Fitting Binomial, Quasi – Binomial, Poisson and Quasi - Poisson Models in Analyzing In Vivo Micronucleus Assay Data
Keywords:binomial, Poisson, quasi-Binomial, quasi-Poisson, over-dispersion
AbstractMicronucleus assay is an important analysis for testing the genotoxicity of a compound thus, a careful analysis should be done. The main focus of this study was to perform the recommendations of Hothorn and Gerhard (2009) on analyzing the number and proportion of micronucleus polychromatic erythrocytes for a given number of polychromatic erythrocytes. The compound TO1 was evaluated using the Binomial, Poisson, quasi-Binomial and quasi-Poisson models. Results showed that most doses of TO1 had no significant effect on the proportion of micronucleus. The result of quasi- Binomial and quasi-Poisson model revealed that most of the inferences agree with the results of the traditional models. However, some inferences can lead to a false positive decision if over-dispersion was not taken into account. The analysis also showed that TO1 was found to be a negative compound.
Bretz F and Hothorn, LA. 2002. Detecting dose-Response using contrasts: Asymptotic power and sample size determination for Poisson data. Stat Med. 21(22): 3325-3335.
Hothorn LA and Gerhard, D. 2009. Statistical evaluation of the In Vivo Micronucleus Assay. Arch Toxicol. 83:625–634.