Instrument performance in bone density testing at five Australian centres.

AIMS To assess the in vitro precision and accuracy of bone mineral densitometry (BMD) within and between locations at five Australian centres. METHODS Using a multicentre reliability study the accuracy and short- and long-term precision of dual-energy X-ray absorptiometry (DXA) in vitro was compared on five instruments. Measures were performed using pencil beam mode on four Hologic QDR-2000 densitometers and one Hologic QDR-1000/W (Hologic Inc, Waltham, MA). RESULTS Short-term precision of bone mineral density measurement was less than 0.5% for spine phantoms (n = 10 for each centre, mean intrasite coefficient of variation [CV] 0.39 +/- 0.09% [SD]) and for hip phantoms (n = 10 for each centre, mean intrasite coefficient of variation [CV] 0.34 +/- 0.10% [SD]). Between-centre measurement (n = 10 for each phantom) of a single spine phantom and a single hip phantom (specified mineral contents-58.5 g and 38.6 g, respectively) revealed ranges of bone mineral content of 57.7-58.1 g (all-point CV = 0.52%) and 37.1-37.8 g (all-point CV = 0.70%), respectively. When results from pairs of machines were compared there were statistically different mean BMD results for the majority of the ten possible pairings for both spine and hip measurements. Each study centre measured in vitro stability of phantom BMD measurements over a one year period (n = 45-283, median 157 for spine; and n = 0-262, median 38, for hip); CVs ranged from 0.38 to 0.53% for the spine measurements and from 0.38 to 0.54% for the hip measurements. The mean all-point accuracy of the spine phantom measurements was 99.1% and the hip phantom measurements was 96.7%. CONCLUSIONS Across a number of instruments DXA demonstrates in vitro all-point precision of 0.5% for the spine phantom and 0.7% for the hip phantom. The instrument demonstrates accuracy of greater than 99% at the spine and 96% at the hip. This finding has clinical, research and quality control implications.