Vertebral fracture risk and impact of database selection on identifying elderly Lebanese with osteoporosis.

The International Osteoporosis Foundation recommends using a universal database i.e. the NHANES database for the diagnosis of osteoporosis. Population-based databases for T-score calculation are still debated in terms of clinical and public health relevance. The current study aimed at estimating the prevalence of vertebral fractures in the Lebanese elderly, determining BMD-fracture relationship, and assessing the effect of database selection on osteoporosis prevalence and fracture risk assessment. Apparently healthy subjects were randomly selected from the Greater Beirut area - one-third of the Lebanese population at large - using a multilevel cluster technique. Subjects with medical conditions likely to affect bone metabolism i.e. history of major chronic disease, intake of medications that affect bone metabolism were excluded. Presence of vertebral fracture was estimated by a semi-quantitative assessment. Bone density was measured by central DXA. Clinical risk factors included age, gender, height, weight, body mass index, smoking, exercise, falls, previous fragility fracture and family history of fragility fracture. Impact of database selection was assessed by: (1) Comparison of sensitivity and specificity for prevalent vertebral fractures of the T-score <or= - 2.5 threshold using local versus NHANES database. (2) Comparison of estimates for fracture risk (RR/SD decrease in BMD) using local versus NHANES database. Prevalence of vertebral fractures was estimated at 19.9% [15.4-25.0] in women and at 12.0% [7.3-18.3] in men. Prevalence of osteoporosis by DXA using total hip was 33.0% [27.5-38.8] in women and 22.7% [16.2-30.2] in men. The NHANES database provided higher sensitivity for vertebral fracture than our population-specific database. RR of vertebral fracture per SD decrease in BMD remained unchanged across the two databases. In women, RR/SD were 1.61 [1.17-2.23] and 1.49 [1.14-1.95] in the NHANES and the local database, respectively, and in men 1.59 [0.94-2.72] and 1.43 [0.95-2.16]. In conclusion, our findings were in concordance with the IOF recommendations for the use of a universal database and could be used for the implementation of a unified fracture risk assessment paradigm along with the WHO initiative.