Clinical Case: Leg Pain (Part 2)
Part 1 is here.
(As before, I wrote this at the same time as the presentation, so none of the comments are taken into account.)
The neurologist ordered an MRA (magnetic resonance angiogram) of the aorta and iliac arteries which showed:
...Complete occlusion of the right common iliac artery. Distal disease involving the trifurcations bilaterally in the lower leg, with no discernible flow within either the left dorsalis pedis or posterior tibial.The translation is that I was absolutely correct in my original diagnosis: she had severely compromised blood flow to her right leg. She has been seen by a vascular surgeon (of great interest, it appears to be the same individual who read her original non-invasive study) and is scheduled for surgery.
A note to those who may feel duped by the fact that a neurologist ordered a "non-neurological" test: Yes, Virginia, there is such a thing as a smart neurologist. Specialists are capable of making diagnoses outside their areas of expertise (though it doesn't happen as much as they like to think it does); that's what happened here. Out in the real world (as opposed to the artificial realm of "case presentations") doctors aren't limited to certain types of testing based on their specialties, and on occasion, they save my ass by doing so.
My mistake was relying too heavily on the negative results of the first test, the doppler flow study.
The medical literature routinely discusses the statistical concepts of sensitivity and specificity, false negatives and false positives. All of these characteristics are highly dependent on the "pre-test probability" that the condition is present. That "probability" is essentially an estimate based on the history and physical; ie, the clinical impression. In this case, my clinical impression was extremely strong. She was a setup for vascular disease: a smoker with exertional pain relieved by rest. Was it incumbent upon me, in retrospect, to pursue a more definitive study before considering the diagnosis ruled out?
The ideal doctor would be completely familiar with each one of these characteristics for every single test ordered. I admit that as a busy clinician, I function more with a general sense of how reliable the test is. For example, I know that the specificity of a rapid strep test is better than its sensitivity. If it is positive, I can believe it; if negative, I send a regular throat culture to confirm it. But if pressed, I'm not sure I could quote the actual percentages for either of those figures.
Cursory web searches reveal sensitivity and specificity figures for pulse volume recordings to be in the mid- to high-90% range for detecting severe vascular disease. I did find an article about the limitations of measurements at rest. Although it recommended repeating the test with exercise to enhance specificity, it was still felt to be more an issue of identifying less severe disease.
However you spin it, it turned out I was looking at a false negative test. The patient told me that the vascular surgeon explained to her that it was likely due to the development of extensive collaterals. Because the blockages had developed slowly over a long time period, the surrounding arteries were able to dilate and carry more blood than usual; sort of like widening the surrounding network of neighborhood streets when there is long-term construction on the main highway.
The teaching point here is what to do when a test shows results that differ from a strong clinical impression.
I think the universal first reaction is to doubt oneself. This is probably more common in training, while clinical judgment is just beginning to develop. It continues well after that, though; no matter how strongly we may think we know what's going on with a given patient, we continue to be surprised on a regular basis. That's why they call it "practicing" medicine. The question here is when do you stick to your guns, and at what point do you admit you were wrong and go looking elsewhere?
This is a case illustrating just how hard it can be to strike that balance.
Afterword: Apparently my error wasn't all that obvious or uncommon. Only MSG and Artemis explicitly entertained the notion that the initial diagnosis was correct. I suspect that -- despite all the warnings I tried to give -- this was approached as a "clinical puzzle" type of case; ie, claudication was considered "ruled out" by the first test. Then again, it is a pretty darn good test; it doesn't often miss severe vascular disease. But this was real life, where these things happen on occasion.
In retrospect, or "the next time I'm in this situation," what would I do differently? That's an extraordinarily difficult question; I'm not at all sure I'd do anything differently. Chances are I'll order a couple more MRAs or CTAs that will probably be negative. Perhaps order "Stress PVRs" to enhance the test's sensitivity. I really don't know. Any thoughts?