MON-799 - Race-Neutral FRAX with BMD Outperforms the Original FRAX in A Multiethnic Cohort with Diabetes

Background: The Fracture Risk Assessment Tool (FRAX) is commonly used in clinical practice. FRAX includes race and predicts lower risk of fracture by factors of 0.43-0.64 for Asian, Black, and Hispanic people compared to White people with otherwise equivalent risk factors. We have previously demonstrated that FRAX without Bone Mineral Density (BMD) substantially underestimates fracture risk in Black and Hispanic people with diabetes. However, it is not clear if these biases would also be present with FRAX with BMD, which is more commonly used in clinical practice. Thus, the purpose of this study was to examine the performance of FRAX with BMD in Black, Hispanic, and White patients with diabetes. We also sought to examine how removing the impact of race in FRAX would change performance.

Methods: This is a retrospective cohort study using routinely collected data from the electronic health record (EHR). Patients must have been seen for diabetes, on average, once per year during follow-up. Patients who had dual energy X-ray absorptiometry (DXA, GE Prodigy) were followed for incident major osteoporotic fractures (MOFs) using diagnosis codes. FRAX with BMD was calculated with inputs coming from the EHR for clinical risk factors and the DXA for BMD. We also calculated variations of FRAX, including using rheumatoid arthritis (RA) as a proxy for diabetes-related fracture risk (RA-FRAX) and the use of a race-neutral (rn-) FRAX that removes the impact of the race correction factors. The rn-FRAX was created using data from the 2022 census. FRAX calculations were adjusted for years of follow-up. Area under the curve (AUC) was used as a measure of discrimination (AUC of 1.0 representing perfect performance), and observed to predicted fracture ratios (O/P) were used as a measure of calibration (O/P of 1.0 representing perfect performance).

Results: Among 1,743 patients (83% female), there were 1336 Black, 109 Hispanic, and 298 White patients with mean 4.8 yrs ± 2.4 yrs follow-up. MOF rates were 4.5% in Black, 5.5% in Hispanic, and 7.1% in White patients. All versions of FRAX had similar discrimination (e.g. original FRAX AUC 0.67). In Black patients, the original FRAX underestimated fractures by two-fold (O/P 2.0), while RA-FRAX had O/P of 1.6. When using rn-FRAX, we found that O/P improved to 1.1; this became 0.8 when adding an adjustment for RA. In Hispanic patients, the original FRAX underestimated fractures by 80% (O/P 1.8), while RA-FRAX had O/P of 1.4. The rn-FRAX demonstrated O/P of 1.2, which became 1.0 when adding an adjustment for RA. In White patients, the original FRAX had O/P of 1.4, while RA-FRAX had an O/P of 1.1. The rn-FRAX had O/P of 1.6, which became 1.3 when adding an adjustment for RA.

Conclusion: We found that the addition of BMD did not alleviate biases in FRAX with 80-100% more observed fractures than predicted in Black and Hispanic patients with diabetes. A race-neutral version of FRAX performed better overall.

*Unless otherwise noted, all abstracts presented at ENDO must not be released to the press or the public until the date and time of presentation. For oral presentations, the abstracts are embargoed until the session begins. The Endocrine Society reserves the right to lift the embargo on specific abstracts that are selected for promotion prior to or during ENDO.*