follow-up among postmenopausal women age 50-64 years.

Setting and design: This was a prospective study conducted between 1993-2008 at 40 US Centers.

Participants: We analyzed data from participants of the Women's Health Initiative Observational Study and Clinical Trials, age 50-64 years, not taking osteoporosis medication (n = 62 492).

Main outcome measures: The main outcome was 10-year (observed) incidence of MOF.

Results: For identifying women with incident MOF, sensitivity of the strategies ranged from 25.8-39.8%, specificity ranged from 60.7-65.8%, and AUC values ranged from 0.52-0.56. The sensitivity of the USPSTF strategy for identifying incident MOF ranged from 4.7% (3.3-6.0) among women age 50-54 years to 37.3% (35.4-39.1) for women age 60-64 years. Adjusting the thresholds to improve sensitivity resulted in decreased specificity.

Conclusions: Our findings do not support use of the USPSTF strategy, Osteoporosis Self-Assessment Tool, or SCORE to identify younger postmenopausal women who are at higher risk of fracture. Our findings suggest that fracture prediction in younger postmenopausal women requires assessment of risk factors not included in currently available strategies.

Figure 1.

Analytic Sample Derivation. There were 90 764 participants age 50–64 years. Of these, information was complete regarding osteoporosis risk factors for 82 247 participants who were not taking osteoporosis-related medications. Ten-year fracture incidence data were available for 62 492 of these 82 247 participants.

Figure 2.

A, AUC for identifying major osteoporotic fracture by FRAX. The white line indicates a reference AUC value of 0.50 (no better than chance alone). B, AUC for identifying major osteoporotic fracture by OST. The white line indicates a reference AUC value of 0.50 (no better than chance alone). C, AUC for identifying major osteoporotic fracture by SCORE. The white line indicates a reference AUC value of 0.50 (no better than chance alone).