Doctors will often say that the medical malpractice system requires that they be perfect, a standard that no doctor can actually reach. What we are actually talking about in this setting, though, are sensitivity and specificity.

Sensitivity is how often you find something that is actually present – a true positive. Specificity is how often you don’t call something when nothing is actually there – a true negative. In other words, if you are 100% sensitive you will find every condition and if you are 100% specific then every patient you find no problem in will be disease-free.

What worries doctors more are false positives, the over-calls that end up causing excess utilization that the doctor is then dunned for, and, most critically from a liability perspective, false negatives, the misses of what is actually present.

Being, well, human, every doctor will have their own mix of inclinations towards these four potential outcomes, only some of which will be conscious. How this manifests in each decision is what statisticians call Receiver-Operating Characteristics.

The cross-over with liability happens in the expression of the decision that the doctor eventually makes. The question is how to accommodate both sensitivity and specificity in a way that can stand as good evidence in a lawsuit or a medical board inquiry.

This takes us back to the essential duty of care because, as a physician, you are there to do two things: detect a medical fact, which goes to sensitivity, and evaluate that fact, which goes to specificity.

Documentation that shows that you met that duty should therefore include

  • What you examined or reviewed
  • Positives that you consider immediately important
  • Positives that you consider to be not of immediate importance
  • Your differential diagnosis, with the most likely condition(s) indicated as such
  • Any recommendations or anticipated actions to narrow the differential
  • A final impression

Considering that list in terms of sensitivity and specificity, we can see that 1, 2, and 3 are about sensitivity, verifying that you did a complete assessment and identified deviations from normal that could then be separated by standards that a reasonably prudent physician would use, and 4,5, and 6 are about specificity, demonstrating that you did not ignore less likely possibilities but that you ranked them as such and that you generated an actual decision.

What this comes down to is that there really are no “incidental” findings – there are just findings.  It is the extent to which they overlap with the current concern that prioritizes them, but not excluding that they may relate in other settings. For example, in a head CT performed on a patient who has just been in a car accident it is clearly necessary to exclude the possibility of a bleed or a skull fracture but a notation of multiple chronic pinpoint infarcts would be still significant guidance to a clinician who has been noting confusional episodes in that patient over the last year.

So, let’s return to where we started, the concern by doctors that they are expected to never make an error.  The sensitivity/specificity paradigm actually provides a clear framework here because what will be at issue will come down to whether you acted reasonably under the circumstances as they were at the time.  Your records will validate that because they will show that your interrogation of the case was adequately sensitive and that your grounds for specificity were appropriate, even if you were ultimately in error.

For example, suppose that a gastroenterologist consults on a case of likely diverticulitis and takes a full GI history, and records that the patient reports severe heartburn but then concentrates only on the colon-related issues. This report is fully sensitive to the fact of gastroesophageal reflux. However, when, a year later, the patient presents with esophageal carcinoma arising in the setting of a Barrett’s esophagus, that the doctor had made a note of a significant symptom but then passed by it without recommending the appropriate further evaluation that would underlie specificity would not be seen as reasonable.

On the other hand, when a radiologist reads a screening mammogram and identifies a small nodule and then evaluates it by ultrasound and sees no suspicious features and so recommends a 6-month follow-up on the basis that this is most likely benign, that report, which is correct in its sensitivity but actually incorrect in its specificity, will still be very defensible when the lesion is later identified to be a medullary carcinoma.

In summary, perfection in sensitivity and specificity is not the standard that you are held to. Rather, doing what a reasonably prudent physician would do under the circumstances to seek relevant findings – sensitivity – and to evaluate them – specificity – and documenting that you did so is what makes you defensible.