Comparison of Nested Case-control and Cohort Analysis Methodologies using a District TB Registry Data

The nested case-control study has become popular as an efficient alternative to the full-cohort design. This study compares the results of a nested case-control analysis approach with the full cohort analysis. A cohort of 276 subjects (new cases from a TB registry) was used for this study. Cox Regression model was used for the full cohort analysis. In order to do the nested case-control analysis, for each death, three random controls were selected from those who did not suffer from the outcome at the time of the outcome took place. Case control data was analyzed by the conditional logistic regression model. Results from both cohort and nested case-control analyses show that treatment group is the only variable that affects on the outcome. Gender, place of residence, and age has no effect on the outcome. For binary exposure variables with trivial effects (e.g. Gender and place of residence), the relative efficiency of nested case-control study design is approximately 75%. Results of this study show that nested case-control study is not only an easy and cost-effective method for data analysis but also is as robust as cohort analysis in rate ratio and its variance estimation.