Comparison of bone mineral density and quantitative ultrasound of the calcaneus: site-matched correlation and discrimination of axial BMD status.

The performance of quantitative ultrasound (QUS) and dual energy X-ray absorptiometry (DXA) bone densitometry of the calcaneus have been compared, both in terms of site-matched correlation and their discriminatory ability to identify osteoporotic and osteoporotic or osteopenic subjects. 91 female subjects (aged 56.9 +/- 9.6 years, 31-84 years) who were routinely referred for axial BMD assessment of the lumbar spine and femoral neck by DXA (Lunar DPX-L), consented to have additional measurements of QUS (McCue CubaClinical Mk II) and DXA (Lunar PIXI) of their left calcaneus. The site-matched correlation between calcaneal BMD with QUS parameters were: (a) broadband ultrasound attenuation (BUA) alone, adj-R2 = 62.7%, p < 0.0001; (b) velocity (VOS) alone, adj-R2 = 48.4%, p < 0.0001; and (c) BUA and VOS combined, adj-R2 = 65.2%, p < 0.0001. The site-matched correlations are towards the higher end of data reported by other researchers, indicative of the exacting measurement protocol implemented here. 30 subjects were categorized as normal, 38 being osteopenic and 23 being osteoporotic. Optimum accuracies and odds ratios were obtained using logistic regression. The differences in accuracy between calcaneal BMD and calcaneal QUS parameters were statistically insignificant, with zero included within the confidence intervals, for the identification of both (a) osteoporotic and (b) osteoporotic or osteopenic subjects. The odds ratios for the discrimination of subject status achieved with calcaneal BMD were higher, although statistically insignificant, than achieved with the QUS parameters. Receiver operator characteristic (ROC) analysis for the identification of subjects into the categories (a) and (b) above was performed. Areas under the ROC curve (AUC) (95% confidence intervals) for the logit of the probability that a subject would be osteoporotic were: 0.814 (0.700, 0.928) for calcaneal BMD; 0.791 (0.673, 0.909) for BUA; 0.717 (0.588, 0.846) for VOS; and 0.793 (0.675, 0.911) for BUA and VOS combined. For the identification of osteoporotic or osteopenic subjects, the ROC areas were: 0.851 (0.774, 0.928) for calcaneal BMD; 0.773 (0.678, 0.868) for BUA; 0.783 (0.690, 0.877) for VOS; and 0.778 (0.685, 0.871) for BUA and VOS combined. Again, calcaneal BMD provided a higher, yet statistically insignificant, AUC than any QUS parameter. In conclusion, for the identification of subjects defined by World Health Organization criteria for axial BMD, the performance of BMD and QUS calcaneal parameters were statistically comparable. The choice of peripheral bone densitometry modality should therefore be made upon factors external to their discriminatory performance.