Abstract
Dual energy xβray absorptiometry is a technique advocated for the measurement of bone mass throughout the skeleton, and recently it has been used to measure changes in periprosthetic bone mass after joint replacement. The accuracy and precision of the method in clinical patient populations have not been firmly established. This study sought to establish the shortβterm reproductibility of measurements made with dual energy xβray absorptiometry of multiple sites in a large sample of elderly patients with rheumatic disease. Reproductibility was assessed in the lumbar spine and in three femoral sites in 69 patients participating in a longitudinal clinical trial. In each patient, absorptiometry was performed twice in the same day at as many as five time points over a 2βyear period. The mean (Β±SD) baseline bone density was 0.783 Β± 0.128 g/cm^2^ for the femoral neck and 1.015 Β± 0.218 g/cm^2^ for the lumbar spine. The correlations between the duplicate baseline measurements of the spine were excellent (r = 0.9936, p < 0.001) and were stable over the 2βyear period; the mean difference between the duplicate baseline measurements was 1.82 Β± 1.54% and the mean coefficient of variation was 1.29%. Measurements in the femur were much less precise: these values were 3.61 Β± 3.14% and 2.55% in the femoral neck, 3.66 Β± 4.35% and 2.59% in the greater trochanter, and 5.28 Β± 5.61% and 3.73% in Ward's triangle. This study evaluated the shortβterm reproductibility of dual energy xβray absorptiometry in a clinical population. The reproductibility of a technique must be taken into account when it is applied longitudinally to follow patients over time. The reproductibility of bone densitometry of the spine in large clinical populations with orthopaedic and rheumatic disease was similar to that reported for in vitro measurements and for very small sample studies of healthy normal subjects. However, the reproductibility of densitometry of femoral sites was much poorer than that for spinal sites. Reproducibility is a significant methodological issue for the use of dual energy xβray absorptiometry to assess the effect of osteoporosis therapies on the hip as well as the change in periprosthetic bone mass around a hip prosthesis. The poorer reproductibility of the measurements in the femoral sites suggests that a much wider margin of error is required in the interpretation of withinβpatient change in bone mass over time.