A comparison of single photon and dual X-ray absorptiometry of the forearm in children and adults

Abstract

We compared single photon absorptiometry (SPA) to dual x-ray absorptiometry (DXA) for determination of bone mineral content (BMC), bone mineral density (BMD), and bone width (BW) of the forearm. The SPA and DXA measurements were done on the same subjects, using Lunar densitometers. The measurements were performed over the proximal radius (1/3 shaft) of the nondominant arm in 285 healthy, Caucasian females and males, ages 9–53. Correlation, linear, and split regression analyses for all subjects, and for subgroups (adults und children), were performed to compare SPA and DXA measurements. Corresponding measurements performed on two densitometers were highly correlated: r = 0.987, 0.975, and 0.943 for BMC, BMD, and BW, respectively. The corresponding measurements were also very similar in value, ranging from 0.9% to 4.1% difference, although they were different statistically. Correlations dropped slightly when subjects were separated into adult and children subgroups, and therefore, split regression analysis was performed resulting in R 2 (adjusted) values of 97.6%, 95.5%, and 89.0% for BMC, BMD, and BW, respectively. Because the group indicator was statistically significant ( p < 0.001) only for the BMC measurements but not for BMD and BW, linear regression of the whole sample was done as well. The difference in fitted values between the two regression methods was insignificant; therefore, we concluded that linear regression was sufficient for description of the relationship between SPA and DXA measurements. The precision study showed that the DXA had better reproducibility than SPA. The DXA precision in vivo (CV%) for BMC, BMD, and BW was 1.06, 0.83, and 0.95, respectively; and the SPA precision for same variables was 2.08, 2.12, and 0.95, respectively. Based on this comparison, SPA of the forearm can be replaced with DXA measurements, and the results from SPA can be extrapolated to DXA equivalents, using linear regression equations for either young or adult subjects when groups are concerned.