e17 Predicting pelvic fractures using bone mineral density in the lumbar spine and femoral neck: a case-control study

Abstract

Background: Low bone mineral density (BMD) is classically associated with an increased risk of fracture. We have previously demonstrated that the femoral neck is a better predictor of fractures than the lumbar spine. It is not known, however, whether BMD from the lumbar spine or femoral neck is better at predicting fractures in the pelvis. Methods: Using a case-control study approach to investigate the likelihood BMD in the lumbar spine, femoral neck and femur had on predicting pelvic fractures in patients referred for BMD estimation using dual X-ray absorptiometry (DEXA). A cohort of patients attending for BMD estimation using DEXA in a district general hospital between 2004 and 2011 were identified. The patients who were referred with a pelvic fracture were identified. Suitable controls matched for age, gender and body mass index were selected from those patients who were referred without a fracture. The risk factors assessed were femoral neck BMD, total femur BMD and lumbar spine BMD (L1-L4). Comparisons between the fracture and non-fracture group were identified using the student's t-test. An ordinal logistic regression model was used to determine the best predictor of pelvic fracture. Results: 316 patients were referred with a pelvic fracture in the study period. Mean age was 67.2 years (SD 12.3). 280 patients (89%) were female. In the pelvic fracture cohort, males were found to have a significantly higher BMD in the right femoral neck (mean difference 0.09g/cm2, p value 0.005), left and right femurs (mean difference left 0.06g/cm2 p value 0.049, right mean difference 0.10g/cm2, p value 0.0072) and in the lumbar spine (mean difference 0.07g/cm2, p value 0.0374). BMD was significantly lower in the pelvic fracture cohort in the femoral neck (mean difference left and right 0.09g/cm2), left and right femurs (mean difference left 0.08g/cm2, right 0.1g/cm2) and the lumbar spine (mean difference 0.1g/cm2), all with a p value of<0.0001. Our Ordinal logistic regression model demonstrated that the BMD of left and right femoral necks are the strongest predictors of pelvic fracture (both p value <0.0000), followed by the BMD of left and right femurs (p value <0.00029 and <0.00004, respectively). The lumbar spine was deemed a non-significant predictor (p value 0.175). Conclusion: In this cohort of patients, we have demonstrated that the BMD of the left and right femoral necks are the strongest predictors of pelvic fracture. The BMD of left and right femurs are also statistically significant predictors of pelvic fracture. The lumbar spine was found not to be relevant. These results support our previous findings that the femoral neck is a better predictor of fractures than the lumbar spine. Validation for other fractures is required.