Standardised quantitative morphometry: a modified approach for quantitative identification of prevalent vertebral deformities

Abstract

Our approach for the quantitative identification of vertebral deformity (standardised quantitative morphometry, SQM) reduces problems associated with obtaining reference intervals from populations with high prevalence of fracture. In women with osteoporosis, agreement with radiological diagnosis (surrogate gold standard) was better for SQM than QM using the Eastell-Melton method. Use of reference intervals for quantitative vertebral morphometry (QM) derived by statistical trimming can be problematic in reference populations with high prevalence of deformity. We have developed a modified approach known as standardised quantitative morphometry (SQM), whereby vertebral height is standardised to eliminate variation between individuals. The aims of this study were to compare SQM to QM (Eastell-Melton method) for identification of prevalent vertebral deformities, using qualitative radiological diagnosis as the gold standard, and automate the process. Our study populations were a clinic-based sample of 80 women ages 48 to 87 years with a high prevalence of vertebral deformity and a general practice (GP)-based sample of 372 women ages 50 to 85 years. Agreement with the gold standard was tested for SQM and QM. Agreement was better for SQM (kappa = 0.80) than for QM (kappa = 0.14) in the clinic sample using clinic-based reference data. The agreement was improved for QM using the GP-based reference data, kappa = 0.63. In the GP population, agreement was good for both SQM and QM (kappa = 0.59 and 0.54 respectively). In our population with a high prevalence of vertebral fracture, SQM performs better than the Eastell-Melton method.