New equation with five variables helps predict hip fracture risk in type 2 diabetes
Age, sex, body mass index, peripheral sensory neuropathy, and estimated glomerular filtration rate were the included variables.
A new equation uses commonly available variables to help estimate risk for hip fracture in patients with type 2 diabetes.
Researchers in Australia used data on 1,251 participants from the Fremantle Diabetes Study Phase I (FDS1) to develop the equation. The mean patient age was 65 years, and 48.8% were men. Median diabetes duration was 4.0 years; median HbA1c level was 7.4%. Patients were followed until first incident hip fracture, death, or 10-year census, with a total follow-up time of 10,306 person-years. The study results were published Nov. 19 by Diabetes Care.
Fifty patients (4.0%) had at least one hip fracture during follow-up, and 415 (33.2%) died, 37 after incident hip fracture. Of these, seven had “hip fracture” listed on their death certificates. The researchers compared characteristics of patients with hip fracture and those without and found that older age, lower body mass index, peripheral sensory neuropathy, and an estimated glomerular filtration rate below 45 mL/min/1.73 m2 were independent risk factors, while male sex appeared to have a protective effect. These five variables were included in the final risk equation. The equation predicted a mean 10-year risk of 3.3% for incident hip fracture versus an observed risk of 4.0%. For a cutoff of 3% for 10-year predicted incident hip fracture risk, the equation had a sensitivity of 76.0%, a specificity of 71.9%, a positive predictive value of 10.1%, and a negative predictive value of 98.6%.
The risk equation was validated in 286 patients with type 2 diabetes from the Busselton Health Study. The validation cohort and the development cohort were similar in age and sex distribution, but body mass index, fasting serum glucose levels, and estimated glomerular filtration rate were all significantly lower in the former than in the latter. Peripheral sensory neuropathy was not measured in the validation cohort. Over 10 years of follow-up, six patients (2.1%) in the validation cohort had an incident hip fracture, and the FDS1 equation would have predicted a risk of 2.4%, assuming no patients had peripheral sensory neuropathy. Using the same 3% cutoff as in the development cohort, sensitivity was 66.7%, specificity was 80.0%, positive predictive value was 6.7%, and negative predictive value was 99.1%. The authors also used the FDS1 cohort to test the ability of the QFracture and FRAX equations to predict hip fracture risk in patients with type 2 diabetes and found that while QFracture (which includes 25 variables) performed well, FRAX substantially underestimated risk.
The authors noted that missing data may have biased their results, that bone mineral density was not measured in all FDS1 participants at baseline, and that the development and validation samples were relative small, among other limitations. However, they concluded that the FDS1 risk calculator will be useful in clinical practice because it is the only such tool that has been developed and validated specifically in this population. “It can be readily implemented as part of the routine assessment of the person with type 2 diabetes,” the authors wrote. “It could be easily incorporated in future studies of fractures complicating type 2 diabetes with a view to potentially wider validation and implementation.”