Precision of bone densitometry measurements: when is change true change and does it vary across bone density values?

The precision error of the bone densitometer is used to interpret significant change in bone mineral density (BMD) in serial studies. The precision error can be expressed as standard deviation (SD) or coefficient of variation (CV). The aims of this study are to determine the precision error over a range of BMD values and to demonstrate the application of the precision error in clinical practice. A bone phantom was used consisting of a perspex block with eight compartments containing varying amounts of hydroxyapatite powder to simulate a range of bone densities. The block was scanned 21 times and manual regions placed over each compartment to measure the BMD in each compartment. There were no significant differences in the variances or SD for all eight compartments, that is, over the range of BMD normally encountered in clinical practice. However, the calculated CV show a progressive fall in values as the BMD rises. Therefore, the SD should be used to calculate significant BMD change. In a practise with quality control procedures in place to detect calibration drift and with appropriately trained personnel, a change of approximately 0.05 g/cm2 is generally regarded as being a significant change at a 95% confidence level.