Background: Fracture rates are increased in HIV+ individuals. FRAX is a web-based fracture risk calculator used with or without bone mineral density (BMD) that estimates absolute 10-year risk of major osteoporotic (hip, spine, forearm, shoulder) and hip fracture. It is widely used for decision making in screening and treatment for fracture prevention. Since FRAX may underestimate fracture risk in HIV+ individuals; some experts recommend adding ‘secondary osteoporosis’ as a surrogate for HIV infection when using FRAX in an HIV+ patient.
Methods: From the Veterans Aging Study Virtual Cohort (VACS-VC), 26,037 HIV+/HIV- 50-70 year-old men were selected for whom complete data were available in 2000 to approximate all but 2 factors for FRAX calculation without BMD (history of secondary osteoporosis and parental hip fracture). Sum of estimated rates by this modified FRAX calculation was compared to observed 10-year fracture rates at similar body sites. Calibration (agreement between observed outcomes and predictions) by observed to estimated ratios (O/E) and discrimination (ability to discriminate subjects with and without outcome) by area under the curve (AUC) were compared by HIV statusResults: In 2000, HIV+ men were similar in age (56±5 vs 56±5, p=0.11) to HIV- men, but had lower weight (79±15 vs 89±16 kg, p<0.01), were more likely to report previous fracture, alcohol and glucocorticoid use, and less likely to smoke or have rheumatoid arthritis. More fractures occurred in HIV+ than HIV- men at major osteoporosis sites (4.61 vs 3.50%, p<0.01) and hip (1.32 vs 0.85%, p<0.01). Estimated rates by the modified FRAX were also higher for HIV+ than HIV- men at major osteoporosis sites (2.85 ±1.5 vs 2.71±1.4%, p<0.01) and hip (0.29 ±0.4 vs 0.24±0.3%, p<0.01). Calibration for major osteoporotic fracture was worse in HIV+ than HIV- men (1.62 vs 1.29, p=0.028). Discrimination for modified FRAX was poor, but did not differ by HIV status (Table). Adding ‘secondary osteoporosis’ as a surrogate for HIV-infection risk in FRAX calculation improved calibration to level of HIV- men (O/E=1.20), but did not improve discrimination (AUC=0.600).
Conclusions: In this study of older men, the modified FRAX score underestimates fracture rates more in HIV+ than HIV- men, and does not discriminate well between those at risk and not at risk for future fracture. Including ‘secondary osteoporosis’ as a surrogate risk factor may improve the performance of the modified FRAX in HIV+ patients.