Medical errors #3 cause of US deaths

NPR story here
Actual journal article here By Martin Makary and Michael Daniel of JHU

I saw a summary on slashdot and wondered “has Marty seen this?” Considering he’s the corresponding author, that’s likely. Also wrote Unaccountable: What Hospitals Won’t Tell You and How Transparency Can Revolutionize Health Care.

I haven’t dug into their methods yet, which will be interesting because their estimate is much greater than numbers I’ve seen before IIRC.

Not to downplay the report, and I only read the NPR article, but I wonder how much of an effect this really has.

The article mentions that many of these patients had cancer or heart disease. The doctor made an error in treatment and the patient died. What wasn’t clear to me is if this error merely hastened the inevitable (i.e., patient comes in with a heart condition and has 3 weeks to live, doctor makes an error and now the patient dies in 2 weeks) or if this was a situation where an easily treatable condition was botched (i.e., patient comes in with a heart condition and has 3 weeks to live, proper treatment would mean they could live another 20 but the doctor screwed up and they die in two) or even if it was a lack of action (i.e., , patient comes in with a heart condition and has 3 weeks to live, doctor misreads the chart and does nothing, patient dies in 3 weeks).

Obviously, none of the above situations are good, but there’s a difference between an oops on the operating table in what was a risky surgery anyway and an oops in a routine checkup where the doctor prescribes cyanide instead of amoxicillin*.

How many Quality Adjusted Life Years are being lost by these medical errors?

  • I realize that the pharmacist should catch that but I wanted an example.

From the full BMJ article:

So yes, they’re counting someone with weeks to live who undergoes a treatment which shortens it.

I hope they’re separating fully informed & consented “This procedure may give you an extra 6 months, or you may die on the table” procedures from “Oops, we gave you the wrong drug and you died today instead of next week” of things, but this quote doesn’t give me much confidence:

“Does not achieve its intended outcome” - does that mean that any surgery where someone dies anyway, or a course of chemo that doesn’t kill the cancer, becomes a medical error death in their terms?

This is largely an indication of how safe our live are. It’s also poor use of statistics that doesn’t allow us to distinguish between an error that shortens a person’s life by a day vs. one that takes years away. Better statistics may not make the medical profession look any better, however we should know what the numbers really do show. And then we also have to consider that without the level of medical care we do have, however you want to rate it, the same people plus many more would be dying of disease instead of medical error.

In conclusion, we waste shiploads of money on both good and poor medical care no matter what these particular statistics show.

Well, if you accept the plain reading of the definition, then no. The surgery is not unintended. A well-performed and intentionally-performed surgical procedure in which the patient still dies would not qualify. A well-performed surgical procedure with a bad outcome but performed unintentionally (such as, on the wrong patient) probably would.

It’s not just the outcome: it’s that the outcome came about because of a preventable error.

I think you are missing the “or” in the definition …

Emphasis added.

In short, taking this definition literally, a “well-performed and intentionally-performed surgical procedure in which the patient still dies” would in fact qualify, because death was not the “intended outcome” the surgery was supposed to achieve.

I see what you’re saying, but I’ll just have to assume that the use of the word “or” was improper. Because the entire article disintegrates into meaninglessness otherwise. (“The third leading cause of death in the United States is the lack of miraculous healing powers in medical professionals.”)

I suspect it hinges on the citation I can’t follow (“Leape LL. Error in medicine. JAMA1994;272:1851-7.”) I’m betting on a misquote here from Leape’s original.

I don’t know if it is a misquote or not. One of the reasons I would critique the article as overly alarmist is that its (apparently) overly-expansive definition of what constitutes “error” does, indeed, make the conclusions in the article (medical error third leading cause of death!!) totally meaningless.

It appears, on its face at least, to be an unfortunately common way of lying with statistics: define a concern overly-broadly, so that it includes things that range from the extremely serious to the relatively trivial or commonplace; then allow your reader to assume that the high incidence of your chosen concern happening means that relatively serious problems may be happening - a lot; then call urgently for “more study”.

It is worth reading the responses (particularly the last one entitled “contribution is not causation”):

I haven’t looked at the study, so maybe this is a dumb question, but how do we know that the original doctor’s decision is erroneous and not the researcher’s later reevaluation?

It’s not a misquote - here’s a GIF of Leape’s original article. The sentence appears at the end of the 1st column.

I agree that as stated, it’s an utterly terrible definition of “error”.

Haven’t reviewed the source, but if ‘omission’ includes missing something that’s only obvious after the fact when you know where to look, it will skew the data.

I’ve had a couple of recent medical procedures (two ER visits and a scheduled surgery) and I can tell you at my hospital, the nurses confirmed who I was every time they talked to me, had me check my wristband for accuracy several times, and my doctor wrote on me where she was supposed to operate.

I can’t speak to misdiagnoses, except to say the ER doc said one thing and the nurse said something else (the nurse was right, the doctor wasn’t.)

Mistakes happen.

During my spouse’s most recent hospital stay despite the alert bracelets, sign on his door, the notes in his chart, and so on and so forth a mistake managed to trigger his latex allergy. Fortunately, it was caught quickly and no permanent harm done, but Stuff Happens. Since we’ve driven down many other causes of death it doesn’t entirely surprise me that “medical error” is rising up the list of top causes.

Some of the examples they give in the article are patients subjected to unnecessary invasive diagnostic tests that directly caused the death of the patient. There’s a lot of unnecessary diagnostic testing going on in the US.

I kind of believe the figure, although I’m not so sure about their methods. They did have to do a lot of guessing and hand-waving. The best thing to come out of this report is perhaps in future people will more accurately track these kinds of deaths and we can A) get some real data, and B) figure how how to eliminate medical errors as much as possible.

To come up with this conclusion they must be using a very liberal definition of medical error. As a physician that specializes in the care of patients in nursing homes, I have had a fair number of patients die. The thing is that everyone is going to eventually die. I can’t think of any of the top of my head that I would say died due to medical error, whether they be my own patients or those of my colleagues. My definition is probably a lot more restrictive than than that of the person who wrote the article in the OP. I would count things like a death resulting from an antibiotic being administered to a patient with a documented allergy, an operation on the wrong body part, or someone having an obvious heart attack being sent home from the emergency room with a diagnosis of indigestion. I would not count someone with cancer on hospice who dies due to taking too much morphine, someone that is already very ill and that has a complex surgical procedure during which everything is done correctly but none the less dies due to a complication, or someone who dies at home due to noncompliance with the appropriate treatment regimen. Deaths in the latter categories are far more common in my experience than those in the former categories.

Anecdotal story:

My wife and I know a couple who lost a son to a medical error. He was 15 years old, as fit and healthy as they come. He went to the hospital to have a pin set into a cracked bone in his wrist (he’d cracked it playing football, so, healthy kid). They put him under general anesthesia (a breathing tube was apparently part of this), everything went just fine, and they moved him to recovery.

Here’s what I understand happened, and I freely admit that I don’t know everything: in recovery, a nurse apparently hooked up his breathing tube wrong. Something like, the “inhale” supply tube was hooked up to his “exhale” tube, so that he couldn’t get rid of the carbon dioxide, and it just built up in his bloodstream and body. Under sedation, he couildn’t do anything to help himself, and a healthy 15-year-old kid laid there and quietly smothered to death.

I’ve never seen such grief as I saw at his funeral.


The definition – the basis of statistical selection – is so broad that any patient death could be called medical error. That probably leads to the dramatic inflation in cause-of-death ranking.

Sheesh. Kinda makes me want to speculate what they’re after with that particular flawed analysis.

Anything less would not get headlines.

Actually, there may be a small use to such a broad definition, and that’s to get the public to think more critically about voluntary procedures with small chances of negative outcomes. A 99% chance of survival sound good, but that’s still a high enough chance of death to make it possibly the most dangerous thing you do all year.

He’s made decent money writing and talking about medical mistakes. Whether that factors in, I don’t know.