Or: Do 30 million people really take NSAIDs every day?

I wanted to know how many people in the world took non-steroidal anti-inflammatory drugs (NSAIDs).

You know, aspirin, ibuprofen, diclofenac, naproxen. Those things. Common drugs to tackle pain, fever, inflammation, and a handful of other things. Prescribed by clinicians, and bought over the counter.

I wanted to know: how many people, each day, take one or more of those?

Well, if I’m being totally honest: I didn’t want to know that.

I actually wanted to look up NSAIDs to learn more about their history, and in doing so, chose a few articles to read.

There are, to be fair, lots of interesting things about the history NSAIDs, including a fascinating “Account of the Succefs of the Bark of the Willow in the Cure of Agues” by Reverend Stone in 1763. It’s well worth a read if you can tolerate “f”s as “s”s: it really, really makes it sound like Reverend Stone should be an Igor.

But, instead of being able to concentrate on the history of NSAIDs, I became distracted by a common piece of information, repeated in a good proportion of the articles and websites I was reading through.

The claim is this, and I wish to write about:

30 million people take NSAIDs worldwide every day

Do they, indeed?

I kept seeing it.

Time and again.

In both journal articles and on websites.

So, is it true?

In this rant, I’m going to take you through my journey into trying to find out whether this claim is, indeed, true.

The tldr; version is that, in 2025, I would be incredibly surprised if this claim was remotely true. It could be true, but only really in the same way that if I threw a random calendar from a box labelled “assorted calendars: years 2000 to 2100” at a clock in the broken clock warehouse, there would be a chance the clock would be showing the correct time, and the minute hand would have pierced the correct day and month on a calendar with the correct year.  The claim has continued to propagate because journal articles are lazily referencing claims without checking if they’re justified. What with the advent of AI bullshit proliferating, is we also allow people to generate and propagate bullshit, we’ll no longer be standing on the shoulder of giants. We’ll be standing in bullshit.

Referencing in journal articles

A quick brief for anyone not well versed in epidemiological journal articles (most articles talking about NSAIDs will fall under the umbrella of epidemiology): if you make a claim in an article, you reference it. That is, if you make a claim, you need to justify it. The exception is for all but the most obvious of claims (although it’s worth noting that obvious claims can sometimes be incorrect).

If an article claimed, for instance, that “malaria has killed more people than any other illness in human history”, I would expect a reference to wherever that claim came from, justifying the claim. Who estimated that? What methods did they use? Were they considering all illnesses, or just a handful? How do you know what people died of thousands of years ago? How do we know they died of malaria and not something else?

A reference should be able to answer questions like that, so we can avoid any cases of “well, I HEARD that malaria has killed the most people. Not sure where I heard it. But it’s probably true, so I’m going to put it into my article.” Or, in some ways more problematically: “someone said it was true, so I’m going to write it in my article and reference them. I haven’t checked to see whether it IS true, and I don’t even know if what I’m referencing justified the claim, but… I’ll do it anyway.”

It’s not necessary to reference claims that an article makes based on the evidence presented in the article itself, but it should be clear whether claims are made based on previous or current evidence. It is also wise to make claims that are supported by your own evidence, rather than make the claim you wanted to make, even though your evidence doesn’t quite support it.

When an article makes a claim without referencing it, or using the evidence from the article to justify the claim, that claim can be ignored.

Saying “trust me” is not good enough.

Equally, if the reference for a claim doesn’t justify the claim, or reference something else that justifies the claim (or references something the references something that references something that… justifies the claim), then the claim can be ignored. There needs to be a justification for a claim. Evidence that says: “this claim is valid”.

Otherwise, anyone could make up anything, and pass it off as the truth.

I’m fully aware that, in 2025, one doesn’t need it pointed out to them that people could just make things up and pass them off as truth, and have millions of other people believe them.

But there has to be a higher bar for science.

Otherwise, science would be faith.

We tried that.

It didn’t work.

Note: I made up the claim about malaria as an example because I heard something about it years ago, then promptly forgot all about it. I checked google to see if it was true. Specifically, I asked “has malaria killed the most people” (without the quotes). The AI overview said:

Gemini now links to its claims: it’s referencing, good!

It may appear from the above that malaria has killed fewer people than TB, smallpox, bubonic plague, and influenza. But the website Gemini took that evidence from said malaria has killed up to 5 billion people, according to Brian Faragher (though how he arrived at a figure of 5% to 6% of all people is hazy).

So, a) Gemini’s claim of “other diseases have killed more people” above isn’t right, based on the evidence it’s referencing, and b) although I’m not going to write about how the claim “malaria has killed half of all humans who have lived” was started and propagated, and how even the “debunking” claim made by Brian Faragher was taken as truth (which it might be, I stopped trying to find how he estimated those figures when I came up with nothing after a few minutes), I imagine I could have done.

Anyway, malaria sucks. Good thing there’s a vaccine now.

Journal articles

Let’s move on to the juicy filling of this rant.

Journal articles are repeating this claim (and so am I, now):

30 million people take NSAIDs worldwide every day.

How many articles make the claim?

I’m not sure.

I tried to keep track while I was reading, but there were a lot, and I wasn’t documenting them the entire time. And I can’t find every journal article that has this in the introduction or discussion: because journal publishers are greedy, parasitic, anti-scientific (see below) entities that care only for profit, quite a substantial proportion of the world’s research into human health is behind paywalls, so I can’t just search for the claim in the bodies of articles, only abstracts.

Nonetheless, here are the claims made a few selected articles, along with information about their references listed for the claims. All articles are fully referenced, but, apologies, many are behind paywalls: hence the screenshots.

I’ve put together flowchart to make it easier to keep track of all the articles and websites I’m going to talk about:

1. Ren, Li MD; Yi, Jie MD; Yang, Jing MD; Li, Peng MD; Cheng, Xueyan MD; Mao, Peixian MD^*. Nonsteroidal anti-inflammatory drugs use and risk of Parkinson disease: A dose–response meta-analysis. Medicine 97(37):p e12172, September 2018.

Reference 1 (Ascherio 2016) doesn’t mention how many people take NSAIDs, and reference 7 (Shin 2017) doesn’t mention the prescription volume of NSAIDs or the proportion of people over the age of 60 taking NSAIDs in the UK.

Reference 6 (Raskin 1999), which I talk about below, does have the claim for 30 million NSAIDs daily. But, so far as I can see, nothing about the annual prescription of NSAIDs in the US. So maybe just a case of getting references mixed up?

Either way, stellar work.

2. Gunaydin C, Bilge SS. Effects of Nonsteroidal Anti-Inflammatory Drugs at the Molecular Level. Eurasian J Med. 2018 Jun;50(2):116-121. Epub 2018 Jun 1. PMID: 30002579; PMCID: PMC6039135.

Reference 1 (Bhala 2013) doesn’t mention how many people take NSAIDs.

Again, just fantastic work.

3. Conaghan, P.G. A turbulent decade for NSAIDs: update on current concepts of classification, epidemiology, comparative efficacy, and toxicity. Rheumatol Int 32, 1491–1502 (2012).

Reference 2 (Singh 2000) DOES make the claim about more than 30 million people taking NSAIDs each day (good!). Reference 1 (Laine 2001) is also fine (good!) – I won’t write about this as the article only talks about US over-the-counter NSAID use, though their references are not available online (they seem like industry data, that’s all I know).

Still, don’t confuse prescriptions written with people taking drugs: there are plenty of repeat prescriptions, and plenty of people don’t take their prescriptions. Good start though.

4. Zoltán S. Zádori, Kornél Király, Mahmoud Al-Khrasani, Klára Gyires. Interactions between NSAIDs, opioids and the gut microbiota – Future perspectives in the management of inflammation and pain. Pharmacology & Therapeutics, 241, January 2023.

The reference is (Singh 2000) again. Note this paper was published in 2023. So only 23 years out of date.

Wonderful.

5. Ourania S. Kotsiou, Sotirios G. Zarogiannis, Konstantinos I. Gourgoulianis. Prehospital NSAIDs use prolong hospitalization in patients with pleuro-pulmonary infection. Respiratory Medicine, 123, February 2017.

Reference 1 is our dear friend (Singh 2000) again. Note this paper was published in 2017.

Better than 2023, but still not good.

6. Singh, Gurkirpal. Gastrointestinal Complications of Prescription and Over-the-Counter Nonsteroidal Anti-inflammatory Drugs: A View from the ARAMIS Database. American Journal of Therapeutics 7(2):p 115-122, March 2000.

Ah, here it is. Singh 2000. For a long time, this was the OG reference: I couldn’t find anything referencing an earlier paper, so thought this might have been the original claim!

But… The article is behind a paywall, even for me.

Great.

That’s why there’s no reference or supporting information in the screenshot above: this is taken from the first line of the abstract (which do not include references, they go into the body of the article).

I figured if I couldn’t find anything else, I’d just buy the paper and see if it has evidence for the claim. I did think that the ARAMIS database didn’t look like it has the capacity to know how many people take NSAIDs worldwide: the Arthritis, Rheumatism and Aging Medical Information System is a chronic disease databank, started in the 1970s. It contained, in 2005, data on about 14,000 rheumatoid arthritis patients, osteoarthritis patients, and individuals from normal and healthy aging populations. That’s not enough people or information to make any kind of guess at the number of people worldwide taking NSAIDs daily.

So, it was entirely possible Singh is referencing someone else.

Before 2000.

Wonderful.

We’re already 25 years out of date with Singh, how much further are we going to go?

7. Singh G, Triadafilopoulos G. Epidemiology of NSAID induced gastrointestinal complications. J Rheumatol Suppl. 1999 Apr;56:18-24. PMID: 10225536.

After a bit more searching, I found another reference from Singh, this one published in 1999. I’m guessing (Singh 2000) is just repeating the claim made here. You may notice the link above is to PubMed. It’s because the Journal of Rheumatology only has archives back to 2001, so doesn’t contain this article. A quick look on the Wayback Machine tells me they never had articles from before 2001. Awesome.

However, the Internet Archive (same site as the Wayback Machine – it’s a good site), has a great repository of old journal articles, including the article! Except I can’t check it out. Likely because, as the website says, 500,000 books were made unavailable after publishers sued the Internet Archive.

Great.

Thanks publishers.

Make us pay to publish, do nothing to earn that money beyond typeset and host content, make us pay to see our own work, then remove anything too old from the internet.

I have no legitimate way to see this article now.

All I can see if that the article is published in a Supplement covering a meeting on the 12^th November 1997. So, the claim is actually from 1997. Or before 1997. The article title makes me think their referencing something else in any case: “The epidemiology of NSAID induced gastrointestinal complications” almost certainly doesn’t include a detailed analysis of prescriptions and over the counter use of NSAIDs worldwide.

If it’s a conference abstract, it also almost certainly won’t contain references, so maybe not being able to take a look at it isn’t so bad…

—

For a time, I was stuck here.

Then, a few days later, I hit upon a rich vein of references.

This was by luck, as I was trying to find earlier references by Singh: article authors often work in the field for a long time, so maybe Singh has earlier articles I could access, and so find out where the claim came from.

Instead, I found this:

8. Raskin, Jeffrey B. Gastrointestinal effects of nonsteroidal anti-inflammatory therapy. The American Journal of Medicine, Volume 106, Issue 5, 3S – 12S, 1999

Reference 2 (Paakkari 1994) seems legit, let’s see what’s there.

9. Paakkari, I. Epidemiological and Financial Aspects of the Use of Non-Steroidal Anti-Inflammatory Analgesics. Basic & Clinical Pharmacology & Toxicology, Volume 75, Issue s2, October 1994

Stymied again by not having access. None of my usual tricks worked to find it.

But, what do you know, the reference list is available for free. Maybe I can find the right reference for the claim just by going through the list?

It was the fourth reference down (Hollingworth 1991).

That was lucky.

10. Hollingworth, J., & Alexander‐williams, J. (1991). Non-steroidal anti-inflammatory drugs and stercoral perforation of the colon. Annals of the Royal College of Surgeons of England, 73 6, 337-9; discussion 339-40 .

Reference 15 (Bjarnson 1987) also seems legit.

We’re getting there!

“There” in this case is a 1987 paper, but nonetheless, we’re getting there!

11. Bjarnason I, Zanelli G, Smith T, Prouse P, Williams P, Smethurst P, Delacey G, Gumpel MJ, Levi AJ. Nonsteroidal antiinflammatory drug-induced intestinal inflammation in humans. Gastroenterology. 1987 Sep;93(3):480-9. doi: 10.1016/0016-5085(87)90909-7. PMID: 3609658.

I linked to PubMed, because the full text link takes you straight to the PDF. The full text is accessible though: enjoy.

This is where my search has seemingly come to an end.

The claim is made in the discussion of the article.

The content of the article is actually about NSAID use in 90 people with rheumatoid arthritis and 7 people with osteoarthritis.

It at no point attempts to estimate the number of people worldwide using NSAIDs daily.

The claim is not referenced.

It is, to all appearances, a figure plucked from the air.

I checked through the reference list, and couldn’t find any mention of the claim. I didn’t check all references, as most related to intestinal problems in rats, and I couldn’t access or find others. I also wasn’t massively hopeful about finding another reference: other claims are referenced in the article, just not this one.

I checked for mentions of the claim from before 1987, but honestly, it’s not easy to find papers from that long ago, let alone those specifically mentioning a specific claim. I searched for a while, then gave up.

However, it is worth mentioning here that the claim in Bjarnson 1987 IS NOT that 30 million people take NSAIDs every day.

It has morphed from the number of patients receiving NSAIDs on a regular basis to the number of people receiving NSAIDs daily. You can see it by scrolling up through the screenshots:

Bjarnson 1987: >30 million patients receive NSAIDs on a regular basis worldwide

Hollingworth 1991: NSAID medications are consumed by 30 million patients worldwide

Paakkari 1994: [Unknown]

Raskin 1999: [NSAIDs] are used by 30 million persons daily

Not only has the “more than” disappeared, “daily” has replaced “on a regular basis”. It’s possible they mean the same thing, of course, but “regular basis” could mean “over the course of a year, 30 million patients received a long-term prescription for NSAIDs”, which may equate to 10 million patients daily for 4 months each, or 2.5 million patients daily for one month each. Or it could mean, if we asked everyone in the world: “have you taken NSAIDs in the last year”, 30 million people would have said “yes”.

It very much depends what the evidence was, if any, for the claim in Bjarnson 1987. That’s one reason why we need claims to be referenced…

It’s fairly moot if it’s a number plucked from thin air, of course…

Note that a different paper referencing Paakkari 1994 (Hawkey 1998) also claimed that NSAIDs were being taken “by about 30 million people each day”, so I think Paakkari 1994 was the article that changed “on a regular basis” to “daily”.

In fairness, because I couldn’t see Paakkari 1994, they may be referencing something other than Hollingworth 1991, and the numbers just line up really, really well, but I think probably not.

Let’s move on to websites for a bit, before I conclude that a claim from 1987 about how often medications are taken worldwide is utterly irrelevant in 2025.

Websites

Websites, particularly medical websites, aren’t known for providing robust referencing for their claims. This website does (or, at least, I try to), but most don’t.

Which is a huge shame, because people probably use websites more than journal articles when trying to find out things about health.

Let’s see how they do regardless:

1. WebMD. Pain Relief: How NSAIDs Work

There’s no clear link to a reference here, and the claim is now specific to Americans. The “SOURCES” are a bunch of people, one of whom is the spokesman for the “American Gastroenterological Association” mentioned above, so I’m guessing the reference for the claim is “Byron Cryer, MD”.

Note: Byron Cryer the same person as Brian Faragher, who was estimating how many people have died of malaria. Clues to that include the fact their names are different. I only mention this because for a moment I thought it might have been the same person. To be fair, there’s been a lot of word since then.

It might be coincidence that the claim is for 30 million Americans (rather than people worldwide), and Byron Cryer may have estimated this figure using robust evidence that they just aren’t referencing.

But without referencing something, there’s nothing I can do to fact-check it.

Good start.

2. London Pain Clinic. What Are NSAIDs & What Are They Used For?

There is only one reference in the entire article (MacGill 2021, Medical News Today), so let’s assume the claim comes from there.

3. Medical News Today (2021). Everything you need to know about NSAIDs

The hyperlink reference (Gunaydin 2018) takes you to the second article listed in the above section, which referenced (Bhala 2013), which in turn didn’t say anything about how many people take NSAIDs.

So… Yeah. That doesn’t work out well.

However, if you click on the Medical News Today (2021) hyperlink I’ve given, it’ll take you to a Web Archive 2021 version of a web page that still exists, last updated in 2024. The current version doesn’t mention how many people NSAIDs daily, but still references (Gunaydin 2018) as a source.

No idea why the claim for 30 million people was removed, it doesn’t say.

Probably a good thing though, given all this.

4. U.S. Pharmacist (2016). NSAIDs: Balancing the Risks and Benefits

Reference 2 (Conaghan 2011) is the third article listed in the above section, which referenced (Singh 2000).

A few more websites make the claim with no references:

5. Gastroenterology Specialties P.C. Nonsteriodal Anti-Inflammatory Drugsn (NSAIDs) <- not my typo

6. Science Daily. Study Shows Long-term Use Of NSAIDs Causes Severe Intestinal Damage

7. EurekaAlert! Non-steroidal anti-inflammatory drugs inhibit ovulation after just 10 days

But at this point I’m just searching Google for the claim in quotes, and it feels fairly pointless.

Actually, one more, because it pissed me off.

8. SESAME care (2023). A complete guide to NSAIDs

The “recent reports [sic]” is a single link to Gunaydin 2018, published 5 years before the SESAME care article was published. It wasn’t “recent” even if it did the original research to justify the claim. Which it didn’t, of course.

“Recent reports”…

Eurgh.

Moving on.

Does any of this matter?

It matters to me?

I would love to know how this story began, but I doubt I ever will.

But in terms of justifying the claim, no, it doesn’t matter.

It simply does not matter, in 2025, whether a paper in 1987 justified its claim that 30 million patients receive NSAIDs on a regular basis worldwide.

Because 1987 is 38 long, long years ago.

It is, in fact, 2 years before I was born, and a lot has happened since then:

1. The total world population was 5 billion in 1987, compared with over 8 billion in 2024
2. The median global age was 22 years old in 1987, compared with almost 31 years in 2024
3. The proportion of people globally aged 65 years or older was 6% in 1987, compared with 10% in 2023
4. The global obesity rate for women was 8.8% in 1990, compared with 18.5% in 2022, and for men was 4.8% in 1990, compared with 14.0% in 2022
5. The percentage of people earning under $5.50 per day decreased from 69% in 1987 to 45% in 2022

I tried to find a similar dashboard for chronic health conditions globally, but didn’t find one: I suspect it would show an increase in chronic health conditions, as people are living longer worldwide (excepting lately in some countries what with COVID-19 and everything).

The number of NSAIDs, their availability in different countries, and their use, have also changed (most of this evidence comes from the respective wikipedia pages for each NSAID):

1. Ibuprofen was patented in 1961, became available over the counter in 1983 in the UK and 1984 in the US, and the patent expired in 1985, just 2 years before Bjarnson 1987 was published
2. Naproxen was patented in 1967 and first approved for use in the US in 1976
3. Diclofenac was patented in 1978, but only came into use in the US in 1988
4. Aspirin, in fairness, had been around since the late 19^th century, but paracetamol was replacing aspirin by the 1960s, as it caused fewer gastrointestinal problems
5. While NSAIDs have been used to train pain, fever, and inflammation for ages, other uses have been found for them over time, such as for cystic fibrosis in 1995
6. Some NSAIDs, for example rofecoxib and valdecoxib, have also been withdrawn.
7. I suspect that the global population becoming at least slightly richer may also change how many people have access to NSAIDs (I haven’t checked this, but it seems reasonable)
8. Equally, manufacturing processes will have advanced, and drugs, particularly those off patent, may well be much cheaper (proportionately) to make (again, I haven’t checked this, but it also seems reasonable)

My point is that, whatever the original claim, it is now wildly and hopelessly out of date.

The world has changed.

It doesn’t matter now if the claim was correct at the time or not (beyond satisfying my personal desire to know something).

It could easily be an order of magnitude out by now.

The original claim can’t even be used as a reference point: so far as I can tell, it is a figure plucked out of the air.

No reference, no legitimacy, no claim.

But, honestly, that’s not really the story I’m trying to convey here.

We’re 4,000 words in, what’s the damn story?

The story here is that, time and again, science has failed (in this one specific instance).

Science, at its best, is transparent, honest, and robust.

But this has been a far cry from that.

1. A claim that is time-dependent has been changed and repeated since 1987. The original claim no longer has any bearing on reality, but is still being claimed is if it were so. It’s like Weekend at Bernie’s (which came out 2 years after the original claim): the claim is dead, but people have slapped sunglasses on it and are waving its arms around.
2. Referencing in a couple of articles was simply wrong: the articles referenced made no claims about the use of NSAIDs. This is obviously bad, because it seems like they’re justifying a claim, but you’d need to waste time checking to find out, actually, there’s no justification at all.
3. The earliest claim wasn’t referenced. As far as I know, it’s a wild guess. I mean, c’mon.
4. Most articles did not reference the earliest claim. References should always be to evidence that supports the claim (although since the claim wasn’t referenced, I guess this point is a little moot). Otherwise, you end up with the current situation: articles endlessly referencing articles that are only a few years out of date. It lends credibility to time-dependent claims, especially if you don’t check the referenced article.
5. I ran into insurmountable barriers when trying to find the original claim. I may not have found the original claim. Many articles don’t exist online. Many articles don’t exist on journal websites but do exist on the Internet Archive, and still can’t be accessed, because journal publishers are, as stated, greedy parasites that contribute nothing while gatekeeping research that can, all being well, improve lives. And I am incredibly privileged: I am a researcher at a university that has the funds to subscribe to a bunch of journals. If I were reliant on open access articles, as is the case for almost all of humanity, I wouldn’t have been able to get as far as I did. Sure, I might have had $50 to spend on each of the dozens of articles I’ve now been through, hoping to find justification for a claim, but honestly, who has the money for that? And why should people have to do that?
6. I also have the skillset required to do this detective work. In many respects, doing this draws on the skills I’ve been using for over a decade. How would it be possible for a member of the public to know when and how to fact check claims that, in all honesty, shouldn’t need fact checking? What is the point of journals if not to guarantee the quality of the work they publish? If they aren’t the guarantors, why do we pay them so much?! These aren’t easy skills one can just pick up, and the financial and technical issues above magnify the problem. It’s not hard to see why websites are a much easier place from which to source information…
7. As with everything else in 2025, this series of events will be made infinitely worse by artificial intelligence (AI). AI, as it stands, has no capacity to understand, being a glorified autocomplete engine. And yet AI is being used to write papers, and then used by AI to regurgitate that same information (see: Gemini, ChatGPT, Llama, etc.). With no critical thought, it will become substantially easier for scientific claims to originate with no evidence. That is, AI will create some bullshit, which will become accepted as truth because other AIs will regurgitate it. I can’t overstate how terrible it would be to have to go through every reference of an article to check whether it’s human bullshit, AI bullshit, or legitimate justification of the claim being made. Instead of standing on the shoulders of giants, we’d be standing on bullshit.

To go back to my statement that science, at its best, is transparent, honest, and robust.

None of this process has been transparent: I’m still not sure I have the original claim, and getting there has been immensely difficult and frustrating, even for someone with the time (I haven’t mentioned it before, but this took soooo long), access to articles, and skills that I have. Journals have made this process substantially less transparent by paywalling articles, never putting articles online, and blocking Web Archive from letting people see existing articles.

Articles referencing articles that don’t provide the justification of the claim is not honest. It’s not saying: “this is my claim, and here’s the evidence to back it up”. It’s saying: “this is my claim, and these other people claimed it too”. That’s not good enough.

Equally, the original claim didn’t even have a reference. Honesty would dictate that, if they didn’t have a reference for the claim, prefaced it with “we believe…” or “we pulled this number out of a hat and ran with it…”: whichever was more honest.

This whole experience has shown how fragile the original claim was (I didn’t want to write “non-robust“). It initially seemed robust, given it was repeated endlessly in articles and websites, but was in fact originally the unreferenced claim of people writing an article that was published 38 years ago. For something that changes over time, that’s fragile.

If you asked a scientist: “how many people take NSAIDs every day, worldwide”, and they answered: “well, 38 years ago, someone said it was 30 million”, I suspect your response wouldn’t be to say: “great, thanks, that’s totally answered my question in a way I find absolutely satisfying”.

Ok, great, you’ve pointed out a problem. So, what’s the answer? How many people take NSAIDs worldwide in 2025?

Fucked if I know.

You could always try asking an AI.

They’re very confident.

…That’s it? Kinda unhelpful, don’t you think?

Yep.

It’s actually really hard to estimate something like that. It’s why I was interested in the first place.

Take the UK.

We have prescription records for both primary and secondary healthcare in the NHS, but they’re patchy because our IT infrastructure is bonkers. If you ask really nicely, private healthcare might give you their prescription records too. You might be able to get all prescription data for a given year this way.

But a prescription doesn’t tell you whether a patient is actually taking the drug. My dad was prescribed naproxen. After reading the list of side effects, he didn’t take it. I can’t imagine he’s alone. People stop being in pain. They’re worried about side effects. They never collect the prescription. They die. Lots of things happen.

And it’s not always clear whether a person is being prescribed half a tablet a day, or 2 tablets 3 times a day. I suspect this would matter when trying to work out how many people take NSAIDs each day, because the regularity of their prescriptions would matter. It might not, you might be able to work around it, but it might.

And that’s just prescriptions.

NSAIDs can be bought over the counter. It’s possible most NSAIDs that are taken, are bought over the counter. I don’t know.

The only ways I can think of to measure how many over the counter NSAIDs people are taking is to ask them, or have them keep a diary. Maybe tag them all with a radioactive tracer? The problems are that, a) these methods are imperfect, and b) the type of people who would answer these things honestly and correctly are not generalisable to the entire population. Random samples of populations for surveys exist, but response rates are typically low and non-random: you can’t force people to respond, and the people who don’t respond, may well use NSAIDs entirely differently to people who do.

And that’s just the UK!

Every country implements healthcare differently.

Records are ok in the UK, but I would doubt rural China for example, has particularly good prescription records. It might, of course! I don’t know, I haven’t tried to get prescription data from there. But I think I probably don’t need a reference to state that some parts of the world don’t have great medical records, or accessing them is very difficult.

Plus, you’d need surveys everywhere. Every country. Every part of every country. In every language. For all age groups. To have any chance of getting a reasonable answer for over the counter use of NSAIDs.

You could ask manufacturers how many tablets they make, and make some back-of-the-envelope calculations for how many people that would equate to if all doses were taken, but it’d likely be wildly wrong. How many NSAIDs in each country are exported? Imported? How many get dumped when they expire? How many are warehoused indefinitely? And you run into the problem of people not taking them even if they get them.

The sheer scale of the task of answering this question well is incomprehensible.

And it probably doesn’t even matter!

We can’t do anything with it!

Individual countries may want to know for various healthcare planning purposes, but the worldwide number isn’t relevant. Drug companies presumably only want to know whether they can continue to make a profit.

It’s only people like me, with an axe to grind on the altar of pedantry, who care about whether this number is right.

For most people, “millions of people take NSAIDs daily” is enough.

It’s almost certainly true, and we don’t need to be more specific than that.

So what’s the issue?

If you are going to make a claim in a scientific journal article, it needs to be justified.

If people are allowed to make any claim they like and justify it based on “it sounds right” or “I want this to be true”, then the scientific record will be corrupted.

Scientific progress requires evidenced claims, or you have to start from scratch all the damn time.

As with AI creating bullshit, if humans can create bullshit and pass it off as evidence, we will no longer be standing on the shoulders of giants.

We’ll be standing in bullshit.