The Estimation Of Chronic Disease Prevalence In Small Area By Hierarchical Model

For the past few years,it is the exploratory analysis to chronic non-infectious disease especially that has attracted much more attention from the researchers of the epidemiology,for example,high blood pressure or diabetes and Cardiovascular disease(CVD)and so on.In point of medical,the main content of exploratory analysis to the epidemiology is to make the comparison in different parts for finding the geographical differences in number which represents the levels of having the disease.It is known that high incidence of the disease characteristics may be just the risk factors disease,so to seek for the high incidence of disease is one of the main research content in exploratory analysis of the epidemiology.However,the most suitable indicator to the situation of incidence of disease is prevalence rate.The meaning of prevalence rate is the rate of the actual number of who having disease during one time and the total number.On the other hand,any other study about disease is based on prevalence rate.In practice,many countries get the disease data by sampling from small areas.However,such illness data got by this way may be very few and very scattered,at the same time,it has the spatial homogeneity and difference.In this situation,using the simple rate(the ratio obtained through dividing umber of cases by investigation of the total directly)as the indicator to reflect the current state of the illness in the exploratory analysis may cause larger bias.Therefore,to find the right way to a more accurate disease prevalence estimation become the basic problem to be solved of epidemiology.The purpose of this paper is to provide the theoretical basis for the application of bayesian hierarchical model in chronic diseases prevalence estimation.In the terms of research methods,this paper takes the most common cardiovascular disease—heart disease as the research object.using the data published by China health and retirement survey(in short of CHARLS)in 2015 to construct bayesian hierarchical model,using one of most wonderful MCMC simulation algorithm—Gibbs algorithm to estimate heart disease prevalence in the elderly aged 45 and above in partial provinces and cities in China.In addition to estimating the prevalence also compare the results of bayesian hierarchical estimation with the traditional estimates on the value.The result showed that there are certain difference between the results from Bayesian hierarchical model and the traditional estimation method.Model test and steady analysis shows that the results obtained from the bayesian hierarchical model the are closer to the observed values;And,eventually conclude that bayesian hierarchical model helps to deal with the issue of regional sampling data such as too scattered,and bayesian hierarchical model can take full consideration to the spatial homogeneity and difference of the data,making the prevalence estimates is more robust,proving the applicability and feasibility of the bayesian hierarchical model in the field of prevalence estimates and providing theoretical basis and new research idea for the use of bayesian hierarchical model in the area of disease prevalence estimation.