Dual X-ray absorptiometry of the proximal femur: normal European values standardized with the European Spine Phantom.

Measurements of the proximal femur by dual X-ray absorptiometry (DXA) are assuming increasing importance in clinical and epidemiological studies. However, different DXA manufacturers have adopted varying approaches to measuring the femoral neck and trochanter regions with the result that there is as yet no agreement on normal ranges and how to cross-calibrate between different machines. In this Concerted Action (of the EU's 2nd Framework Programme), 12 clinical centers in eight countries recruited a total of 855 female and 517 male Caucasian subjects who were judged to be free of conditions known to cause secondary osteoporosis and who had not been treated with drugs known to affect bone mass. The DXA machines used were made by Hologic, Lunar, and Norland. Each machine was cross-calibrated with the European Spine Phantom prototype designed by Kalender. All femoral neck and trochanter results were expressed as standardized values and were first examined for possible bias due to between-brand differences in choice of areas for measurement. On average, trochanter areas were similar between brands but Norland chose a narrower neck region for analysis. Bone mineral density (BMA or BMD g/cm2) was log-normally distributed for all groups at the femoral neck but normally distributed at the trochanter in men and postmenopausal women. There were substantial between-center differences after cross-calibration in age-adjusted mean values at both sites and also in the rates of apparent bone loss with age and the amounts of scatter within populations about mean age-adjusted values. With the proviso that locally derived normal ranges would allow more accurate comparisons for clinical purposes, European normal ranges were derived for both neck and trochanter. Data obtained with the three individual brands of machine fitted these overall ranges well. Bone loss was apparent in premenopausal women at the femoral neck but not the trochanter. Postmenopausal women's data was somewhat better fitted by reference to years since menopause than to chronological age. This work provides a basis for conducting future epidemiological and clinical studies with more than one brand of machine as long as each individual subject, if measured consecutively, is measured on the same machine throughout.