A recent case highlights the utility of using our framework for understanding the relationships between multiple diagnoses to avoid diagnostic and therapeutic error.
To wit: When multiple diagnoses are being considered, there is a principal diagnosis that explains the chief complaint and the clinical presentation, as well as one or more of the following:
- An incidentaloma. This is an accidental discovery of something asymptomatic that is unrelated to the clinical presentation and the principal diagnosis. The "coincidence" with the chief complaint is a spurious one. The incidentaloma was only discovered (causally, it turns out) because of investigation of the chief complaint! Related to this is a false positive result. Some testing, especially that which was not informed by a prior probaility, is highly likely to represent a false positive. Together, indicentalomas (real but asymotomatic findings, viz, a thyroid cancer on CTPE) and false positives, represent the so-called "red herring" - a scent that leads you off the trail to the principal diagnosis. Or one that inverts your management priorities.
- Pre-existing disease upon which the principal diagnosis is superimposed. The guy with systolic heart failure who gets admitted for decompensation every few months may have pneumonia this time around; no surprise there, unless you're inured (anchored) to the heart failure
- Findings and diagnoses that are causally related to the principal diagnosis: underlying causes (endpoint diagnosis; multiple myeloma previously unknown, that is causing the pneumonia with an encapsulated organism), or complications and epiphenomena of the principal diagnosis.
- Rarest of all, if it even occurs: another symptomatically coincident, unrelated disease along with the principal diagnosis. One must believe in a very wrathful God if he thinks that that God strikes a man with two independent diseases at nearly the same time. Disease onset is stochastic (random) so this should be a very rare (co)occurrence. The closer the onset (judged by symotoms) and the younger the patient, the more unusual would be the coincidence, pointing the diagnostician to one of the other categories, 1,2,3.
An elderly man with advanced prostate cancer presented with mental status changes, not readily explained by a basic workup. During rounds, because of the unrevealing nature of the data collected thus far, a lumbar puncture and an MRI were planned. As per usual, the MRI was delayed and the LP was performed. There were 3 WBC, but the protein was 108, ULN about 55. Does this represent paraneoplastic or infectious encephalitis? (Experience teaches that CSF protein is very frequently nonspecifically elevated in critically ill patients of all stripes.)
The MRI shows infarcts in the bilateral basal ganglia and thalami in a watershed pattern, explaining the altered consciousness. Should we also pursus an encephalitis of some sort? Heavens no!
Now that we have confidently arrived at a principal diagnosis that explains the clinical presentation, and we don't believe God's wrath is so great as to strike a man twice at one time, we can avoid the temptation of #4. #2 is also excluded as this was not a pre-existing disease. So, we have either an epiphenomenon of the principal diagnosis, or an indentaloma or non-specific finding. We need only to consult Open Evidence or a traditional source of medical information to learn that 50% of stroke patients will have non-specific eledation of CSF protein, as do garden variety critically ill patients. We are dealing with an epiphenomenon or a non-specific result, and we can stop right there. God is not so vengeful after all.
The frequency with which I see this error of basic diagnostic reasoning is stunning, and can lead to far worse diagnostic and therapeutic misadventure. In another case, a previously healthy 30-something man suffered a severe TBI in an MVA. CSF protein was elevated and he got diagnosed with autoimmune encephalitis and was given so-called pulse-dose solumedrol and started on cytoxan. I did not delve deeper into the details of the case because the dark horse was already out of the barn. But I know that, given these two scenarios:
- 30-ish male gets TBI and shortly thereafter gets autoimmune encephalitis (AIE) from an ireful God
- that male has antibodies detected in CSF outside of a clinical context compatible with AIE and they represent a false positive result
#2 is order(s) of magnitude more likely than #1. To confidently diagnose him with AIE, I need to build a case with more than just the lab finding; determine that any clinical findings are incompatible with TBI, or at least more compatible with AIE; and have a very clear treatment and monitoring plan. Exactly what do I expect to improve if I treat him with immunosuppression, and how will I determine improvement? If there is no improvement based on these parameters, I must be prepared to stop the immunosuppression, lest I commit him to a lifetime of therapy with clear harms but no benefits.
Unfortunately, false positive diagnostic labels stick like stink on a monkey. You can't wash it off.
Medicine is hard. Especially when you think reflexively, rather than reflectively.

No comments:
Post a Comment
Note: Only a member of this blog may post a comment.