Instructions: The use of statistics in epidemiology to helps determine differenc
ID: 53979 • Letter: I
Question
Instructions: The use of statistics in epidemiology to helps determine differences in risk due to exposures and allows for the evaluation of associations with respect to validity in order to strengthen study conclusions. Imagine that you are an epidemiologist investigating the development of leukemia associated with occupational exposures to benzene. Employment records identify a total of 9,024 individuals who have worked in a facility where benzene is present. Of the 9,024 individuals, you are able to locate 3,249 individuals. Of those persons located, 487 have leukemia. Your investigative team also decides to conduct an occupational surveillance program at a facility to identify those who are subjected to excessive benzene exposures. You screen a total of 1,832 workers using a new diagnostic test, and 214 workers with excessive benzene exposure test positive. A negative test finding was observed in 29 workers who had excessive exposure to benzene, and 90 workers without excessive benzene exposure tested positive. 1. Identify potential sources of bias in this study. 2. Describe how these potential sources of bias could be addressed in future studies. 3. Based on the results of the screening test, construct a 2 x 2 table and calculate the following measures: a. Sensitivity b. Specificity c. Prevalence of excessive benzene exposure d. Positive predictive value e. Negative predictive value f. Accuracy 4. Based on the results of the screening program, what conclusions would you draw regarding the new diagnostic test?
Explanation / Answer
1. The sources of bias could be in
a. planning
b. Data collection
c. Analysis
d. Publication phases of research
Spectrum or random bias can also take place in the study
Some of the sources explained:
Data collected for prevalence of the target condition. This prevalence can vary with the setting and can also affect diagnostic performance estimates. In settings of higher prevalence, context bias can also take place.
Error in estimation due to an imperfect reference standard can lead to underestimation or overestimation of the performance of index test.
2. The potential sources of bias can be addressed in future studies is by setting inclusion criteria in future studies.