Calibration and standardization of bone mineral densitometers

Measurements of bone mineral density (BMD) by four 153Gd and two x-ray bone densitometers were compared utilizing spine phantoms that simulated the human lumbar spine. The six instruments provided BMD values that differed by as much as 16%, due to differences as large as 8% in bone mineral content and as large as 7% in bone area. Instrument calibration curves, determined by measuring thin, medium, and thick hydroxyapatite blocks, were linear (r = 0.99) but had different (p less than 0.0001) slopes and intercepts. Serial measurements of spine and total body phantoms were employed to evaluate the long-term stability of 153Gd bone densitometry. These measurements of spine phantom BMD increased 1.0-2.6% (p less than 0.0001) following a software change, and measurements of total body bone density increased 4-6% after 153Gd source replacement. The changes occurred at a time when serial measurements of cylindrical calibration standards were stable, indicating that such simple standards are unable to detect and correct for changes in instrumental response. We conclude that investigators, manufacturers, and government regulatory agencies must develop and implement the following: (1) effective calibration procedures that would assure comparability among instruments, and (2) appropriate quality control phantoms that would allow the confident interpretation of serial patient measurements.