The claim
“Medical error is the third leading cause of death. Hospitals kill more people than they save, so stay away from them.”
The short answer
Medical errors are real and they do kill people. But the famous 250,000-a-year figure was an extrapolation from a handful of small studies. Studies that review actual hospital deaths find that roughly 3% were probably preventable, and avoiding care has its own body count.
Medical error isn't the third leading cause of death. Staying away from the hospital is the bigger danger.
A 2016 estimate said medical error kills 251,454 Americans a year, more than anything except heart disease and cancer. Here's how that number was built, what careful studies of hospital deaths found, and why it can be dangerous to believe hospitals are worse than staying home.
Why the number stuck
In May 2016, Johns Hopkins surgeon Martin Makary and a colleague published an article in The BMJ with a title designed to be remembered: "Medical error—the third leading cause of death in the US."
Their estimate was 251,454 deaths a year. According to the CDC figures Johns Hopkins quoted at the time, heart disease killed 611,105 Americans in 2013, cancer 584,881 and chronic respiratory disease 149,205. Medical error, the authors argued, slotted in at number three, ahead of chronic respiratory disease.
It's easy to see why the claim spread. It came from a famous hospital and a respected journal. It confirms something many families feel after a bad hospital stay. And it carried a fair point: US death certificates have no way to record that a medical mistake caused a death. Makary argued the coding system "was designed to maximize billing for physician services," not to count national causes of death.
Since then, the number has travelled a long way from the paper. Online it often turns into something the authors never claimed: that hospitals kill more people than they save, so you're safer at home.
That conclusion is wrong, and as an emergency physician I think it can get people killed. To see why, you have to look at how the number was made.
Where 251,454 came from
The 2016 article didn't count deaths. No one went through death records or hospital charts to produce it.
According to Johns Hopkins' own announcement, the authors took four earlier studies that had looked at death rates in hospital data from 2000 to 2008. They took a rate from those studies and applied it to the 35,416,020 hospital admissions in the US in 2013. That produced 251,454 deaths, which they said would be 9.5% of all US deaths each year.
That method has three big weaknesses.
The studies were small. One of the four was a well-known study of 10 North Carolina hospitals. According to a review by the federal Agency for Healthcare Research and Quality (AHRQ), it found only 14 deaths related to adverse events among 2,341 admissions. The AHRQ reviewer put it bluntly: it is "problematic to extrapolate from such small numbers of deaths." Multiply a handful of deaths up to 35 million admissions and a small difference in the starting count becomes tens of thousands of deaths at the end.
The studies weren't designed to answer this question. They were built to find harms, things that went wrong in the course of care. They were not designed to work out whether a death was caused by a preventable error. The AHRQ review notes that they reported deaths "sometimes without explicitly stating if the death was due to the error."
The pieces weren't combined properly. In a 2017 critique in BMJ Quality & Safety, patient-safety researchers Kaveh Shojania and Mary Dixon-Woods made this point. As AHRQ summarises it, the estimate combined error rates from earlier studies without a rigorous statistical synthesis and without accounting for uncertainty.
There's also a simple sanity check. In 2010, according to the CDC's National Center for Health Statistics, about 715,000 people died in US hospitals. If medical error really killed about 251,000 hospital patients a year, it would mean roughly one in every three hospital deaths was caused by a mistake. (The two numbers come from different years, 2010 and 2013, so treat this as a rough check, not a precise ratio.) That's not what studies of actual hospital deaths find.
A death with an error is not a death caused by one
This is the heart of the problem, and it's easy to miss.
The sickest people in a hospital get the most treatment: more drugs, more procedures, more lines and tubes, more days in bed. Every one of those is a chance for something to go wrong. So the sickest patients have the most errors and complications, and they are also the patients most likely to die.
That means you will always find errors in the charts of people who died. The question is whether the error caused the death, or whether the person died of the cancer, sepsis or heart failure that brought them in, with a mistake happening along the way.
Shojania and Dixon-Woods made exactly this point. The earlier studies used "trigger tools," which flag possible harms in a chart, and those tools can't reliably show whether a harm contributed to a death. Critically ill patients rack up more adverse events because they get more medical care. Their deaths may still be due to their illness.
A missed dose of antibiotics in a 92-year-old with widespread cancer is an error. It should be caught and fixed. But it isn't the same thing as an error that kills a healthy 40-year-old who would otherwise have gone home.
What happens when doctors review the actual deaths
There is a more direct way to answer the question: take the people who died in hospital, have doctors review each chart, and ask whether better care would probably have saved them.
That has been done many times. In 2020, a team of US researchers pooled the best of these studies in the Journal of General Internal Medicine.
- They found 16 studies that met their criteria. Eight that reviewed consecutive or randomly chosen deaths, 12,503 deaths in all, were pooled.
- A death was counted as preventable if most reviewers judged there was a greater than 50% chance it could have been prevented.
- The pooled rate of preventable death was 3.1% of hospital deaths.
- Applied to the US, that works out to about 22,165 preventable deaths a year, and about 7,150 among patients who were expected to live longer than three months.
- The authors concluded that most hospital deaths are due to underlying disease, and that most preventable deaths happen in patients with less than three months to live.
That last point matters. An earlier study of Veterans Affairs hospitals, published in JAMA in 2001, made it vivid. Internists reviewing hospital deaths rated 22.7% as at least possibly preventable, and 6.0% as probably or definitely preventable. But when the reviewers also estimated how long patients would have lived with perfect care, only 0.5% of those who died would have lived three months or more in good cognitive health. That came to roughly 1 patient per 10,000 admissions.
The same study showed how subjective these judgements are. Two reviewers looking at the same death agreed only modestly (a reliability of 0.34). Its subtitle says it all: preventability "is in the eye of the reviewer."
These studies have limits. The eight pooled ones came from the UK, Norway, Denmark, the Netherlands and Canada, not the US. They depend on doctors judging other doctors' care after the fact, which brings hindsight bias. But they are at least asking the right question, about the right deaths.
Other ways of counting give smaller numbers too
Every method of counting has blind spots, so it helps to look at several.
Death certificates. A 2019 analysis in JAMA Network Open, using the Global Burden of Disease study, counted US deaths where the death certificate listed adverse effects of medical treatment as the underlying cause. In 2016 that was about 5,180 deaths. Adverse effects appeared anywhere on the certificate, as a cause or a contributing factor, in 2.8% of all US deaths. The age-adjusted death rate from these harms fell 21.4% between 1990 and 2016.
That figure is surely too low, and its authors say so: death certificates are unreliable at recording medical harm, and many such deaths are probably never captured. Makary was right that death certificates miss errors. But the truth lies somewhere between "too low" and "too high," and the direct reviews of hospital deaths put it far closer to the low end than to 251,000.
Other expert estimates. The same AHRQ review mentions an estimate by a former AHRQ director of about 75,000 preventable in-hospital deaths a year. The famous 1999 Institute of Medicine report To Err Is Human said "as many as 98,000" Americans die each year from errors in hospitals; the 2020 review describes the report's range as 44,000 to 98,000. Even those higher estimates leave medical error well outside the top three causes of death.
Medical error is real, and it is serious
None of this means hospitals are safe enough. They aren't.
To Err Is Human launched the modern patient-safety movement for good reason. And a 2023 study in the New England Journal of Medicine, often called the SafeCare study, shows the problem hasn't gone away. Researchers reviewed a random sample of 2,809 admissions to 11 Massachusetts hospitals in 2018:
- At least one adverse event happened in 23.6% of admissions, nearly one in four.
- About a quarter of those events were preventable. A preventable adverse event occurred in 6.8% of all admissions.
- Most common were drug-related harms, followed by surgical or procedural problems, falls and pressure sores, and hospital-acquired infections.
That's a lot of harm, much of it avoidable. But notice what the same study found about deaths. Among those 2,809 admissions there were seven deaths, and one was judged preventable.
That is the honest picture: hospital harm is common, mostly non-fatal, and worth fighting hard to reduce. Shojania and Dixon-Woods made the point that an inflated death count can actually distract from that fight. As AHRQ summarises their argument, a focus on headline deaths pulls attention from harms such as pressure ulcers and medication errors, which rarely kill but matter a great deal to patients.
Johns Hopkins' own announcement made a point that often gets lost, too. The researchers said most errors aren't due to bad doctors, and reflect problems in systems, like poorly coordinated care.
What happens when people stay away
We don't have to imagine what happens when large numbers of people decide hospitals are too dangerous. It happened in spring 2020, when fear of catching COVID kept people at home.
- Emergency department visits fell 42% in the early weeks of the pandemic compared with the same weeks in 2019, according to a CDC analysis. Visits for heart attacks fell too. The CDC authors warned that people "could be delaying care for conditions that might result in additional mortality if left untreated."
- A second CDC report found that in the 10 weeks after the national emergency was declared, ED visits for heart attack fell 23%, for stroke 20%, and for dangerous high blood sugar 10%, compared with the 10 weeks before. The authors said the drop was too large to reflect a real change in how many people were having heart attacks and strokes. People were either unable to get care or were delaying or avoiding it.
- By June 2020, a CDC survey found that 40.9% of US adults had delayed or avoided medical care because of COVID worries, including 12.0% who had avoided urgent or emergency care. People with two or more chronic conditions, who had the most to lose, were about twice as likely to have skipped urgent care.
Heart attacks and strokes didn't stop happening. People just didn't come in. And deaths followed:
- A study in the Journal of the American College of Cardiology found that deaths from ischaemic heart disease rose nationally after the pandemic began, compared with the same period in 2019, as did deaths from high-blood-pressure-related disease. The rise was far steeper in New York City than in the rest of the country. The authors concluded the pandemic may have had an indirect toll on people with heart disease.
- In New York City, the rate of out-of-hospital cardiac arrests in which paramedics attempted resuscitation was three times higher in March and April 2020 than a year earlier, and on the peak day there were nearly ten times as many. Some of those deaths were COVID itself, so this number can't all be blamed on people avoiding hospitals. But it shows what happens when serious illness plays out at home rather than in an emergency department.
The lesson isn't about COVID. It's about what happens when fear of the hospital outweighs fear of the illness. For heart attacks and strokes in particular, treatment works best when it's fast, and every hour at home narrows the options. (A separate article covers the symptoms that shouldn't wait.)
Why the narrative is dangerous
"Hospitals kill more people than they save" doesn't follow even from Makary's own figure. More than 35 million hospital admissions happen in the US each year, and according to the CDC, about 2 in every 100 hospital stays end in death. The other 98 go home, many of them because of treatment they could not have received anywhere else. As a physician, I can tell you that for sepsis, heart attacks, strokes, major trauma, diabetic emergencies and difficult births, staying home is far more dangerous than any hospital.
The people most likely to act on the slogan are often the ones with the most to lose: older adults, people with several chronic illnesses, people who already distrust the system. Those were the groups in the CDC survey most likely to skip urgent care in 2020.
As an emergency physician, I'd much rather see someone who turns out to have indigestion than miss someone who stayed home with a heart attack because they'd read that hospitals were the third leading cause of death.
Being a careful patient is the better answer. Bring a list of your medicines. Ask what each new drug is for. Ask staff to wash their hands. Have a family member with you when you can. Speak up if something seems wrong. That's how patients help make hospitals safer. Staying away just moves the risk somewhere with no doctors in it.
The bottom line
The claim that medical error kills about 250,000 Americans a year came from an extrapolation, not a count: a rate from four earlier studies, some with very few deaths, multiplied across every US hospital admission in 2013. Those studies counted deaths where errors happened, not deaths that errors caused, in patients who were often already very sick.
When doctors review actual hospital deaths, about 3% look probably preventable, most of them in patients with weeks or months to live. That still adds up to thousands of deaths a year, and preventable harm short of death is far more common. Patient safety deserves serious attention.
But "stay away from hospitals" is the wrong lesson. When Americans did exactly that in 2020, visits for heart attacks and strokes fell and heart-disease deaths rose. If you think you're having a medical emergency, go.
Sources
- Makary MA, Daniel M. Medical error—the third leading cause of death in the US. BMJ. 2016;353:i2139.
- McMains V. Johns Hopkins study suggests medical errors are third-leading cause of death in U.S. Johns Hopkins University Hub, 3 May 2016.
- Shojania KG, Dixon-Woods M. Estimating deaths due to medical error: the ongoing controversy and why it matters. BMJ Qual Saf. 2017;26(5):423–428.
- AHRQ Patient Safety Network. Summary of Shojania KG, Dixon-Woods M. Estimating deaths due to medical error: the ongoing controversy and why it matters.
- Ranji S. Measuring and Responding to Deaths From Medical Errors. AHRQ Patient Safety Network, Annual Perspective.
- Landrigan CP, et al. Temporal trends in rates of patient harm resulting from medical care. N Engl J Med. 2010;363(22):2124–2134.
- Rodwin BA, et al. Rate of preventable mortality in hospitalized patients: a systematic review and meta-analysis. J Gen Intern Med. 2020;35(7):2099–2106.
- Hayward RA, Hofer TP. Estimating hospital deaths due to medical errors: preventability is in the eye of the reviewer. JAMA. 2001;286(4):415–420.
- Sunshine JE, et al. Association of adverse effects of medical treatment with mortality in the United States: a secondary analysis of the Global Burden of Diseases, Injuries, and Risk Factors Study. JAMA Netw Open. 2019;2(1):e187041.
- Hall MJ, Levant S, DeFrances CJ. Trends in inpatient hospital deaths: National Hospital Discharge Survey, 2000–2010. NCHS Data Brief No. 118. March 2013.
- Institute of Medicine. To Err Is Human: Building a Safer Health System. Washington, DC: National Academies Press; 2000.
- Bates DW, et al. The safety of inpatient health care. N Engl J Med. 2023;388(2):142–153.
- Hartnett KP, et al. Impact of the COVID-19 pandemic on emergency department visits — United States, January 1, 2019–May 30, 2020. MMWR Morb Mortal Wkly Rep. 2020;69(23):699–704.
- Lange SJ, et al. Potential indirect effects of the COVID-19 pandemic on use of emergency departments for acute life-threatening conditions — United States, January–May 2020. MMWR Morb Mortal Wkly Rep. 2020;69(25):795–800.
- Czeisler MÉ, et al. Delay or avoidance of medical care because of COVID-19–related concerns — United States, June 2020. MMWR Morb Mortal Wkly Rep. 2020;69(36):1250–1257.
- Wadhera RK, et al. Cardiovascular deaths during the COVID-19 pandemic in the United States. J Am Coll Cardiol. 2021;77(2):159–169.
- Lai PH, et al. Characteristics associated with out-of-hospital cardiac arrests and resuscitations during the novel coronavirus disease 2019 pandemic in New York City. JAMA Cardiol. 2020;5(10):1154–1163.