From @brxad.bsky.social et al, a reminder global ID collaborations do and should continue to exist and strengthen in the current political landscape. doi.org/10.1017/ash....

Special thanks to Kevin Schwartz @kevinschwartz.bsky.social for his leadership and Kevin Brown, PhD for his expertise and teaching me about instrumental variable analysis!

This highlights the importance of the analytical method to reduce risk of bias when measuring abx benefits and harms in observational studies. This also supports the fact that there is no additional harm to using shorter courses and are still consistent with #shorterisbetter.

In the instrumental variable analysis we found no difference in HARMS or SAFETY outcomes across three antibiotics - cephalexin, ciprofloxacin, and amoxicillin. But interestingly when using a propensity matched analysis we did find an increase in SAFETY outcomes with long duration

Then we evaluated 117,682 community dwelling patients of those prescribers and looked at their HARMS outcomes (abx overuse) - such as AMR, side effects, C. diff, and SAFETY /balancing outcomes (abx underuse) - such as repeat Rx, hospital visits and mortality

There is a lot of of inter-prescriber variability in antibiotic duration prescribing, even after accounting for patient factors. Duration prescribing is pseudo-random! So we classified MD prescribers based on their proportion of long-duration antibiotic prescriptions - the "instrumental variable"