ABSTRACT
Objective:To evaluate and compare the clinical value of unadjusted fracture risk assessment tool(FRAX) and adjusted FRAX in predicting the risk of hip fracture in patients with type 2 diabetes(T2DM).Methods:In this 10-year retrospective cohort study, 1 730 patients with T2DM were collected from August 2009 to July 2013. The 10-year risk of hip fracture was calculated using the China FRAX model. Hip fracture events during the follow-up period were collected through electronic medical records and telephone interviews. The value of FRAX and adjusted FRAX in predicting the risk of hip fracture in T2DM patients was evaluated from two aspects of discrimination and calibration. Cox regression model was used to investigate the relationship between diabetes related factors and hip fracture.Results:A total of 39 participants(2.3%) experienced hip fracture during a median follow-up of 10 years. The area under the curve of unadjusted FRAX was 0.760, but the calibration ability was poor [calibration χ2: 75.78, P<0.001; calibration ratio(observation/prediction): 3.97(95% CI 2.76~5.17)]. There was no significant improvement in calibration ability of adjusted FRAX. After adjustment for unadjusted or adjusted hip fracture probability calculated by FRAX(FRAX-HF), duration, estimated glomerular filtration rate, insulin use, cerebrovascular diseases, and diabetic peripheral neuropathy were significantly associated with an increased risk of hip fracture( P<0.05). Conclusion:The FRAX tool significantly underestimated the risk of hip fracture in T2DM patients, and there was still significantly underestimation after adjustment due to the failure to eliminate the influence of diabetes-related factors such as disease duration and peripheral neuropathy.