Precision and accuracy of measuring changes in bone mineral density by dual-energy X-ray absorptiometry

Long-term precision of two Hologic DXA scanners was derived from repeated clinical measurements. With typical subjects, the long-term coefficients of variation were about twice the short-term. The accuracy of the measurement of changes was compromised by anomalies, but this did not seriously affect clinical conclusions.

Long-term precision and accuracy of BMD measurements need review.

Long-term precision was examined by selecting, from Hologic databases, subjects who had had four scans over a period of 2 to 5 years and was calculated from the SEE of a regression of BMD against time. Accuracy was assessed from relationships between changes (delta) in BA, BMD and BMC.

For one group of subjects, the long-term precision was 2.4% for lumbar spine, 2.3% for total hip and 2.7% for femoral neck when expressed as CVs. These values were nearly twice the short-term CVs of 1.3% for spine, 1.2% for total hip and 1.4% for femoral neck. For another group, a negative exponential regression gave a better fit, leading to CVs of 1.3% for the spine, 1.4% for total hip and 2.1% for femoral neck. Significant correlations between deltaBA and deltaBMC were found. These led to an underestimate of deltaBMD in spine and hip by 25%.

The poorer long-term precision for typical patients should be borne in mind in monitoring progress. The underestimate of changes could account for only a part of the underestimation by BMD measurements of the anti-fracture effects of anti-resorptive drugs.