Inconsistency in filling in the bottom of the spine bone map affects reported spine bone mineral density

We hypothesized that variability from year to year in how much of the bone map was filled in at the bottom of the spine region of interest (ROI) contributes substantially to variability in measurement of spine bone mineral density (BMD).

A total of 110 spine BMDs with defects in the bone mapping at the bottom were reanalyzed, with the only change being manually drawing a straight line across the bottom of the ROI and filling in the bone map.

The mean (SD) change in area, bone mineral content, and BMD for total spine when the bottom of the bone map was filled in was 0.919 (0.411) cm2, 0.201 (0.121) g, and -0.0098 (0.0043) g/cm2, respectively, and all changes were significant (P<.0001). The largest individual change in total spine BMD with reanalysis was 0.0238 g/cm2, close to the least significant change (LSC) of 0.026 g/cm2 in our center. To quantify variability due to this change in analysis, we calculated an LSC(fill), in which the pairs of scans consisted of the same scan before and after filling in the bottom of the spine bone map, without any other change. The LSC(fill) attributable just to the reanalysis of missing bone map at the bottom of the spine was 0.021 g/cm2, suggesting substantial variance due to variability in mapping the bottom of the spine.

When there is a noticeable defect in the bottom of the spine bone map, filling this defect in consistently eliminates a significant source of variability in analysis of spine BMDs and might allow us to achieve smaller LSCs.