HINDSIGHT BIAS
Jury Trial
Clinical Practice
Standard of Care Issues
Medicolegal Practice Implications

Folk wisdom holds that hindsight is 20/20. It is not so self-evident that hindsight is a force to be reckoned with in daily medical practice and in medical malpractice litigation.

Jury Trial

Hindsight bias is insidious and not readily appreciated.

As we have previously noted, radiologist error rates are fairly steady at about 30%.

However, a Mayo Clinic study ⁽¹⁾ found that, of lung cancers that were eventually diagnosed on chest radiographs, 90% could be seen on earlier radiographs that had been interpreted as normal.

Defence counsel quoted this Mayo Clinic study in a recent US medical malpractice action.

Medical experts from both sides acknowledged that there were changes on the original radiographs that the defendant radiologist in this case had interpreted as normal.

However, defence and plaintiff experts differed on whether the abnormalities were so subtle that they personally would have spotted them without having previously seen the later, diagnostic radiographs.

The jury, having been taken through this backwards process, were unable to accept the defence expert opinion that plaintiff experts were exhibiting hindsight bias, and found for the plaintiff ⁽²⁾.

"In general people expect others to see what they themselves can see." ⁽³⁾

Clinical Practice

Hindsight can interfere with the exercise of clinical judgment.

The term hindsight bias was coined in 1975 and the first research topics were historical events and psychotherapy case histories.

Six years later, groups of practising physicians were given medical case history material; one "foresight-only" group were given no diagnosis, and each of the other four "hindsight" groups was told that a different one of the 4 possible diagnoses was correct.

The "hindsight" groups that had been given each of the least likely diagnoses judged that diagnosis much more probable than did the "foresight-only" group.

These test subjects, it seems, were attempting to make sense of the "true" outcome by working backwards, rather than analysing the data forwards ⁽⁴⁾. Knowing the supposed outcome interfered with the clinicians' judgment and diminished their diagnostic accuracy.

Similar problems occur in medical education, both undergraduate⁽⁴⁾ and postgraduate ⁽⁵⁾.

Consulting physicians who are providing second opinions are likewise biased: when told of previous diagnoses, they tend to confirm rather than refute them.

It is as if the physician who has hindsight will slip into the ski-tracks of his predecessors, rather than blaze a new trail.

Just as personal injury claimants unconsciously have revisionist memory for symptoms that favours Causation, physicians tend to adjust the facts to fit a "known" diagnosis.

Standard of Care Issues

Medical malpractice plaintiffs are favoured by hindsight bias.

The phenomenon of hindsight bias creates a common problem with medical malpractice litigation arising from misdiagnosis: other treating physicians - the colleagues of the defendant physician - tend to be uncritical of the faulty diagnosis.

This folie à deux encourages two alternative misinterpretations: either 1) multiple members of the treatment team provided negligent assessment; or 2) misdiagnosis in this particular situation is the community standard, against which the defendant physician is to be judged and will not be found to have failed.

However, the correct test is whether a reasonably conscientious physician would have made the faulty diagnosis "from scratch".

Some physicians have acknowledged how difficult it is when providing expert opinion to assess retrospectively routine clinical diagnosis^{(3, 6)} and laboratory interpretation ⁽⁷⁾.

For example, when Pap smears showing equivocally abnormal features were inserted, without special designation, into the routine workload of experienced technicians or pathologists ⁽⁸⁾, only 4% picked up on the abnormalities.

Seeing the same specimens designated as "special test slides", the same laboratory staff identified the suspicious features in 90% of the specimens.

Medicolegal Practice Implications

When radiologists or anatomical pathologists are asked to comment on Standard of Care in misdiagnosis, they should be given no more - and no less - clinical information about the case than was the defendant physician.

In particular, they should be given no information about outcome.

As a practicality, such clinical information is generally not available in the "complete medical record" of a hospitalisation. By Canadian hospital convention, the requisition forms are filed only in the diagnostic imaging department or laboratory, and must be explicitly requested from those locations.

Even so, the experts will inevitably be biased by the hindsight knowledge that diagnostic error has occurred.

A more realistic alternative to retrospective review, in the face of hindsight bias, is for the medical expert to submit the radiographs or slides to a small group of colleagues for blind rescreening.

That is to say, those reading the test material should be unaware that it is other than routine.

Ideally, any clinical specialist providing medical malpractice expert opinion on misdiagnosis should be given no more information than was available to the defendant physician at the time of the alleged negligence.

This similarly does not represent current medicolegal practice.

Copyright © Handbook of Legal Medicine