Advice on statistical analysis of in vivo bacterial data

#1
Hey everyone!
Cattle are the main natural reservoir of STEC (Shiga toxin-producing E. coli) and excrete these bacteria in their feces. Human transmission occurs through consumption of fecally contaminated food or water, resulting in serious diarrhea and kidney damage. Currently there is no specific treatment for human infection. Therefore, limiting bovine carriage (by vaccination) is a viable option.

I am doing research on an experimental bovine vaccine. I performed a RCT (randomised controlled trial) with 3 treatment groups (control group n=6, twice vaccinated n=6, thrice vaccinated n=6). These animals are divided into 3 separate pens (2 animals of each treatment group in every pen, mixed design). We then infected all animals with the challenge strain, and then vaccinated them (either 3 adjuvant injections, 2 vaccinations +1 adjuvant or 3 vaccinations) with a 3-week interval. We took fecal samples to see how much of the challenge strain is being excreted, which gives a CFU value (colony forming units). So we basically have 2 samplings per week per animal with a certain CFU value. (This is EXP-1)

After the last vaccination, we reinfected all animals again to see if the vaccination would have an advantageous effect. We followed up again for a number of weeks (sampling thrice a week now), again giving CFU values per individual animal. (This is EXP-2)

I am currently struggling with the statistics.

My concrete questions are:

i) I want to investigate if there is a significant difference in fecal excretion between vaccinated and non-vaccinated animals. In EXP-1, all animals in the 2-dose and 3-dose group can be considered equal right until the 3rd vaccination (so 12 animals all received 2 doses of the vaccine). I want to compare the fecal excretion of these animals with the control group (no vaccine). So 6 control animals vs. 12 vaccinated animals. Which test do I use? Mann-Whitney test?

ii) the same for EXP-2, only here we compare control group with 2-dose group and with 3-dose group so 6, 6, 6 every group. Which test do I use? one-way ANOVA?

Currently, I am using area under the curve (AUC) to calculate the mean AUC of each treatment group and compare these values. However, this doesn't really take into account the longitudinal aspect of the trial. Anyone has advice on this?

iii) I took fecal samples but also rectal swabs. I want to investigate which is the more sensitive sampling method to confirm the presence of the bacteria (swabs or feces). What is the best way to do his?

Advice would be highly appreciated! I hope the above description is somewhat clear.

Many thanks in advance!
KR,
F
 
#2
I believe that it is common to take the log10(cfu) (the 10 logarithm of the cfu). And then use that as the dependent variable in an anova. I would use that both in comparing the 6 versus the 12 and the 6 versus 6 versus 6.

I did not understand the AUC stuff. What was that?