Monday, August 6, 2018

More proof medical error is not the third leading cause of death

Over the last 20 years, estimates of the number of deaths caused by medical error have risen from 44,000-98,000 in 1999 [1] all the way up to 440,000 [2] and 251,000 [3]. Despite my efforts [4, 5] and those of others [6, 7] to debunk these guesses, they continue to permeate the lay press. If you Google “third leading cause of death,” you will find countless headlines naming medical error.

The papers claiming medical errors cause so many deaths assume that all complications result from errors and all complications are preventable. They extrapolate their final numbers from small studies not designed to or capable of estimating deaths due to medical error nationwide.

The most recent figures available from the National Hospital Discharge Survey [8] state that the number of hospital deaths dropped from 776,000 in the year 2000 to 715,000 in 2010. It is simply not plausible that 251,000 (35%) or 440,000 (61%) inpatient deaths are due to medical error.

A recent study [9] from Norway found that of 1000 consecutive in-hospital deaths reviewed, only 42 (4.2%) were judged to be probably (greater than a 50% chance) to definitely avoidable.

The investigation took place at a 900-bed university hospital and excluded patients who died on psychiatric or obstetric services and those 16 years of age and under. Patient records were reviewed by five clinicians with experience in patient safety and quality improvement. The most likely place for a possibly preventable death to have occurred was during an invasive procedure. However, the authors point out that “medical errors, both of omission and commission, are more easily discerned in surgery than elsewhere.”

Retrospective case record review can be difficult due to problems with inter-rater reliability, hindsight bias, and the difficulty in deciding what might have happened to a patient had an error not occurred. The median age of patients in the Norwegian study was 77. The 42 patients whose deaths might have been avoidable were significantly more likely to have had trauma, poisoning, or other external causes of hospitalization and more likely to have been admitted to a higher level of care.

Since extrapolations are part of most papers regarding deaths due to medical error, here is mine. Assuming there are approximately 700,000 patients per year die in US hospitals, a rate of 4.2% yields 29,400 deaths. That is an unacceptable number and should be reduced, but it is far less than 251,000 or 440,000.

As far as I know, the Norwegian study received no media coverage. Why not? Because it is good news and would not have generated any sensational or clickbait headlines.


1. Institute of Medicine. To Error Is Human: Building a Safer Health System, 1999
2. James, J. A new, evidence-based estimate of patient harms associated with hospital care. J Patient Saf, 2013
3. Makary M, Daniel M. Medical error—the third leading cause of death in the US. BMJ, 2016
4. Skeptical Scalpel. Medical errors and deaths: Is the problem getting worse? September 23, 2013
5. Skeptical scalpel. Are there really 250,000 preventable deaths per year in US hospitals? May 5, 2016
6. Hogan H. The problem with preventable deaths. BMJ Qual Saf 25:320-3, 2016
7. Shojania K, Dixon-Woods M estimating deaths due to medical error: The ongoing controversy and why it matters. BMJ Qual Saf 26:423-8, 2017
8. National Hospital Discharge Survey
9. Rogne et al. rate of avoidable deaths in a Norwegian hospital trust as judged by retrospective chart review. BMJ Qual Saf Published Online First: 19 July 2018

Thanks to @MaryDixonWoods for alerting me to the Norwegian paper


Unknown said...



Bharani Padmanabhan MD PhD
Brookline MA

artiger said...

Clickbait or not, the problem is that real money, time, and resources are being spent on sensationalism. To think of how much preventive care and wellness could be paid for and administered instead...oh dear.

Anonymous said...

Anonymous Europe: What patients and families need to understand is, that anyone who works makes mistakes at some point. We are also human. We work under the daily threat of lawyers snapping at our heels, whereas our free time(when we may recharge) is reduced to almost nil...
Besides the training we receive is more like training bureocrats than surgeons. We should increase training time, decrease documentation time and increase time for recharging. With healthy, rested, well trained doctors the amount of mistakes will also decrease. At least that is my opinion.

Anonymous said...

I never believed the hype around medical errors. Anecdotally, I can think of very few patients who've died from true medical errors, even fewer who died from actual doctor error.

Anonymous said...

I would believe the figure of 440000 as a minimum. This research was first conducted by a NASA scientist who lost his 19 year due to medical error. His book Sea of broken hearts explains how this happens. Also having unfortunately been in bad health, I have unfortunately had to endure the horrors of negligent detached incompetent medical professions in my own life and have been left with a chronic condition which has ruined my life. But at least I am not dead. My sister who is a nurse told me a recent story of a man brought in by his wife and 4 year old daughter, long story short he was not treated as he should have been and died. I could go into details but bottom line if the doctors operated odds are he would be alive. Also a lot of covering up by doctors for other doctors. My sister witnessed this as well. In short the day arrogant overpaid ultimately useles doctors are replaced with machines who can be programmed with empathy (the first thing a doctor loses) and superior medical knowledge the better off we all will be.

Anonymous said...

From my experience in a western country, the charts are completely unreliable. Most of them are written retrospectively after the patient is discharged/dead.

Just a small example: During a two months hospitalization that resulted in dead, the patient refused to eat and suffered from severe cachexia that definitely contributed to his final demise. However, in the charts you won't find a shred of evidence, because the medical team never took his weight, did not keep tabs on his meals and never ordered the appropriate blood tests (albumin, total protein).

So I'm interested to know how the Norwegian auditors would rate the inevitability of this patient's death, for example.

I wonder if the Norwegian auditors would have spoken with the families of the deceased instead of relying on the unreliable charts, perhaps their assurance that 958 of those 1000 deaths were unavoidable would not be so high.

Much more objective research is autopsy research, and in those studies up to 50% of diagnoses (and therefore treatment) are found to be completely wrong or at least partially inaccurate.

Skeptical Scalpel said...

Anon, what evidence is your first paragraph based on? In the US, most hospitals have electronic medical records that time chart entries to the second. Most notes are written contemporaneously. Anything added after discharge is clearly identifiable.

Anecdotes are not proof. Autopsies find lots of things that may or may not be relevant to the patient’s death.

Post a Comment

Note: Only a member of this blog may post a comment.