Longitudinal evaluation of morphometric X-ray absorptiometry for the identification of vertebral deformities.

Abstract

Morphometric X-ray absorptiometry (MXA) has not been evaluated for the identification of incident vertebral deformities. The reliability of longitudinal measurements in quantitative vertebral morphometry is influenced by the precision of the technique. Long-term precision in vitro (weekly MXA phantom scans) assessed by retrospective cumulative sum (Cusum) analysis, detected one event during a 6-month period when the measurement process was 'out of control'. Inspection of service records revealed that repair work was performed around this timepoint. The coefficient of variation (CV) for long-term precision (vertebral heights) in a population-based sample of postmenopausal women ages 56 to 83, mean 65+/-6 years (n = 48), was 4.0% for morphometric radiography (MRX), 2.9% for MXA using the compare facility for analysis of serial scans, and 3.2% when 'compare' was not used. In women with osteoporosis ages 49 to 87, mean 67+/-9 years (n = 50). the CV was 5.0% for MRX, 4.1% for MXA using 'compare' and 8.5% without 'compare'. Precision errors for height ratios (MRX and MXA) were greater than for vertebral heights. Incident deformities were identified by MRX and MXA in the women with osteoporosis, using point prevalence, 20% minimum reduction in vertebral height, and percent least significant change (LSC) in vertebral heights and height ratios. Semiquantitative analysis of radiographs by a radiologist (Genant method) was used as the gold standard. The results were similar for MRX and MXA, and all morphometric criteria identified a similar proportion of true incident deformities, although the false positive rate was generally greater for the height ratio approaches. MXA has good long-term precision and is comparable to MRX for the identification of incident deformities when scans are analyzed with the compare facility.