How Many People Die From Medical Errors Every Year?

Picture a hospital ward at four in the morning. A nurse is three hours into a twelve-hour night shift, looking after eight patients. One of them, an 84-year-old man admitted with a chest infection, has a slightly faster heart rate than at midnight and a temperature that has crept up half a degree. On their own, neither number sets off an alarm. Together, and spotted early, they are the first whisper of sepsis.

On most nights, someone notices. The doctor is called, antibiotics are changed, fluids go up, and by breakfast he is sitting up asking for tea. On some nights, the whisper is missed until it becomes a shout. That gap between "noticed in time" and "noticed too late" is what the world of patient safety is really about.

It is also one of the most argued-over numbers in medicine. You may have read that medical errors are the third leading cause of death in America. You may also have read that this claim is badly wrong. Both views come from serious researchers. This article walks through the evidence, fairly, so you can see where the real numbers sit.

The headline number

The broadest estimate comes from the World Health Organization. According to WHO's patient safety fact sheet, more than 3 million people die every year as a result of unsafe care worldwide. Around 1 in 10 patients is harmed while receiving health care, and more than half of that harm is considered preventable.

Time period	Deaths linked to unsafe care (approx.)
Per year	3,000,000
Per month	250,000
Per week	57,700
Per day	8,200
Per hour	340
Per minute	5.7

To put that in context, roughly 170,000 people die every day from all causes, which you can watch on our live death counter. If WHO's estimate is right, unsafe care is linked to something like one in every 20 deaths on the planet.

One important caveat before going further. "Unsafe care" is not the same as "a doctor made a mistake". WHO's definition covers harm from many sources: infections caught in hospital, medicines given wrongly, missed diagnoses, falls on wards, blood clots, and failures of whole systems such as staffing or record-keeping. Many of the people counted were already seriously ill. Unsafe care may have shortened their lives rather than ended a long healthy one.

How that compares with other killers

Even with those caveats, the global figure is large enough to sit alongside some of the best-known causes of death.

Cause	Approx. deaths per year	Approx. per day
Unsafe care (WHO estimate)	3,000,000+	8,200
Tuberculosis	1,250,000	3,400
Road injuries	1,190,000	3,260
HIV/AIDS	630,000	1,730
Malaria	597,000	1,640

You can explore the others on our tuberculosis, road injuries, HIV/AIDS and malaria pages. Unlike those, though, unsafe care is not a single disease with a death certificate code. That is exactly why the numbers are so contested.

Where the harm happens

WHO's fact sheet breaks down the main kinds of harm patients experience. These are the ones that most often contribute to deaths.

Type of harm	What WHO reports
Medication errors	Affect about 1 in 30 patients; around half of all avoidable harm is medication-related
Diagnostic errors	Occur in 5% to 20% of doctor-patient encounters
Healthcare-associated sepsis	23.6% of hospital-treated sepsis began in a healthcare setting; about 24.4% of those patients died
Blood clots (venous thromboembolism)	About a third of complications linked to a hospital stay
Patient falls	3 to 5 per 1,000 bed-days, over a third causing injury
Pressure ulcers	More than 1 in 10 adult hospital patients
Surgical harm	About 10% of preventable harm; over 300 million operations a year

Sepsis deserves a special mention, because it links so many of these threads. It kills around 11 million people a year in total, and a significant share of cases start in hospitals or care homes. We cover it in depth in how many people die from sepsis every day.

Falls are another. Older patients in unfamiliar wards, on new medicines, getting up at night to use the toilet, are at particular risk, as we explain in how many people die from falls every day.

Most of the harm is not dramatic. It is not the wrong leg amputated or a surgical sponge left inside. It is a dose that was slightly too high, a test result that was not chased, a warning sign that was written down but not acted on.

The American argument: 22,000 or 250,000?

The most famous medical error figure comes from the United States. In 2016, Dr Martin Makary and Michael Daniel of Johns Hopkins published an analysis in the BMJ estimating that 251,454 Americans die from medical error every year. If true, that would make medical error the third leading cause of death in the US, behind only heart disease and cancer.

The claim travelled around the world. It is still repeated constantly. But it was never a new study. Makary and Daniel took death rates from four earlier studies and extrapolated them to all US hospital admissions in 2013. Here is how their estimate compares with others.

Study	Year	Estimated US deaths per year	Method in brief
Institute of Medicine, "To Err Is Human"	1999	44,000 to 98,000	Extrapolated from hospital record reviews in two states
James, Journal of Patient Safety	2013	210,000 to 400,000	Pooled four studies of adverse events
Makary and Daniel, BMJ	2016	251,454	Extrapolated from the same four studies to all 2013 admissions
Rodwin et al., Journal of General Internal Medicine	2020	about 22,000	Meta-analysis of 12,503 hospital deaths reviewed by doctors
Newman-Toker et al., BMJ Quality & Safety	2023	371,000 deaths following misdiagnosis	Disease-by-disease model of diagnostic error, all settings

That is a spread of more than tenfold between serious, peer-reviewed estimates. How can experts disagree so much?

Why the estimates differ so much

The critics' main objection to the 250,000 figure is simple. The studies it was built on counted deaths in which an error or adverse event occurred, not deaths that were caused by one. A frail 88-year-old with advanced heart failure who gets the wrong dose of a medicine and then dies a week later might be counted, even if she would have died that week anyway.

The critics also pointed out that the underlying studies were small. One analysis of the source papers noted they rested on just 35 actual deaths. Extrapolating from so few cases to the whole of America creates huge uncertainty.

The 2020 Yale-led meta-analysis by Benjamin Rodwin and colleagues tried to answer the question more directly. Doctors reviewed the records of more than 12,500 people who died in hospital and judged whether each death was preventable. They found that 3.1% were. Scaled to the US, that gave about 22,000 preventable hospital deaths a year, and only about 7,150 of those were in patients who would otherwise have been expected to live for more than three months.

Issue	Why it changes the number
"Error happened" vs "error caused death"	Counting any death with an error inflates the total
Underlying illness	Many hospital deaths are from disease that was always likely to be fatal
Small samples	A few dozen deaths extrapolated to millions creates wide error bars
Hospital-only vs all settings	Ignoring GP surgeries, A&E and care homes lowers the total
Error vs adverse event	Some harms (e.g. some infections) happen even with perfect care
Reviewer judgement	Doctors often disagree on whether a given death was preventable

The 2023 Johns Hopkins study on diagnostic error, led by David Newman-Toker, is measuring something different again. It estimated that about 371,000 Americans die and 424,000 are permanently disabled each year after a misdiagnosis, across all settings, not just hospitals. Its own authors acknowledge the precision is limited, because many misdiagnoses are never detected at all.

So which is right? Probably none of them exactly. A fair summary is that tens of thousands of Americans die each year from clearly preventable hospital errors, and the wider toll of harm from unsafe care, including missed diagnoses outside hospitals, is substantially larger. "Third leading cause of death" is a contested claim, not an established fact. For context, around 8,500 Americans die every day from all causes, as we cover in how many people die in the US every day and the leading causes of death in the US.

What about the UK?

The best-known British study took a more cautious approach. In 2012, Helen Hogan and colleagues had trained doctors review the case notes of 1,000 adults who died in ten acute hospitals in England.

Finding (Hogan et al., 2012)	Figure
Deaths judged at least 50% likely to be preventable	5.2%
Estimated preventable adult hospital deaths in England per year	about 11,859
Main problem: poor clinical monitoring	31.3% of preventable deaths
Diagnostic errors	29.7%
Inadequate drug or fluid management	21.1%
Preventable deaths in frail elderly patients with under a year to live	60%

That last line is crucial. Most preventable hospital deaths happened in elderly, frail people with several illnesses, who were judged to have had less than a year to live anyway. That does not make those deaths unimportant. It does mean the "years of life lost" are far smaller than the raw number suggests. You can see how death risk climbs with age on our 85 and over page and our United Kingdom page.

Why so much harm is preventable

WHO's view is that more than half of patient harm can be prevented, and up to 80% in primary and outpatient care. The most effective fixes are often unglamorous.

Measure	What it does
Surgical safety checklists	Confirm patient, site and procedure before the first cut
Early-warning scores	Turn small changes in heart rate, breathing and temperature into a trigger for a doctor
Electronic prescribing with alerts	Catch wrong doses and dangerous drug combinations
Hand hygiene and infection control	Cut infections caught in hospital
Blood clot risk assessment	Ensure at-risk patients get preventive medicine
Safe staffing levels	Give nurses and doctors time to notice deterioration
Patient and family involvement	WHO says engaging patients can cut harm by up to 15%

The last one is worth dwelling on. Patients and relatives are often the first to notice that something is not right. In England, "Martha's Rule" now gives patients and families the right to request an urgent review from a different team if they are worried someone is getting worse. It is named after Martha Mills, a 13-year-old who died of sepsis in 2021 after her family's concerns were not acted on.

What you can do as a patient

None of this is a reason to avoid medical care. The overwhelming majority of treatment helps, and the risk of not being treated for a serious illness is far higher than the risk of harm from treatment. But a few habits genuinely improve safety:

Keep an up-to-date list of your medicines, including doses, and bring it to every appointment.
Ask what a new medicine is for, and whether it interacts with anything you already take.
Ask "what else could it be?" if a diagnosis does not seem to fit.
Chase test results. No news is not always good news.
Speak up if someone is deteriorating, especially an elderly relative who cannot speak for themselves.
Bring someone with you to important appointments to listen and take notes.

If you want a broader sense of the everyday risks you actually face, our risk calculator and our guide to the odds of dying from everyday causes put them in proportion.

The honest bottom line

Healthcare is one of the most complex things humans do, carried out on people who are already ill, often in a hurry, often at night. Some harm is inevitable. A lot of it is not. The exact death toll depends heavily on how you define "error", and the most dramatic headline figures overstate the case. But even the most cautious estimates describe thousands of avoidable deaths a year in a single country, and WHO's global figure runs into the millions.

The good news is that this is one of the few major causes of death where the solutions are well understood: checklists, staffing, good systems, and listening to patients.

While you were reading

This article takes about twelve minutes to read. In that time, roughly 1,400 people died around the world, and around 68 of those deaths, on WHO's estimate, were linked in some way to unsafe care. You can watch the global count rise in real time or see today's running total.

Sources: World Health Organization, Patient Safety fact sheet (September 2023) and Global Patient Safety Report 2024; Makary MA and Daniel M, BMJ 2016; Rodwin BA et al., Journal of General Internal Medicine 2020; Newman-Toker DE et al., BMJ Quality & Safety 2023; Hogan H et al., BMJ Quality & Safety 2012; Institute of Medicine, To Err Is Human (1999); James JT, Journal of Patient Safety 2013; NHS England, Martha's Rule. Estimates of deaths from medical error vary widely depending on definitions and methods; figures here are approximate. This article is general information, not medical advice. If you are worried about someone's care, speak to the clinical team, and in an emergency call 999 (UK) or 911 (US).