KMID : 0380819930150010085
|
|
Korean Journal of Epidemiology 1993 Volume.15 No. 1 p.85 ~ p.95
|
|
A Study on the Nondifferential Misclassification in Case-Control Studies -A Mathematical Approach for Correcting the Estimates
|
|
|
|
Abstract
|
|
|
This study is conducted to suggest a method for correcting the biased estimates due to nondifferential misclassification of polytomous exposure and confounder. The basic idea for correcting the bias is a rearrangement of misclassified data
structure
using misclassification probability matrices. We present here a linear relationship between the misclassified and true data.
Simulation studies were also tried to investigate the magnitude and direction of the bias mentioned above. In simulation studies, we focused on the misclassification patterns in three circumstances, misclassification of exposure,
misclassification
of
confounder, and joint misclassification of both exposure and confounding variables. The simulation results show that the direction of exposure or confounder misclassification biases are heavily dependant on the misclassification patterns.
The proposed method is applied to an empirical data on the presence of medical utilization and smoking history where corrected odds ratio are examplified considering plausible ranges of misclassification probaility patterns.
|
|
KEYWORD
|
|
|
|
FullTexts / Linksout information
|
|
|
|
Listed journal information
|
|
|