Perspective: Are radiologists blind? They should be.

By Charles A. Pilcher MD FACEP
September, 2015

Last year in the Journal of the American College of Radiology, Durand et al. published a thorough review of the biases inherent in experts’ review of radiologic images during the course of med mal litigation. They then discussed strategies that might be helpful in mitigating those biases. Most of the information in the review applies to biases likely to be present in testimony of expert witnesses in all areas of medicine, as I have written about in a previous “Perspective.”

Durand et al. divide the biases into the following categories:

Informational biases:

  • Contextual bias: The unavoidable changes in attention level and diagnostic threshold when a reviewer knows that a case is in litigation. This increases sensitivity at the expense of specificity.
  • Hindsight bias: The tendency to overestimate the conspicuous-ness of a finding once one knows that it is present.
  • Outcome bias: The tendency to impart greater negligence to a worse outcome.

Litigation biases:

  • Selection bias: Picking the most favorable opinion from a number of experts consulted early
  • Undersampling bias: Presenting to a jury/judge only 2 opposing opinions for a 50/50 weighting, when the larger universe of experts might be 90/10.
  • Compensation bias: The  shift in opinion toward the sponsor paying for the opinion of the expert.
  • Affiliation bias: The inevitable subtle manipulation/embellishment of opinion resulting from multiple and repeated contacts and conversations between the sponsor and the expert.

Suggested remedies to reduce inherent biases:

  • Case history blinding: Withholding information from an expert regarding the relevant finding and case outcome. (Addresses “hindsight” bias)
  • Context blinding: Embedding a vector case into normal radiology workflow. (Addresses “contextual” bias and is sometimes used currently as part of radiology quality control.)
  • Panel review: Having multiple blinded experts review images and present the results of the review as a percentage or fraction. (Addresses “selection” and “undersampling” biases)
  • Use of intermediaries: This prevents direct contact between attorneys and experts until a formal opinion has been rendered. Experts would be selected randomly from a pool of experts, who would agree to testify for either side. Attorneys then choose whether or not to use the opinion. (Addresses “selection,” “affiliation” and “compensation” biases)

All of these strategies apply to experts in all fields of medicine to one degree or another. All have varying degrees of simplicity, practicality, effectiveness and cost. The goal remains to bring the most accurate, objective information available into the litigation so as to eliminate manipulation of the actual facts or “standard of care” in contemporary medical practice.

To read the entire original article click here:

[Editor’s Note: I strive to minimize these inherent biases as much as humanly possible. My approach has always been to receive as little input up front as possible, review records in as virgin a state as possible and then ask myself “Whose side can I best represent?” The result is that I support the position opposite that of the attorney contacting me approximately 70% of the time in plaintiff cases and 50% of the time when contacted by defense counsel. This leads to fewer lawsuits, less wasted time and money and earlier settlements. My attorney clients seem to appreciate this because they keep calling me back. In all cases I am happy to explain my position to the client of the attorney seeking my advice. In plaintiff cases, that explanation may provide the answers and reassurance the plaintiff was seeking in filing a lawsuit. I have nothing to sell besides my opinion and will never do that at the expense of my integrity.]

Leave a Comment