April 2, 2018

Malpractice by a Computerized Decision-Support Tool?

Holland & Knight Healthcare Blog
Charles A. Weiss

Mrs. Skounakis died of a coronary artery occlusion after being prescribed an unusual combination of drugs (phendimetrazine and liothyronine) by Dr. Sotillo for weight loss. Her husband sued Dr. Sotillo for malpractice. What makes this case interesting to those other than the parties is the claim Mrs. Skounakis' husband also brought against Dr. G's Franchising Companies LLC ("Dr. G"), which provided Dr. Sotillo with a proprietary software package that was said to have endorsed the combination of drugs alleged to have caused Mrs. Skounakis' death.

The case was filed in the Superior Court of New Jersey, which held that the plaintiffs' expert witness was not qualified to render his opinions and accordingly granted summary judgment in favor of both Dr. G and Dr. Sotillo. In March 2018, the Appellate Division reversed, holding that the trial court erred in excluding the plaintiffs' expert and hence erred in granting summary judgment to the defendants.

The plaintiffs' expert, Dr. Decter, a board-certified internist and cardiologist, opined that Dr. G's program deviated from the standard of care by including a combination of phendimetrazine and liothyronine because of their well-known combined negative effects. The trial court held that he was not qualified to render this opinion because he was not a computer software expert:

"And certainly as to the software itself, ...there is no expert opinion as to any – from anyone about the software being grossly defective – or defective whatsoever. There's no testimony about it, there's nothing in the record about it, and Dr. Decter cannot possibly opine about the efficacy of the software."

In reversing the trial court's decision, the Appellate Division held that it was erroneous to have focused on the operation of Dr. G's software, as opposed to the question of whether the combination of drugs that it recommended was appropriate for Mrs. Skounakis:

"We, however, find no reason to conclude Dr. Decter was unqualified to render an opinion regarding Dr. G's program because he was not a computer software expert. The thrust of his concern about Dr. G's program was its endorsement that [Mrs. Skounakis] take the combination of medications that he believed led to her death. His opinion offers no – and need not offer any – insight about the details of Dr. G's computer software; he only discourses on whether the medications were appropriate for Melissa's quest to lose weight. Considering his clinical practice experience, Dr. Decter is qualified to opine regarding the propriety of the medications recommended by Dr. G's program."

As framed in this appeal, the proposition that the appropriate expert to opine on the propriety of medications prescribed to a patient is a physician is straightforward enough. That observation holds without regard to how or by whom the medications are identified, i.e., by a physician, a computerized decision-support tool such as that provided by Dr. G's, clinical practice guidelines, standard order sets or a text book. But three related questions that go beyond this anodyne observation were not addressed in this appeal.

First, does a "standard of care" apply to a computer program operating as a clinical decision-support tool? It is one thing to opine – as did the plaintiffs' expert Dr. Decter in this case – that a particular combination of drugs was inappropriate for a specific patient and that the physician who prescribed them did not meet the applicable standard of care for physicians of that practice in that locality. As the Appellate Division held, those questions are medical one to be answered by a physician. But the question of what degree of accuracy and precision are reasonably expected from a computerized decision-support tool such as the one supplied by Dr. G's here does not seem so clearly to be a medical one expected to be within the expertise of practicing physicians. The differences between what is expected from physicians treating patients and computerized decision-support tools becomes particularly clear if one views the potential liability of the former through the lens of medical malpractice and that of the latter through the lens of products liability.

Second, what is the required showing of causation for plaintiffs' claims against the supplier of a decision-support tool that provides its recommendations to a physician and not directly to the patient? Although not expressly stated in the Appellate Division's decision, it appears that Dr. G's program recommended the combination of phendimetrazine and liothyronine to Dr. Sotillo, not to Mrs. Skounakis. But even if the recommendation had been conveyed directly to Mrs. Skounakis as part of Dr. G's weight-loss program, both drugs are available only by prescription, and could not have been dispensed to her without a prescription from Dr. Sotillo. Regardless of the combination recommended by Dr. G's program, the courts would presumably hold that Dr. Sotillo owed a nondelegable duty to Mrs. Skounakis to exercise proper medical judgment, and not prescribe a particular combination of drugs because they were recommended by Dr. G's program.

Third, New Jersey (like almost all states) applies the learned intermediary doctrine, under which the manufacturer of a prescription drug satisfies its duty to warn of potential dangers by providing an appropriate warning to the physician (the learned intermediary) via the drug's prescribing information. Here, for example, the package insert for liothyronine cautions against its use for weight loss and includes a black-box warning against combining it with drugs such as phendimetrazine. The learned intermediary doctrine would undoubtedly apply if the plaintiffs had alleged that the manufacturer of liothyronine had failed to warn Mrs. Skounakis of these dangers; to the extent such a warning was necessary, the manufacturer's duty to provide it was satisfied by the warnings on the label. If Dr. G's recommendations were accompanied by an instruction to consult the prescribing information for the recommended drugs for further information, Dr. Sotillo would have found (if she had checked the labels) these warnings. If she decided to proceed with prescribing the combination despite them, would the learned intermediary doctrine bar the plaintiffs' claim against Dr. G's on the basis that even though its program recommended a potentially dangerous combination of drug, it directed Dr. Sotillo to review the prescribing information that would have warned her of the potential danger?

As medical protocols and treatment guidelines migrate from the familiar pocket reference guides stuffed into the lab coats of residents to programs that offer at least the appearance of individualized recommendations generated for a specific patient based on his or her relevant characteristics, watch for more cases like this one.

Citation: Skounakis v. Sotillo, A-2403-15T2 (N.J. Super. Ct. App. Div. Mar. 19, 2018).

Related Insights