Wikipedia:Reference desk/Archives/Science/2023 August 28

Welcome to the Wikipedia Science Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

August 28

Letter to the Editor Being Cited

I'm watching Painkiller (TV series) on Netflix and it covers how a letter to the editor called Addiction Rare in Patients Treated with Narcotics published in The New England Journal of Medicine played a major role in the Opioid epidemic in the United States. According to our article:

“ The article has been cited extensively as evidence that addiction was very rare among patients who were prescribed narcotics (more specifically, opioids)...the New England Journal of Medicine published a bibliographic analysis of the letter showing that it had been cited 608 times since it was published. ”

First, as far as I know, letters to the editor don't undergo the normal peer-review process that a normal article in a scientific journal undergoes. As a layperson, it seems shocking that a letter to the editor would be cited for anything, let alone hundreds of times.

Second, even if the letter had undergone normal peer review, it's only a single source. Unless other scientists are able duplicate the results in their own studies (again, which should be peer reviewed), the letter's conclusions should be treated with a high degree of skepticism.

How can such obvious violations of the scientific process have happened? Pealarther (talk) 11:59, 28 August 2023 (UTC)

On the first point, what are called "letters" are often short publications too short for a full article, but still peer reviewed. It's also not uncommon for such "letters" to be responses to an existing publication or article, either in support or refutation or just critical of one point. For example, an article discussion the vibrational spectra of a heme extracted from a protein might have a letter responding to it pointing out that one of the vibrational band assignments is incorrect, and is actually evidence that at least some of the heme was not cleaved and was still in the protein pocket (oftentimes the letter authors being very experienced with the same subject as the article). Such a letter will generally still be peer reviewed. I'm not sure on the terminology in terms of "letter to the editor;" it is possible that such terminology differs somewhat from journal to journal and my field hasn't seen me regularly looking at NEJM.

As to the second point, there's kind of a flaw in the current journal system in that you don't often get to publish an article that is simply duplicating the work in another article as the peer-reviewers will reject your article for not bringing anything original. Annoying, I know, just like how negative results (i.e. you conducted an experiment based on a hypothesis that your results ended up refuting) often don't get selected for publication either, despite a negative result being just as (if not more) informative than a positive result. Often, the main way to get duplication of results into scientific literature is to conduct a similar experiment, or an experiment building upon that already published. Since it is building upon the previous work, that the new experiment still proceeded as expected from the previous work is the duplication.

In the case of the specific letter you are discussing, the issue that has led to it being part of the opioid epidemic seems not to have to do so much with its lack of duplication so much as its misrepresentation by those citing it. The letter only discussed low doses of opioids and only when administered in hospital (and therefore under strict control of the medical staff), and was never about post-hospital use of narcotics as painkillers. That is has been cited so many hundreds of times to support the idea that narcotics prescribed for pain in general use has a low incidence of addiction is utter misrepresentation of the letter contents and that is the core problem, not lack of duplication per se. I would love to see a study done on these "citations" that examines conflicts of interest and who the authors citing this letter worked for.

--OuroborosCobra (talk) 14:01, 28 August 2023 (UTC)

I have taken the liberty of inserting paragraph breaks for legibility. Feel free to revert if you disapprove. --142.112.221.184 (talk) 02:36, 29 August 2023 (UTC)

Thanks! That's completely fine. I don't have a ton of experience with this "reply" button function, so wasn't sure how it would handle paragraphs. Your help is appreciated. --OuroborosCobra (talk) 15:33, 29 August 2023 (UTC)

Thanks for the response. I have another question. If the letter in the New England Journal of Medicine was being misrepresented, it should have been easy for anyone to cross-reference the claims with the source. How is it possible that, apparently, nobody fact-checked this? Pealarther (talk) 10:57, 30 August 2023 (UTC)

Many have fact checked it. See our article on the letter that you linked to, which has such fact checking and also has the full text of the letter. --OuroborosCobra (talk) 12:32, 30 August 2023 (UTC)

OxyContin was released in 1995 and this letter was cited as proof that less than 1% of people who take it would become addicted. I'm not sure when it became common knowledge that Purdue Pharma was misrepresenting the letter, but our article says:

“ In 2017, the letter attracted renewed interest because the New England Journal of Medicine published a bibliographic analysis of the letter showing that it had been cited 608 times since it was published. ...Of these 608 citations, the analysis also showed that 72.2% of them cited it in support of the claim that patients treated with opioids rarely developed addiction. ”

It is strange to me that it took so long for people to notice that the cited source did not support Purdue Pharma's claim regarding addiction risk. Pealarther (talk) 15:31, 30 August 2023 (UTC)

It’s called “conflict of interest.” Do you think the makers of OxyContin were interested in pointing out that the study they cited was limited to hospital settings and low doses only? That’s why I noted in my initial response to you that I would love to see a follow up study examining COIs of those ~600 studies. You seem to be repeating the same question which was already answered in our Wikipedia article and by me. Is there some detail you specifically don’t understand from these answers? —OuroborosCobra (talk) 17:20, 30 August 2023 (UTC)

Not all of the 439 citations that misrepresented the study and its conclusions were due to Purdue Pharma or its puppets; I expect these were in fact a minority. So the question remains, why did so many who were not beholden to Purdue Pharma, including major popular media, continue to parrot the misrepresentations? Without further detailed study it is hard to tell, but a reasonable assumption is that much of it was a form of laziness that is all too common: not actually consulting the sources one cites but relying on the summaries of others. Possibly, the propensity to blame the victims has also played a role. And why did it take 37 years before the NEJM (or anyone) drew attention to this widespread misrepresentation?  --Lambiam 18:22, 1 September 2023 (UTC)