You see it often because it’s a standard error

Congrats!!!

Woo Laura!!!

Will the panel include fixed effects?

So sorry! It’s so much work! Best of luck with whichever direction you take!

To be clear, we use the ED because it’s a clean sample largely free of prior influences from prior docs. This phenomenon of variation across docs in same facility can likely be found in length of stay for hospitalists, pneumonia read rates for radiologists, etc.

We agree that SDoH are important. This is why we take care to make comparisons within ED while also controlling for such things as time of arrival, ESI, and location within ED. We suspect there aren’t large differences in SDoH across docs in same ED after controlling for time, location, ESI, etc.

We use mortality because it’s an important measure and it’s largely non-contestable how to measure it. How would one measure an indicated vs not indicated admission? We also find that admitted patients of higher admitting docs more likely to be discharged before 24 hrs.

Thanks. Having more docs not trained in EM at the VA is something we acknowledge in the limitations of our paper, as is our inability to include doc characteristics such as training. But other lit using Medicare data show similar level of admit variation. www.healthaffairs.org/doi/pdf/10.1...

We didn’t get that granular but that’s a great question to explore.

Whoops, tagging Stephen’s bluesky account, not his Twitter account. @coussens.bsky.social

But higher admission rates do NOT ⬇️ important adverse outcomes like mortality. Given high costs of admission ($, provider & facility capacity, pt well-being), better understanding how such variation arises could be fruitful for pts, docs, and healthcare system. 8/ jamanetwork.com/journals/jam...

In sum, there is much variation in admit practices, likely due to diffs in skill & risk aversion. This mirrors variation in other doc specialties, who also greatly differ in their decisions. Of note, results do NOT argue for high-admit docs to indiscriminately ⬇️ admit rates. 7/

But seeing a higher-admitting doc does NOT reduce your likelihood of dying (either within 30 days [shown here], 7 days, 14 days, 90 days, or a year). 6/

Higher-admitting docs also order more radiology and laboratory tests in the ED. This suggests that admission rates may also be reflective of practice pattern intensity more generally. 5/

With data on over 2 million pts across 100 hospitals nationwide, we find that patients treated by docs with higher admission propensities are more likely to be discharged from the hospital within 24 hours when admitted, suggesting a lower clinical need for their hospitalization. 4/

We use rich VA EHR data with info not available in claims data such as ⌚️ of arrival, location within ED, and ESI (a # based on pt severity). This allows us to demonstrate that variation in docs’ admission rates is attributable to docs themselves, not to diffs in pt health. 3/

The decision to admit or discharge a patient is one of the most important decisions an ED doc makes. By how much do ED docs vary in this decision? A lot! Being treated by a doc in top 10% vs bottom 10% can nearly double your probability of admission. 2/

Wow, amazing news. Congrats to you both!