Background Coeliac disease and other malabsorptive diseases have been associated with a low bone mineral density secondary to a variety of factors including calcium absorption. We have previously shown that there is a fivefold propensity for having Osteoporosis (OP) compared to controls with a predilection for spinal osteoporosis (1). However this does not translate completely to predicting fragility fractures. The FRAX™ tool uses the femoral neck to predict fractures on a population basis and ignores the lumbar spine.
Objectives We set out to determine the predictors of fragility fractures in a large observational cohort. Specifically we set out to determine if the lumbar spine bone mineral density (BMD) is a predictor for fragility fractures.
Methods Patients referred for bone mineral density (BMD) estimation in a scanner in the North west of England between 2004 and 2014 with a history of coeliac disease were identified from a dual X-ray absorptiometry (DEXA) database. Demographics and other risk factors were also recorded. Initially those that sustained a fracture were compared ot those that had not sustained a fracture using the chi squared test for categorical variables and Students T test for continuous variables. Fragility fractures were also recorded. Initially univariate logistic regression models were modelled predicting fractures in the cohort and a multivariate model was subsequently fitted. Variables included age at scan, gender, body mass index (BMI), family history of fracture, alcohol, smoking, glucocorticoid exposure, rheumatoid arthritis in addition to BMD in the lumbar spine (L1-L4) and the femoral neck.
Results 788 patients were scanned in the referral period. Mean age at scan was 55.4 year (SD 14.4), 576 (73.1%) were female, 159 (20.2%) had sustained a fracture. Patients who had sustained a fracture were more likely to be female 127/576 (22%) vs 32/212 (15%) (p=0.03), older 59.2 years (SD14) vs 54 years (SD 14) (p<0.001) and have concomitant rheumatoid arthritis 5/159 (3%) vs 6/629 (1%) (p=0.04). They also had lower BMD in both the lumbar spine and the femoral neck. Results of the univariate analysis are shown in table 1. Significant predictors are denoted with an asterisk (*).
In the multivariate model, the only variables that predicted fractures in this cohort were the BMD in the lumbar spine (OR 0.09 95%CI 0.03,0.25) and age at scan OR 1.19 (95%CI 1.04,1.37) per decade.
Conclusions In univariate analysis, many risk factors are associated with fracture in this cohort, but in multivariate analysis, the best predictor of fracture was the BMD in the lumbar spine. This is not included in the FRAX™ tool and would make estimating the fracture risk in this cohort difficult. Further modelling in other cohorts is needed to validate this finding.
Oldroyd et al. Ann Rheum Dis 2011;70(Suppl3):225.
Disclosure of Interest None declared