Résumé
ABSTRACT Objective Vertebral fracture is the most common osteoporotic fracture, affecting quality of life and increasing mortality. Epidemiological data on incidence of vertebral fracture are scarce in Brazil and throughout Latin America. Our aim was to determine vertebral fracture incidence and risk factors in a female Brazilian population. Subjects and methods Postmenopausal women with low bone mass were studied from the Brazilian placebo group of Arzoxifene Generations Trial (n = 974), followed for up to 5 years. The primary endpoint was new vertebral fractures, detected by X-Ray. Experimental design defined two strata: A. Osteoporosis or previous vertebral fracture with osteopenia; B. Osteopenia without previous fracture. Previous fracture, T-score, ionized calcium, alkaline phosphatase, creatinine and glucose were analyzed at baseline. Crude and adjusted incidence rates of vertebral fractures were estimated and Poisson regression model was used. Results Incidence rate was 7.7 (95% CI of 5.4 to 10.9) per 1,000 person-years (PY), increasing as a function of age. Women with new vertebral fractures had higher prevalence of previous nonvertebral fracture after menopause, were older and had lower lumbar spine (LS) T-score. Fracture risk increased by 46% for each unit reduction in LS T-score. Variables correlated with new vertebral fracture were age (p = 0.034), LS T-score, stratum A (p = 0.001 for both) and previous nonvertebral fracture after menopause (p = 0.019). In the final model, LS T-score was the strongest predictor. Conclusions Incidence rate of vertebral fracture of 7.7 per 1,000 PY. Age and previous fractures were associated with new vertebral fracture, but LS T-score was the most important predictor.