A simple method for determining the probability a new vertebral fracture is present in postmenopausal women with osteoporosis

Abstract

The vertebral fracture status of women with osteoporosis has strong prognostic implications that may influence clinical decisions. We developed a simple method for estimating the probability that a new vertebral fracture has occurred in postmenopausal women with osteoporosis. Data was from the placebo groups of the Fracture Prevention Trial (median observation =21 months) and the MORE Trial at 2 years. A logistic regression analysis identified prior vertebral fracture (yes/no), new or worsening back pain (yes/no), and height loss (> or =2 cm, yes/no) as significant predictors for the presence of a new vertebral fracture. The actual probability of a new vertebral fracture in patients without these predictors, over the median observation period of 23 months, was 2.1%. Presence of back pain increased this probability fourfold; prior vertebral fracture increased this probability threefold, and height loss > or =2 cm increased this probability threefold. The predicted probabilities of a new vertebral fracture being present for each subgroup representing each of the eight possible combinations of back pain, prior vertebral fracture, and height loss were highly correlated with both the multivariate logistic regression-derived probabilities (r=0.98, p <0.001) and with the actual probabilities (r=0.99, p <0.001). The validity of this simple method was confirmed in patients from the MORE trial at both 2 years and 3 years, and in the Fracture Prevention Trial alone. This simple method provides clinicians with an estimate of the probability that a new vertebral fracture has occurred in postmenopausal women with osteoporosis.