A new method for vertebral fracture diagnosis

Abstract

A number of methods have been proposed for estimating the prevalence of vertebral fractures. Most methods are based on the distribution of normal vertebral dimensions in the population. However, these methods fail to identify a significant proportion (20–30% or more) of fractures documented on serial radiographs. This may occur because vertebral size varies between individuals as a result of differences in body size (and possibly other factors), and a normal range based on population reference data may be too large. In this paper, a new method is described for identifying vertebral fractures that are missed using diagnostic criteria based on vertebral dimension distributions of the population. This new method is based on calculating the average vertebral size, and statistical confidence limits, for the individual. The average vertebral size method was evaluated by testing its ability to identify incident fractures (which were identified from changes in dimensions compared to previous radiographs), using only the final film. The new method correctly identified most (81% of crush and 83% of wedge) incident fractures on the final radiograph. In contrast, criteria based on population distributions correctly identified only 53% of crush and 72% of wedge incident fractures. Using prospective data, prevalent fractures identified using both population-based and individual size-based criteria predicted the risk of incident fractures. Furthermore, incident fractures identified using both methods (population- and individual-based criteria) were associated with increased back pain. These data suggest that both types of prevalent deformities are important indicators of disease. We conclude that the new diagnostic method, which is based on an individual's vertebral size, will identify prevalent vertebral fractures missed by other methods and improve the accuracy of classification by fracture type (wedge, crush, and so on).