KSE
Articles
- Page Path
- HOME > Epidemiol Health > Volume 7(2); 1985 > Article
- Original Article Analysis of selection bias in a case-control study
- Yoon Ok Ahn
-
Epidemiol Health 1985;7(2):195-201
DOI: https://doi.org/
- 5,387 Views
- 16 Download
- 0 Crossref
- 0 Scopus
Abstract
Several situations in case-control studies, such as using hospital controls, prevalent cases, special surveillance groups, and low participation rate, necessitate evaluation of selection biases. A case-control study of renal adenocarcinoma which has been conducted in 100 towns around Boston, U.S.A., showed low participation (interviewed) rate, 41.1% and 62.0% among cases and controls, respectively. By comparing medial abstract data of interviewed and not-interviewed cases, interviewed data of controls and mail-questionaire data of not-interviewed controls, it was possible to estimate selection odds ratio for selected 8 variables and to assess selection bias associated with smoking, obesity, renal caluli, and hypertension. The conclusions drawn were as follows; 1) It is certain that a considerable number of factors served as actual selection factors in case-control study. 2) The interpretability of certain association resulted from interviewed data was strengthened and weakened by analysis of selection bias. For the valid explanation of study results of interviewed data, the evaluation of selection bias should be quite valuable and necessary. 3) The simplified schematic presentation of theoretical aspects of selection bias can be useful to analyse it afterward, while, there are no standardized methods of data collection for the evaluation of selection bias.
Cite
