I have spent more than 25 years in operating rooms, and in every one of them, a pulse oximeter has been clipped to my patient's finger. Surgeons live by that number. When it holds steady, we keep working. When it drops, everything stops.
For most of my career, I never questioned it. The pulse oximeter is one of the most basic tools in medicine: inexpensive, fast and, we believed, reliable for every patient. Until we discovered it wasn't.
During the worst of the COVID-19 pandemic, physicians caring for critically ill patients noticed something troubling. Pulse oximeters were showing normal oxygen levels in some Black patients whose clinical condition suggested otherwise. That bedside observation prompted a landmark study—funded by the National Institutes of Health (NIH) and the Veterans Health Administration—that examined tens of thousands of paired oxygen measurements from patients at 178 hospitals nationwide.
The findings were unambiguous. Pulse oximeters missed dangerously low oxygen levels in Black patients nearly three times as often as in white patients. Calibrated decades earlier using volunteers with predominantly lighter skin, the devices sometimes reassured physicians when patients were, in fact, becoming dangerously hypoxic.
That discovery changed medicine. It reshaped how physicians interpret pulse oximeter readings. It prompted updated guidance from regulators. It accelerated improvements in how manufacturers evaluate these devices. Most importantly, it helped patients receive safer care.
None of that happened because someone set out to prove a political point. It happened because researchers were free to ask a simple scientific question: Does a medical device that physicians trust work equally well for every patient?
That's not just how medicine advances; it’s how society advances.
There was a time in this country where we believed any number of things that science later proved to be false. We once believed that human flight was impossible, or that the Earth’s continents were fixed in place. What changed?
We question assumptions. We test accepted wisdom against evidence. And when the evidence tells us we've been wrong, we change.
That is why I am concerned about a proposed rule from the White House Office of Management and Budget that could make it more difficult to support exactly this kind of research.
Obscure rule, big impact
Most physicians have never heard of the proposed revisions to the federal government's “Uniform Guidance,” the regulations governing how billions of dollars in federal research grants are awarded and managed. But these rules influence the kinds of questions researchers pursue, how grant proposals are evaluated, and ultimately the evidence physicians rely on to care for patients.
Buried within the proposal is language prohibiting federal funding from supporting theories related to “disparate-impact liability.” No one disputes that taxpayer dollars should never support unlawful discrimination. As currently drafted, however, the provision risks blurring an important distinction between advancing a legal theory and conducting scientific research that asks whether a treatment, diagnostic test, device, or clinical algorithm performs differently across patient populations. That distinction matters because some of medicine's most important corrections have come from asking exactly those questions.
Pulse oximeters are hardly the only example. For more than two decades, the equation physicians used to estimate kidney function included a race coefficient based on assumptions that later proved unfounded. As a urologist, I relied on those estimates constantly—deciding whether a patient's kidneys could tolerate surgery, adjusting medications, weighing when to involve a nephrologist.
Federally funded researchers demonstrated that the adjustment delayed referrals for dialysis and kidney transplantation for many Black patients. They then developed race-neutral equations that are now used nationwide. A recent analysis found the change expanded access to kidney transplantation for thousands of patients while reducing reliance on years of costly dialysis.
Again, the goal wasn't ideology. It was better medicine.
What happens when we fail to ask
I know the value of asking these questions because I spent years asking them myself. As an NIH-funded researcher, I studied whether men diagnosed with prostate cancer—a disease I would later confront as a patient—received the same quality of treatment regardless of who they were.
Too often, the data showed they did not. Research like that wasn't about checking demographic boxes. It produced stronger science, and it gave physicians greater confidence that the care they deliver works safely and effectively for the people in front of them.
We have seen the consequences when medicine fails to ask. For years, many clinical trials did not adequately include women. A government review found that eight of the ten prescription drugs withdrawn from the U.S. market between 1997 and 2000 posed greater health risks to women than to men. Better evidence doesn't simply improve scientific understanding. It prevents avoidable harm.
This is exactly what the administration has described as “Gold Standard Science”—testing assumptions, following the evidence wherever it leads, and changing course when the evidence demands it.
Medicine is now entering another period of profound transformation. Artificial intelligence (AI), predictive algorithms, wearable technologies and continuous monitoring are rapidly becoming part of everyday clinical practice. These innovations have enormous potential to improve care. They also introduce new opportunities for hidden bias, unexpected blind spots, and uneven performance across different patient populations.
The lesson of the pulse oximeter has never been more relevant. Somewhere in American medicine today, a device, algorithm, laboratory test or AI system is failing a group of patients in ways we do not yet recognize. We don't know what it is. We don't know whom it affects. But history suggests it exists.
Finding it will require researchers to ask difficult questions about the tools physicians trust most. It will require physicians to remain willing to challenge accepted wisdom. And it will require a research environment that encourages rigorous inquiry rather than making scientists wonder whether certain questions have become too risky to ask.
Medicine has earned the public's trust not because we claim to have all the answers, but because we are willing to question them. Every major advance begins with someone discovering that something we accepted as true wasn't.
The next pulse oximeter-level failure is out there. We should be making it easier to find—and fix.
Comments on the proposed rule are due July 13, and anyone can submit one.
