Background Low BMD taken from a single non-dominant hip is typically used to predict fracture; especially in FRAX. However, the average BMD from both hips could be used to predict fracture. No studies to date have questioned whether an average value from both hips predicts fracture better than using the lower single hip value. It is therefore important to consider whether combined hip BMD is superior at predicting fracture compared to the lowest BMD in a single hip. This may be important for clinical practice.
Objectives To discover whether low BMD in a single hip predicts fracture better than the aggregate average BMD of both hips.
Methods Dual X-ray Absorptiometry scan results of patients referred to a DGH between 2004 and 2010 were used. The following were recorded: age, sex, BMI and BMD. Patients were divided into a fracture and non-fracture group. Baseline characteristics were compared using students T-test for continuous and chi2 test for categorical variables. Logistic models were fitted to determine the predictors of fracture, one using the non-dominant hip, one using the dominant hip and one using the mean BMD of both hips. Fit of the model was assessed by fitting receiver operating characteristic (ROC) curves
Results 14,584 patients were included in the study and 27% had sustained a fracture. Baseline demographics are summarised in Table 1. Linear regression analysis comparing left and right mean hip BMD showed coefficient of 0.948 with 95% confidence interval 0.943-0.953 (p<0.05). Logistic regression analysis is summarized in Table 2.
Conclusions There is little difference between using the lowest, highest or combined hip BMD to predict fractures. When the total left hip BMD is lowest, this predicts fracture just as well as the higher right or total combined (OR 11.86 vs 11.77 vs 11.49 respectively). A similar pattern is found using the right total hip BMD as lowest (OR 10.72 vs 12.81 vs 12.63 respectively). Linear regression analysis shows that there is a tendency for patients to have similar BMD in both hips; therefore low BMD on one side means it is likely there will be low BMD in the contralateral hip. This may partially explain some of the findings. This is the first time this has been demonstrated, further studies are required to validate this adjusting for potential confounders.
Disclosure of Interest None declared