Quantitative ultrasound measurements of bone: measurement error, discordance, and their effects on longitudinal studies.

The performance of quantitative ultrasound (QUS) for the assessment of osteoporotic fracture must be fully defined if it is to be used in the clinical setting. We have examined the precision and reproducibility of two ultrasound instruments in 1,267 institutionalized elderly men and women (mean age 82.1 and 85.9 years, respectively). Measurement of BUA and VOS was reliable, least significant change for BUA being 2.4% and for VOS 0.3% using the CUBA McCue instrument and 2.7% for BUA using the Metra QUS-2. Importantly, age was not found to influence the precision of either parameter. Comparison of the instruments showed the QUS-2 measures BUA higher than the CUBA by 1.9 dB/MHz (+/-9.55 dB/MHz). Moreover classification of individuals as osteoporotic by machine-derived T-score was different between instruments (kappa = 0.44 in men and kappa = 0.62 in women). Fifty-six residents were remeasured by CUBA after 2.2 years; BUA was observed to decrease by 5.2% (+/-16.5%, p = 0.02) without significant change in VOS, suggesting BUA is more useful in longitudinal studies. In conclusion, ultrasound measurements were highly reliable and sensitive to longitudinal change even in the very elderly. However, diagnosis by T-score criteria was instrument specific, and such criteria should be interpreted with caution in the assessment of fracture risk.