Endocrine Abstracts

Aims and objectives: This study was undertaken to look at the agreement between the hologic database (HD) based on NHANES bone mineral density (BMD) data in Caucasians and the ICMR database (ICMRD) published in 2010 in defining normal and subnormal BMD in subjects with or without hip fracture and to arrive at a BMD cut off which has a high sensitivity of predicting fracture.

Materials and methods: A cross sectional study of 3098 subjects (men-341 and women-2757) (mean age+S.D.=60.1+7.6 years), which included 314 subjects with low impact hip fracture, 2321 from the hospital database and healthy postmenopausal women (n=461) from community who underwent (DXA) scanning between 2007 and 2012. Recalculated T-scores from ICMRD were used for the diagnosis of osteoporosis and compared with HD. The threshold of hip BMD which best predicts the fracture risk was assessed using ROC curve.

Results: Almost a perfect agreement existed between two databases for the diagnosis of osteoporosis at the hip (Kappa-0.82, P<0.0001) in overall subjects and substantial in subjects with hip fracture (Kappa-0.65, P<0.0001). Seventy-three of 314 (23.5%) defined as osteoporosis according to HD were classified as osteopenia according to ICMRD. The ROC derived BMD cut-off of 0.681 gm/cm² had a 82% sensitivity and a 71% specificity to detect a fracture which corresponded with a T-score of −2.1 using a HD and −2.0 with ICMRD using hospital database as controls. When healthy postmenopausal women from community were taken as controls, BMD of 0.681 gm/cm² had 82% sensitivity and 71% specificity in predicting fracture risk which corresponded to a T-score of −2.1 using a HD and −2.0 with ICMRD.

Conclusions: The threshold of hip BMD T-score for treating osteoporosis may have to be redefined if the ICMR reference database is used. This study also highlights the importance of assessing risk factors other than BMD affecting bone health for making therapeutic decisions.