| 1,190 | 7 | 194 |
| 下载次数 | 被引频次 | 阅读次数 |
老年失能风险是人口老龄化背景下值得关注的重要问题。已有研究往往忽略失能数据的删失属性,因而无法充分利用数据信息。本文提出了一种新的老年失能风险建模方法,在考虑失能数据区间删失和右删失的情况下,对中国老年健康影响因素跟踪调查(CLHLS)数据按个体从健康状态转移到首次失能状态经历的时间进行分类,同时利用CLHLS死亡调查补充个体死亡前的失能信息。在此基础上,基于生存分析中的半参数转换失能时间模型,实现对老年失能率更准确的估计和预测。与传统模型相比,新模型将年龄、性别、教育水平等变量纳入模型,能够对不同特征人群的失能风险做更细致和精准的分析,为完善我国长期护理保险制度提供实证依据。
Abstract:The disability risk among elderly people is an important problem in the context of population aging. Existing research often ignores the censoring properties of disability data and cannot make full use of micro-data information. This paper presents a new method for modeling the disability risk among old people. First, considering the interval censoring and right-censoring of disability data in the Chinese Longitudinal Healthy Longevity Survey(CLHLS), this paper classifies the data according to the transition time from the healthy state to the first disability state. This paper also uses the CLHLS death survey to add information on disability prior to an individual's death. This paper uses the semi-parametric transformation model in survival analysis to construct a model of disability time, which estimates and predicts the elderly disability rate more accurately. Compared with the traditional model, the new model can incorporate variables such as age, gender, and education level into the model, making a more detailed and accurate analysis of the disability risk for people with different characteristics. Our model provides an empirical basis for improving China's long-term care insurance system.
[1]胡晓宁,陈秉正,祝伟.基于家庭微观数据的长期护理保险定价[J].保险研究, 2016(4):57–67.
[2]焦开山.健康不平等影响因素研究[J].社会学研究, 2014, 29(5):24–46, 241–242.
[3]金蛟,李瞳辉,徐帅帅,等.测量误差模型稳健方法及应用研究[J].统计研究, 2021, 38(11):150–160.
[4]王德文,叶文振,朱建平,等.高龄老人日常生活自理能力及其影响因素[J].中国人口科学, 2004(S1):93–97, 177.
[5]王晓军,陈惠民,赵晓月.我国男女两性老龄人口死亡率联合建模与预测[J].统计研究, 2021, 38(10):151–160.
[6]王雪辉,彭聪.我国老年人口群体特征的变动趋势研究[J].人口与社会, 2020, 36(4):29–45.
[7]吴炳义,李娟丽,刘文浩,等.生活方式对老年人日常活动能力影响[J].中国公共卫生, 2019, 35(7):881–884.
[8]闫伟,何梦娇,路云,等.基于CLHLS的我国老年人失能现状及其影响因素研究[J].护理研究, 2021, 35(10):1807–1811.
[9]张琳,汤薇.基于非齐次Markov模型的长期护理保险定价研究[J].保险研究, 2020(7):108–121.
[10]张文娟,魏蒙.中国老年人的失能水平和时间估计——基于合并数据的分析[J].人口研究, 2015, 39(5):3–14.
[11] Austin P C, Brunner L J. Type I Error Inflation in the Presence of a Ceiling Effect[J]. The American Statistician, 2003, 57(2):97–104.
[12] Czado F C. Application of Survival Analysis Methods to Long-term Care Insurance[J]. Insurance:Mathematics and Economics, 2002, 31(3):395–413.
[13] Fuino M, Wagner J. Duration of Long-term Care:Socio-economic Factors, Type of Care Interactions and Evolution[J]. Insurance:Mathematics and Economics, 2020, 90:151–168.
[14] Hanewald K, Li H, Shao A W. Modelling Multi-state Health Transitions in China:A Generalised Linear Model with Time Trends[J]. Annals of Actuarial Science, 2019, 13(1):145–165.
[15] Li D, Hu X J, Wang R. Evaluating Association Between Two Event Times with Observations Subject to Informative Censoring[J]. Journal of the American Statistical Association, 2021:1–13.
[16] Lindsey J C, Ryan L M. Methods for Interval-censored Data[J]. Statistics in Medicine, 1998, 17(2):219–238.
[17] Rigobon R, Stoker T M. Bias from Censored Regressors[J]. Journal of Business&Economic Statistics, 2009, 27(3):340–353.
[18] Sun T, Ding Y. Copula-based Semiparametric Regression Method for Bivariate Data under General Interval Censoring[J]. Biostatistics, 2021,22(2):315–330.
[19] Wang Q, Hanewald K, Wang X. Multistate Health Transition Modeling Using Neural Networks[J]. Journal of Risk and Insurance, 2022, 89(2):475–504.
[20] Zhou Q, Hu T, Sun J. A Sieve Semiparametric Maximum Likelihood Approach for Regression Analysis of Bivariate Interval-Censored Failure Time Data[J]. Journal of the American Statistical Association, 2016, 112(518):664–672.
(1)数据源自国家统计局公布的第七次全国人口普查和第六次全国人口普查数据,网址为http://www.stats.gov.cn/zt_18555/zdtjgz/zgrkpc/dqcrkpc/。
(1)涵盖我国辽宁、吉林、黑龙江、河北、北京、天津、山西、陕西、上海、江苏、浙江、安徽、福建、江西、山东、河南、湖北、湖南、广东、广西、四川、重庆、海南,共23个省份。
(2)6项日常生活活动能力包括洗澡、穿衣、如厕、室内活动、大小便和吃饭,每一项ADL都对应三个备选答案,分别为“独立”“部分独立”“依赖”。
(1)因篇幅所限,从健康到失能的时间分布图以附图1展示,见《统计研究》网站所列附件。
(1)基准风险不会随着时间而递减。
(1)性别变量估计标准误为0.0276,P值为0.0031;年龄变量估计标准误为0.0014,P值为0.0000。
(1)四分位数加上1.5倍的四分位数间距之外的数据通常被视为潜在的离群点(金蛟等,2021)。
基本信息:
DOI:10.19343/j.cnki.11-1302/c.2024.10.009
中图分类号:F842.6;D669.6;C81
引用信息:
[1]李云龙,王晓军,孙韬.我国老年失能风险研究:基于删失数据半参转换模型[J].统计研究,2024,41(10):122-133.DOI:10.19343/j.cnki.11-1302/c.2024.10.009.
基金信息:
国家自然科学基金青年科学基金项目“复杂数据下半参数转换模型及其在老年慢性病发展中的应用研究”(72101261);国家自然科学基金面上项目“面向老年失能风险管理的复杂多状态生存数据建模及应用”(72471229); 教育部哲学社会科学研究重大课题攻关项目“健康中国2030背景下的健康老龄化体系优化研究”(20JZD023); 国家社会科学基金重点项目“我国基本养老保险制度研究”(20AZD075)
2024-10-25
2024-10-25