ORLANDO, Fla. — “Harmful errors are often not disclosed, and when disclosure does take place, it often falls short of the expectations of the patient and the family and the clinical care team,” Suzanne M. Dintzis, MD, PhD, noted in an ethics session on Saturday at the College of American Pathologists (CAP) 2025.

When the fault lies within the lab, disclosure questions emerge for pathologists, including whether the concern should be reported to the patient and, if so, who should tell the treating provider or the patient, said Dintzis, breast pathologist at the University of Washington Medical Center in Seattle.

‘Never Enter a Disclosure Unprepared’

Dintzis offered advice on initial steps in reporting an error involving actual or potential patient harm. First: “Never, ever enter a disclosure unprepared.” Conduct a root cause analysis and determine whether care was reasonable. Dintzis was a coauthor on a paper with guidance for pathologists on how to respond after an error in Advances in Anatomic Pathology.

She cited an example of a clear laboratory error involving patient harm and inappropriate care, as opposed to care that a reasonably competent healthcare provider with the same training or experience would give. Two kidney biopsies were taken close together on the same day. One slide showed severe clinical rejection, and the other showed normal kidney function. The slides were accidentally switched and the reports mislabeled, resulting in clinical harm to both patients.

When patients have been harmed or care has been unreasonable, clinicians should involve a risk management team and notify their malpractice providers, too, Dintzis said. They also should disclose the error to the treating physician, rather than directly to the patient, she said. “They have a relationship with the patient that pathologists don’t.”

“You might offer to the treating clinician that you can be available or present should an error disclosure to the patient or patient’s family be required,” Dintzis added.

In almost every case, the patient will ask what’s being done to prevent the problem from happening again, so “make sure you have an answer,” or follow up if the answer is not immediately clear, Dintzis said.

She also cautioned against giving any answers related to the case over the phone, as callers might not be who they claim to be.

Ethics Around De-Identified Data

The growth of artificial intelligence (AI) in pathology has expanded ethical considerations for laboratory clinicians.

Brian R. Jackson, MD, MS, laboratory director and director of clinical pathology at the University of Maryland Medical Center in Baltimore, posed a hypothetical scenario, reflective of pitches colleagues have received: A start-up, looking to develop diagnostic software, offers a hefty payment for digitized copies of a pathology practice’s slides along with diagnostic reports — all of the personal information de-identified and Health Insurance Portability and Accountability Act (HIPAA)-compliant. The practice agrees to the deal.

Later, that same start-up may also purchase de-identified data from the hospitals in the area and buy claims data from insurance companies via data brokers, all de-identified.

“But with the magic of AI, you put those datasets together,” and data becomes reidentified, Jackson said.

Companies can and have used the information for marketing or other purposes, he said. “It’s not only possible, it’s commonplace.”

“A growing number of companies specialize in gathering longitudinal information from hundreds of millions of hospitals’ and doctors’ records, as well as from prescription and insurance claims and laboratory tests,” Scientific American has reported.

“This is one of the big loopholes in HIPAA law,” Jackson said. “Our hospitals and health systems and those of us at academic centers have long been operating under the idea that if we use de-identified data, we’re OK.”

His message for pathologists: “If you’re in a position to make a decision about sharing your practice’s data with any third parties, at a bare minimum, come up with contractual terms to ensure these downstream actions are prohibited.”

Who Checks the Quality of Diagnostic Software?

In another scenario Jackson offered, a software company develops diagnostic software, and the FDA approves the technology. A private equity firm then acquires your pathology group. It promises not to interfere with pathologists’ practice of medicine. But the firm also has a financial interest in the software, which it installs on all of your computers and encourages the practice to use in their diagnostic process.

“If you have software in your setting that is going to play a part in your diagnostic process, who should be overseeing the quality and the accuracy of that software? I think pathologists need to be officially and practically in charge of any software that’s going to be used in diagnosis,” Jackson said. “A regulatory approach needs to be developed, and that will probably require legislation,” he said, adding that pathologists’ advocacy will be important.

He acknowledged the role of chief medical information officers includes quality control in hospitals. But “most of them don’t have the same kind of sign-off responsibility on quality that a lab director would have for a clinical lab,” he said. “AI is a powerful tool, but AI, for that reason, needs competence in quality management. Frankly, the field of medical AI is really immature right now.”

Dintzis and Jackson reported having no relevant financial conflicts of interest.

Marcia Frellick is a Chicago-based, independent healthcare journalist and a frequent Medscape Medical News contributor.