Cognitive Bias Is Influencing Forensic Pathology Decisions

The decedent is a young woman found in bed, in pajamas, with vomit on the pillow under her head. She is a known drug user. You do an autopsy and find nothing substantial to explain her death — some superficial injuries but no fatal trauma. No neck injuries. Her face is red, maybe with a few petechiae, findings consistent with being face-down on a pillow. Nothing else.

What would you do if you were given each of the following three scenarios?

A. The room was locked and secure. The death was reported by the decedent’s sister, who says she found the body, and no one else was in there.

B. The room was locked, and the body was found by her boyfriend. He admits to using drugs with her the night before. He says he found her face-down and flipped her over.

C. The room was unlocked, empty except for the body. The neighbors heard the decedent arguing with her boyfriend the night before. He has a prior history of domestic abuse. The police track down the boyfriend. He says they were using drugs together, and he is still intoxicated when questioned by the police. He confesses that he must have done something to her.

Will your cause and manner of death be different depending on the scenario presented? What if the toxicology comes back with a lethal combination of drugs and alcohol? If your findings are different, how? And would you change your conclusions if the case was first presented as scenario C, but then you later found out that what really happened was B…?

I was first presented with the case of C turned to B as a legal consultant after the fact, and I was asked to comment not only on the cause of death but also on methodology. Did the pathologist who performed the autopsy follow correct forensic procedure? It was my professional opinion that no, they did not. They did not do a posterior back or neck dissection, even though they were informed before the autopsy of the suspicious circumstances. Their documentation of findings was sparse and insufficient to the case. I suspected that the autopsy pathologist had performed a substandard exam as a result of cognitive bias: that being informed of the boyfriend’s arrest and putative confession in advance made this doctor less likely to pursue other possible causes of death, or even to wait for the toxicology results to come back, before they informed the police that they thought the victim had been fatally asphyxiated.

What is cognitive bias, and how do we address or minimize it in the practice of forensic pathology? I don’t have all the answers, but I know this: Denying that bias exists is not the way to start. Many of my colleagues disagree. They have made their objections to the examination of bias in our profession known in a series of letters-to-the-editor of the Journal of Forensic Sciences, following the recent publication of a peer-reviewed research paper, “Cognitive Bias in Forensic Pathology Decisions.” I am one of the seven authors of that study.

Another one of the authors, University College London cognitive scientist Itiel Dror, PhD, developed a protocol called Linear Sequential Unmasking (LSU) as an approach to minimize bias in other forensic disciplines. LSU has a specific utility and method of application. Protestations from my confused colleagues notwithstanding, LSU does not in any way dictate that coroners or police should restrict what forensic pathologists get told about a death scene.

Cognitive scientists and forensic pathologists concerned about bias might, however, apply LSU to a more solutions-based approach to the problem of bias, to find protocols that could fit offices of all sizes. Forensic pathologists need their patient’s pertinent medical histories, including the circumstances of death, in order to do their job properly. But to examine if scene information is a source of bias, we need to first examine whether there is evidence of bias in our current investigation procedures and in the death certification rulings that follow. For example, are we more likely to assign a manner of death as undetermined if the decedent is poor, of a certain race, died while in police custody, or has a history of substance abuse? Are we more likely to limit the postmortem on a hanging to an external examination if the family is wealthy and objects to an autopsy? Would that decision change if we knew there was prior history of domestic abuse and a large insurance payout to the decedent’s spouse? Are we more likely to call an exit wound an atypical entrance wound or an entrance wound a shored exit wound if we were told different witness accounts of how the shooting occurred?

Instead of denying the existence of cognitive bias in our profession, as a community we might be proactive and consider ways we can minimize extraneous influences, and test to see if those strategies actually work. Can the pathologist determine anything salient from visiting a death scene first, before being told about equivocal eyewitness information? Is the pathologist more or less likely to pick up on subtle clues that can help make the correct death determination by going to the scene in person, rather than getting a secondary account in police reports? What about old or cold cases that present a suspicion of bias — is there a way to perform an unbiased or less-biased retrospective analysis? Could the reviewer be given the investigative materials in a certain order, like microscopic slides first, then autopsy photos, and then autopsy report, rather than the other way around? Does the order in which forensic pathology experts review information change whether they catch certain medical clues? Order effects have proven very influential in other fields, as initial information is not only remembered well, but it generates hypothesis, and drives cognitive attention and testing strategies.

Let’s explore these questions as a professional community in the spirit of scientific inquiry, continuous reflection, and improvement. We first must acknowledge that cognitive bias has an impact on our work, as it does in clinical medicine and forensic science. We definitely do not want to implement any policies to mitigate bias that might instead have the opposite effect — that make our determinations more likely to be wrong. The last thing we need is an outside agency restricting information in a misguided attempt to “unbias” us. Every year, U.S. medical examiners and coroners offices become more understaffed and more underfunded. We don’t need new policies and procedures that make our work more complicated and burdensome.

But we can’t ignore the problem, either. The Journal of Forensic Sciences cognitive bias paper has got forensic pathologists talking about bias. It may not be an easy or a welcome discussion, but it is needed if we are to advance our field in the pursuit of — literally — truth and justice.

Judy Melinek, MD, is a forensic pathologist and CEO of PathologyExpert Inc. She is currently working as a contract pathologist in Wellington, New Zealand. Her New York Times bestselling memoir, co-authored with her husband, writer T.J. Mitchell, is Working Stiff: Two Years, 262 Bodies, and the Making of a Medical Examiner. The duo have also embarked on a medical-examiner detective novel series with First Cut, available from Hanover Square Press.