Prognostic utility of a semiquantitative spinal deformity index

Abstract

The semiquantitative spinal deformity index (SDI) is a summary measure of the vertebral fracture status of the spine incorporating both the number and severity of vertebral fractures. For each vertebra, a visual semiquantitative grade of 0, 1, 2, or 3 is assigned for no fracture or mild, moderate, or severe fracture, respectively, and the SDI is calculated by summing the fracture grades of all vertebrae (T4 to L4). We investigated the effect of prevalent vertebral fracture number and severity, as integrated by the SDI, on 3-year vertebral fracture risk by performing logistic regression modeling with data from the MORE trial. There was a striking linear relationship between baseline SDI and the model-based vertebral fracture risk estimates, with a near-perfect correlation (r = 0.98, P < 0.001). However, the SDI may be overly simplistic, as a given SDI value can be attained through differing vertebral fracture scenarios (i.e., an SDI of 3 can be realized three ways), each corresponding to potentially different vertebral fracture risk. To address this issue, a second, more complex model was constructed that included individual predictor variables for number of mild, number of moderate, and number of severe prevalent vertebral fractures. The model-based risk estimates for vertebral fracture using the SDI and the more complex model were highly correlated (r = 0.91, P < 0.001), giving almost identical values up to an SDI of 5. Thus, for most clinical scenarios, it is not necessary to consider the particular fracture configuration that led to a given SDI score for predicting a patient's future vertebral fracture risk. These results validate the SDI as an accurate tool for assessing future vertebral fracture risk; patients with greater baseline SDI had greater future risk for vertebral fractures.