A few months ago, I cared for an older patient that I will call Mr. Jones. He lived alone and one afternoon, after a neighbor found him disoriented, was brought into the emergency department. Diagnosed with pneumonia, he was admitted and started on standard guidelines-based antibiotics. By next morning, he had taken a turn for the worse and was sent to the intensive care unit where he struggled and eventually died 72 hours later. Troubled by the death of a seemingly healthy man, I spoke to his daughter who lived in California. She mentioned he had been admitted to another hospital two months prior, also for pneumonia. The lab at that hospital told me his pneumonia had been caused by an unusual bacterium susceptible to only a few antibiotics—none of which we had used. And days later, when our lab confirmed that Mr. Jones’ current pneumonia was caused by the same bacteria, I knew his death had been wholly preventable. Despite our best efforts, we all missed it.

A recent report in the British Medical Journal (BMJ)estimates that 250,000 Americans die each year due to medical errors. This report sparked a firestorm. Skeptics carefully documented why that number might be wrong. While other estimates have put the number of deaths from medical errors at 100,000 , many in the medical community believe the number of cases is much smaller. The truth depends largely on the most important question of all: what is a medical error?

Some instances of medical errors are obvious. A patient gets 50 times the standard dose of a medication because of a typographical error, or has the wrong kidney removed. These cases add up to only a small fraction of the deaths that are attributed to medical errors. Most cases that contribute to those eye-opening numbers are cases of hospital-acquired infections (come in for knee replacement, leave with pneumonia) or medication errors (get a drug that interacts adversely with another drug). Critics of the medical error studies argue that many of these are not preventable and should not be thought of as errors. Indeed, they argue, this is the cost of providing technically advanced care to sick patients. They have a point, of course, but only to a degree.

In 2016, we live in the world of complex medicine—we have thousands of treatments that save lives that would have been lost just a few decades ago. In this world, saying that medical errors are the third leading cause of death doesn’t seem quite right. It implies individual sloppiness—careless doctors and nurses walking around, performing wrong-site surgery or not paying attention to what medications they give. Nothing could be further from the truth. Doctors and nurses work diligently to manage the deluge of information they face. I routinely care for acutely ill patients who are on 20 medications at home—and struggle with how to manage their multifaceted chronic diseases while getting them through their acute one. Healthcare today is about managing complexity and the system isn’t performing well. The problem is not about individual sloppiness, it is about system sloppiness. From that angle, the 250,000 number seems far more plausible. But it still depends on what we call a medical error.

Which brings us back to Mr. Jones. Was his death a medical error? The admitting physician didn’t know he had been hospitalized for pneumonia elsewhere. Even if he had known, he probably could not have gotten laboratory results from that hospital in the evening. So he made the best, evidence-based choice—and that hardly feels like a medical error. Yet, saying his death was not preventable is also wrong. The system failed him. A system designed for complexity would have alerted us that he had gotten care at another institution. It would have allowed us to look up the microbiology results, even in the middle of the night, so we could have made a better antibiotic choice up front – a choice that was customized to him, not to the generic patient.

Mr. Jones had never been admitted to our hospital before, so his medical record was thin. But I often have the reverse problem, where I see a patient with literally 50 previous hospitalizations. It would take me days to carefully go through all of those records, so I skim, praying that I don’t miss anything important. A smart system designed for the 21st century would use algorithms ensuring I saw information that was relevant for my patient. These algorithms exist everywhere outside of healthcare—Google knows what I’m searching for before I do—but in our $3 trillion healthcare system, they haven’t been prioritized.  

All of this begs the question why—why is our system so sloppy? The answer: because it can be. Because the costs of medical errors are hidden. No senior hospital administrator or Washington policymaker heard about Mr. Jones’s death. Even when the errors are more obvious, the financial consequences to institutions are small. And until we fundamentally shift the incentive structure of the system so that the most careful and safe systems get rewarded handsomely (and the poor performers get punished), none of the sloppiness goes away.

So we come back to the BMJ study—are there 250,000 deaths each year due to medical errors? We don’t know, but all the data suggests that the true number is very large. Our system fails so many people every day—through errors of commission and errors of omission—that 250,000 is not outside the realm of possibility. The general rule is that the harder you look, the more you find.  In most studies, Mr. Jones would not have been flagged as a medical error. But he deserved better. So whether we call it medical errors, system failures, or just failing to live up to what we know is possible—the bottom line is that hundreds of thousands of Americans suffer due to these failures. And we know how to do better.