A Risk Assessment Tool (OsteoRisk) for Identifying Latin American Women with Osteoporosis

• Published in: Journal of general internal medicine : JGIM 2005-03, Vol.20 (3), p.245-250
• Main Authors: Sen, Shuvayu S., Rives, Vincent P., Messina, Osvaldo D., Morales‐Torres, Jorge, Riera, Gregorio, Angulo‐Solimano, Juan M., Neto, João F.M., Frisoli, Alberto, Sáenz, Ricardo C., Geling, Olga, Ross, Philip D.
• Format: Article
• Language: English
• Subjects: Aged; Area Under Curve; Biological and medical sciences; Bone Density; bone mineral density; Bones; Diseases of the osteoarticular system; European Continental Ancestry Group; Female; Femur Neck - physiopathology; General aspects; Humans; Internal medicine; Investigative techniques, diagnostic techniques (general aspects); Latin America; Least-Squares Analysis; Medical sciences; Middle Aged; Multivariate Analysis; Older people; Original; Osteoarticular system. Muscles; Osteoporosis; Osteoporosis - diagnosis; Osteoporosis - physiopathology; Osteoporosis. Osteomalacia. Paget disease; Radiodiagnosis. Nmr imagery. Nmr spectrometry; Regression Analysis; Reproducibility of Results; Retrospective Studies; risk assessment; Risk Assessment - methods; Sensitivity and Specificity
• ISSN: 0884-8734; 1525-1497
• DOI: 10.1111/j.1525-1497.2005.40900.x

Objective: To develop a simple and easy‐to‐use tool for identifying osteoporotic women (femoral neck bone mineral density [BMD] T‐scores≤−2.5) in Latin America. Design: Retrospective study involving review of medical records. Setting: Osteoporosis clinics in 6 Latin American countries. Patients: Postmenopausal women ages ≥50 in Latin America who had femoral neck BMD measurements. Measurements And Main Results: A risk index was developed from 1,547 patients based on least square regression using age, weight, history of fractures, and other variables as predictors for BMD T‐score. The final model was simplified by reducing the number of predictors; sensitivity and specificity were evaluated before and after reducing the number of predictors to assess performance of the index. The final model included age, weight, country, estrogen use, and history of fractures as significant predictors for T‐score. The resulting scoring index achieved 91% sensitivity and 47% specificity. Simplifying the index by using only age and weight yielded similar performance (sensitivity, 92%; specificity, 45%). Three risk categories were identified based on OsteoRisk, the index using only age and body weight: high‐risk patients (index