Complex Sampling Estimations Of PHE And Prevalence Of Chronic Diseases Under Different Sampling Frames

ObjectiveEstiblishing different sampling frames of The4th National Health Service Survey inSh’aanxi province in order to utilizes The6th National Census of populations imformationas efficency sampling weight material. Using sample random method and two complexsampling methods with different sample weights to analysis personal health expenditureand prevalence of chronic diseases of Shaanxi province in2008. Discussing the affectcaused by various analytical methods and various sampling frames on the results ofcomplex sampling statistical analysis methods based on the comparison of three methods.MethodThe4th National Public Health Service Sampling is a four-stages stratified complexsample survey with unequal-probability，using community committee as primary samplingunits (PSUs) and family as ultimate sampling units(USUs). Since The6th National Censusof Population only contains gender-specific and age-specific information of each county，this article established two different sampling frames used county as it’s PSUs and person asit’s USUs. Weighted samples according to different sampling frames. Using simple randommethod and two complex sampling methods with different sample weights to analysis the influence factors of personal health expenses and chronic disease situation of Shanxiprovince in2008.Estimating the fpc results and DEFF of complex sampling.ResultsIn this study, two sampling frames were established for the4th National Health ServiceSurvey of Shaanxi Province. The first sampling frame included6strata and40clusterswhile the second frame included204strata and1360clusters after considering gender andage class. Samples were weighted according different sampling frames. The comparisonamong three methods showed that the point estimation value became lower in complexsampling methods. The DEFFs of the first frame were larger than those of second frame.For the standard error, the SRS was the largest and the first frame was the smallest.ConclusionSample weighting methodology is of essential to complex sampling estimations. In theresults of the estimations of personal health expenditure and rate of chronic diseases usingthree methods, it was clear that the standard error of the two complex methods were higherthan that of SRS method and the standard error of the first frame was higher than that ofthe second frame.