Background High alcohol intake is a known risk factor for fractures.¹ This association is mediated by a combination of factors including muscle atrophy, unsteadiness and loss of bone mineral density (BMD). Chronic alcohol consumption is common, and predictors of fractures in these patients have not been the object of many studies.

Objectives We used data from a large observational cohort to determine predictors of fractures among chronic heavy drinkers.

Methods A dual X-ray absorptiometry (DEXA) database from a district general hospital in Northwest England was used to identify patients drinking more than the maximum recommended amount of alcohol. Patient demographics, presence of other risk factors and bone mineral density (BMD) at the femoral neck, total proximal femur and L1-L4 lumbar vertebrae were recorded. Univariate logistic regression models predicting fracture risk were fitted using traditional FRAX risk factors for osteoporosis. These include age at scan, female gender, total hip BMD (left and right), femoral neck BMD (left and right), L1-L4 BMD, presence or absence of smoking, rheumatoid arthritis, family history of fracture, steroid exposure and body mass index (BMI). Subsequently a multivariate model was fitted with a cut off value of p<0.05 for individual variables.

Results 1794 patients were referred between 2004–2014; of these, 737 (41.1%) had sustained a fracture. Median age at scan was 59.6 years, and 1221 (68.1%) were female. Results of univariate analysis are shown in Table 1.

In the multivariate model, age at scan, total hip BMD (left), lumbar spine BMD, BMI and steroid exposure were predictors of risk. As in the univariate model, steroid exposure in this cohort appears to reduce the risk of fractures.

Conclusions In the univariate analyses, bone mineral density measured at any site, age and smoking status were predictors. In the multivariate model, the strongest predictors were lumbar spine BMD, BMI and steroid exposure (p<0.001), with age and total left hip BMD also significant. In this population, lumbar spine BMD may be a better predictor than hip BMD used in the FRAX risk assessment tool. There may well be an as yet undetermined role of unmeasured confounders, and further research is needed to guide the development of a more accurate risk assessment tool in this population.

Steroid exposure reduces the risk of fractures in both models, which is difficult to explain. Steroid exposure indicates engagement with medical care, which may be less likely amongst the heaviest drinkers, thus creating confounding. Our data does not quantify alcohol consumed, and it is possible that those with significant steroid exposure consume less alcohol overall. Further work is needed to clarify the role of steroid exposure in fracture risk prediction in chronic drinkers.

1. Kanis J, Johansson H, Johnell O, Oden A, De Laet C, Eisman J et al. Alcohol intake as a risk factor for fracture. Osteoporosis International. 2004;16(7):737–742.

Disclosure of Interest None declared