I believe that the percentage of drug cases that involve meth should be 93%. I think your calculation doesn't take into account the one case where the processing code set the meth flag but not the drugs flag.

Since the number of pregnancies (which likely tracks the population) varies by county, we should probably compare counties by a rate of deaths where meth was mentioned out of all pregnancies, or at least compare the percentage of meth deaths out of all zero-day deaths.

I think this needs to be updated to https://s3.amazonaws.com/tmp-gfx-public-data/meth-moms20220628/oklahoma-meth-moms-data.csv