Apparent “Rapid Loss” After Short-Interval Bone Density Testing in Menopausal Women Is Usually a Measurement Artifact

Abstract

Abstract Context Medication may be considered when bone mineral density (BMD) loss is reported as “excessive.” Objective We hypothesized that the rate of BMD change between 2 serial tests demonstrates higher random variability at shorter vs longer intervals, misclassifying some women as “rapid losers.” Methods This retrospective observational cohort study in Manitoba, Canada included women aged > 55 years without osteoporosis medications or glucocorticoids. Using paired baseline (1998-2016) and repeat (2001-2018) BMD measurements, we estimated the distribution of annualized change (first to second BMD) at spine, hip, and femoral neck stratified by testing interval (2-2.9, 3-3.9,...9-9.9, ≥ 10.0 years). “Rapid annual bone loss” was defined as exceeding the 95th percentile for decreases from all measurement pairs. Odds ratios (OR) for rapid loss were estimated using regression models adjusted for age and clinical covariates. Results From 7126 paired BMD measurements, mean annualized change was constant yet standard deviations in BMD change were > 2-fold greater with intervals of 2 to 2.9 years vs ≥ 10 years(P < 0.001). “Rapid annual loss” was seen in ~10% of short-interval tests vs < 1% of long-interval tests. ORs for “rapid loss” progressively declined with increasing testing interval (spine 15.3 [4.8-48.9], total hip 9.3 [4.4-19.5], femoral neck 18.7 [6.8-51.3] for a 2- to 2.9-year testing interval; referent ≥ 10 years). Conclusion There is a wider apparent range in annualized BMD loss with short-interval testing which greatly attenuates over longer intervals. BMD reports of “rapid loss” across shorter testing intervals likely reflect an artifact of BMD measurement error and should not be used as an indication for antifracture medication initiation.