QCT permits a direct measure of bone and muscle size and assessment of bone/muscle relationship. We have studied cross-sectional QCT variables in mid-thigh as predictors of incidental limb fractures in the AGES-Reykjavik Study, a cohort of 66–96-year-old men (n=2160) and women (n=2385) drawn from an established population based cohort and not taking medications affecting bone metabolism. We used 4-detector Siemens CT system, a single axial section through the right mid-thigh (10 mm slice thickness). The variables included in the Cox’s proportional hazard model were; total cross-sectional cortical area (CSA), derived cortical thickness, shaft BMD, shaft bending strength index (BSI), medullary area and buckling ratio, total cross-sectional muscle and quadriceps area and bone/muscle area ratio. All low trauma limb fractures (including proximal end of femur but excluding toes, foot, hand and finger fractures) during mean 3.5 years of follow-up were validated by medical and radiological records, altogether 170 in women and 61 in men, including 87 hip fractures.

Results: with sex as a confounding variable the most significant risk factor was the buckling ratio (ratio of bone radius to cortical thickness) with more than twofold greater risk in the top compared to the lowest tertile. This variable was mostly independent of muscle area which was however a significant protective factor independent of bone variables. Bone/muscle area ratio and BSI were not significant in multivariate analysis. The area under the ROC curve, using these QCT predictors and age, was 0.70 (CI 0.66–0.73).

We conclude that cortical instability associated with decreasing cortical thickness due to medullary expansion is a significant risk factor for limb fractures including hip fractures in old age. Further knowledge on factors affecting buckling ratio might be of importance in the prevention of these fractures in the elderly.