|Malladi, Sasidhar - University Of Minnesota|
|Ssematimba, Amos - University Of Minnesota|
|Stepehns, Chris - Orise Fellow|
Submitted to: Journal of Veterinary Diagnostic Investigation
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/10/2019
Publication Date: N/A
Interpretive Summary: Determining the quantity of avian influenza virus (AIV) in a sample requires the use of a large number of chicken eggs. Essentially the virus can grown in chicken eggs and by seeing how far it can be diluted and still be detected one can determine the concentration in a sample. For accuracy, several egg needs to be treated with each dilution; usually five eggs for each dilution and six dilutions are tested; so 30 eggs per sample. Because eggs are expensive and take a lot of room, we investigated whether using only 3 eggs would provide enough accuracy. Data from experiments quantifying AIV was applied to several statistical models to determine how using only 3 eggs would affect the accuracy of the test. It was found that there was a minor effect on accuracy and precision, but overall results with 3 eggs instead of 5 were valid. This will result in substantial monetary savings when AIV needs to be quantified because fewer eggs (18 instead of 30) need to be used.
Technical Abstract: The standard methods for titrating avian influenza viruses (AIV) in order to calculate their log10 mean infectious dose specifies using 5 embryonating chicken eggs (ECE) per dilution. Specific pathogen free or specific antibody negative ECE are expensive and ECE require more space during incubation than cell culture. If the accuracy of results could be maintained, it would reduce cost and improve efficiency if the number of ECE could be reduced from 5 to 3 per dilution. Using titration data from 18 AIV isolates (8 low pathogenicity and 10 highly pathogenic), Beta Poisson and Exponential dose response models were used in a simulation study to evaluate the effect of using 3 ECE per dilution instead of 5. Additionally, the reproducibility was evaluated with one AIV isolate titrated by 3 replicates with the Beta Poisson and Exponential dose response models and also the Weibull dose response model. Good fit was observed with all of the models. Reducing the number of ECE per dilution from 5 to 3 resulted in a slight decrease in the precision of the virus concentration estimates where the standard deviation of the error increased by 0.06 to 0.09 log EID50 under different scenarios. Although there was an increase in the width of the 95% confidence interval, the precision of the result using 3 ECE per dilution (95% C.I., +- 0.64 to +-0.75 log10 EID50 in different scenarios) should be adequate for most applications.