Outcome Of Alzheimer’s Disease Based On Transition-Specific Cox Model

Objective:The aims of this study were(1) to explore related factors for every transition in outcome of Alzheimer’s disease(AD) so as to develop measures to delay AD progression based on a cohort study among community-dwelling elderly;(2) to show the advantages of transition-specific Cox model in outcome of chronic diseases with transient states so as to provide a new reference method for such diseases follow-up studies. Methods:Based on AD progression, six states were constructed: mild cognitive impairment(MCI), AD, loss-to-follow up(LTF), AD/LTF(LTF after AD/ AD after LTF), death for others, and death for AD. Nine wave’s data of follow-up came from 3 communities(Laojunyin/Xigong/Xikuang) in Taiyuan during Oct 2010 and Oct 2014. Transitionspecific Cox model was fitted and model parameters were estimated to analyze the influencing factors for every possible transition for AD progression. Results:At the significance level of 0.05, the transition-specific Cox model showed that: age group, smoking, sex, hypertension, high cholesterol, and diabetes were statistically significant for transition from MCI to AD; former profession, drinking, and hypertension were statistically significant for transition from MCI to LTF; age group, monthly income, and hypertension were statistically significant for transition from MCI to DE(death for others); relatives with dementia was statistically significant for transition from MCI to DEATH(death for AD); age group and education level were statistically significant for transition from AD to DE; relatives with dementia and hypertension were statistically significant for transition from AD to DEATH; marital status, drinking, and age group were statistically significant for transition from LTF to DE; marital status, age group, and smoking were statistically significant for transition from AD/LTF to DE. The results indicated that older age, smoking, female, hypertension, high cholesterol, and diabetes had higher risk for MCI�'AD; former manual profession, drinking, and hypertension had higher risk for MCI�'LTF; older age, lower monthly income, and hypertension had higher risk for MCI�'DE; first-degree relatives with dementia had higher risk for MCI�'DEATH; older age and lower educational level had higher risk for AD�'DE; first-degree relatives with dementia and hypertension had higher risk for AD�'DEATH; single, drinking, and older age had higher risk for LTF�'DE; in married state, older age, and smoking had higher risk for AD/LTF�'DE. No related factors for AD�'AD/LTF, LTF�'AD/LTF, LTF�'DEATH, and AD/LTF�'DEATH were found. Conclusions:There are different related factors about every possible transition in outcome of AD. Transition-specific Cox model has more advantages over other traditional methods in longitudinal data analysis for some states or some potential outcomes.