How Good is Our Best Guess? Clinical Application of the WHO FRAX Tool in Osteoporotic Fracture Risk Determination and Treatment Decisions.

Historically, treatment decisions for osteoporosis were based on bone mineral density. However, many fractures occur in patients with T-scores outside the osteoporotic range, emphasizing the importance of multi-factorial risk assessments. The World Health Organization Fracture Risk Assessment Tool (FRAX) predicts 10-year risk of osteoporotic fracture. We hypothesized that physicians' clinical estimates of osteoporotic fracture risk would differ significantly from that calculated by FRAX. Thus, treatment decisions would differ depending whether or not physicians used FRAX. A survey consisting of five clinical scenarios was administered to 76 endocrinologists, family physicians, internists, and internal medicine residents. They were asked to estimate the osteoporotic fracture risk and decide whether they would offer preventative treatment. Their estimates were compared to the risk predicted by FRAX and national treatment threshold guidelines. The primary outcome was the difference between the participant's estimate and the FRAX-based estimate of the 10-year risk of osteoporotic fracture for each scenario. In each scenario, physicians statistically significantly over-estimated fracture risk compared to that predicted by FRAX. Estimates for hip fracture risk were 2-4 times higher than FRAX estimates. The major osteoporotic fracture risk at which participants would offer treatment varied with physician group, with endocrinologists, family physicians, and residents requiring a 10-20 % 10-year risk, while internal medicine physician thresholds ranged from 2 to 20 %. Physicians greatly over-estimated the risk of hip fracture based on clinical information. FRAX is necessary to accurately quantify risk, but because physicians varied in the level of risk required before they would offer treatment, uniform approaches to risk estimation may still not result in uniform clinical treatment decisions.