Standardization of BMD T-Scores in the first five years after the menopause: do femoral neck-equivalent and older normative range T-Scores improve diagnostic agreement?

Calculating T-scores using an older reference population reduces inconsistency between measurement sites when osteoporosis is diagnosed in the elderly. The present analysis in a younger, early postmenopausal cohort examined 5-yr consistency of normalization by local and femoral neck-equivalent T-scores. NHANES (femur) and Hologic (spine and forearm) references were applied to baseline, 1-, 2-, 3-, and 5-yr scans in 925 untreated women in a national cohort study, and alternative local and neck-equivalent scores calculated. The baseline prevalence of osteopenia/osteoporosis was 35.5%/4.1% (spine), 31.0%/1.2% (neck), 31.3%/1.2% (total hip), and 37.2%/2.5% (forearm). It increased to 54.6%/7% by combining sites. The prevalences at 5-yr were 57.2%/12.4% (spine), 51.9%/5.0% (neck), 46.6%/3.7% (total hip), 52.5%/7.4% (forearm), and 77.3%/17.8% (any). A T-score cut-off at the lowest of four sites of -1.65 for osteopenia and -3.37 for osteoporosis was equivalent in patient numbers to T<-1 and T<-2.5 at the femoral neck. The proportion of inconsistently classified subjects decreased from 48% to 42% (p<0.05) with neck-equivalent scores. No improvement remained after 5 yr. Kappa scores did not improve by the use of local or femoral neck scores. In conclusion, adjusted thresholds cannot remove the anatomic discrepancy between T-scores. To overcome this problem, risk-based diagnostic cut-offs must therefore be calculated separately for each measurement site and fracture localization.