Interesting, according to the medium article: 1) the numbers used in the study are wrong 2) they are not representative and 3) the numbers do not represent medical error
From the article:
The problem is that this is very subjective and mostly assessable in hindsight — it’s easy to say on reviewing a person’s chart that they might’ve done better on different medication, but whether that really does constitute a medical error is complex and not nearly as cut-and-dried as the BMJ paper suggests.
Yeah, there are medical errors like "patient got the opposite treatment of what they should have been given" or "gross misdiagnosis" but there's also "didn't get the ideal drug for their case - but got something ok"
Not just flawed argument, but wildly inaccurate data that was then grossly misapplied.
Unfortunately a lot of people on HN will just read the headline and think this is true, when it's very far from being the case. I wish we could link this whole thread to the refutation instead of further propagating this misinformation.
Based off the fact that mistakes that lead to death are so high... If we had statistics on errors that didn't lead to death, the results would be an astronomically larger illustration of incompetence then Just mistakes that lead to death.
Mistakes don't sort neatly by causes into fatal and non-fatal. Some of almost any kind are fatal. Anything that prevented fatal mistakes would eliminate most others.
I’m not in the medical industry, and I’m a Brit and don’t know what the estimated stats are here. But the way I think about this is that medicine must be at least as complex and side effects are at a minimum as hard to anticipate as software development.
Bugs are a fact of life in development, nobody thinks software having any bugs at all is normal or that bugs are an irrefutable indicator of incompetence. That’s because anyone with any knowledge of Tyne subject has developed software and spent a huge part of that time fixing their own bugs.
The way you deal with that is through testing, defensive coding, making software maintainable, etc. You minimise the impact bugs have and try to ensure mistakes are recoverable. You also avoid counterproductive measures like punishing bug reporting, because that will just lead to cover ups and sweeping the problems under the rug.
The last thing we want to do is punish the reporting of medical errors. So yes I absolutely agree, and the way to do that is to build an honest and healthy culture around reporting of medical mistakes.
We already know the most effective way to reduce, and often eliminate, medical mistakes: rigorously applied checklists. They are still used only here and there, years after this was well demonstrated. It is well past time to treat not rigorously following checklists as negligent malpractice.
Medicine depends overwhelmingly more on manually applied procedures that would be automated, in a computer, as patients and staff manifestly are not. So, the same mistakes are repeated again and again. A software fix would correspond to changing a checklist.
Still, it would be no bad thing to use checklists for manual processes in software development, but I have never heard of any.
Exactly. Every time I come to a medical thread on HN I realize how bad even intelligent people are at understanding medical statistics and recognizing when to raise an eyebrow of skepticism. It really does take a lot of study and/or professional training to develop those skills.
Interesting. I would be one of those people who can't fully understand the accuracy of these numbers, so I have to take other peoples' word for it.
Are you part of the medical field yourself? If so then I wouldn't trust anything you say given that if the statistics were true, what you're saying has a high chance of just covering yourself up.
I'm not, but I've spent a lot of time over many years cultivating an understanding of epidemiology and med statistics, especially as they relate to study/trial design. The article linked above points out many of the problems with the parent's study. Medical errors are a problem, but we don't know much about it because of these problems, so I worry, as always, that people will take the wrong message away from headlines like this.
From the article:
The problem is that this is very subjective and mostly assessable in hindsight — it’s easy to say on reviewing a person’s chart that they might’ve done better on different medication, but whether that really does constitute a medical error is complex and not nearly as cut-and-dried as the BMJ paper suggests.