You are here:
THE POWER TO DETECT A DIFFERENCE: DETERMINING SAMPLE SIZE REQUIREMENTS FOR EVALUATION OF REPRODUCTIVE/DEVELOPMENTAL EFFECTS FROM EXPOSURE TO COMPLEX MIXTURES OF DISINFECTION BYPRODUCTS
Citation:
SIMMONS, J. E., C. DINGUS, L. K. TEUSCHLER, G. E. RICE, AND M. G. NAROTSKY. THE POWER TO DETECT A DIFFERENCE: DETERMINING SAMPLE SIZE REQUIREMENTS FOR EVALUATION OF REPRODUCTIVE/DEVELOPMENTAL EFFECTS FROM EXPOSURE TO COMPLEX MIXTURES OF DISINFECTION BYPRODUCTS. Presented at Society of Toxicology Annual Meeting, Seattle, WA, March 16 - 20, 2008.
Impact/Purpose:
A novel statistical approach to calculating power was developed for the third phase of the U.S. EPA’s 4Lab study.
Description:
Toxicological assessment of environmentally-realistic complex mixtures of drinking-water disinfection byproducts (DBPs) are needed to address concerns raised by some epidemiological studies showing associations between exposure to chemically disinfected water and adverse reproductive and developmental outcomes. There is a growing emphasis on inclusion of low, environmentally relevant doses when assessing complex chemical mixtures. However, such environmentally relevant doses may cause subtle effects that would not be readily detected in standard animal testing protocols using typical sample sizes, as these assays were developed to have appropriate power for much higher dose levels. Proper study design, and in particular, the characterization of statistical power, is critical to ensuring that experimental results are meaningful at low levels and to proper interpretation of negative data. As methods were not available, a novel statistical approach to calculating power was developed for the third phase of the U.S. EPA’s 4Lab study, centered around a multigenerational rodent bioassay to test for reproductive and developmental effects from exposure to a complex mixture of DBPs in drinking water concentrates. Using data generated in earlier phases of the 4Lab Project and the newly developed approach, the power calculation results show that a study design of two blocks of 100 animals each with a 40:60 ratio of control:treated animals and a significance level of 0.05 provides approximately 90% and 45% power to detect pup weight decreases and prenatal loss increases, respectively. In conclusion, power calculation methods were developed, then used to calculate power at relevant exposure levels of complex mixtures and aid in design of a muligenerational reproductive/developmental study with a highly complex mixture of DBPs. This abstract does not reflect EPA policy.)