Schonberg Mara A, Li Vicky, Marcantonio Edward R, Davis Roger B, McCarthy Ellen P
Division of General Medicine and Primary Care, Department of Medicine, Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, Massachusetts.
J Am Geriatr Soc. 2017 Jun;65(6):1310-1315. doi: 10.1111/jgs.14805. Epub 2017 Feb 21.
Extended validation of an index predicting mortality among community-dwelling US older adults.
DESIGN/SETTING: Examination of the performance of a previously developed index in predicting 10- and 14-year mortality among respondents to the 1997-2000 National Health Interview Surveys (NHIS) using the original development and validation cohorts. Follow-up mortality data are now available through 2011.
16,063 respondents from the original development cohort and 8,027 respondents from the original validation cohort. All participants were community dwelling and ≥65 years old.
We calculated risk scores for each respondent based on the presence or absence of 11 factors (function, illnesses, behaviors, demographics) that make up the index. Using the Kaplan Meier method, we computed 10- and 14-year mortality estimates for the development and validation cohorts to examine model calibration. We examined model discrimination using the c-index.
Participants in the development and validation cohorts were similar. Participants with risk scores 0-4 had 23% risk of 14-year mortality whereas respondents with risk scores (13+) had 89% risk of 14-year mortality. The c-index of the model in both cohorts was 0.73 for predicting 10-year mortality and 0.72 for predicting 14-year mortality. Overall, 18.4% of adults 65-74 years and 60.2% of adults ≥75 years have >50% risk of mortality in 10 years.
Our index demonstrated excellent calibration and discrimination in predicting 10- and 14-year mortality among community-dwelling US adults ≥65 years. Information on long-term prognosis is needed to help clinicians and older adults make more informed person-centered medical decisions and to help older adults plan for the future.
对一项预测美国社区居住老年人死亡率的指数进行扩展验证。
设计/背景：使用最初开发和验证队列，检验先前开发的指数在预测1997 - 2000年国家健康访谈调查（NHIS）受访者10年和14年死亡率方面的表现。随访死亡率数据现可获取至2011年。
来自最初开发队列的16,063名受访者和来自最初验证队列的8,027名受访者。所有参与者均为社区居住且年龄≥65岁。
我们根据构成该指数的11个因素（功能、疾病、行为、人口统计学特征）的有无为每位受访者计算风险评分。使用Kaplan - Meier方法，我们计算了开发队列和验证队列的10年和14年死亡率估计值，以检验模型校准。我们使用c指数检验模型辨别力。
开发队列和验证队列中的参与者相似。风险评分为0 - 4的参与者14年死亡风险为23%，而风险评分（13分及以上）的受访者14年死亡风险为89%。该模型在两个队列中预测10年死亡率的c指数为0.73，预测14年死亡率的c指数为0.72。总体而言，65 - 74岁的成年人中有18.4%、75岁及以上的成年人中有60.2%在10年内有超过50%的死亡风险。
我们的指数在预测65岁及以上美国社区居住成年人的10年和14年死亡率方面显示出良好的校准和辨别力。需要长期预后信息来帮助临床医生和老年人做出更明智的以患者为中心的医疗决策，并帮助老年人规划未来。
