Risk factors for hip fracture in men from Southern Europe: The MEDOS study

The aims of this study were to identify risk factors for hip fracture in me n aged 50 years or more. We identified 730 men with hip fracture from 14 ce nters from Portugal, Spain, France, Italy, Greece and Turkey during the cou rse of a prospective study of hip fracture incidence and 1132 age-stratifie d controls selected from the neighborhood or population registers. The ques tionnaire examined aspects of work, physical activity past and present, dis eases and drugs, height, weight, indices of co-morbidity and consumption of tobacco, alcohol, calcium, coffee and tea. Significant risk factors identi fied by univariate analysis included low body mass index (BMI), low sunligh t exposure, a low degree of recreational physical activity, low consumption of milk and cheese, and a poor mental score. Co-morbidity including sleep disturbances, loss of weight, impaired mental status and poor appetite were also significant risk factors. Previous stroke with hemiplegia, prior frag ility fractures, senile dementia, alcoholism and,gastrectomy were associate d with significant risk, whereas osteoarthrosis, nephrolithiasis and myocar dial infarction were associated with lower risks. Taking medications was no t associated with a difference in risk apart from a protective effect with the use of analgesics independent of co-existing osteoarthrosis and an incr eased risk with the use of antiepileptic agents. Of the potentially 'revers ible' risk factors, BMI, leisure exercise, exposure to sunlight and consump tion of tea and alcohol and tobacco remained independent risk factors after multivariate analysis, accounting for 54% of hip fractures. Excluding BMI, 46% of fractures could be explained on the basis of the risk factors sough t. Of the remaining factors low exposure to sunlight and decreased physical activity accounted for the highest attributable risks (14% and 9% respecti vely). The use of risk factors to predict hip fractures had relatively low sensitivity and specificity (59.6% and 61.0% respectively). We conclude tha t lifestyle factors are associated with significant differences in the risk of hip fracture. Potentially remediable factors including a low degree of physical exercise and a low BMI account for a large component of the total risk.