Practical operationalizations of risk factors for fracture in older women: results from two longitudinal studies.

Several guidelines on osteoporosis have proposed algorithms to identify persons at high risk of fractures. Although these algorithms include well-known risk factors, it is not clear how they can best be operationalized for use in general practice. The aim of this study was to compare the predictive performance of different operationalizations of four categories of risk factors for fractures that can be used in general practice. This study included 4157 women of > or =60 yr of age (mean +/- SD: 74.1 +/- 9.1 yr) with a median follow-up of 8.9 yr of the Rotterdam Study and 762 women of > or =65 yr of age (mean +/- SD: 76.0 +/- 6.7.yr) with a median follow-up of 6.0 yr of the Longitudinal Aging Study Amsterdam (LASA). At baseline, information on four categories of risk factors was obtained, including (1) family history of hip fractures, (2) type of prior fractures, (3) low body weight/body mass index (BMI), and (4) mobility impairment. The occurrence of fragility fractures, including hip, pelvic, humerus, and wrist fractures, was used as outcome measure. We quantified the predictive performance of each risk factor by a chi(2) statistic, calculated as the difference in -2 Log likelihood attributable to the risk factor, with adjustment for age and other risk factors. In the Rotterdam Study, 399 fragility fractures occurred during 31,472 person-years (PY) of follow-up. In this study, any prior fracture in the past 5 yr (chi(2) = 6; p = 0.02), body weight < 64 kg (versus > or =64 kg; chi(2) = 6.7; p = 0.01), BMI < 22 kg/m(2) (versus > or =22 kg/m(2); chi(2) = 8.7; p = 0.003), and use of a walking aid (chi(2) = 7.5; p = 0.004) were the most practical operationalizations of the risk factor categories, after adjustment for age and other risk factors. In LASA, 52 fragility fractures occurred during 3935 PY of follow-up. Associations were similar as in the Rotterdam Study, except that low body weight and BMI were not associated with fragility fracture. None of the usual operationalizations of family history of hip fractures was independently associated with fragility fracture in either study. Prior osteoporotic fracture, body weight <64 kg, a BMI <22 kg/m(2), and the use of a walking aid are practical operationalizations of risk factors for fragility fractures. On the basis of the results of this study, a simple, practical algorithm can be developed for use in general practice.