In 1993, the U.S. Supreme Court established a standard in Daubert v. Merrell Dow Pharmaceuticals, Inc. under which judges can evaluate the extent to which expert testimony is based on scientifically valid reasoning that can be applied to the facts in a specific case.1
While the Daubert rule has been useful in excluding junk science from the courtroom by requiring a factual basis for expert testimony, even facts are subject to interpretation.2 Notwithstanding experts’ best efforts, it has become increasingly obvious that they may often display bias of one type or another.
Regardless of whether the bias is informational or litigation-based, the inadvertent skewing of the “facts” at issue results in unpredictable and sometimes unjust outcomes in radiology-related
MPL litigation.
Four types of cognitive bias
Cognitive biases in MPL cases can be broken down into four subcategories.
Framing bias occurs when the way data is presented affects decision making. For example, in The Adventures of Tom Sawyer, the protagonist frames the chore of whitewashing the fence in positive terms, thereby convincing his friends to pay them for the “privilege” of doing his work.
Retrospective bias—also known as hindsight bias—describes people’s natural tendency to use current information to shape judgments of past events. The seminal example is, of course, the Monday-morning quarterback.
Outcome bias is a particular form of retrospective bias in which the result of subsequent events colors judgment made about the original event. Under this scenario, a reviewing expert who knows that an error results in actual damage is likely to judge that error as being more significant
than in the case where no actual damage resulted from the error (consider the saying, “No harm, no foul”).
Visual hindsight bias, almost unique to radiology, refers to the retention of visual information in the mind, either consciously or subconsciously. In the popular children’s book series Where’s Waldo? the reader tries to find Waldo in a visually cluttered picture. Initially, it can be a
very difficult task, but once found, viewers are more likely to find Waldo more quickly when returning to the picture, even after many months. Once seen, something can never be “unseen.”
These informational biases all operate in the MPL setting. When a lawyer contacts an expert to review any medical records, the expert knows that the story does not end well, because lawyers do not send experts records of unremarkable patients.
Once the review has been completed, the event at the heart of the litigation is brought into particular focus by the patient’s poor outcome. Finally, once noticed, this event remains prominently noticeable whenever the records are subsequently reviewed by anyone familiar
with the event.
Four types of litigation bias
Litigation biases in MPL cases can also be broken down into four subcategories.
Selection bias refers to the way experts are retained. When preparing for litigation, an attorney may consult with several experts to understand the case. When it comes to selecting an expert for depositions and trials, the attorney is most likely to select the one whose opinion best suits the attorney’s needs.
Undersampling bias results when a small subset of a population is not representative of the larger population yet is presented as being representative. At trial, each side is allowed an equal number of experts. This implies that expert opinion is evenly divided. Compounding the problem, the population of all possible experts is not known, and their opinions cannot be verified.
Compensation bias simply acknowledges that experts may naturally adjust their opinions nearer to alignment with those of the people who are paying them.
Affiliation bias, which is closely related to compensation bias, occurs when experts affiliate with one side and then become more likely to share the opinions and outlooks of their litigants’ team. This bias occurs independent of compensation. For example, at trial an attorney may
state his own point of view and then ask the expert to agree or disagree with the hypothesis presented. In this setting, the expert’s agreement may suppress nuances in opinions that they might otherwise have stated had they been given the opportunity of an open-ended question.
How to address biases
Given that expert testimony can be subject to such a variety of possible biases, MPL suits sometimes yield unexpected, if not unjust, results.
Several methods can address the problem.
The “no-fault” administrative model removes MPL claims from the tort system altogether. So-called “health courts” were proposed in the 1970s to operate under a “no-fault” administrative model, in which preset compensation is offered for injuries that are either avoidable or preventable. No-fault administrative systems for resolving malpractice cases are in use in Denmark, New Zealand, and Sweden, but have not found favor in the United States.
Evidence-based medicine (EBM) guidelines offer a way to eliminate the need for expert testimony altogether. Such guidelines are drawn from the scientific literature and claim to represent sound medical reasoning. Unfortunately, in many areas of medicine, EBM guidelines
do not exist. In others, there are an abundance of EBM guidelines that only an expert can sort through.
Some guidelines are general, intended for all patients, but others apply to specific subsets of patients who have additional restrictions. In this setting, expert testimony may be required to justify the application of a particular EBM guideline to a particular patient’s condition, reintroducing the problem of bias that use of the guidelines sought to eliminate.
Given that neither “no-fault” nor guidelines-based rubrics adequately address the problem of expert bias, other methods must be used to maximize objectivity. Attorneys Christopher Robertson and David Yokum performed an experiment with mock juries to test the effect of expert blinding on jury decisions in an MPL setting.<sup>3</sup> Some participants were told that the defense expert had been hired by an intermediary and was blinded as to which side of the suit they were testifying. Others believed that the plaintiff ’s expert had been so blinded. Robertson and Yokum concluded that juries found blinded experts to be significantly more credible than non-blinded ones.
Expert blinding methods
Because of the nature of medical records, expert witness blinding is impractical in most settings. However, radiology litigation is uniquely suited to expert blinding because the matter at issue is usually the interpretation of an exam that is itself available for review. An expert can be blinded by techniques of deception or concealment.
The “sub-rosa” approach disguises the exam at issue as a routine current exam and inserts it into the daily workload. The expert witness is thus deceived as to the contentious nature of the exam itself. While intellectually appealing, this approach presents several practical
challenges.
Foremost among these is the conflict between engaging an expert and having that expert remain naïve to the litigation exam when it is presented. The exam must be disguised, at least to the extent that it would need to appear to be a current exam, from a facility served by the
expert in the course of daily work.
Medical enterprise security requirements generally do not allow knowingly false patient information to be stored on their systems. Additionally, a cooperating actor must be available by telephone, should the expert want more information or to discuss the exam with the “ordering” provider.
The other two approaches involve an intermediary to deidentify and conceal the exam at issue in a small set of similar deidentified exams drawn from a preexisting database. These exams do not have to be identical in nature to the exam at issue, but similar in type. The review set exams can contain both normal and abnormal findings. The expert is told that one or more of the exams in the set is the subject of litigation.
In the “second-read” approach, the expert reviews each exam in the set and offers an interpretation of that exam to the intermediary.
The “oversight” approach presents the original radiologists’ report along with the exams in the review set. In this scenario, the expert reviews the exam along with the report and reports to the intermediary whether or not the report met the standard of care as an interpretation of that particular exam.
In both the second-read and oversight approaches, the expert knows that one or more of the exams in the set is the subject of litigation. This knowledge forces the expert to evaluate each exam as an independent event. The expert’s assessment of the exam at issue is then presented to the attorney.
Compensation bias can be addressed in these settings by having the expert initially contacted by an intermediary. In this way, the expert does not know which side of the litigation has engaged them.
Both second-read and oversight methodologies are commercially available<sup>4</sup> and have been applied in more than 130 lawsuits in several states by plaintiff and defense attorneys. Both attorneys and the radiologist experts retained to do the reviews have found the process reliable
and the subsequent testimony compelling.
A final note
Informational and litigation biases magnify the natural differences of opinion between medical experts in MPL litigation. Since expert testimony is likely to remain critical to these cases, blinding techniques offer a method to minimize these biases.
Appropriate blinding requires that the expert be unaware of the contentious nature of the particular issue in dispute while developing an opinion. Deception and concealment techniques represent two useful blinding methods. Such blinding is impractical in many medical settings but can be effectively accomplished in radiology litigation.
True objectivity in expert witnesses may be impossible to achieve in the MPL realm. However, blinding the expert can eliminate some biases and mitigate others.
References
1. “Daubert Standard,” Legal Information Institute, Cornell University Law School,
https://www.law.cornell.edu/wex/daubert_standard678723q3.
2. Daubert v. Merrell Dow Pharmaceuticals, Inc., 1993. 509 U.S. 579.
https://supreme.justia.com/cases/federal/us/509/579/.
3. Robertson, C.T.. Yokum, D.V., 2012. The effect of blinded experts on juror
verdicts. Journal of Empirical Legal Studies 9 (4), 765-794. https://scholarship.
law.bu.edu/faculty_scholarship/1147/.
4. www.cleareview.com.
Jeffery D. Robinson, MD, MBA, FACR, is the Founder and President of Cleareview.
Neerav Mehta, MD, MS, is the CEO of Cleareview.