Determination of sub-groups in reference interval calculations and an application

Abstract

Objective: The main objective of this study is to determine the methods used for dividing reference data into sub-groups and propose a new solution. Therefore, the newly proposed method is used to determine age sub-groups while calculating the reference interval of AST variable.Methods: The data of this study are composed of the registered patients who applied only once to the polyclinics at the Health Research and Application Hospital, Clinical Bio-Chemistry Laboratory of Mersin University. The main variables selected for the study are sex, age and aspartate transaminase (AST). MARS 2.0 package software was used for the statistical analysis. Harris&Boyd method was used for sex and age while the MARS method was used for the determination of age sub-groups.Results: According to the Harris&Boyd method, it was concluded that it is not necessary to create sub-groups for sex and age variables. Age sub-groups are determined with the MARS method by taking into account the correlations between AST variables and age. Age sub-groups are calculated as 5-9, 10-18 and 19+. Correlation between the age and AST variable is investigated for these sub-groups. It is obtained that the correlation is statistically not significant for the 5-9 age group (p=0.3189), significant for the 10-18 age group (p<0.001) and significant for 19 and above (p<0.001).Conclusion: The selection of transformation and outlier value elimination method and especially determination criteria for separation of sub-groups are made according to researcher's subjective desicions in the application of many studies. Because of this reason, MARS can be used as an alternative and objective method for determining subgroups. In addition, results of MARS method being parallel with the findings of the producing company supported the usage of this method for determination of sub-groups.