Spectral Evidence and Other Poor Foundations For Decision-Making in Radiology

Do our assumptions and “satisfaction of search” prevent us from digging deeper on imaging reads?

I spent an extended Halloween weekend in Salem, Ma. this year. It had been on our "bucket list" for a while and one pretty much has to book a year in advance to get decent accommodations.

Lest your history classes were deficient or simply too far in the past for proper recollection, Salem's claim to infamy was a witch-hunting craze that seized the place in the early 1690s. One can take the subject as lightly or seriously as one wants when visiting Salem, from buying goofy merchandise to visiting historical sites.

We did both and were reminded of various reasons why the witch-hunting trials went as far as they did. One factor was the admissibility of "spectral evidence" in court. A witness could offer testimony that the accused's spirit ("spectre") had appeared to them as a result of supernatural powers granted by the devil.

It doesn't take much imagination to see how this stacked the deck in favor of the prosecution. When the jury is ready to believe accusatory "evidence" that is 100 percent subjective (if not entirely made up), what can one do in the name of cross-examination?

Most treatments of the subject matter acknowledge that, yes, folks are much more sophisticated nowadays. However, we shouldn't pat ourselves on the back too proudly over how we have evolved. There are other much more recent events thatv echo the witch-hunting trials. The Salem Witch Museum points out the "red scare" and McCarthyism just a few decades ago.

Everyone is capable of rushing to unfounded judgments. We may not take it as far as hanging people or otherwise ruining their lives. However, I have noticed how eagerly folks will allow themselves to consider flimsy or non-existent evidence as justification to form opinions and take actions.

One might hope, if not expect, that a good education would help counteract this. For instance, most folks who've gone through medschool have learned about things like confirmation, hindsight, and anchoring biases (amongst others). Being aware of common ways that reasoning goes awry should help us avoid drawing conclusions based on faulty evidence, right?

Unfortunately, I have seen that such education can also be a liability, and scientific pursuits including radiology are not at all immune. It is dangerously easy to assume one's learning and experience are nigh-impenetrable defenses. If I decided that X were true -- passing the muster of my education and experience -- I might consider X as proven as anything needs to be.

Some efforts are made to teach residents to remain humble in this regard. Consider, for example, the potential impact of "satisfaction of search" (aka "happy eyes"). Identifying an abnormality on an imaging study, especially one that answers the clinical question, has a way of dialing down our efforts to make other important diagnostic finds. In a way, we can be tempted to accept the "spectral evidence" that we have found the salient abnormality, render our judgment, and prematurely move on.

Another spectral evidence mimic is a leading clinical history that can bias us toward certain diagnoses (or lack thereof). For instance, suppose you receive a full-body CT for "mechanical fall," "found down," or the like in a nursing home patient who has had a dozen similar negative scans in recent years. You might be predisposed that CT image will also be a completely negative waste of time (or you might consider it quick and easy RVUs for zero injuries).

Alternatively, suppose you get the same scan, but the history tells you the patient fell out of a second-story window, has an obvious hip deformity, and is unable to bear weight. Convinced you will see a fracture or dislocation before you even see the images, you might have trouble believing your eyes when there is absolutely nothing there, not even a hematoma. In this situation, one might be tempted to hedge rather than rendering a normal report.

(If any referrers are reading this, please, please do not think this makes it okay to withhold pertinent history on that grounds that it "might bias" the radiologist. For anybody who thinks that kind of behavior is okay, please do everyone a favor, and go find another line of work.)

If we're vulnerable to spectral evidence when doing the image analysis for which we have been rigorously trained, is there any chance we're immune when dealing with the ancillary stuff surrounding it?

Suppose, for instance, you receive the aforementioned scan with no clinical history whatsoever. You might be ready to jump to all kinds of summary judgment about the referrer (thinking he or she is lazy, doesn't care, or is disrespectful of the radiology consultation ordered), the clerical staff who handled the referral, the CT tech who received the case and didn't get you an actual reason for exam, the order-entry system, or the hospital, etc.

Any of these judgments might turn out to be right or you might just assume them to be since you move on with your day and never get any evidence countering your appraisal of the situation. This may especially be the case if you have made similar unchallenged judgments before.

One susceptibility to faulty decisions is if they suit an established personal narrative. If you have already accepted the notion that we get lousy clinical histories because of laziness, incompetence, crummy infrastructure, etc., any new referral can get thrown into the same mental bucket. It is the same deal if you have a gut instinct that all "altered mental status" brain scans are going to be negative.

Such narratives (and faulty conclusions based on spectral evidence) may also factor into perceptions of higher order things, whether they are policies and politics in your rad group, or the overarching hospital system affiliated with the rad group.