The use of the FRAX® tool and its adjustments in women living with diabetes: a cohort study in primary care in Brazil

The FRAX® algorithm showed good accuracy in women living with DM followed in primary care. There were no differences between the ROC curve with and without adjustments for major and hip fractures. The FRAX® 10-year and FRAX® AR were better calibrated in this population.

Purpose: An increased risk of fractures in people living with diabetes has been described. Screening instruments to calculate this risk have been proposed, including the FRAX® algorithm. Some studies suggest that minor modifications to this instrument can improve its performance. These modifications work well in other countries, but we do not know if they work in Brazil. The objective of our study was to evaluate the performance of the FRAX® algorithm with and without adjustments for women living with DM (WLDM) in primary care in Brazil.

Methods: A cohort study that included post-menopausal women attending primary care in Santa Maria, Brazil, was conducted from 2013 to 2018. The risk for major and hip fractures was calculated using the FRAX® tool. The FRAX® risk was calculated: (1) without adjustments (unadjusted FRAX®); (2) increasing the entered age by 10 years in individuals with DM (FRAX® 10 years); and (3) inserting the diagnosis of DM as rheumatoid arthritis (FRAX® AR).

Results: The accuracy for major fracture was 0.948 (unadjusted FRAX®), 0.947 (FRAX® 10 years), and 0.946 (FRAX® AR). For hip fractures, the accuracies were 0.989 (unadjusted FRAX®), 0.988 (FRAX® 10 years), and 0.988 (FRAX® AR). Furthermore, there were no differences between the area under the ROC curve with and without adjustments for major and hip fractures. Conversely, the FRAX® 10 years and the FRAX® AR were better calibrated, presenting a lower Chi-square.

Conclusion: The FRAX® algorithm showed good accuracy in WLDM followed in primary care. The FRAX® 10 years and FRAX® AR were better calibrated in this population.