Enhanced Precision of the New Hologic Horizon Model Compared With the Old Discovery Model Is Less Evident When Fewer Vertebrae Are Included in the Analysis

The International Society for Clinical Densitometry guidelines recommend using locally derived precision data for spine bone mineral densities (BMDs), but do not specify whether data derived from L1 L4 spines correctly reflect the precision for spines reporting fewer than 4 vertebrae. Our experience suggested that the decrease in precision with successively fewer vertebrae is progressive as more vertebrae are excluded and that the precision for the newer Horizon Hologic model might be better than that for the previous model, and we sought to quantify. Precision studies were performed on Hologic densitometers by acquiring spine BMD in fast array mode twice on 30 patients, according to International Society for Clinical Densitometry guidelines. This was done 10 different times on various Discovery densitometers, and once on a Horizon densitometer. When 1 vertebral body was excluded from analysis, there was no significant deterioration in precision. When 2 vertebrae were excluded, there was a nonsignificant trend to poorer precision, and when 3 vertebrae were excluded, there was significantly worse precision. When 3 or 4 vertebrae were reported, the precision of the spine BMD measurement was significantly better on the Hologic Horizon than on the Discovery, but the difference in precision between densitometers narrowed and was no longer significant when 1 or 2 vertebrae were reported. The results suggest that (1) the measurement of in vivo spine BMD on the new Hologic Horizon densitometer is significantly more precise than on the older Discovery model; (2) the difference in precision between the Horizon and Discovery models decreases as fewer vertebrae are included; (3) the measurement of spine BMD is less precise as more vertebrae are excluded, but still quite reasonable even when only 1 vertebral body is included; and (4) when 3 vertebrae are reported, L1 L4 precision data can reasonably be used to report significance of changes in BMD. When 1 or 2 vertebrae are reported, precision data for 1 or 2 vertebrae, respectively, should be used, because the exclusion of 2-3 vertebrae significantly worsens precision.