Validation of FRAX without BMD: An age-related analysis of the Fifth Korean National Health and Nutrition Examination Survey (KNHANES V-1, 2010)

- We compared FRAX fracture probabilities calculated with and without BMD data.
- There was strong correlation between FRAX probabilities calculated with and without BMD.
- The difference of average 10-year predicted fracture probability varied according to age and sex.

Although the Fracture Risk Assessment Tool (FRAX) is widely used to evaluate probabilities of fractures, there is no consensus regarding whether it is accurate when bone mineral density (BMD) is not included. This cross-sectional study aimed to compare the 10-year predicted fracture probabilities calculated using FRAX with and without BMD. Data were collected from the 2010 Fifth Korean National Health and Nutrition Examination Survey, and 2706 participants (1260 men and 1446 women) aged 50-90 years were analyzed. Ten-year predicted probabilities for major osteoporotic and hip fractures were calculated using the FRAX model. In men, the 10-year probabilities without BMD were 3.9 ± 1.8% and 1.3 ± 1.4% for major osteoporotic and hip fractures, respectively. In women, the 10-year probabilities without BMD were 7.7 ± 4.4% and 2.6 ± 2.9% for major osteoporotic and hip fractures, respectively. These probabilities were significantly correlated with the probabilities calculated using FRAX with BMD (all, p < 0.001). When participants were divided into 10-year age groups and compared with the 10-year predicted fracture probability with BMD, the 10-year predicted fracture probability without BMD was lower in men 50-59 years old, similar to men 60-69 years old, and higher in men ≥ 70 years old. The FRAX scores without BMD were generally lower for all women. The FRAX model without BMD appears to be a slightly lower fracture probability compared to that calculated with BMD, especially in younger participants. Although these results have important clinical implications for areas with limited ability to evaluate BMD, they must be confirmed by a large prospective study.