Wikipedia talk:Manual of Style/Medicine-related articles/Archive 13

Archive 10

Archive 11

Archive 12

Archive 13

Archive 14

Archive 15

→

Archive 17

NOTE: Continued from Wikipedia talk:Manual of Style/Medicine-related articles/Archive 11 and Wikipedia talk:Manual of Style/Medicine-related articles/Archive 12

Discussion at the Administrative Noticeboard

I have notified several individual editors but also noting here that I have posted at the Administrative Noticeboard. Barkeep49 (talk) 03:45, 7 January 2020 (UTC)

Link to archive: Wikipedia:Administrators' noticeboard/Archive317#An update on and a request for involvement at the Medicine MOS SandyGeorgia (Talk) 14:21, 30 January 2020 (UTC)

User:Levivich/Drug prices RfC draft 3

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

---

User:Levivich wondering your thoughts on adding some more expensive ones aswell:

• Sofosbuvir "As of 2016 a 12-week course of treatment costs about US$84,000 in the United States, US$53,000 in the United Kingdom, US$45,000 in Canada, and about US$500 in India."
• Onasemnogene abeparvovec "It carries a list price of US$2.125 million per treatment, making it the most expensive medication in the world as of 2019."
• Delamanid "As of 2016 the Stop TB Partnership had an agreement to get the medication for US$1,700 per six month for use in more than 100 countries."

Doc James (talk · contribs · email) 20:55, 13 January 2020 (UTC)

@Doc James: Adding these to which RfC draft/question? (Or all of them?) – Levivich 02:16, 14 January 2020 (UTC)

@Levivich: my though is to have one section of examples rather than three sections. Doc James (talk · contribs · email) 11:05, 14 January 2020 (UTC)

@Doc James: Oh, I meant those (3A, 3B, 3C) as alternative RfCs, not as three questions in one RfC–i.e., we'd only have ran one of those three, with one set of examples. Originally it was just 3A, but I added 3B and 3C in response to WAID's concerns above, and then 3D as an entirely different format. To answer your question, about adding these three examples to the others, I guess it would depend on what the question was. As I understood it, the issue in dispute, at least the one to be handled by an RfC, was just about how to use IMPPG as a source. There are a lot of other issues (and other sources) we could discuss about prices and pricing in general, and those three seem like good examples for that conversation (for some of the reasons pointed out below). – Levivich 19:08, 14 January 2020 (UTC)

Ah okay User:Levivich. I am happy with either of the 4 suggestions. 3D gets to the specifics which may be useful. There are many potential ways to present pricing information and I personally am not particularly attached to any single one. 3C does the best with respect to a high level question IMO. Doc James (talk · contribs · email) 11:09, 15 January 2020 (UTC)

References

I'd like to have that conversation, but it doesn't fit any of the questions that Levivich has written. This is because none of those prices are cited to the IMPPG or any other drug database, and all of the questions are about IMPPG or drug databases.

I think that the conversation about prices that are not sourced to drug databases needs to be held separately. WhatamIdoing (talk) 05:42, 14 January 2020 (UTC)

Indeed, in fact an RFC on this doesn't need to happen at all, based on current evidence I've seen. Those are exactly the sort of drug-cost comments that are supported by policy and we generally get right. They are mentioned because they are exceptional and notable. Folk can argue about the merits per weight, reliability of source, etc, and no sign that is a problem with editors. The only concern with some of them is that some price changes may be dated and folk need to keep an eye on the article to ensure we are no longer saying the price is $XXX,XXX when it is now just $X,XXX. -- Colin°^Talk 09:11, 14 January 2020 (UTC)

Agree; these appear to be examples that are compliant with WP:NOTPRICE, so are not in question. We need a database example that meets WP:V. SandyGeorgia (Talk) 09:14, 14 January 2020 (UTC) Strike, update; fail WP:V per Colin. SandyGeorgia (Talk) 09:41, 14 January 2020 (UTC)

I agree as well. These are not examples of what is generally being disputed. They might be used as examples that are far better than those under dispute. If in later discussions we consider what is proper in article ledes, then we could bring these up again. --Ronz (talk) 18:19, 14 January 2020 (UTC)

Ronz, SandyGeorgia, Colin, WhatamIdoing, focus on the locus of the dispute is great. Apologies for asking this but how are the examples in 3A - 3C not being sourced to a drug database? Best, Barkeep49 (talk) 15:57, 15 January 2020 (UTC)

With discussion in multiple sections now, and after six weeks at this (while trying to move on to regular editing), I may be losing the plot. I am looking at and responding to the samples just above, in this section, that are not to a database, rather are sourced in ways that appear to be in accordance with NOTPRICES. My concern with the (good) work done by Levivich continues to be that we are not giving enough information to draw in the reader, encourage them to participate, or understand what question is being asked, all of which WAID's draft does well. SandyGeorgia (Talk) 16:16, 15 January 2020 (UTC)

Barkeep49, 3A, 3B, and 3C in Levivich's draft are about sourcing content to drug price databases. Doc James' three suggested additions at the top of this section, however, are sourced to a journal article, a British newspaper, and a non-profit organization's website. It would not be logical to ask "Can we use drug price databases?" and then give an example of something that is sourced to The Guardian.

IMO the locus of the immediate dispute is not "Can we include any prices at all?" or even "Can we use drug price databases at all?" The locus of the dispute is much closer to "Can editors pick any record they want in a drug price database, and generalize the contents to a statement about the entire developing world?" WhatamIdoing (talk) 18:03, 15 January 2020 (UTC)

If that's the locus for which the first RfC is intended to gather more viewpoints, then perhaps the main question in your proposed draft could focus more attention on this aspect of the issue? isaacl (talk) 18:22, 15 January 2020 (UTC)

isaacl, have a look at #Questions not Opinions. There are lots of questions people might ask themselves when looking at the source, the text and the half-a-dozen polices and guidelines linked at question at Wikipedia:Manual of Style/Medicine-related articles/RFC on pharmaceutical drug prices. The one WAID mentions is closed to #3 in my list and is certainly a very important one about whether the source supports the "in the developing world" claim in the text. However, that is also a question unique to one database and one world area, and doesn't help when we consider US, UK or other regions. Another important question is the degree to which original research is permitted to calculate a daily dose or treatment cost, especially when the source (nor other similar databases) does not say what strength of tablets to pick, how many to take a day and how long to take them for. Or whether picking one pill or one vial or one tube of cream, out of multiple options, is acceptable and representative. And so on. There are lots of questions and I think the beauty of WAIDs RFC is it allows editors to think about however many they want. They can concentrate on WP:V or WP:OR or WP:WEIGHT or WP:NOT or WP:LEAD. As soon as we ask a specific question and just that question, then it isn't so powerful, and I don't think a "should" question is as helpful. Compare to 3D "Should articles state any pricing information cited to the International Medical Products Price Guide". That invites a yes/no response, or some vague "Yes, sometimes or "No, only sometimes", and isn't actually being contested by anybody in the dispute. It leads away from looking for policy answers and rather more towards some advocacy answer and wishful optimism. -- Colin°^Talk 21:00, 15 January 2020 (UTC)

I disagree that concentrating on a specific point of contention makes the discussion less powerful. Generally a better discussion ensues from focusing on one thing at a time, with more in-depth follow-up occurring. It's not like there's only one opportunity to gather information, so it isn't necessary to take comments on everything at once. I appreciate, though, that it's been difficult to try to reach a consensus on a specific focus. isaacl (talk) 23:14, 15 January 2020 (UTC)

isaacl, yes there are positives to being focused. But there are degrees of being focused. At the outer extreme, there's the "Medication prices can be included when appropriate sources are avaliable" question that James offered a while back, which considers only the big "should we do this" but not "can we do this while following policy as Wikipedians" and leaves out the question of what sources are appropriate and when and what kind of statements we can make from them. At the other end of specific we could get a question about whether and when records in a database can be considered representative of the developing world. And that's a good question to discuss but only one of many and very specific to one database. If we agree, for example, that the median supplier price of a record with three or more suppliers is representative of the developing world, that doesn't permit all the other crimes against policy and guideline with that database, or other databases. It is perhaps too small a step, though one that needs to be taken.

We had a discussion before about whether to resolve all this in one go, or to have followup RFC(s). Tryptofish was very keen to do it in one go. WAID's RFC explicitly says this will not necessarily be resolved in one go. I think it is too hard to resolve in one go, because there are more than enough possible questions to discuss about this one source, without also considering problems with NADAC, BNF, Drugs.com, etc. And it would be tedious to have an each RFC for all these. I wonder, though, if there is a way to structure discussion in the RFC so we have sections that look at different aspects. Perhaps sections for each policy? What do you think? -- Colin°^Talk 09:24, 16 January 2020 (UTC)

I appreciate that different editors want to look at different questions, and so that's how the current draft RfC has ended up how it is. I have similiar concerns as Tryptofish regarding the quality of feedback that may be received (though I don't share the worry about having just one RfC) and it's unfortunate the editor is no longer participating. I understand, though, that others are not as worried about the quality of feedback. If the key participants can agree upon the critical issues to discuss and they can be put into separate sections, great! isaacl (talk) 17:14, 16 January 2020 (UTC)

I've thought about pre-structuring response sections, by way of sorting it. I'll set up a section and ping you in the edit summary (so you'll know when it's done), so you can see what one approach would look like. (Feel free to revert it after you've read it, if you don't think it will work well.) WhatamIdoing (talk) 20:14, 16 January 2020 (UTC)

@WhatamIdoing:, I am concerned that we avoid anything looking like Wikipedia:Requests for comment/Genetically modified organisms. When that RFC came up, I took one look, and ran the other direction. I never participated. With too many options, people won't engage. (Obviously, I went too far the other direction with the singular "is it in sync" question at the MEDLEAD RFC.) The lengthy GMO RFC is what leads me to believe that the approach you took is best-- in between the two extremes, and written in a way that it draws the reader in and encourages participation rather than voting. SandyGeorgia (Talk) 20:26, 16 January 2020 (UTC)

In that vein, I think we are moving backwards here. The reader is now presented with too many options, and a lengthy TOC before they are drawn into your effective writing. SandyGeorgia (Talk) 20:32, 16 January 2020 (UTC)

I prefer the revised section headings by Ronz. The previous ones seemed to imply that commenters should comment only in one section. I would omit the sentence starting with "You can put all of your ideas in one section..." I think it makes more sense to have people separate their comments on different areas into different sections.

I'm not exactly sure what SandyGeorgia is referring to by "..presented with too many options, and lengthy TOC". However as I discussed before, if it were solely up to me, I'd move the "In the real world" section up and expand it a little to lay out the specific key issues that Colin and WhatamIdoing described. I'd probably trim the discussion around each example. For the section headings for the comments, I would select headings that target the key issues, like geographic scope and timeliness of the data, and let commenters discuss how the different Wikipedia principles should be weighed against each other in context of an issue. For instance, how should the encyclopedic value of having some information be balanced against the age of the data, and how many geographic regions does it cover?

Nonetheless, I appreciate that my opinion is just one person's, and a lot of work has gone into building a consensus for the current draft. If consensus holds for the present format (or a slightly tweaked version), that's great! I don't feel the RfC should be held up if most people are happy with the current one. isaacl (talk) 01:16, 17 January 2020 (UTC)

The 23 separate proposals at the GMO RFC look like an anti-pattern.

User:Isaacl, the reason that I took the "mostly about" and "comment anywhere" approach is that I can foresee someone trying to be "helpful" by re-factoring other people's comments (e.g., splitting off a sentence or paragraph that is on a different subject). That would be a bad outcome. WhatamIdoing (talk) 07:22, 17 January 2020 (UTC)

I too like the change Ronz made. Brevity is good. In that vein, I'd drop the whole "We're going to try to.. in separate sections" bit. People will see the sections, and the "other" bit, and work it out for themselves and and nobody reads the instructions anyway. I'd drop the leading "The thing I really like" and "The problem I see" and just leave space for comments. It would be good to wikilink policy pages for convenience (and specifically WP:NOTPRICES which may even merit being quote boxed). I think section headings should drop the "Comments about" prefix, as the whole thing is in a discussion section. I agree that having lots of options is really bad per the GMO one and because options force people to pick one and vote. These section headings aren't options and I think the discussion will need some structure lest it just be a sprawling mess. For example, the question of when/whether the source becomes representative of developing world is "statistics" and less about any wiki policy, but the question about pricing "per dose" when the source doesn't mention which strength and how many tablets to take each morning and night, is WP:V mixed with WP:OR.

WAID, one concern about these prices is whether the way they are stated can have no meaning for the reader. A price "per treatment" has a clear meaning, and a price "per month" (for a drug you take long-term, which again, our source doesn't say) has a meaning. But a price "per dose" or "per xx tablet" or "per vial" ... I can't really see how the reader can extract meaning from that unless we state that you take two 50mg tablets three times a day for 7 days, or that the vial is 10ml and a shot is 7ml, etc. And our MEDMOS policy has always been to avoid giving dose information. Perhaps that's MoS but it is also "is this encyclopaedic" -- a fact that one can't use meaningfully. Not sure where that goes. -- Colin°^Talk 09:06, 17 January 2020 (UTC)

This is a mess. If I am understanding correctly, we are now discussing WAID's proposed RFC in Levivich's draft section, so who knows where we stand and who's on first. I could have it completely wrong; if we are in fact still discussing Levivich's drafts, could someone clue me in? It is time to archive off everything here, and start over, ala cot-cob per Barkeep49. THIS is the version I support; no convoluted TOC, no overwhelming the reader, simple but not over-simplified. SandyGeorgia (Talk) 16:33, 17 January 2020 (UTC)

Note the table of contents can be suppressed, if desired, in any version. isaacl (talk) 16:34, 17 January 2020 (UTC)

Agree that it is confused mess. I tried to restart discussion of the WAID draft at #Polishing the draft. Perhaps we can continue this there. -- Colin°^Talk 16:56, 17 January 2020 (UTC)

Actually, James the first example, which was added to the article by you, has a problem. The source says "The cost of a 12-week course of sofosbuvir is as high as US$84,000 in the US". Our article might also say "as high as" rather than "about", because they mean different things. The date of the US price is "accessed Dec 2015" so that price isn't "as of 2016". "As-of" date inflation is the norm for drug prices, and I don't know why. The paper says "Branded sofosbuvir is available at a suggested price of US$900/12-week course in 101 named low-income countries" but the $900 figure isn't in the article sentence. The prices for UK, Canada and India claim to be sourced to this article, but are not. So of the first four drug prices you suggest could be included in an RFC, three of them are simply unsourced, and a chance to include the price for 101 low-income countries is not taken. --- Colin°^Talk 09:37, 14 January 2020 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Defined daily dose (DDD)

There has been a fair bit of discussion on whether or not DDD is appropriate for rough estimates of medication prices. We have a number of sources which support this use. Specifically the government of Canada says the DDD can be used to provide "a rough idea of the daily cost of utilizing a drug in a specific formulation" and "provides a rough idea of the cost differential between the two formulations of the same drug". We are using it for the first purpose. Doc James (talk · contribs · email) 11:19, 6 January 2020 (UTC)

James, I think our priority right now should be to launch an RFC. There's a whole section at WHO on DDD and specifically on its use and abuse. It very much argues against the use we are doing, and why their sole focus on DDD is for utilisation. And yes at a crude population level, it could give a rough idea of the cost of utilising that drug in a specific formulation. That's up to researchers to judge if it meets their needs, not Wikipedians. What it doesn't represent, is the dose that an individual patient might take and thus cost them per day or month. Repeatedly MSH and WHO have to remind us it is not a therapeutic dose nor does it represent average prescribed doses (which themselves vary from country to country and year to year) . Wrt your second example, yes WHO give that as an example of how it could be used to compare the cost of e.g. 5mg tablet vs 10mg/5ml syrup. But, if you think about it for a moment, all you are doing with that is agreeing on an arbitrary dose to compare two formulations: the actual dose need not be representative of any therapeutic value. It's just a number.

Furthermore, there are no sources saying what indication the DDD was calculated for or what choice they made if their own source lacked a clear "maintenance dose" value (they sometimes pick initial and sometimes max). When we have a drug for multiple conditions (epilepsy, neuropathic pain, mental health disorders), the DDD is totally useless because the cost to treat really depends on what indication you are treating, and you don't know that. DDD is a red herring. Our only source that offers it (MSH) warns about its careful use and does not themselves use it to give a cost-per-day, which should be a clear warning sign that this is inappropriate original research. The other sources (Drugs.com, BNF, NADAC) do not give a DDD or indeed any one dose, so any attempt to use them to give a cost per day/month/treatment is both original research and synthesis of sources. -- Colin°^Talk 11:50, 6 January 2020 (UTC)

Yes the DDD only provides a "rough idea"[1] of the daily cost. Yes it does not provide "detailed" description of the cost. That is why we use the term "about". Doc James (talk · contribs · email) 12:14, 6 January 2020 (UTC)

This source converts the MSH data into price per year with "Carvedilol is listed on the Management Sciences for Health International Drug Price Indicator Guide with average price per tablet of $0.20 or $144 per year for twice daily treatment"[2] Doc James (talk · contribs · email) 12:21, 6 January 2020 (UTC)

James, it is so "rough" because it is for population studies. The kind where someone says we might spend £1.5 million on a drug, but it probably doesn't change the point if the actual cost is £1 million or £2 million. Wikipedia is giving a price in dollars and cents for one patient, and then waving the word "approximately" about as if that absolves all sins. As for the paper from 2011, it is wonderful what Google can turn up. I tried to find the record they quote. Carvedilol in 2011 gives three doses. The 6.25mg tablet has a median (only) supplier price of 0.0414. At the DDD of 37.5, is six a day x 365 = $90. The 12.5mg tablet has no suppliers and a median buyer price of 0.1669. That's three a day x 365 = $182. The 25mg tablet has no suppliers and a median buyer price of 0.2041 which is pretty close to the "average price per tablet of $0.20" they mention. But wait, 25mg doesn't go into 37.5mg, and they mention "twice daily treatment" so I guess they mean 50mg per day. That price is twice a day x 365 which is $149. Not quite the $144 they give but close.

What does this tell us? The paper mentions "per tablet... twice daily" and yet there is no tablet dose that is 18.75mg. This is because DDD is not actually a therapeutic dose that any patient might take. Their maths mostly work at 50mg. Looking at Drugs.com we see the 6.25, 12.5 and 25mg tablets "twice a day" being the initial and then tritrated "if tolerated" up to the maximum, but lots of other dose options too. That really isn't the paper to convince anyone that DDD is a useful measure, when they don't use the DDD of 37.5mg but instead 50mg. Further, we see that if we did try to use the DDD, we get yearly prices of $90, $149 and $182 depending on which tablet size we chose. All three tablet sizes make sense for individual patients and indications and stages of treatment. This is why we don't allow original research: the numbers are effectively random. -- Colin°^Talk 13:25, 6 January 2020 (UTC)

The argument here is similar to saying, "doses of medications are random". Please note they are not. There are well accepted dosage ranges. Do we need a RfC to ask "are dosages of medications random" as that is the argument you are making? Doc James (talk · contribs · email) 13:32, 6 January 2020 (UTC)

No, I'm saying that original research produces random numbers. The source gives three pill sizes and none of them are 37.5mg or 18.75mg. You say we should use DDD and then you cite a paper that looked at the DDD of 37.5mg and went ??? that's not gonna work and picked 50mg instead. Decisions decisions decisions and each time a different result. If the researchers had searched in 2010 rather than 2011, they'd only have the 6.25mg tablet, and nobody is taking six a day unless they have no other choice. -- Colin°^Talk 13:59, 6 January 2020 (UTC)

On the big picture, we have (mis)spent a lot of bandwidth on this talk page trying to sort out the various problems (SYNTH, NOR, WEIGHT) in price text in our drug articles. We went down that path when we (I?) were (was?) seeking one good example of drug pricing information from these databases that did not have SYNTH problems, and did not find one. We can save a good deal of time by having you actively engaged in those discussion, James, and I'm not sure forbidding pings will help us move forward. Do you read all posts here, or do we have to ping you to each post? I ask because I really hate this pingie-thingie myself; when I come to a page I participate in and follow, I pull up a diff of everything since I last read, and the extra pings are just an irritation. Do we need to ping you, and if we are trying to sort out the price data in our drug articles, how can we assure you see the discussions if Colin is forbidden from pinging you? SandyGeorgia (Talk) 12:25, 6 January 2020 (UTC)

I guess the question is do we have a fundamental disagreement? Do you believe it is possible to provide a rough estimate of the cost of a medication in LMIC or other region of the world? This can be broken down into two parts:

1) Do you believe it is possible to provide a typical dosage range for a medication in adults used for a specific purpose?

2) Do you believe it is possible to provide a rough price for an amount of medication in LMIC?

The popular press manages to determine the cost for a course of treatment "Sovaldi treatment cost $1,000 a pill, or $84,000 over 12 weeks." The Guardian managed to determine the typical dose, the typical duration of treatment, and the rough cost per dose than do the math. [3]. What you call SYNTH and NOR is simple WP:CALC. WEIGHT is clear per sources such as Doctors Without Borders.[4]

The next question than becomes one of knowledge parity. The popular press generally just writes for wealthy people in the developed world. Do those in LMIC deserve to have pricing information for medications they may care about? My position is yes, and we have excellent sources such as MSH that provides these details. Doc James (talk · contribs · email) 12:55, 6 January 2020 (UTC)

I am thinking these could be two useful questions for the RfC. User:WhatamIdoing User:Tryptofish wondering your thoughts? If either of these are false than we would all agree that it is impossible to list a price for a medication for an area.

Other questions could be is multiplying a dosage range for a purpose by the rough cost estimate by a time period WP:CALC or WP:SYNTH and are prices of medications WP:DUE.Doc James (talk · contribs · email) 13:17, 6 January 2020 (UTC)

James, these aren't yes/no questions that apply in all cases. We can find drugs with one standard dose and we can find drugs with many suppliers in the MSH. But in the general case, no you can't. There are multiple indications, multiple dose ranges, multiple patient ages, weights and liver function and existing medications.

What matters isn't so much right now these abstract questions, which are complicated to answer, but actual sources and actual texts in actual articles, which despite months of discussion have not changed. If you believe these things are possible, and possible generally for most drugs (over 500 have these prices), then argue the case when the RFC is posted. There is a reason why WHO/HAI have only 14 core global medicines and survey at most 50 (at specific strengths and formulations for specific indications and patient age and specific treatment duration or daily dose). The MSH database is way too sparse to be a reliable source for 500+ medicines. It is officially not a reliable source for international reference price if there are not many suppliers. That's WHO/HAI official policy. Which trumps any Wikipedian opinion. -- Colin°^Talk 13:34, 6 January 2020 (UTC)

James, it is helpful to have you fully engaged in understanding that these are complex questions even for those of us who understand the sources and understand SYNTH. The problems we have now in more than 500 articles go way beyond what an average reader, or even RFC respondent, can understand. Diverting energy now towards addressing those problems is a distraction from the RFC. But, we must continue to make sure you are fully engaged with us in sorting out these problems. We have this split now between the sample price discussion above of chlorthalidone, and this section; let's continue here. Do you now see that what we have presented to our readers for chlorthalidone, in the lead of an article, is neither useful nor accurate nor in accordance with NOR ? Sorting out that we have a big problem in 500+ articles is a very different matter than deciding via RFC whether drug prices from databases should even be in our articles at all. SandyGeorgia (Talk) 13:55, 6 January 2020 (UTC)

After our discussion what I see is that what we have in our article on chlorthalidone is a perfectly reasonable estimate of the price per month of the medication in question. Looking at the references yes 12.5 mg can be used, 25 mg can be used, and 50 mg can be used. Could the price range from 6.75 to 27 USD? Sure. I prescribe medications that range in price from pennies to 10,000s per dose (a million fold difference). A 2 or 4 fold difference fits well within the range of "about".

Additionally I believe our readers are smart enough to realize this. Doc James (talk · contribs · email) 14:14, 6 January 2020 (UTC)

I chose chlorthalidone as a starting example because it is the one of the simplest I have seen in these discussions; the text we have presented on other drugs has far bigger problems. It is your opinion that our readers can sort it out. It is my opinion (and others) that what we have here is a problem not only of LEAD, NOTPRICE and WEIGHT, but a problem of SYNTH. Do you see why we need a separate RFC on that matter before we move on to the wider RFC? How can we ask Wikipedia editors whether price information should be included, when we do not even have price information that we all agree conforms with other policy ? SandyGeorgia (Talk) 14:31, 6 January 2020 (UTC)

James, mathematically, you cannot say "the wholesale cost is about US$13.50 a month" if you feel the price is only accurate to within 2x or 4x approximation and you don't care if 12.5mg, 25mg or 50mg tablets are used. Our readers trust what we write and really no amount of "about" or "approximately" covers us if we give a price to four significant figures. It is one thing if our sources use that language, but it isn't a sticking plaster for dodgy maths. They may look at treatment X and treatment Y and conclude that treatment Y is 2x or 4x more expensive than X whereas in fact the difference in price is down to the random chance of original research. Btw, at Carvedilol we give a "wholesale cost per dose". What does "per dose" mean? The dictionary tells me it is how much you take at one moment of time. So a DDD of 50mg but taken twice a day would produce a 25mg dose, but in a once a day sustained-release tablet would produce a 50mg dose. I don't think there's any consensus that we should provide prices in dollars and pence and then excuse our original research random results by saying the prices are only meant to be accurate to two orders of magnitude and our readers are bright enough to know that. But you can try that claim at the RFC. -- Colin°^Talk 14:35, 6 January 2020 (UTC)

On all of this about the "rough estimate": Editors who are responding the RFC are welcome to express opinions about whether the example sentences would be clearer if they included words like "rough estimate". Editors might advise us, for example, that the current examples aren't great, but that if we added words like 'Using the defined daily dose to create a rough estimate of costs' to the start of the sentence, they'd be satisfied. Or they might tell us to use fewer significant figures, so that "about $13.50" becomes "on the order of $10". Or they might say not to combine the DDD with the MSH's price per pill with the Canadian source about using the DDD this way, and tell us to just use the price per pill. They might even tell us to omit the dollars-and-cents and instead search for a source that would let us write "generally considered inexpensive". They might tell us all sorts of things. But again, all this is "answering the question", and the goal on this page today is only to write the question. Answer the question next week, please, and on the other page. WhatamIdoing (talk) 20:31, 6 January 2020 (UTC)

Generally there is one main indication and regardless typically most if not all indications generally use similar doses. Most medication have a fairly narrow dosage range. Some of the sellers offer the medication in more than LMIC 100 countries. Qualifiers applies to all evidence within medicine and most of the time we have no idea how the qualifies affect claims of benefit as they have just not been studied. But these still not prevent use from providing an overview. Yah sure MSH is not as good when there are fewer supplies. Same as a meta analysis is not as accurate when their are fewer RCTs. Doc James (talk · contribs · email) 13:51, 6 January 2020 (UTC)

When our math is doubly wrong (12.5 vs 25 mg for chlorthalidone), we cannot say we are within a "fairly narrow dose range", and that is only one very simple example. SandyGeorgia (Talk) 13:57, 6 January 2020 (UTC)

Are you saying 12.5 to 25 mg is a wide dosage range?

Now back to the question at hand. Do you believe that there is an accepted dose range for a specific indication in a typical adult? We can go with this if you do not want to use DDD. Doc James (talk · contribs · email) 14:09, 6 January 2020 (UTC)

I am saying we don't have sources to back up the information we are giving our readers without using SYNTH, and our personal opinions, what we have experienced or seen as physicians or people interpreting for physicians, should not be coming in to play at all. We do not have sources that back up the information we have presented unless we do synthesis. And we do not have sources that, according to DUE WEIGHT, tell us what to present in the example of chlorthalidone. The problems in other articles are worse. This is probably why we should be respecting WP:NOTPRICE and not presenting these prices at all based on database sources, but that is my opinion.

What would be helpful to see is whether you have any example, for any drug, that discusses drug cost in a way that does not breach, IMO, WP:SYNTH. Then we could more accurately discuss WP:WEIGHT and WP:LEAD issues. SandyGeorgia (Talk) 14:22, 6 January 2020 (UTC)

What I am saying is we do have sources that provided accepted ranges for medications plus we have sources for DDD. It is thus a simple WP:CALC to convert price per dose to price per day for a typical dose or dose range which is more useful than just the price per amount by itself. Doc James (talk · contribs · email) 14:27, 6 January 2020 (UTC)

So do you see that we have a policy disagreement as to whether this is simple math without synthesis that needs to be put forward in an RFC, because we got ZERO feedback on the matter when we posted a question to the NOR noticeboard? And that we need to sort that out before we can solve the bigger question? We have not, throughout these discussions, seen one straightforward example of drug price information from the sources used in over 500 articles that we can all agree does not involve SYNTH. SandyGeorgia (Talk) 14:35, 6 January 2020 (UTC)

We have "As of 2016 a 12-week course of treatment costs about US$84,000 in the United States, US$53,000 in the United Kingdom, US$45,000 in Canada, and about US$500 in India.Hill A, Simmons B, Gotham D, Fortunak J (January 2016). "Rapid reductions in prices for generic sofosbuvir and daclatasvir to treat hepatitis C". Journal of Virus Eradication. 2 (1): 28–31. PMC 4946692. PMID 27482432." Not sure if you consider this SYNTH aswell.

But yes lets put the SYNTH versus CALC discussion to a RfC. Doc James (talk · contribs · email) 14:53, 6 January 2020 (UTC)

it isn't a simple calculation to convert. Firstly the sources don't give a price per "dose" either. They give a price for one tablet size or another. Even if we thought DDD was acceptable (which WHO, who invented it, don't) it doesn't necessarily divide into specific tablet sizes (see above for 37.5 DDD) so no there is not a straightforward calculation. The drug may be once a day or twice a day or some other option, and this may depend on what kind of tablet you take. These are all things a doctor will decide at prescription time. The most you can do, James, is convert a price per day to a price per month, assuming it is obvious the indication requires long-term treatment, of course. -- Colin°^Talk 14:39, 6 January 2020 (UTC)

Yes we known that you think that WHO does not think that it is acceptable. But it does not appear that WHO thinks that it is unacceptable. Doc James (talk · contribs · email) 14:53, 6 January 2020 (UTC)

I think Nil Einne (above, collapsed) has a point. Most of this involves one party making a claim and then another party disagreeing. We can do that in the RFC. I think this should wait till the RFC, when other voices can chip in with support or rejection and hopefully a consensus form. There's zero evidence this discussion is heading towards any consensus, that's why we are having the RFC, to get other voices. -- Colin°^Talk 14:55, 6 January 2020 (UTC)

RfC are we allowed to multiple the "typical dosage range per day of a medication in adults as used for a specific purpose" by the "price per dose" to get the cost per day for a specific purpose? Doc James (talk · contribs · email) 14:58, 6 January 2020 (UTC)

Example article text + source please, which gives either of those things, singular. -- Colin°^Talk 15:15, 6 January 2020 (UTC)

We probably don't have an example of an existing article that talks about "typical dosage range per day of a medication in adults as used for a specific purpose" because MEDMOS has said "Do not include dose or titration information except when they are extensively discussed by secondary sources, necessary for the discussion in the article, or when listing equivalent doses between different pharmaceuticals" for years and years and years, and the appetite in the community for putting easily vandalized dosage numbers into articles has basically been zero. Let's please not try to change that rule today. (I'm willing to talk about that later, and I even have some ideas about how to manage vandalism, but whether that information is encyclopedic is a discussion for another time.) WhatamIdoing (talk) 20:20, 6 January 2020 (UTC)

User:Doc James, my thoughts on your second suggested question ("Do you believe it is possible to provide a rough price for an amount of medication in LMIC?") is that the question is too general for other editors to grapple with. I can tell you my own answer (i.e., I believe it's possible to do this at the wholesale but not retail level, for some but not all drugs, and that most articles could be improved), and I believe that at this point, my answer has more factual and policy basis than the beliefs of >99% of Wikipedians, no matter what their beliefs are. But I don't think that asking about uninvolved, non-expert editors' beliefs helps those editors help us. The current RFC draft is essentially three worked examples of how the most common source has been used. We're inviting editors to look at that and tell us how much better we can make it. That's more likely to produce informed comments than merely asking editors what they believe ought to be possible hypothetically. WhatamIdoing (talk) 20:46, 6 January 2020 (UTC)

Reboot

Allright, we are spinning our wheels. (Not complaining, since this is much better than the alternative, which was silence.) Let's start over.

James, on this page we have been discussing a SYNTH problem. You believe this is simple math, others do not. You believe using other sources supports the math you are using to add text cited only to a database. Let Colin pick a typical example, and you write text that incorporates all of those other sources you are using to support your math, and shows how you have used them. Then we can talk SYNTH vs. CALC. And from there may emerge an example to be used for an RFC. SandyGeorgia (Talk) 15:07, 6 January 2020 (UTC)

I dont' want to be accused of cherry picking. James cites Carvedilol above. Try that. -- Colin°^Talk 15:15, 6 January 2020 (UTC)

Sure lets. "In the United States, the wholesale cost per dose is less than 0.05 USD as of 2018."NADAC as of 2018-12-19". Centers for Medicare and Medicaid Services. Retrieved 22 December 2018."

Per the reference all doses at that point in time were less than 0.05 USD per tablet. Doc James (talk · contribs · email) 15:47, 6 January 2020 (UTC)

Your source does not say what string you searched with. If I search with "Carvedilol" I get prices less than $0.05 per tablet, though the source does not say if one "tablet" equals one "dose", nor that I need to take that twice a day (so costing me twice as much). If I search with "Carvedilol ER" I get prices of $6.44, $6.61, $7.08 and $6.57 for each 10, 20, 40 and 80mg extended release tablet. Again the source does not say that I only take that once a day. See also Drugs.com Carvedilol Dosage and Carvedilol Prices. So 5 cents or 7 dollars? -- Colin°^Talk 16:03, 6 January 2020 (UTC)

OK, the first example missed by a factor of 100. (Imagine our readers trying to sort that ?!?!?!) James got to choose that example, now Colin gets to choose one. Next. SandyGeorgia (Talk) 16:36, 6 January 2020 (UTC)

Can I ask all involved (Sandy, Colin, James) how this is helping us move forward with the RfC? If you all agree it is helpful I don't want to stand in its way and I'm glad for James' involvement as well but this seems to be continuing the conversation that has been ongoing for as long as this dispute. If it's not helpful maybe it's best put aside for now. Barkeep49 (talk) 18:02, 6 January 2020 (UTC)

Barkeep49 My reasoning: there is resistance on this page to launching WAID's RFC, and I am working to build understanding that there is a logical reason why we need that RFC before a broader one. Realizing the possibility that James was not following all of the discussion here earlier (possibly because of disallowed pings), I tried to start over and go through an example with him. We can probably close this off now; I hope we can see that the first (WAID) RFC is not without merit, and there really is a need to consult the community on how to use these sources, since we got no response from the NOR noticeboard. SandyGeorgia (Talk) 18:06, 6 January 2020 (UTC)

I agree per comment I made above, though this has at least provided an excellent example of when I say many of the prices are incorrect, we aren't just talking rounding errors from choosing a 30-day month! "Some of our article prices are incorrect by a factor of 100". Original research -> random numbers. Anyway, back to the RFC. -- Colin°^Talk 18:20, 6 January 2020 (UTC)

I think it's been helpful in clarifying the NOR concerns, and thank Doc James for participating. --Ronz (talk) 19:33, 6 January 2020 (UTC)

Barkeep49, to be honest, what's the most frustrating here is that a fait accompli was apparently rewarded. Generally speaking, when it becomes apparent that a large-scale change was not clearly supported by consensus, especially when it was also against policy, it should be reversed, not maintained, during the discussion on it. WP:NOPRICES is currently policy. If we want to have an RfC on whether it should or shouldn't be, we should have that, at WT:NOT (not here). Policy can change. But until and unless it does, the status quo ante should have been restored, and that was that prices are almost never included. Seraphimblade ^{Talk to me} 01:56, 7 January 2020 (UTC)

Seraphimblade, I hear you. When I first started trying to mediate this, I thought similarly and looked for the status quo version to roll this back to you. However, I have become convinced through education by those with institutional memory and lots of reading that this dispute goes back years and, at various times, there has been more support for including pricing information. Knowing that an RfC has been promised it's possible that some number of those who are in favor have not participated, especially as this page grew and grew and grew in length. I don't think anyone, on any side of this disagrees that WP:NOT plays an important role in this discussion. Best, Barkeep49 (talk) 02:11, 7 January 2020 (UTC)

This part of this question has been under discussion in various ways, off and on, since at least 2014. The practical options are:

• leaving it alone, because the difference between having this content in articles for 59 months or for 61 months is basically a rounding error, or
• blanking a few lines from more than 500 articles (and re-blanking it again in some cases, because not everyone will notice edit summaries, etc., and assume it was an accident or otherwise not warranted), and then maybe needing to restore all of those a month or two later.

I prefer not to blow up people's watchlists over this. Let's do it right, once, when we have a solid agreement on what "doing it right" looks like. WhatamIdoing (talk) 02:25, 7 January 2020 (UTC)

@Seraphimblade: and my concern was, in what order do we approach these issues (NOT, DUE, LEAD, WEIGHT) when it became apparent we didn't even have a policy-compliant (NOR) example to put forward (in the opinion of those who believe the samples are not CALC, rather OR). If we put forward a RFC just to determine if the community supports pricing in articles, and find out that the community does support drug prices in articles, does that endorse these databases being used? We have to get this question addressed first.

Also, when I was digging around to sort all of this out, it was quite disconcerting to find that we had redirects away from NOT (WP:PRICE, WP:PRICES, that earlier pointed to NOTPRICE)) to an essay, and the first line of that essay stated that Wikipedia had no policy on prices. That misleading info stood for four years (I corrected it last week). So how do we know how much of the community was misinformed by a changed redirect that pointed at faulty info? SandyGeorgia (Talk) 02:45, 7 January 2020 (UTC)

(edit conflict) SandyGeorgia, well, the junk essay was certainly a problem, saying "There's no policy on prices" when, well, there was one. I'm generally in favor of giving wide latitude on essays, but that stops at blatant factual inaccuracies, and "There is no policy on prices" is factually false when, well, there most certainly is one. But I think that's a bit backwards. If the determination is that we shouldn't include prices at all, the question of sources for them becomes entirely moot. It's only if we determine they should be that we even have to care about how to source them. Seraphimblade ^{Talk to me} 02:48, 7 January 2020 (UTC)

Even the strictest reading of WP:NOT indicates that Wikipedia should include at least some prices. Therefore the question is always "when and how?" rather than "always or never?" WhatamIdoing (talk) 02:54, 7 January 2020 (UTC)

(edit conflict)Well, sure. In exceptional cases, we should include prices. I don't think anyone would argue that, for example, I Am Rich should not include pricing information; that's why it's notable. There was another example I can't recall right to hand of a single treatment that costs over $2 million, and that price has similarly been extensively covered in reliable sources. We should include that there. In the Shkreli incident, price was the main issue, extensively discussed by reliable sources, and so the article must include information about it. But for general articles about products, drugs or otherwise, where the price is mentioned but not especially significant? That's exactly what NOPRICES is meant to exclude. In exceptional cases, we include prices. But most of the immediate cases aren't exceptional, and we don't include them routinely. Seraphimblade ^{Talk to me} 03:02, 7 January 2020 (UTC)

And if other editors agree with you that, e.g., the price of WHO Essential Medicines (whose "essentialness" is partly a factor of their low cost) aren't "exceptional", then eventually we would remove those prices. In between now and then, it's IMO better to leave well enough alone than to guess that your view is the one that the RFC(s) will eventually produce. Wikipedia:There is no deadline for this, and even if there were, the deadline would not be "1497 days after the date was added to diazepam". WhatamIdoing (talk) 19:31, 7 January 2020 (UTC)

@Seraphimblade: Here is what editors possibly saw for years. Yes, in which order to approach this is a dilemma. There was a point that I thought we just needed to go back to ANI, present the data developed by Colin showing how many editors have tried to remove this data over the years, and look for a new directive. That decision is above my paygrade. So, in which order do we proceed now? It is my opinion that WAID has a well-crafted RFC, while the alternative needs considerable work towards refinement, and the experience of WAID to turn it into something less confusing. But yes, there is a risk we are approaching this in the wrong order. I dunno; I am hoping Barkeep has an approach to AN that will help sort all of the different factors, but including that we don't know how many editors were misled for four years. Add to that the OTHERCRAPEXISTS meme that even FAs have prices (some of which are compliant with NOTPRICE, and some of which shouldn't even be FAs, I say with former FAC delegate hat on), and we have lots to sort. SandyGeorgia (Talk) 02:58, 7 January 2020 (UTC)

Barkeep49 did you mean to close this one, too? SandyGeorgia (Talk) 23:26, 23 January 2020 (UTC)

No I left this open because it seemed to range beyond jsut the RfC question. DDD is obviously an essential part of the debate but it is only indirectly covered (by design) at the RfC. Best, Barkeep49 (talk) 01:05, 24 January 2020 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.