Ettinger Bruce, Hillier Teresa A, Pressman Alice, Che Maggie, Hanley David A
Division of Research, Kaiser Permanente Medical Care Program, Oakland, California, USA.
J Womens Health (Larchmt). 2005 Mar;14(2):159-71. doi: 10.1089/jwh.2005.14.159.
To devise, validate, and test a software model that improves how clinicians calculate individual risk for osteoporotic fracture and expected treatment benefit.
We developed a simple model of seven easily ascertained items plus bone mineral density (BMD) that calculates absolute fracture risk and expected absolute risk reduction after treatment. Baseline clinical variables and longitudinal fracture data from two large osteoporosis cohort studies validated the model's accuracy in predicting fracture risk. We then surveyed 298 clinicians to evaluate the likelihood they would prescribe alendronate in three hypothetical cases, first given the clinical data alone and then with model-derived data on fracture risk and expected treatment benefit.
We found a strong linear relationship with the model's predicted fracture risk and observed fracture rates in two large observational cohorts but the model overestimated risk 2-3 fold. The model predicted a 1:200 5-year risk for spinal fracture and a 1:40 risk for nonspinal fracture in an index case of a younger, thin, osteopenic woman. Given this hypothetical history with BMD t-scores, 26% of clinicians were likely to prescribe alendronate; when also given model-calculated 5-year fracture risks with or without treatment, only 13% were likely to prescribe alendronate (p < 0.001). For 2 other osteoporosis patients in whom risk was much higher, further information on fracture risk and expected treatment benefit did not alter prescribing.
Reporting absolute fracture risk with and without treatment promises to be most useful in women with osteopenia, a common clinical dilemma in younger postmenopausal women.
设计、验证并测试一种软件模型，以改进临床医生计算骨质疏松性骨折个体风险及预期治疗获益的方式。
我们开发了一个包含七个易于确定的项目以及骨密度（BMD）的简单模型，该模型可计算绝对骨折风险以及治疗后预期的绝对风险降低值。来自两项大型骨质疏松队列研究的基线临床变量和纵向骨折数据验证了该模型在预测骨折风险方面的准确性。然后，我们对298名临床医生进行了调查，以评估他们在三种假设情况下开具阿仑膦酸盐处方的可能性，首先仅给出临床数据，然后再给出模型得出的骨折风险和预期治疗获益数据。
我们发现该模型预测的骨折风险与两个大型观察性队列中观察到的骨折发生率之间存在很强的线性关系，但该模型高估了风险2至3倍。在一名年轻、消瘦、骨量减少的女性索引病例中，该模型预测脊柱骨折的5年风险为1:200，非脊柱骨折的风险为1:40。对于具有这种假设病史及BMD t值的情况，26%的临床医生可能会开具阿仑膦酸盐处方；当同时给出模型计算的有或无治疗的5年骨折风险时，只有13%的临床医生可能会开具阿仑膦酸盐处方（p < 0.001）。对于另外两名骨折风险高得多的骨质疏松患者，关于骨折风险和预期治疗获益的进一步信息并未改变处方情况。
报告有治疗和无治疗情况下 的绝对骨折风险，对于骨量减少的女性最为有用，这是年轻绝经后女性常见的临床难题。
