Bone mineral density (BMD) measurement with dual-energy X-ray absorptiometry (DXA) guides initial diagnosis of osteoporosis, fracture risk assessment and determination of pharmacologic intervention.1 BMD monitoring in untreated and treated individuals is also widely used.2 Continued BMD loss that exceeds limits for measurement error is unexpected in individuals adherent to effective therapy and places such individuals at higher fracture risk, whereas a significant increase in BMD is associated with lower fracture risk than stable BMD.3 Pooled individual-level clinical trial data suggest that change in total hip BMD is a good surrogate marker for anti-fracture efficacy.4,5 Observational data from the clinical practice setting suggest that BMD monitoring in individuals initiating anti-osteoporosis treatment may be associated with lower fracture risk then when patients are not monitored.6

Given that short-term changes in BMD are usually small, strict attention to quality assurance procedures and standards is required to ensure that true changes can be distinguished from sources of measurement error.7 Random sources of error are usually addressed using test-retest measurements to estimate the 95 % least significant change (LSC) as described by the International Society for Clinical Densitometry (ISCD).2 Systematic sources of error can occur with the DXA scanner, including drift in calibration.8 Application of Shewhart chart rules provides an approach to identify problems with DXA scanner performance. These rules serve to identify major problems that require immediate attention. It is less clear how to approach less severe problems and their potential impact on clinical decision making. Sources have suggested BMD calibration tolerances in the range 0.5 % to 1.5 %, but have not set a formal cutoff.8, 9, 10 Theoretical predictions suggest that uncorrected equipment drift of 1.5 % can result in 18 % (or greater) misclassified patients, but are not based upon a direct assessment of misclassification in clinical patients.8 DXA manufacturers do not specify a tolerance, but propose daily quality control requirements. For example, the GE Healthcare calibration block consists of three chambers simulating BMD values of approximately 0.500 g/cm2, 1.000 g/cm2, and 1.500 g/cm2 and requires that each chamber is within 0.030 g/cm2 of its expected value.

The current study was performed to directly determine how varying degrees of simulated DXA calibration drift would affect misclassification of BMD change in routine clinical practice. Results are applied to a DXA scanner that was observed to be gradually drifting in order to inform corrective action.