Develeopment of a socioeconomic index to be used in health reserches in Turkey

ObjectiveThe purpose of this study is to form a valid and -easy to applyhousehold-based index of socioeconomic status (SES), to be used in theinequalities in health studies and in causality researches.Materials and Methods0-59 month aged children, reproductive aged (15-49) women andhousehold data bases of Manisa Population and Health Surveys (MPHS)1999 and 2005 were used in this methodological type study.The methodology of this study was composed of 5 steps:1- The determination of the potential dependent and independentvariables by literature search.2- The demonstration of the relationships between dependent andindependent (SES) variables by univariate analysis.3- The demonstration of the relationships between dependent andindependent (SES) variables by multi-variate analysis.4- Establishement of SES variable groups (index), and5- To test the relationships of dependent (test) variables with newestablished SES indexes on MPHS data bases by known groups method.During literature search, the SES variables are classified as essential(direct) and indirect SES variables. The possible potential essential variablesare ?social class based on occupation of head of the household?, ?socialclass of woman?, ?best socail class of the household?, and education of manand woman. Among indirect variables, ?social insurance? and ?resistantconsumption goods? are included to the analysis. Univariate analysis wereperformed in order to show the statistical relationships with SES variablesand the predefined dependent variables (such as presence of child mortalityin the household, the place of labour and perceived health of 15-49 agedwomen in the household) and decrease the number of independent SESvariables in the indexes. Following univariate analysis multivariate (LogisticRegression) models were set up.As a result of univariate and multivariate analysis, the most significantvariable that explain the variability on the dependent variables was womaneducation. So, woman education was included in all of the variablecombinations (index models). In addition to woman education, based on bothliterature evidence and univariate and multivariate analysis ?social class?based on occupation of head of the household was also incorporated to theindexes. During the formulation of the indexes, Boratav Social Class modelawas used in addition to a number of international SES measures in theliterature.By utilizing SES variables such as woman education and the socialclass of the household we developed three different index models which hasequal weights of woman education and social class. Literature findings wereused during the development of ?equal weights socioeconomic index A? and?Social Class weighted socioeconomic index A? whereas ?Woman educationweighted socioeconomic index A? were developed based on our findings.During the development of equal weights index V.E.R.İ Socioeconomic index,Hollingshead social condition index and Barratt social status indexes wereused which have beed used in Turkey. In addion to these models, threealternative indexes were developed by substitution of the ?best social class ofthe householed? instead of ?social class of the householed.During the last phase of the study known groups method (criterionvalidity) was used in order to test the expected relationships between thedeveloped indexes and dependent variables by using trend Chi square andMann Whitney U test)Results and ConclusionThe SES variables such as woman education and social class werefound to be sensitive to the woman and child health related dependent116variables. The index model generated from both of these SES variables werealso found sensitive to the sub-categories of test (dependent) variables.By using epidemiological terminology we can say that the newdeveloped indexes gave dose-respose relationships with test variables. So,these generated indexes may be used independently as SED variavles in thefollowing researches. Based on the type of the dependent variables (e.g.women and child health measures) that were used to test the performance ofthe indexes developed in this study, we can say that the proposed indexeshere can be regarded as valid only on the women and child health measures.In other words we can say that these indexes should be interpreteed withcaution when used to discriminate sub-categories of the health measuresother than women and child health indices.Key words: Health surveys, socioeconomic index, socioeconomic variables,socioeconomic status, social class.