Discover and read the best of Twitter Threads about #timeseriesanalysis
Most recents (1)
Let’s say you’re doing a quasi-experimental study of an intervention to reduce antibiotic prescribing in primary care. Then #COVID19 happens and prescribing takes a nose-dive because no one is visiting the doctor. Analysis would have been controlled interrupted time series. 
The study is set to start in a few months. Left is what we expected, right is what we got. How would you change the analysis to accommodate for the #COVID19 effect on the baseline period? #statistics #stats #timeseries #timeseriesanalysis
Since #COVID19 destroyed any possibility of a smooth trend in antibiotic prescribing in primary care (affecting both the pre-intervention and likely post-intervention periods), my thought is to compare the "recovery" in prescribing in the intervention and control groups.
