Measurement imprecision in vertebral morphometry of spinal radiographs obtained in the European Prospective Osteoporosis Study: consequences for the identification of prevalent and incident deformities.

Several algorithms are currently in use for evaluating vertebral deformities from plain lateral radiographs of the lumbar and thoracic spine. However, the effects of measurement imprecision as well as uncertainties over image magnification on the correct identification of prevalent and incident vertebral deformities with these algorithms has been little studied. In a pilot study for the European Prospective Osteoporosis Study (EPOS), plain radiographs were submitted to a single central evaluating centre for measurement of vertebral height from T4 to L4. The thoracic and lumbar spines were imaged on separate films, and we have assessed the precision of measurement of vertebral heights and height ratios. The standard deviation of the differences between films of each of three height measurements ranged from 1.1 to 1.2 mm. A two-stage strategy for identifying incident deformities was devised. This required that the vertebra be a prevalent deformity at the time of the second radiograph and also that at least one of the vertebral ratios should have changed significantly since the first radiograph. The second stage removed all but two of the 18 vertebrae flagged positive in the first stage but not considered to be certain incident fractures by clinical reading of the radiographs.