Wait, you have just three numbers and want to know if there is a drop In counts across time? You have too little data, but normally you’d do a Poisson regression to determine if the slope is negative.

You can use statistical process control (SPC) for such data

https://en.m.wikipedia.org/wiki/Statistical_process_control

But as @Michael said you have too few timepoints. You should count a considerable number of weeks before the intervention and the same number of weeks after.

Jk

The EpiData software can be used:

http://www.epidata.dk/downloads/epidata_flyer_spc.pdf

It's available for free download on (donation/voluntary work):

http://www.epidata.dk/

Best

John-Kåre

Thanks for the information. I have never had a data situation like this but in the off chance there was some statistical test for significance I never heard about, I at least wanted to ask.