The author notes that potential for bias is large for environmental epidemiologic studies that utilize geographic information systems (GIS) mainly because data in available exposure and health effects databases are usually not collected for the purpose of environmental health research. Sources of bias include confounding due to the use of an invalid comparison group, information bias due to measurement or classification error, selection bias, due to invalid methods of selecting subjects or subject non response, follow-up bias due to losses to follow-up, specification error, due to making incorrect assumptions when analyzing data, and ecological bias, due to the use of grouped data for making inferences about individuals. The objective of this proposal is to develop formal methods for evaluating the magnitude of bias in environmental epidemiologic studies that utilize GIS, so that uncertainty about the role of bias will be minimized. It is proposed to develop formal methods for evaluating bias using sensitivity analysis, a technique that can be used to examine how the magnitude of bias varies under different scenarios. It is proposed to develop methods that will allow an investigator to evaluate bias as follows: 1) Calculate the maximum and minimum amount of bias to relative risk (RR) estimates under worst-case (and best case) scenarios, 2) list scenarios for which bias exceeds some specified level, 3) list scenarios for which bias could totally explain an observed relative risk, 4) list scenarios for which bias could have obscured a relative risk of a specified size, 5) calculate a probability distribution for the bias-corrected relative risk, and 6) calculate the probability that the bias-corrected relative risk exceeds a specified threshold. It is proposed to develop these methods for evaluating the magnitude of confounding, bias due to exposure measurement error, and bias due to disease misclassification of some of the more important sources of bias for GIS studies and for epidemiologic studies in general. Methods are proposed that can be used to evaluate the magnitude of these three sources of bias both singly and in combination. It is proposed to evaluate the usefulness of published models for an information-bias-corrected relative risk for environmental epidemiology studies that use GIS, and to generalize them as needed for GIS studies.

National Institute of Health (NIH)
National Institute of Environmental Health Sciences (NIEHS)
First Independent Research Support & Transition (FIRST) Awards (R29)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG7-STA (01))
Program Officer
Collman, Gwen W
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Minnesota Twin Cities
Public Health & Prev Medicine
Schools of Public Health
United States
Zip Code
Maldonado, G (2008) Adjusting a relative-risk estimate for study imperfections. J Epidemiol Community Health 62:655-63
Maldonado, George; Greenland, Sander (2002) Estimating causal effects. Int J Epidemiol 31:422-9