Thread by @sargsyanz

#MedThread on #ClinicalReasoning

@medicalaxioms and co made a great list👇 of routinely missed/delayed diagnoses.

Theme: uncommon conditions that present with common/non-specific symptoms.

How can we make these dx sooner without overtesting?

https://twitter.com/medicalaxioms/status/951840976481394688

X/ Goal is to arrive at the appropriate pre-test probability with the fewest/least expensive/least harmful tests possible, and end up on the right side of the testing threshold of the expensive/risky but more “definitive” test.

X/ By the way: “test” in this thread refers not only to labs and imaging but also to history questions, exam maneuvers, etc

X/ Tip 1: Always construct a differential. “Seems like a lumbar strain, but if it’s not, what else could it be?” Thinking is free, and making a ddx might lead to other free/quick/harmless bedside tests (eg questions or exam) that will refine your degree of suspicion.

X/ Tip 2: Have a sensitive radar for things that don’t fit, eg atypical features of the presentation or other unexplained abnormalities that could be part of the story. “Seems like a lumbar strain, but why is his temperature 99.4?” Or why was his calcium 10.4 last month?

X/ Tip 3: Be aware of (and look up) how good your tests are, and how that should affect post-test probability. “The globulin gap is normal, but what’s the sensitivity of elevated gap for myeloma?” Keep in mind, history elements are also tests, can have good or bad se and sp.

X/ Tip 4: Avoid confirmation bias. Many of our patients are so complex that you can paint a lot of reasonable narratives. But they’re not necessarily the right ones. “Did you eat a hamburger yesterday? Aha! That must be why you’re vomiting.”

X/ Tip 5: Use time, follow-up (including by phone), and response to treatment to your advantage. For us inpatient docs, this means a crystal-clear communication of remaining uncertainties, thought process, and next steps in the discharge summary

Summary: tips to avoid delay of uncommon dx
- always make a differential
- look out for things that don’t fit or red flags
- ask, but how good is this test?
- ask if you’re biased to confirm one of several narratives
- dx is a team- and time-dependent process: follow up key

End/

Thanks for reading! This is just a start, please add other tips, resource links, etc.

@DxRxEdu @StephVSherman @ImproveDX @andrewolsonmd @AshleyNDMeyer @HardeepSinghMD @rabihmgeha