Selection Bias and Wire Coding

Description

Results of studies of the association between magnetic field and childhood leukemia risk have been inconsistent, although recent studies have suggested an association for exposure above 0.3 or 0.4 mT. Selection bias, however, cannot be ruled out as an explanation for the association. This project will examine the potential role of selection bias in the Canadian study published by McBride et al. in 1999. Due to the unique controls selection methods employed in this study (controls were selected from provincial health insurance rolls), it is possible to include first-choice controls in the analyses even if they decided not to participate in the study. Exposure will be assessed using wire coding, which requires no subject involvement or participation and could be completed based on available address information. Effect estimates for the wire code -childhood leukemia association will be calculated and compared in analyses using participating controls only or both participating and non-participating first-choice controls. Any difference between the two analyses using the two different control groups, in addition to random variability, is likely to be due to differential participation of controls in the study.

Principal Investigator

• John Spinelli
• Mary L McBride

Project Coordinator

• Agnes Lai

Research Team

• Gabor Mezei
• Paul Wong

Top of Page	Prev (6 of 7) Next

URL: http://www.bccrc.ca/ccr/spinelli_selectionBias.html
The BC Cancer Agency Research Centre is the research arm of the BC Cancer Agency (BCCA),
and is supported by the BC Cancer Foundation.
This page was last modified at 4:45pm on May 20, 2004
© 1999-2008. BC Cancer Agency. All rights reserved.
About this website...