Spine deformity index (SDI) versus other objective procedures of vertebral fracture identification in patients with osteoporosis: a comparative study

Abstract

Radiologic identification of vertebral fractures is most important in the diagnosis and monitoring of patients with spinal osteoporosis. Different methods, using vertebral height measurements for fracture identification, have therefore been developed. We compared four methods for fracture identification in spinal x-rays of 62 female patients with primary osteoporosis. The methods of Hedlund and Gallagher, Melton et al., and Davies et al. are based on the ratio of heights within one vertebra or of the height ratios of adjacent vertebrae; all three methods result in counting the number of vertebral fractures. The fourth method of Minne et al. relates anterior, middle and posterior heights of the vertebrae between T5 and L5 to the respective heights of T4. The relative vertebral heights of patients with osteoporosis are compared to the respective relative heights (anterior, middle, and posterior) of normal subjects (T5-L5). This allows the identification of fractured vertebrae, as well as a quantification of the extent of deformation due to these fractures (spine deformity index, SDI). The same measurement data of 62 spinal x-rays of anterior, middle, and posterior heights between T4 and L5 were used to detect vertebral fractures by the four different methods. Correlation between the number of identified fractures by the different methods ranged between r = 0.56 and 0.83. On the other hand, we found a remarkable difference in the mean number of identified fractures and a discrepancy in the identification of single vertebrae as fractured or not. All four methods revealed an accumulation of fractures in the midthoracic area and in the region of transition from thoracic to lumbar spine. Vertebral fractures as identified by SDI were not detected by the other three methods in 12-29% of the cases, even if vertebral height reduction was more than 6 mm. The reliability of each method was examined by the determination of "decreasing" number of fractures during follow-up. A decrease in the number of fractures was found in about 25% patients, if using the three methods that count only the number of fractures. We obtained a 3.6% decrease in the number of fractures using the fourth method. Furthermore, the decrease in SDI values in follow-up was within the range of variance. We therefore believe that SDI and related procedures are reliable in quantifying spinal osteoporosis and monitoring during follow-up.