Bayesian Semiparametric Multi-state Models In Relation Between Cognition And Depression Among The Elderly

Objective:Bayesian semiparametric multi-state models were applied in this study to explore the influencing factors and relation regarding cognition and depression in the elderly so as to provide references for control and prevention strategies development of cognitive impairment and depression in the elderly, and to provide model built references for other chronic diseases co-existing studies. Methods:The data came from eight waves of cohort study of 287 community-dwelling elders in Taiyuan from October 2010 to May 2014. Based on five states: only mild cognitive impairment(MCI), only depression, only Alzheimer's disease(AD), MCI and depression co-existing, AD and depression co-existing, Bayesian semiparametric multi-state models were constructed to analyze influencing factors for every transition and the relation between cognition and depression by their scores. Results:Bayesian semiparametric multi-state models fitted the data well. Multivariate analysis showed that aluminum cooking utensils, sports activities, doing housework and heart disease were statistically significant for transition from only MCI to MCI and depression co-existing; age and socially useful activities were statistically significant for transition from only MCI to only AD; dietary restriction was statistically significant for transition from only depression to MCI and depression co-existing; age, dietary restriction, sports activities and hobbies were statistically significant for transition from only AD to AD and depression co-existing; educational level, sports activities and cerebral disease were statistically significant for transition from MCI and depression co-existing to AD and depression co-existing. Therefore higher educational level, frequent socially useful activities and frequent sports activities were protecting factors of cognition; older, more dietary restriction and having cerebral disease were risk factors of cognition; frequent sports activities, more hobbies and more housework were protecting factors of depression; older, more dietary restriction, more aluminum cooking utensils use and having heart disease were risk factors of depression; age, dietary restriction and sports activities were common influencing factors of cognition and depression. Hazard ratio(HR) of cognition scores was 1.017(95%CI: 0.920-1.124) for transition from only MCI to MCI and depression co-existing and 0.769(95%CI: 0.671-0.882) for transition from only AD to AD and depression co-existing; respectively for transition from only depression to MCI and depression co-existing, HR of depression scores was 0.885(95%CI: 0.761-1.030). Conclusions:There are many factors influencing cognition and depression. Traditional factors play different roles in the different stages of natural history of cognitive impairment and depression co-existing. Severe cognitive impairment may lead to depression, while depression may not lead to cognitive impairment. Therefore compared with MCI in the elderly, the elderly with AD should be paid more attention of depression. Bayesian semiparametric multi-state models have more advantages over other traditional methods in exploring influencing factors of chronic diseases co-existing studies.