Abstract: Background　The Fourth Sample Survey on the Living Conditions of China's Urban and Rural Older Persons showed that as of 2015，there were 40.63 million disabled and semi-disabled older persons，accounting for 18.3% of the elderly population.The disability of the elderly has become an important public health problem in China.And studying the disability transition regularity in this population is crucial to solve the problem of aging.Objective　To study the disability transition regularity in Chinese older persons using a multi-state Markov model.Methods　Data stemmed from the China Health and Retirement Longitudinal Study （CHARLS） conducted in 2011 and 2013.The disability status of the involved older people aged 60 or above was classified by the National Long Term Care Survey （NLTCS） disability criteria.Disability transition regularity was analyzed with a multi-state Markov model.Results　A total of 6 615 cases were enrolled.A two-year follow-up showed that，compared with baseline，3 740 （56.54%） of the participants were found to have little changed functional state，837 （12.65%） showed improvement in functional state，1 704 （25.76%） demonstrated declined functional state，and 334 （5.05%） died.The functional state had a possibility to be improved regardless of the baseline status.Mild disability had the highest probability of improvement （0.414 0）．Those who were disabled at baseline had higher possibility of transition in functional state than those who were nondisabled at baseline.However，the possibility of improvement in functional state was negatively correlated with the level of disability at baseline，namely，it decreased successively in participants with baseline assessed mild，moderate and severe disability.The possibility of improvement in functional states was higher than that of deterioration in those with impaired health，mild or moderate disability，but in those of severely disabled，a totally opposite result was observed.Those of severely disabled also showed the highest probability of death （0.355 1）．The mean sojourn time in those with healthy，impaired health，mild disability，moderate disability，and severe disability was 3.33，1.83，1.02，1.02 and 1.52 years，respectively.Conclusion　Functional states being unchanged，improved or deteriorated during a period of time are common in older persons.The transition regularity of severe disability is different from that of impaired health，mild or moderate disability.The latter three functional states have a higher likelihood of improvement，while the former，severe disability，has longer sojourn time and a lower likelihood of recovery，and the highest probability of death.

Key words: Multi-state Markov model, Aged, Disability, Transition regular, Activities of daily living

摘要： 背景　第四次中国城乡老年人生活状况抽样调查显示，截至2015年，我国失能、半失能老年人的人数约为4 063万人，占老年人人口的18.3%。老年人失能已成为我国现阶段的重要公共卫生问题，老年人失能状态转移规律研究在解决老龄化问题中显得至关重要。目的　运用多状态Markov模型对我国老年人失能转移规律进行研究。方法　利用中国健康与养老调查（CHARLS）2011年和2013年数据，按美国长期护理调查（NLTCs）制定的失能分类标准对CHARLS数据中60岁及以上的老年人进行失能状态分类，构建多状态Markov模型对老年人失能不同状态间相互转移的规律进行研究。结果　本研究共纳入6 615例受访者，随访时间为2年，3 740例（56.54%）保持状态不变，837例（12.65%）状态改善，1 704例（25.76%）状态变差，334例（5.05%）死亡。无论老年人的状态处于哪个阶段，均有可能改善，轻度失能状态改善的概率最高，为0.414 0；初始状态为失能的老年人比初始状态为非失能的老年人更具有状态改变的倾向，但初始状态为重度失能的状态改变的概率低于初始状态为轻度失能和中度失能状态改变的概率；健康受损、轻度失能及中度失能的老年人状态改善的概率均高于变差的概率，重度失能的老年人状态改善的概率低于变差的概率；重度失能死亡的概率最高，为0.355 1。健康、健康受损、轻度失能、中度失能、重度失能状态的平均逗留时间分别为3.33、1.83、1.02、1.02、1.52年。结论　老年人状态不变或改善与状态变差一样普遍存在。重度失能老年人的失能转移规律不同于健康受损、轻度失能及中度失能老年人的失能转移规律，健康受损、轻度失能及中度失能仍有较大的恢复的可能性，重度失能的逗留的时间较长，恢复的可能性较小，死亡的概率最高。

关键词: 多状态Markov模型, 老年人, 失能, 转移规律, 日常生活活动