Bone mineral density measures in longitudinal studies: the choice of phantom is crucial for quality assessment. The Tromsø study, a population-based study.

Determination of change in bone mineral density (BMD) requires high-precision densitometry techniques. The purpose of the study is to investigate to what degree different densitometer phantoms reflect observed changes in human BMD and to investigate to what degree fluctuations in densitometers' measurement level influence bone loss estimates. Densitometer influence was assessed using the aluminum forearm phantom (AFP) provided by the manufacturer, the European forearm phantom (EFP) of semi-anthropomorphic calcium-hydroxyapatite, and repeated population measurements on different densitometer combinations. The mean follow-up time was 6.4 years (SD 0.6). Measured population bone loss varied from 4.6%/year to 3.2%/year, depending on densitometer combinations. These variations could not be explained by differences in sex, age, height, weight and baseline BMD. They were predicted by EFP measurements, but not AFP measurements. The EFP measurements indicate that X-ray tube replacement changed the densitometers' measurement level in one of three instances, whereas "wear and tear" did not. We used the EFP data for adjustment of the densitometers' measurement levels. After adjustment, the overall crude bone loss was reduced from 4.14% to 3.92%. Mean annual loss was reduced from 0.64% or 0.61%. We conclude that densitometer performance might influence the accuracy of bone loss estimates. Changes in performance are not detected by aluminum phantoms. Quality control of BMD measurements in longitudinal studies should be performed with anthropomorphic calcium-hydroxyapatite phantoms in order to detect possible differences between the participating densitometers' measurement levels.