Medical errors: what physicians say, in their own words

Last updated September 20, 2026.

The short answer

Read together, the 383 essays and podcast transcripts on this page say one thing consistently: physicians reject the headline number and confess the private one. The public figure is 250,000 deaths a year from medical error, and physicians on KevinMD dismantled it within days of its 2016 publication and have kept dismantling it, while one safety advocate on the site still cites it. The private figure is the mistake each physician carries, usually the first one in training, that could have been fatal and was not, or was. Between the number and the memory sits the argument the record has had since 2010, whether errors are the fault of systems or of people, and since 2019 a newer one, whether the physician who erred is a second victim or whether that label hides the system that produced the error.

This page is a maintained record of what physicians have written about medical errors and patient safety on KevinMD.com, a physician-authored publication founded in 2004 by Kevin Pho, MD, a board-certified internal medicine physician in Nashua, New Hampshire. It draws on 383 essays and podcast transcripts with medical error, patient safety, mistake, misdiagnosis, diagnostic error, near miss, second victim, apology, or wrong-site surgery in the title, published between May 2004 and September 2026 by 215 named contributors, 199 of the posts bylined by physicians, with the rest by nurses, patients, safety researchers, and attorneys. Posts about malpractice as a legal matter belong to a separate record. Every claim on this page is attributed to a named author with the date it was published and a link to the original.

The sections below are organized around the questions physicians, patients, hospital leaders, safety researchers, and journalists ask. Each opens with a direct answer, followed by what named authors have said, in the order they said it. Where authors cite studies, the study is named alongside the author, and the figures are theirs. Physicians who write about their own errors are quoted as they wrote, and the patients who write about errors done to them are cited at the same weight. This is a page about opinion and experience, not clinical guidance.

This record at a glance

383posts

2004 to 2026years covered

215named contributors

42posts cited on this page

Questions this record answers

What do physicians say it is like to make a mistake?
How many people do medical errors kill? Physicians disagree with the headline.
Why do physicians say errors happen: systems or people?
Should physicians apologize and disclose? What the record says about apology laws.
What do physicians say about diagnostic error?
Should errors be crimes?
How has the medical error conversation changed since 2004?

What do physicians say it is like to make a mistake?

Physicians on KevinMD describe their own errors in a small number of essays, most about the first one in training, and they describe the same sequence: the moment of recognition, the reassurance that anyone could have made it, and the years it is carried afterward. Several write that residency taught them to hide it. Posts written in the first person about a mistake have never exceeded 4 percent of the corpus in any period and stand at 2 percent since 2021, while the vocabulary of shame and blame has held at about a fifth of posts since 2009.

Manoj Jain, MD, MPH, wrote in September 2013 urging doctors to take responsibility for their mistakes “even when a patient hasn’t been harmed,” and quoting a nursing professor that “Learning to disclose errors as a team is a terrific way to learn how to work together effectively,” in “Urging doctors to take responsibility for their mistakes.” An anonymous medical student wrote in May 2016 of “killing our simulated patient by causing him to aspirate,” and that “My inability to forgive myself for my mistakes and learn from them is starting to annoy me,” in “It’s OK for medical students to make mistakes. Here’s why.” Rogue Dad, MD, a pediatric oncologist writing under a pen name, wrote in May 2017 of a residency in which he took call every fourth night, worked 30-hour shifts, and covered “a floor with 30+ patients by myself,” and of the mistake that followed, after which he was assured “it was a mistake anyone could have made, the patient came to no harm,” and that was essentially the end of it, in “My first mistake could have been fatal.” Saba Fatima, MD, wrote in September 2017 that “Two months into my pediatric residency, while working on an overnight call, I encountered my first medical mistake,” and that residents need a safe space to talk about them, in “A safe space is needed to talk about medical errors.”

Rebecca E. MacDonell-Yilmaz, MD, MPH, wrote in July 2018 that every July “everyone in every position is one year less experienced at it than the person who held that position” the day before, and recalled her own rookie mistakes, in “A physician’s mistakes as a rookie MD.” Naeem Rahman, MD, a surgeon, wrote in December 2018 of a robotic case in which he needed a colorectal colleague to recover, and that “excellent surgeons and clinicians aren’t those who’ve never made mistakes, but those who are quick to recognize them,” and who seek help when needed, in “Excellent surgeons are those who are quick to recognize their mistakes.” J. C. Sue, DO, wrote in April 2026 that “health care workers commonly feel pressure to be perfect due to the potentially serious consequences of many medical mistakes,” and that a trainee or physician who makes one may understandably feel inadequate, in “Medical mistakes happen and you are still enough.”

Questions on this page · All records

How many people do medical errors kill? Physicians disagree with the headline.

The figure the public knows is 250,000 deaths a year, from a 2016 BMJ analysis by Makary and Daniel that called medical error the third leading cause of death in the United States. Physicians on KevinMD rejected it within days and have kept rejecting it, on the grounds that the analysis extrapolated from small studies, counted deaths in patients who were already dying, and confused error with adverse event. The older figure, the Institute of Medicine’s 44,000 to 98,000 from 1999, drew the same objection from some physicians a decade earlier. Safety advocates on the site continue to cite the larger number. Mentions of the 250,000 figure peaked at 12 percent of posts in 2013 to 2016 and have fallen to 3 percent.

Robert Donnell, MD, wrote in December 2010 that the Institute of Medicine “claimed that 44,000 to 98,000 Americans were killed in hospitals each year,” that its methods “were not transparent and its analysis was suspect,” and that the result was “a culture of blame in which any adverse event is someone’s (and hopefully someone else’s) mistake,” in “A postmortem on the patient safety movement.” Leana Wen, MD, wrote in January 2013 that “Over 100,000 deaths due to medical error occur every year, and the majority of these errors are errors in diagnoses,” in “The culture of ruling out leads to misdiagnosis.” Dr. Saurabh Jha, a radiologist, wrote in May 2016, days after the BMJ paper, that “the problem with Makary’s analysis isn’t the estimation of deaths attributable to medical errors” but the inconsistency with which error was adjudicated, in “Hyperbole is seldom helpful. Especially when it comes to medical errors.” Julie Craig, MD, wrote the next day that the study “suffers from questionable statistics that serve up alarmist conclusions,” in “The alarmist approach to medical errors solves nothing.”

Cory Fawcett, MD, a surgeon, wrote in February 2017 that if told the third most common bird in the United States were a flamingo he would question that statement, and “That is why I know the third most common cause of death is not medical errors,” in “Medical errors are not the third leading cause of death.” Skeptical Scalpel, MD, wrote in February 2020 that Canadian authors reviewing the claim had noted it “enabled supporters of the NRA to say doctors are more harmful than guns,” in “Medical error is not the third leading cause of death.” Barbara L. Olson, RN, a safety advocate, wrote in October 2023 the other side: that “Upwards of 250,000 people in the U.S. die each year because of medical errors,” and that the fix is care systems that “anticipate and correct predictable human errors before they reached patients,” in “A shop teacher’s daughter on transforming patient safety.”

Questions on this page · All records

Why do physicians say errors happen: systems or people?

The word “system” appears in 15 percent of KevinMD error posts before 2009 and in 64 percent since 2021, and the shift is the record’s central argument. The systems view, that errors are produced by handoffs, fatigue, the electronic record, staffing, and design, has been the majority position since 2010. The dissent, that it removed personal accountability and that just culture can become no culture, has been steady at about a fifth of posts. A newer version of the argument, since 2019, is over the second victim: whether the physician who erred deserves that name, or whether the name diverts attention from the system that produced the error. The newest, since 2026, is whether the system now includes the algorithm.

Kevin Pho, MD, wrote in December 2009 that after preoperative checklists were introduced their use rose from 75 percent in 2003 to 100 percent in 2007, and quoted the study’s call for a culture “that focuses on systems issues as opposed to individual blame, maintains personal accountability, and encourages open communication,” in “A cockpit checklist to improve patient safety.” In December 2009 he cited a Journal of Hospital Medicine study in which “Ninety-six percent of the 50 patients surveyed left out at least one drug when they were asked to list their medications,” in “How medication mistakes happen in the hospital.” Robert Wachter, MD, wrote in February 2011 that “If we just tell the workers to use checklists, we will have solved the problem of catheter-related blood stream infections” was quite simply the wrong conclusion from the Michigan Keystone project, and that a JAMA study showed “culture can be modified with a feasible set of interventions,” in “Teamwork helps doctors with patient safety.”

David B. Nash, MD, MBA, wrote in March 2013 that surgical never events occur about one in every 12,000 procedures, and that “simple checklists and protocols have been shown to reduce the occurrence of such mistakes to near zero,” in “Reducing medical errors: What we can learn from the Dreamliner.” Suneel Dhand, MD, wrote in August 2013 that “Many of the leading voices of the patient safety movement are quick to draw a comparison with aviation,” and that aviation does not hold all the answers, in “Aviation doesn’t hold all the answers to patient safety in health care.” Robert Wachter, MD, wrote in October 2013 that hand hygiene in monitored ICUs rose from 7 to 82 percent when cameras were pointed at sinks and “there was someone in Bangalore, India reviewing the video feed every hour and sending back compliance data,” in “How video can reduce medical errors and improve patient care.” Ashish Jha, MD, MPH, wrote in July 2014 that “Tens of thousands of Americans die needlessly because of preventable infections, medication errors, surgical mishaps,” and proposed that certified electronic records be required to “measure, track, and feed back patient safety events,” in “The difficulty of moving the needle on patient safety.”

Harry Karydes, DO, wrote in May 2019 that “the first victim is always the patient while the second victim is any health care provider” involved, in “Second victim syndrome: a doctor’s hidden struggle.” Susan Wilson, MD, wrote in November 2020 that confidential peer support is “confidential, non-judgmental sessions wherein the clinician is afforded a safe space to discuss his/her feelings,” in “Peer-to-peer support and the second victim syndrome.” Arthur Lazarus, MD, MBA, wrote in July 2024 that “lapses in vigilance, miscommunication, or failure to follow protocols” produce medical errors the way they produce security failures, in “When protocols fail: lessons from Trump’s attempted assassination and medical errors.” Olumuyiwa Bamgbade, MD, wrote in August 2025 that health care has approached safety “with a Safety-I lens, emphasizing the prevention of errors and investigating failures,” and proposed a resilience-focused practice instead, in “From errors to resilience: a smarter approach to patient safety.” Timothy Lesaca, MD, answered in April 2026 that the second victim label “is distracting us from the systems that make the distress,” and that “Systems that burn out doctors and nurses are the same systems that produce errors,” in “The second victim label ignores patient safety reality.” Duane Penshorn, MD, wrote in July 2026 that in Kathleen Mosier’s flight simulator studies the pilots with more experience “were more likely to believe the computerized alerts because they had worked side by side for years with the system,” and that automation bias, “the tendency to believe automated cues as a shortcut to vigilant information-seeking,” is now medicine’s problem, in “Automation bias is the new medical error.”

Questions on this page · All records

Should physicians apologize and disclose? What the record says about apology laws.

Physicians on KevinMD have argued for disclosure and apology since 2009, and most say the barrier is fear of litigation rather than pride. The evidence they cite has changed. Early posts assumed apology reduced lawsuits; the recent ones report that apology laws, which shield “I’m sorry” from being used in court, do not reduce claims, while communication-and-resolution programs, in which the institution discloses and offers compensation, do. Apology or disclosure appears in about a fifth of posts across the record.

Kevin Pho, MD, asked readers in February 2009 whether doctors should apologize after a medical error; the post is a short reader poll, counted here but not indexed or linked. Itzhak Brook, MD, an infectious disease physician writing as a patient with cancer, wrote in May 2013 that surgeons “mistakenly removed scar tissue instead of the cancerous lesion,” and that “admission and acceptance of responsibility by the care providers can bridge the gap” errors open, in “Medical mistakes occurred at all levels of my care.” Michael Kirsch, MD, wrote in December 2013 that an apology of the form “I’m sorry for my oversight which wouldn’t have happened if” is not one, and praised institutions that “admitted they screwed up, apologized and didn’t blame anyone,” in “How should a physician apologize after a medical error?”

Lisa Goodman-Helfand, a patient, wrote in March 2016 that “much of what my family and I endured was caused by medical error,” and asked what would happen “if the errors were exposed, examined by colleagues, and protocols were remediated,” in “Medical mistakes happen. It’s what doesn’t follow that is unforgivable.” Skeptical Scalpel, MD, wrote in September 2016 that “It is very easy for a physician to apologize for an error committed by someone else,” and that in a Medscape survey “81 percent of those who responded said they thought an apology would not make a difference” to whether a suit was brought, in “Why don’t more doctors apologize for their mistakes?” Maja Castillo, MD, wrote in October 2016 as a patient that “Only after this confrontation did the doctor and nurse speak directly to me about the error,” and that “our culture of blame has shrouded the subject in fear, guilt, and shame,” in “A physician experiences a medical error. Here’s her story.” Timothy Lesaca, MD, wrote in May 2026 that a 2019 Stanford Law Review analysis “found that apology laws did not achieve their goal of reducing legal risk,” partly because an apology “can provide new information that prompts them to file a claim,” in “Medical apology laws don’t reduce malpractice lawsuits.”

Questions on this page · All records

What do physicians say about diagnostic error?

Diagnostic error is the newest subject in the record and the fastest-growing: 2 to 7 percent of posts through 2020, 16 percent since 2021, with cognitive bias and anchoring rising from 4 to 15 percent over the same period. Physicians on KevinMD trace it to anchoring on the first plausible explanation, to a culture of ruling out that tests for what a symptom is not rather than what it is, and, in the most recent posts, to conditions in women that are dismissed or misattributed.

Jerome Groopman, MD, and Pamela Hartzband, MD, wrote in February 2011 that anchoring “refers to the tendency to latch on, or anchor, to the first symptom or bit of data” and to miss the full spectrum of information, and that it is often triggered “by a mistake in translating the patient’s words into clinical terms,” in “How anchoring leads to misdiagnosis.” Leana Wen, MD, wrote in January 2013 that ruling out often results in misdiagnosis because, in the case she describes, “the focus was never on figuring out the problem to begin with,” in “The culture of ruling out leads to misdiagnosis.” Ashley Meyer, PhD, and Hardeep Singh, MD, MPH, wrote in February 2019 that physicians rarely learn of their diagnostic errors, and that “improve physicians’ calibration by providing them with feedback about their diagnoses” is the fix the psychology literature suggests, in “How should physicians hear back about their diagnostic errors?” Miles J. Varn, MD, wrote in March 2022 that “some studies have found that the problem is more common for women,” partly because their symptoms differ from men’s, in “Strategies to help women advocate for themselves and avoid misdiagnosis.” Susan L. Montminy, EdD, MPA, RN, and Marlene Icenhower, JD, RN, wrote in February 2025 that malpractice claims data suggest “28 percent of all diagnostic error occurs in the ED,” and that 52 percent of those errors result in death or high-severity injury, in “Solving a hidden challenge: 10 tips to reduce diagnostic error in the emergency department.” Kimberly L. Sterling, PharmD, and Audrey Wells, MD, wrote in March 2026 that “women’s sleep struggles are often dismissed, misdiagnosed, or overshadowed,” in “Why women’s sleep struggles are often dismissed, misdiagnosed, or overshadowed.” Neil Panchal, DO, wrote in July 2026 that in perimenopause “A normal result in a symptomatic forty-six-year-old is common and reassures nobody correctly,” in “Perimenopause misdiagnosis: what a normal lab result hides.”

Questions on this page · All records

Should errors be crimes?

Twice in the record, in 2011 after a pharmacist was imprisoned for a fatal compounding error and in 2022 after the nurse RaDonda Vaught was convicted of negligent homicide for a medication error, physicians on KevinMD argued that criminalizing honest mistakes makes patients less safe, because it teaches clinicians to hide them. No physician on the site argued the other side.

Jeffrey Knuppel, MD, wrote in May 2011 of Eric Cropp, “a pharmacist who was sent to prison for a mistake that resulted in the death of a child,” and asked “where do we draw the line on criminalizing mistakes?” in “Where do we draw the line on criminalizing mistakes?” Michael Ramsay, MD, wrote in April 2022 that Vaught’s conviction “caught the attention of the entire global health care community,” and that “76.9 percent of nurses fear the reactions of administrators and colleagues after reporting an error,” in “The criminalization of true medical errors is a step backwards for patient safety.”

Questions on this page · All records

How has the medical error conversation changed since 2004?

The record begins with Kevin Pho tracking the policy response to the Institute of Medicine report, Medicare’s 2007 decision to stop paying for certain errors, checklists, and wrong-site surgery. From 2009 to 2012 the safety movement wrote itself in, Wachter, Jha, and Nash, along with the first dissent from Donnell that the numbers were a myth. 2016 was the year of the 250,000 figure and its rebuttals. From 2017 the first-person essays arrived, and from 2019 the second victim. Since 2021 the subject has turned to diagnostic error, cognitive bias, and, in 27 percent of recent posts, AI.

Term	2004 to 2008 (52 posts)	2009 to 2012 (100 posts)	2013 to 2016 (77 posts)	2017 to 2020 (57 posts)	2021 to 2026 (97 posts)
System or systems	15 percent	47 percent	51 percent	47 percent	64 percent
Blame or shame	2 percent	17 percent	19 percent	21 percent	22 percent
Institute of Medicine or 98,000	4 percent	14 percent	18 percent	5 percent	5 percent
Third leading cause or 250,000	0 percent	1 percent	12 percent	7 percent	3 percent
Apology or disclosure	8 percent	24 percent	22 percent	19 percent	20 percent
EHR or alerts	0 percent	9 percent	22 percent	18 percent	22 percent
Fatigue or hours	0 percent	9 percent	8 percent	7 percent	11 percent
Nurses	8 percent	29 percent	36 percent	32 percent	26 percent
Second victim	0 percent	0 percent	1 percent	9 percent	8 percent
Diagnostic error or misdiagnosis	2 percent	5 percent	4 percent	7 percent	16 percent
Cognitive bias or anchoring	2 percent	5 percent	5 percent	4 percent	15 percent
First person: I made a mistake	0 percent	1 percent	4 percent	4 percent	2 percent
AI	0 percent	0 percent	1 percent	2 percent	27 percent

Term frequencies are the share of posts in each period using the term at least once, computed on the September 18, 2026 corpus of 383 posts.

Kevin Pho, MD, noted in August 2007 that Medicare’s decision to stop paying for certain hospital-acquired conditions would mean cultures on every admission, in a post from before 2009 that is counted here but not indexed or linked, and in November 2009 asked whether cameras would reduce wrong-site surgery, in “Will video cameras in the OR decrease the rate of wrong-site surgery?” The 2010 to 2011 posts are Donnell’s postmortem, Wachter on teamwork, Groopman and Hartzband on anchoring, and Knuppel on Cropp. Jha in 2014 and the 2016 quartet of Jha, Craig, Castillo, and Goodman-Helfand define the middle. Rogue Dad and Fatima in 2017, MacDonell-Yilmaz and Rahman in 2018, and Karydes in 2019 are the first-person and second-victim turn. Since 2022 the record is Ramsay on Vaught, Olson holding the 250,000 line, Lesaca on apology laws and the second-victim label, Sue on being enough, Penshorn on automation bias, and the diagnostic error essays.

Questions on this page · All records

About this record

The KevinMD medical error corpus by the numbers

The figures below describe the set of KevinMD posts this page draws on, as of September 18, 2026. They are counts of what KevinMD has published, not incidence data.

Measure	Value
Posts in the corpus	383
Date range	May 26, 2004 to September 5, 2026
Named contributors	215
Posts bylined by an MD or DO	199
Posts by nurses, patients, researchers, attorneys, and other non-physicians	About 95
Posts with patient safety in the title	107
Posts with medical error in the title	85
Posts with apology or apologize in the title	41
Short posts by Kevin Pho, 2004 to 2011	89, median 117 words
Podcast episodes with full transcripts	17
Posts in the corpus that are not indexed, counted but not cited	86
Total words	About 297,000
Peak year	2010, 33 posts
Most frequent contributors	Robert Wachter, MD (9); Skeptical Scalpel, MD (7); David B. Nash, MD, MBA (6); Miles J. Varn, MD (5); Suneel Dhand, MD (5)

How this page was built and how it is updated

The corpus was assembled from title searches for medical error, patient safety, mistake, misdiagnosis, diagnostic error, near miss, second victim, apology, and wrong-site surgery, with posts about malpractice as a legal matter set aside for the malpractice record and posts about financial, career, and contract mistakes removed. Term frequencies were computed against the full text of each post on the corpus assembled for the September 18, 2026 build; this page was converted to the current standard on September 20, 2026 without rebuilding that corpus, so the counts and figures are as of September 18, 2026. Posts are counted whether or not they are indexed; citations on this page are limited to posts that are indexed and can be verified at the link. Eighty-six posts in the corpus are not indexed: Kevin Pho’s 2004 to 2011 commentaries, six short essays, and four podcast episodes without transcripts. They are counted but not cited. Quotations are taken verbatim from the original posts. Author credentials are as they appeared in the byline at publication; one physician writes under a pen name and is cited under it. Patients who write about errors done to them are cited at the same weight as physicians. Kevin Pho’s posts from 2004 to 2011 were short commentaries on news and studies and are counted separately above. Every source is linked in the sentence that cites it, and the full list appears at the end of the page. The closest archive is the Hospital Medicine tag. Related records: Medical malpractice: what physicians say, in their own words, The electronic health record: what physicians say, in their own words, Artificial intelligence: what physicians say, in their own words, and Nursing: what nurses and physicians say, in their own words.

This page is updated as new essays on medical errors and patient safety are published on KevinMD. When it is updated, the date at the top changes, the counts in the tables are recomputed, and new named claims are added to the relevant section. Nothing is removed unless the original post is removed. An author who believes a quotation on this page misrepresents them can write to Kevin Pho and the page will be corrected.

To cite this page: Pho K. Medical errors: what physicians say, in their own words. KevinMD.com. Updated September 20, 2026. https://kevinmd.com/medical-errors