Medical error—the third leading cause of death in the USBMJ 2016; 353 doi: https://doi.org/10.1136/bmj.i2139 (Published 03 May 2016) Cite this as: BMJ 2016;353:i2139
All rapid responses
I thank the authors for having enlightened and allowed to debate an aspect of modern health system which remains largely misundesrstood and still denied by many doctors, as it has been evidenced in some responses, and ignored by patients.
However, the discussion here was held mostly about the technical side of the problem, resulting in counts ranging roughly from 170 000 to 251 000 deaths caused by medical errors in hospitals in the United States defined by the authors as people dying” from the care they receive rather than from the disease or injury that brings them to care”.
As US healthcare system is a model for the world, a model intended to extend and be imitated by emerging countries this would legitimate, in my opinion, a less technical and more global approach to the problem of medical harms elicited by this model.
At the turn of the twentieth century, leading causes of death in the United States like in european countries were infectious diseases. At this time people died mostly from pneumonia, tuberculosis and diarrhea and life expectancy at birth was about 50 years just like it is still now in poorest countries  . During the twentieth century there was the so called epidemiologic transition in developed countries and degenerative non communicable diseases have replaced communicable diseases as the leading causes of deaths . The same phenomenon is seen in non developed and emerging countries and It appears as the natural consequence of progress and improvement in quality of life standards .
Historically, medicine deals with disease and doctors are committed to cure diseases, essentially with invasive procedures. In medical schools, students are trained to be highly confident in the power of medicine and are taught that there is a precise and effective procedure for every disease that has to be implemented. They are prepared to be heroes and to save lives. As heroes they are armed with potent weapons that is medical devices and sophisticated drugs and the more the weapons a doctor can use are numerous and powerful, the more he uses them the more it gives him prestige and money. Just like heroes doctor’s creed is action and abstention and caution are not part of the vocabulary they are taught. For all the duration of their studies, medical students are trained in a way that will give them a highly distorted vision of the benefit risks balance of medical procedures and of the part played by medicine in improving public health.
But diseases responsible for most morbidity and mortality in developed countries have profoundly changed in nature. As it si for degenerative diseases, when acute or chronic symptoms appear the disease has already been evolving for years or decades and when the hero comes in, the die is cast and the end of the play is close.
It is now widely admitted that degenerative diseases are caused and that their evolution is accelerated by our environment and our lifestyles.
Nowadays, per capita mean refined sugar consumption is ten times higher that it used to be two centuries ago , we spend many more hours a day sitting than our parents and many much more that our grandparents , air pollution in urban areas has a major impact in morbidity and mortality due to COPD  and this has been known for long…
Physicians don’t feel very concerned about these facts as there are not within the reach of traditional therapeutic medicine but are more a matter of public policies, personal behavior and choices and cultural habits . As the epidemiology changed, instead of changing their approach to medicine, they have tried to do the same that they were used to do with symptomatic diseases with degenerative diseases. So they have started screening for them to find these diseases earlier and cure them with the usual invasive methods. During the eighties and nineties there has been, in developed countries, a myriad of mass screening programs concerning prostate cancer, breast cancer, thyroid cancer, colon cancer… Most of these programs require highly technical devices and highly invasive procedures, are used in persons that have no complaints and have been demonstrated to have bad benefit risk balance because of overdiagnosis and adverse effects of the procedures used .
Mathematically, when you use widely invasive procedures and you target millions of healthy non complaining people you sharply increase the opportunity to harm and so to impair public health.
Moreover, more interventionist and tehcnophiles trends in medicine lead to the extensive use of invasive , costly but also rewarding procedures and drugs with marginal benefits and involve the abandonment of common diseases and of those who are more in need of medical care. This has been theorized by Julian Tudor Hart in the early 70 and is called “the inverse care law”. In the modern medical perspective every healthy patient is “un malade qui s’ignore” but really ill patients don’t deserve appropriate care.
So, modern medicine can harm by doing too much on one side and too little on the other that is by misdirecting medical care.
Naturally medical harms are not only about medical errors and medical care misdirection. They are also about adverse effects of drugs and procedures, useless polymedication and drugs interaction and malevolent behavior of pharmaceutical companies , as well as the culpable negligence of patients groups, politicians and medical associations funded by these companies as it was highlighted once again in the opioïd crisis affair  .
So we don’t just need some technical adjustments to end with medical wanderings and with medical harms. We need a revolution. And firstly, medical students should be taught to have a much more humble and skeptical attitude towards medicine, and a more balanced approach of the benefit risk issue.
Competing interests: No competing interests
You report a Johns Hopkins study that said over 250,000 deaths resulted
yearly from medical errors.
This figure is alarming, but the medical profession deserves a statistical
context: the number of mistake-opportunites which occur each year. Taking
into account all the prescriptions written, all the hospital orders made
and implemented, this number must be in the billions, which means that only
a tiny percentage of such prescriptions and orders are wrong. .001
You should calculate this statistical basis and present it concurrently to
put these "scare" data in perspective.
Am I wrong?
Jonathan Wells, Newbury, MA
Competing interests: No competing interests
Trained laboratory pigeons are reported to produce excellent cancer detection rates when confronted with biopsy slides.
Pigeons learn fast, in only 15 days, and manage to efficiently diagnose cancers, with accuracy levels increased to 99%.
The strategy for obtaining second opinions on pathology slides from pigeons should be explored, and compared to that of expert pathologists with a high volume of diagnostic work.
Competing interests: No competing interests
I read this article with interest since I am interested in quality measures in health outcomes; and, as well, I am interested in data science methods. I don't doubt the statistics cited by the authors; however, how those statistics were reported, in their raw form, is not useful; and, possibly misleading. Medical care delivery is complex encompassing many independent and dependent variables and even more complex intervening variables and certainly many confounding variables. Thus it seems that a great deal of multivariate and logistic regression data science methodology must be applied before we hit the scientific or popular press with a shock value number to report. If this level of data were presented to me by a doctoral candidate as a proof for the Ph.D. dissertation defense, I would not find it at all acceptable. Of course, the raw data intrigues me and I would certainly begin asking many questions about the data; and what data was not included in the report. Certainly, there was no sophisticated analysis that sheds either descriptive or inferential analysis of the raw data.
Dr. Michael W. Popejoy, Ph.D., Ed.D., M.B.A., M.P.H., M.H.S.A., FRSPH, FAMEPPA
Orlando, FL USA
Competing interests: No competing interests
Response to Aubrey Milunsky:
Research by the Betsy Lehman Center showing that 23% of adults in Massachusetts reported experiencing a diagnostic error in the last five years and the recent Institute of Medicine report “Improving Diagnosis in Health Care” have helped raise awareness about the prevalence of the problem and established a new priority for health care improvement. While medical school curriculum, like our own at Johns Hopkins, are expanding to teach the science of safety, an important part of that culture change is a departure from a blame the individual approach and an evaluation of the how systems can be designed safer. We appreciate the non-technical skills of doctoring that Dr. Milunsky suggests.
Response to Tjaard U. Hoogenraad:
Thank you for this case study demonstrating how an outcome can be non-fatal, yet significant. While we did not focus on non-fatal injury and disability due to medical care gone awry, examples such as these point out the patient-centered approach that is central to any patient safety strategy.
Response to Kaveh G. Shojania:
The very real problem of medical care gone wrong should be measured rigorously with more autopsies and a data collection infrastructure as strong as the one used to generate national cancer statistics each year. But currently that does not exist. In the mean time, we rely on scientific estimates from scientific studies in the literature. We applaud the prior contributions of the Canadian physician and sociologist, but reject their primary argument that they “find it very hard to believe” as a scientific argument. The purpose of science is to extend the validated body of knowledge, which inherently is expected to challenge and even controvert deeply held beliefs. In their own articles, they have frequently cited the 1999 I.O.M. To Err is Human estimate that uses data from 1984 and 1992 (1). In our article, we challenge this establishment estimate, and point out its limitations. While no estimate is perfect, we respectfully submit a more updated estimate that uses better methodology.
The Canadian physician previously wrote that the studies from which the 1999 I.O.M. estimate was derived may not have detected diagnostic errors which in he reported as being found in 8.4% of hospital deaths, or as he extrapolates in his article, 71,400 deaths of the 850,000 deaths in the year of his research study (2). Simply adding his own estimate to the 1999 I.O.M estimate would rank the incidence of death from medical error well above the #3 cause of death that year.
While the Canadian physician and sociologist cite studies to arrive at their current estimate of 25,200 annual deaths from medical error (far below the estimate of the I.O.M. To Err is Human Report), the rates described in the our analysis are based on rigorous scientific studies published in the New England Journal of Medicine, Health Affairs, and the independent Medicare OIG report, all of which have been cited and affirmed by several recent Institute of Medicine Reports, and also in the recent study by John James that suggests our estimate may be low. When there is a pattern of literature supporting the incidence we describe, it should be considered in good faith. While the media is known to polarize people, it is our hope that we can be constructive. We can learn from your primary care and sociology vantage points respectively, and invite you to attend our surgical morbidity and mortality conference as we do each week. Given the number of lives ruined by disability and death due to medical care gone wrong, we believe a more scientific conversation, rather than a territorial one, helps put the focus on solutions.
Today in the U.S., there are more medications, diagnoses, procedures, and handoffs performed than ever in the history of medicine. Moreover, overtreatment is now an endemic problem in some areas of medicine. With more medical care being delivered, there are naturally more opportunities for things to go wrong. In fact, harm may be associated with complexity. The Commonwealth Fund reported that the U.S. leads the world in medical errors, observing that 34% of patients with health problems in the U.S. report experiencing medical, medication, or test errors—the highest rate of any nation (3). We believe a systems approach, rather than a blame the doctor or shoot the messenger approach, is the best path to a safer health care system. An honest conversation without obvious and apparent bias about the best available science is an important prerequisite to addressing the problem.
1. To err is human: building a safer health system . Washington, DC: National Academy Press, Institute of Medicine; 1999.
2. Shojania KG, Burton EC, McDonald DM, Goldman L., Changes in rates of autopsy-detected diagnostic errors over time: a systematic review. JAMA 2003;289(21):2849-56.
3. The Commonwealth Fund: http://www.commonwealthfund.org/publications/press-releases/2005/nov/int...
Response to Daniel J Baldor
In our study, we reported a range in addition to a point estimate. We agree with medical students Baldor and Kravietz in their final conclusion that after their self-described adjustment, “medical errors remain the third leading cause of death medical error”. The proportion of adverse events that are preventable depends on several factors and for that reason there is no quick fix to “correct” for differences in study methodology when reconciling different studies.
Response to John T. James
We agree with Dr. James and consider his 2013 article in the Journal of Patient Safety, cited in our analysis, to be an important contribution to the field. While it suggests that our estimate may be low, we all recognize that none of the studies mentioned captured outpatient surgery deaths or deaths at home due to medical error. Thus we recognize that the true death toll from medical care gone wrong is likely higher.
If the house is on fire, we should ask “How can we put out the fire?” Discussing differences about what is the best scientific estimate of the fire’s temperature can be paralyzing for some. We appreciate Dr. James’ scientific work, detailed in his comment, coupled with his tremendous advocacy work in the field.
Response to Linda Williams:
We applaud the your efforts and those of the Director of the National Center for Patient Safety and agree with your response pointing out that the vast majority of errors are system errors. In defining medical error as we did in our article to include both a system error and an individual error, it created a territorial discussion around nomenclature. Our definition is best summarized in the notion that people can die from the care they receive rather than from the disease or injury that brings them to care. We believe the data is best discussed with a patient-centered lexicon as you explain from your experiences.
Response to Dr. Gallie:
Anesthesiology has long been at the forefront of patient safety. We agree and recognize that there is a lot of work that needs to be done to address the many contributing factors to the problem of medical care gone wrong. Patient safety tools are one step. Addressing unwarranted clinical variation around best practices in the context of a culture of safety is another. Recognizing that we have a problem is an important prerequisite.
Response to Dr. Ghirardini:
The Bristol events highlight both the ubiquitous and compartmentalized nature of the problem. More globally, a discrete two-tiered health care system is forming. Patients are often referred to as “private patients” or “public hospital patients”. Disparities in quality, safety, overtreatment, and undertreatment can vary markedly within this these two health care systems. Similarly, one’s perspective on patient safety can be skewed by the tier in which they practice. We agree with the physician’s strategy towards a better health care system for all.
Response to Dr. Cohen:
We agree that working towards a culture of safety and teamwork is an important context for many of the solutions mentioned. Thank you for sharing your clinical and leadership experiences.
Competing interests: No competing interests
It is medical student graduation season and the air is filled with their choral intonation of the Hippocratic Oath, Declaration of Geneva, or the Oath of Maimonides. Thrilled with gratitude, they are about to embark on a lifetime journey with the primary goal of helping people. None would argue with the expressed ethical guidelines, humanitarian goals, and the exaltation “First do no harm”. Their departure, however, is heralded by glaring headlines in the New York Times that and elsewhere about this startling report of the awesome lethal consequences of medical error in the U.S. Beyond the awesome toll of deaths, are many patients with serious to devastating life-time injuries.
In a recent survey, commissioned by the Betsy Lehman Center in Boston, 23% of surveyed adults in Massachusetts reported experiencing a diagnostic error in the last five years, with 75% of these errors occurring during treatment at a hospital. The National Practitioner Data Bank reports that from 2004-2014 there were 154,621 medical malpractice payments and 392,100 adverse actions taken against physicians.
Given the flow of catastrophic data, isn’t it time to get real? There is an urgent need to focus the attention of our medical students and graduates on the continuing crisis and lack of patient safety. I suggest it is time to add a supplementary invocation, that could include the following:
• I will listen to my patients.
• I will document key information and recommendations.
• I will recognize my duty to warn patients of risk.
• I will seek informed consent.
• I will be cognizant of my limitations.
• I will not be arrogant.
• I will take a family history.
• I will communicate with physicians who care for the same patient.
• I will refer and confer.
• I will see the laboratory report for the test I order.
• I will not visit my religious, racial, or eugenic views upon my patients.
• I will adhere strictly to surgical checklists and time-outs.
• I will carefully check prescription dosage, prescribed and administered.
• I will seek help if I become dependent on drugs/alcohol.
• I will admit and apologize for an error.
• I will say “I don’t know”, when I don’t know.
Medical students need to be better schooled about patient safety and the causes and prevention of medical errors, instead of becoming part of the killing fields. A final invocation could remind them of their profound duty.
Competing interests: No competing interests
We thank Makary and Daniel(1) for their very essential alert to the health system to tackle medical errors as the third leading cause of death. However, we believe medical errors are also among the top leading causes of iatrogenic clinical deteriorations, which decline the quality of life of our patients and lead them towards a gradual death. In the following, we discuss a dramatic error in the medical decision making in Wilson’s disease, which is well treatable, but fatal without treatment. There are situations in which aggressive approaches are regarded as the only options to save the life of the patients. According to the Hepatology guidelines(2) for treatment of acute liver failure in Wilson’s disease, an urgent liver transplantation is regarded as the only lifesaving therapeutic option. However, this approach is in contrast to the guidelines we developed for the treatment of Wilson's disease,(3-4) in which aiming at normalisation of free copper by zinc is advised. By knowing the aetiology of the disease as being, not the accumulated copper, but free copper intoxication, causal treatment with oral zinc therapy is available, which favours safety and affordability, and thus patients can enjoy a normal healthy life as it is exemplified in the following story.
A young patient with acute liver failure and in a very poor clinical condition was diagnosed to have Wilson’s disease, and thus was put on the waiting list for an urgent liver transplantation. While awaiting the transplant operation, biochemical results indicated extremely high copper values in the urine and blood. However, based on the Dutch guideline3 for treatment of free copper intoxication in Wilson’s disease, treatment was started with oral zinc therapy in order to stimulate the excretion of copper via the stools and to stop the copper intoxication of the liver, which had led to acute liver failure. The response to zinc therapy was exceptional. After a few days, the clinical situation had improved to such an extend that the patient could leave the hospital in perfect health continuing with zinc treatment. Liver function tests and urinary copper had normalised in a few months, and the patient was able to live a normal life.(5)
How big is the problem?
1. The error is extremely complicated and global. The idea that an impaired copper metabolism that results in copper intoxication in Wilson's disease should be treated by liver transplantation is unscientific and false.
2. The idea that Wilson's disease can be treated with zinc to normalise raised copper levels was discovered by a Dutch neurologist, Schouwink, in 1961. Although, it was based on the best available evidence, zinc therapy was not taken seriously and was underreported, which led to much clinical deterioration and unneeded deaths.
3. An unneeded transplantation is not only financially devastating (about $600,000 USD in the US and about £65,000 and £260,000 in the UK), but also puts the patient to an enormous loads of consequences of the immunosuppressive medications (about $10,000 USD for the first year with many complications), exigent limited life style, and regular followup tests. However, zinc therapy would cost about $5 USD a months with no major side effect.
4. Putting our patients under an unneeded preventable risk of liver transplantation and its consequences is unethical and illegal.
In conclusion, the presented story was not just an anecdotal successful treatment of acute liver failure in Wilson’s disease by zinc; in fact, it was the result of a scientific rational decision making. If a doctor in charge of a Wilson’s disease patient knows about the possibility of zinc therapy and does not inform the patient and family about it in order to make a patient-centered decision, then he or she is superimposing the patient to the risk of death or a poor quality of life, which is an unethical practice. We believe liver transplantation for Wilson’s disease is one of the biggest global errors ever in medicine and it should be prevented.
1 Makary MA, Daniel M. Medical error—the third leading cause of death in the US. BMJ 2016;353:i2139. doi:10.1136/bmj.i2139
2 European Association for Study of the Liver. EASL Clinical Practice Guidelines: Wilson’s disease. J Hepatol 2012;56:671–85. doi:10.1016/j.jhep.2011.11.007
3 Hoogenraad TU. Wilson’s disease. In: major problems in neurology. Editor: Jan van Gijn. London: W.B.Saunders; 1996. vol 30. ISBN: 0702018422
4 Hoogenraad TU. Paradigm shift in treatment of Wilson’s disease: Zinc therapy now treatment of choice. Brain Dev. 2006;28:141–6. doi:10.1016/j.braindev.2005.08.008
5 Avan A, Kianifar HR, Hoogenraad T. Initial zinc therapy in a Wilson's disease patient with acute liver failure and copper intoxication: A clinical observation [abstract]. Movement Disorders 2013;28 Suppl 1 :990
Competing interests: No competing interests
Makary and Daniel recently gained widespread publicity with their estimate that medical error constitutes the 3rd leading cause of death in the US. As editors of BMJ Quality and Safety and researchers of long-standing in patient safety and quality of care, we appreciate the urge to draw attention to this area. But it is critical that the claims made to secure attention are well-founded. We worry that this estimate is not.
First, the estimate fails the plausibility test. Of around 2.5M deaths in the US each year, approximately 700,000 occur in hospital. We – and many clinicians and researchers - find it very hard to believe that one in 10 of all US deaths, or a third of inpatient deaths (the 251,454 estimated by Makary and Daniel) result from “medical error”.
Second, the authors of the article do not provide any sort of formal methodology. Their estimate seems to rely on extrapolating preventable death rates from those reported in other studies. They then place the estimate derived from these heterogeneous studies in a “ranking” of causes of death in the US to make their argument that it is the third leading cause. These two steps are both precarious. The four studies on which they appear to base their estimate on use different methodologies and wildly varying definitions that Makary and Daniel collapse into their vividly-titled construct of “preventable lethal adverse event”. It is not clear how the “point estimates” they derive were calculated, but it is notable that the denominators across the studies are not comparable and no confidence intervals are reported.
The authors call for death certificates to include an extra field asking whether a preventable complication stemming from the patient’s medical care contributed to the death. The practical details of how this might be achieved are scant. Causes listed on death certificates already represent educated guesses much of the time, as not many patients die of diagnoses supported by gold standard tests during life (or autopsy results after death) Moreover, the doctor who pronounces death (and thus fills out the death certificate) may be ill-placed to know whether the patient experienced a preventable complication in care. But, suppose we sidestepped these practical issues (and put aside questions of resources) to implement a system whereby at least two clinicians not based at the hospital where death occurred undertake an independent medical record review and then discuss the case in order to reach consensus about whether or not medical error had likely contributed to death.
As it turns out, this approach has been implemented in research settings on at least three occasions.[4-6] In all of these studies, the authors sampled deaths from multiple institutions and asked trained reviewers to look over the cases to identify possible quality of care problems and to make a judgment about the preventability of death. In all three studies, reviewers estimated that around 3% to 5% of deaths were ‘probably preventable’ (a greater than 50% chance that optimal care would have prevented death). The largest and most recent of these studies reported that trained medical reviewers judged 3.6% of deaths to have at least a 50% probability of avoidability. Applying this rate of preventability to the total number of hospital deaths in the US each year produces an estimate of about 25,200 deaths annually that are potentially avoidable among hospitalized patients in the US—roughly 10-fold lower than the estimate advanced by Makary and Daniel.
The Makary and Daniel claim that medical error accounts for more than 250,000 deaths per year therefore stands in contrast to the results of several robust studies performed using the type of review they say is needed for medical error to be listed on a death certificate. Moreover, two of these studies [4, 6] make the point that roughly half of patients who probably had a preventable death were in their last six months of life – unlike the tens of thousands of people each year in the US who die in car accidents, shootings, or commit suicide (some of the causes of death that Makary and Daniel argue account for less mortality than does medical error).
A further problem with the estimate is more subtle. Making the field of patient safety all about death has risks. Just as most deaths do not involve medical error, most medical errors do not produce death—but they can still produce substantial morbidity, costs, suffering and distress. Drawing attention only to death as the focus of patient safety efforts risks drawing resources away from many settings of care – including almost all non-hospital environments – where death is not the most relevant outcome.
As people who care deeply about patient safety, we are troubled by at figures that produce lurid headlines but distract from areas where harm may be most amenable to interventions. As researchers, we fear for efforts to engage with clinicians when they are confronted by headline-grabbing numbers that fly in the face of their clinical experience. And, finally, as concerned citizens, we would rather not have medical care characterized as more dangerous than firearms or motor vehicles.
On one point, we agree with Marky and Daniel: they say that “sound scientific methods, beginning with an assessment of the problem, are critical to approaching any health threat to patients.” Sadly, their paper does not exemplify such sound science.
1 Makary MA, Daniel M. Medical error-the third leading cause of death in the US, BMJ 2016;353:i2139.
2 Hall MJ, Levant S, DeFrances CJ. Trends in inpatient hospital deaths: National Hospital Discharge Survey, 2000-2010, NCHS Data Brief 2013;(118):1-8.
3 Shojania KG, Burton EC. The vanishing nonforensic autopsy, N Engl J Med 2008;358:873-5.
4 Hayward RA, Hofer TP. Estimating hospital deaths due to medical errors: preventability is in the eye of the reviewer, JAMA 2001;286:415-20.
5 Hogan H, Zipfel R, Neuburger J, et al. Avoidability of hospital deaths and association with hospital-wide mortality ratios: retrospective case record review and regression analysis, BMJ 2015;351:h3239.
6 Hogan H, Healey F, Neale G, et al. Preventable deaths due to problems in care in English acute hospitals: a retrospective case record review study, BMJ Qual Saf 2012;21:737-45.
Competing interests: No competing interests
I welcome Makary and Daniel’s call for better data on the burden of patient harm.(1) I strongly agree with the need for epidemiology to serve as one of the basic sciences of patient safety, just as it is for public health.(2)
While few would dispute that primary thesis, many commenters have taken issue with the authors’ analysis that “If medical error was a disease, it would rank as the third leading cause of death in the US.”
Some critiques focus on the term “medical error,” which is perhaps more blame-focused than a term like “avoidable patient harm.” Others have argued against the conclusion on epidemiological grounds, focusing on the assumptions used to arrive at the estimate, or on the use of point estimates alone, without any accounting for statistical uncertainty.
What seems to have been missed by the authors and critics alike is that, even if the numbers themselves are correct, they do not support the headline conclusion.
As I have argued elsewhere,(2) there is an important difference between “deaths from avoidable patient harm” and “deaths from avoidable patient harm in hospitals alone.” Conflating the two seriously underplays the importance of patient harm as a public health issue.
Our ignorance of the epidemiology of patient harm only grows more profound when we turn our gaze to care outside the hospital environment. But data from paid physician malpractice claims suggests that the burden of serious harm from outpatient care may be similar to that from inpatient care.(3)
And even that does not present a complete picture. It excludes much of the healthcare system, including long-term care, behavioral healthcare, home healthcare, community pharmacy, and outpatient care provided by non-physicians. To equate the death toll from avoidable harm in hospitals with the overall impact of patient harm is to contribute to the longstanding invisibility of patient harm in these environments.
If we aim to assess avoidable patient harm as a public health issue, it is also important to note that deaths (whatever their exact number may be) are only the tip of the iceberg. In hospitals, James found the rate of serious (physical) harm was 10-20 times higher than the number of deaths.(4) And, while it remains almost wholly uncounted at this writing, there is also an emerging recognition that avoidable psychological harm is an important part of the public health burden.(5)
Despite the lack of hard numbers, it is clear that the pandemic of patient harm is a major public health issue. It is time for governments and other funders to begin treating it as such, and to support patient safety improvement efforts with the same level of research funding and infrastructure that have allowed us to assess and address similar threats, such as smoking, heart disease, and cancer.
1. Makary MA, Daniel M. Medical error — the third leading cause of death in the US. 2016;2139(May):1–5.
2. Card AJ. Patient Safety: This is Public Health. J Healthc Risk Manag. 2014;34(1):6–12.
3. Bishop TF, Ryan AM, Ryan AK, Casalino LP. Paid malpractice claims for adverse events in inpatient and outpatient settings. JAMA. 2011 Jun 15;305(23):2427–31.
4. James JT. A New, Evidence-based Estimate of Patient Harms Associated with Hospital Care. J Patient Saf. 2013 Jul 15;9(3):122–8.
5. Card AJ, Klein VR. A New Frontier in Healthcare Risk Management: Working to Reduce Avoidable Patient Suffering. J Healthc Risk Manag. 2016;35(3):31–7.
Competing interests: No competing interests
In the article, “Medical Error – The Third Leading Cause of Death in the US”, Makary, et al. average the outcomes from four major studies to identify a rate of lethal medical errors for extrapolation to the 2013 US population. Their outcome is over-estimated due to (1) errors in operationalizing variables when combining studies, and (2) the authors’ decision, in the absence of data in the original studies, to classify the adverse event rates as 100% preventable. Third, this method was applied to the two studies with the highest sample population. These errors bias the resulting estimated medical error rate in the United States to be greater than it should be. After correcting for these biases, we found there to be 174,901 preventable medical error deaths per year in the United States, which is 30.5% lower than the published 251,454 deaths. The implications for this difference are enormous, with numerous media outlets reporting the finding that medical errors are much higher than they might be. Nonetheless, after our correction, medical errors remain the third leading cause of death.
The first error is due to improper operationalization of the term ‘adverse events’ within the studies that the authors examined. In the Health Grades study, a lethal adverse event rate of 0.71 was identified (263,864 lethal adverse events/37,000,000 hospitalizations). This data came from an assessment of patient safety incidents (PSIs) documented by the Agency for Healthcare Research and Quality (AHRQ). Unlike the definition of ‘adverse events’ in the other three studies, AHRQ includes ‘Failure to Rescue’ as a lethal adverse event, which incidentally contributes to 70% of lethal adverse events in the Health Grade’s study (=187,289 Failure to Rescue deaths/263,864 lethal adverse events), as outlined in Appendix F of that article.
The other three studies’ that Makary et al. examine define ‘adverse event’ without including ‘Failure to Rescue’. This is appropriate considering that the formal operationalization of ‘Failure to Rescue’ “is intended to identify patients who die following the development of a complication… Failure to Rescue may be fundamentally different than other [PSIs], as it may reflect … effectiveness in rescuing a patient from a complication versus preventing a complication…”. Included in that definition is also an evidence-based discussion on the construct validity of the principle, “…[F]ailure to rescue was independent of severity of illness at admission, but was significantly associated with the presence of surgical house staff and a lower percentage of board-certified anesthesiologists. The adverse occurrence rate was independent of this hospital characteristic.” In short, 70% of adverse lethal events in the Health Grades study may be better classified as resource and staffing issues rather than preventable medical errors, especially when compared to the inclusion criteria of ‘adverse events’ in the other three studies (which used the National Quality Forum’s Serious Reportable Events, CMS hospital Acquired Conditions, and the Institute of Medicine’s Global Trigger Tool.). This is a clear example of heterogeneity of results, and thus cannot be combined in a valid way. When Failure to Rescue is removed the definition of ‘adverse events’ in the Health Grade’s study, it overlaps well with the other studies inclusion criteria for ‘adverse events’ and this homogeneity of results allows valid compilation for broader interpretation.
The second error is due to the manner in which the authors’ assign the two studies with the highest rates of lethal medical errors to be 100% preventable without evidence to support this. In their calculations, Makary et al. applied a 100% preventable adverse event rate to the Health Grades study outcome. We previously showed that in this study, 70% of reported fatalities were characterized as failure to rescue -- a classification that includes a contention over preventability in its definition. The second study assigned a 100% preventable adverse event rate, the study by Classen, et al., was done by Makary et al. based on the following statement by Classen et al.: “Because of prior work with Trigger Tools and the belief that ultimately all adverse events may be preventable, we did not attempt to evaluate the preventability or ameliorability …of these adverse events.” This explicit statement proves that their definition of which adverse events were truly preventable was fundamentally different than the other studies, another example of misclassification bias leading to heterogeneic results in the Makary et al. study. Importantly, the Classen et al. study had the highest reported overall lethal event rate (1.13%), which was almost double and triple the event rates in the other two studies,. Moreover, many studies using the Global Trigger Tool (GTT) report a percent of adverse events that were considered preventable[5–9], suggesting that the statement ‘all adverse events may be preventable’ by Classen et al. is not in line with other studies seeking to identify those events using the GTT.
We have recalculated Makary et al.’s outcomes table, substituting a 30% preventable adverse event rate to the Health Grades Study because we showed that 70% of these events were subject to misclassification bias. This assumes that all deaths not including those due to failure to rescue are considered preventable. This therefore eliminates heterogeneic results and allows for a more valid approach to combining study outcomes. To resolve the issue of a 100% preventability rate applied to the Classen et al. outcome, we performed a literature search on other studies using the GTT in which preventability rates were calculated, calculated a weighted average of these values (68%, range: 45-72%), and applied them to the Classen et al. outcomes.
Our final estimate of 174,902 annual deaths due to preventable lethal adverse events in the United States is 30.5% lower than the original finding of 251,454 annual deaths due to preventable medical errors.
1 Makary MA, Daniel M. Medical error-the third leading cause of death in the US. BMJ 2016;353:i2139.http://www.ncbi.nlm.nih.gov/pubmed/27143499 (accessed 6 May2016).
2 Crages H. HealthGrades Quality Study; Patient Safety in American Hospitals. 2004.
3 Agency for Healthcare Research and Quality. A Guide to Patient Safety Indicators. 2003. http://www.qualityindicators.ahrq.gov/downloads/modules/psi/v31/psi_guid...
4 Classen DC, Resar R, Griffin F, et al. ‘Global trigger tool’ shows that adverse events in hospitals may be ten times greater than previously measured. Health Aff (Millwood) 2011;30:581–9. doi:10.1377/hlthaff.2011.0190
5 Landrigan CP, Parry GJ, Bones CB, et al. Temporal trends in rates of patient harm resulting from medical care. N Engl J Med 2010;363:2124–34. doi:10.1056/NEJMsa1004404
6 Stockwell DC, Bisarya H, Classen DC, et al. A trigger tool to detect harm in pediatric inpatient settings. Pediatrics 2015;135:1036–42. doi:10.1542/peds.2014-2152
7 Kennerly DA, Kudyakov R, da Graca B, et al. Characterization of adverse events detected in a large health care delivery system using an enhanced global trigger tool over a five-year interval. Health Serv Res 2014;49:1407–25. doi:10.1111/1475-6773.12163
8 Kennerly DA, Saldaña M, Kudyakov R, et al. Description and evaluation of adaptations to the global trigger tool to enhance value to adverse event reduction efforts. J Patient Saf 2013;9:87–95. doi:10.1097/PTS.0b013e31827cdc3b
9 de Wet C, Bowie P. The preliminary development and testing of a global trigger tool to detect error and patient harm in primary-care records. Postgrad Med J 2009;85:176–80. doi:10.1136/pgmj.2008.075788
Competing interests: No competing interests