Special Topic: The patient counts

Show Notes

Theranos is remembered for defrauding investors. For doctors the lesson is what happened to patients. Products that change care still reach them on their makers' word, before the evidence.

Read the special topic and its sources: https://physicianintheloop.org/specials/theranos-lessons/

The voice on this episode is generated by AI. It reads a special topic from Physician in the Loop as written; every fact in it links to its source.

What is Physician in the Loop?

How technological innovation is changing medicine for patients and the people who practice on the front lines.

This is a Physician in the Loop special topic.

The patient counts

Theranos is remembered for defrauding investors. For doctors the lesson is what happened to patients. Products that change care still reach them on their makers' word, before the evidence.

On Oct. 5 the Supreme Court turned away the appeal of Ramesh Balwani, who was president of Theranos. That leaves standing a federal appeals court opinion that reads in places like a case report. One woman testified that an HIV test she took through Theranos came back positive. Two later tests showed she was negative. Another testified that a Theranos blood test gave her incorrect results about her pregnancy.

A jury convicted Elizabeth Holmes of defrauding investors and acquitted her on the counts that concerned patients. Mr. Balwani, tried separately, was convicted on every count, patients included. For doctors, the patient counts matter most. What hurt those patients was an inaccurate test, and an honest company can sell one of those. Theranos sold its tests in pharmacies for two years before inspectors found that its laboratory endangered patients. A product that changes a patient's care should prove itself on patients before it is sold at scale, and the people who build it need a clinician with the power to stop it.

Ms. Holmes founded Theranos in 2003, after leaving college in her second year, according to the Securities and Exchange Commission. The company claimed it could run “fast, accurate, and affordable tests with just a drop of blood.” Of 300 small-sample tests it developed, the court wrote, only twelve were ever run on its Edison device. Inspectors from the Centers for Medicare and Medicaid Services arrived unannounced in late 2015 and concluded that the lab's practices posed “immediate jeopardy” to patient health. Theranos voided every patient result from the Edison. The appeals court said the trial judge reasonably treated this as an admission that the results “were, and always had been, unreliable.” When the agency imposed sanctions, Ms. Holmes said the company accepted “full responsibility for the issues at our laboratory in Newark, California” and had begun “comprehensive remedial actions.”

Dr. Adam Rosendorff, one of Theranos's laboratory directors, had trained in medicine and in laboratory medicine and had overseen testing at a children's hospital for five years. His job, the court wrote, was to make sure the tests leaving the lab were fit for patient care. He came to worry about “the systemic inaccuracy of the Theranos device.” He worried about proficiency testing that federal rules required and that management was unwilling to do properly, and about the pressure to account for bad results when patients and physicians complained. He had the duty but, in practice, not the power to stop the testing.

It is happening again, and it does not take a fraud. Cue Health's co-founders hold degrees in mathematics and in retail and consumer science, according to its 2023 proxy statement. The FDA authorized its COVID-19 test for use at home in 2021, the year the company reported $618 million in revenue. In May 2024 the agency told patients, caregivers and health care providers “not to use Cue Health’s COVID-19 Tests due to an increased risk of false results.” An inspection had found that the company changed the tests without the agency's authorization, and the agency said the changes reduced their reliability. The company said it was “evaluating the letter and determining its response.” Cue had begun a review of strategic alternatives, including a possible sale, by April. Two weeks after the warning, “after considering all strategic alternatives,” it filed for liquidation.

The Epic Sepsis Model, which scores a patient's sepsis risk every 15 minutes, was running in hundreds of American hospitals when Michigan researchers found, in a retrospective study of more than 38,000 hospital stays, that, at the alert threshold the hospital used, it missed two-thirds of patients with sepsis. It performed, they wrote, “substantially worse than the performance reported by its developer.” Epic said in an emailed statement, reported by Fierce Healthcare, that the study left out the tuning required before deployment and used a relatively low threshold, suited to “a rapid response team that wants to cast a wide net to assess more patients.” It has since rebuilt the model. When the FDA warned in 2022 that prenatal genetic screens, known as NIPS, can give false results, it had authorized none of them. It wrote that “pregnant people have ended pregnancies based only on the results of NIPS tests.” The screens were marketed as laboratory-developed tests, and a federal court has since struck down the FDA's rule to regulate those. The agency did not appeal.

Founders will call this unfair, and they have a case. The Theranos crimes were punished, with prison terms of more than 11 and nearly 13 years and $452 million in restitution. A sepsis score that misses cases and alerts too often is not a blood test sold by fraud, and a demand for trials before every launch could bury useful tools. But verdicts come after the harm. What protected patients at the time was an inspection, and it came late. The changed home tests, the sepsis model and the prenatal screens reached patients the ordinary way, on their makers' word.

Start with the buyer, usually a hospital full of doctors. Before a health system switches on software that changes care, it should do what the Michigan researchers did afterward: test it on its own patients' records and publish the results. Founders and their investors have the harder job. A company whose product touches patients needs evidence from patients before it sells at scale. It needs a clinician who can stop a launch and answers to the board, not the chief executive. Theranos had a laboratory director who saw the problem, and the testing went on.

The Supreme Court's order closed an appeal. It did nothing for the doctor deciding this week whether to trust a new score or a new screen. Courts count what founders took from investors. Somebody has to count what products do to patients, and count it before the product is everywhere. That job falls to doctors.

This special topic was read by an AI voice. Its sources are linked at physicianintheloop.org.