Wikipedia:Manual of Style/Medicine-related articles/RFC on pharmaceutical drug prices

These examples of drug prices for generic pharmaceutical drugs have been taken from the leads of articles. Do you think that this content complies with Wikipedia's standards for verifiability, due weight, no original research, what Wikipedia is not, and how to write a lead section? 23:02, 23 January 2020 (UTC)

The source:

All of these examples use the same source (chosen for simplicity and because it appears in many articles). The source, the International Medical Products Price Guide, is a database of prices voluntarily reported by some organizations and government agencies that buy or sell drugs in low and middle income countries. For example, a "supplier" price may be reported by a non-profit or a for-profit organization that sells drugs only to health organizations and government agencies in a single country, or one that sells drugs to buyers in many countries. The "buyer" price is most often reported by a government, and sometimes by a non-profit or medical missions organization. The "supplier" prices are usually more relevant than the buyer prices, as buyer prices do not include shipping, handling, insurance, or other costs, and some buyer prices involve unusual circumstances (such as atypical prices from local manufacturers). The database's website recommends the median supplier price as the most reliable option, assuming that enough data points are available. Read more about prices on the database's website.

Each database entry also includes some other information, such as the strength and dosage form of the specific drug for that entry (most drugs are available in different strengths; some are available in both liquid and pill forms; each strength and form gets a separate database entry, search here to find examples) and the defined daily dose (read the article). This database has not been updated since 2015. You can look at some statistics about entries in the MSH database.

Three examples:

• Ethosuximide:
  • What the lead says: The wholesale cost in the developing world is about US$27.77 per month as of 2014.[1]
  • What the source says about:
    • Suppliers: The cited database entry lists one supplier, which offered 250 mg tablets of ethosuximide for sale for US$0.1845 each (100 tablets per package) in 2014, only to organizations in the Democratic Republic of Congo.
    • Buyers: The database has no information about buyer prices.
    • The drug: Each pill contains 250 mg of ethosuximide. The defined daily dose is 1.25 grams. There are no other entries in the MSH database for this drug.
• Carbamazepine:
  • What the lead says: The wholesale cost in the developing world is about US$0.07 to US$0.24 per day as of 2015.[2]
  • What the source says about:
    • Suppliers: The cited database entry lists twelve suppliers that offered 200 mg tablets of carbamazepine for sale in packages ranging from 100 to 1,000 pills each in 2014. The lowest price reported by these 12 organizations was US$0.0138 per pill, and the highest price was US$0.0372 per pill, with a median price of US$0.0185 per pill, down 5%. The High/Low Ratio was 2.70.
    • Buyers: The cited database entry lists three governments plus the Central American Integration System. The lowest price reported by these four organizations was US$0.0155 per pill, and the highest price was US$0.0480 per pill with a median price of US$0.0202 per pill, down 9%. The High/Low ratio was 3.10.
    • The drug: Each pill contains 200 mg of carbamazepine. The defined daily dose is 1 gram. There are five other entries in the MSH database for this drug for 2015.[3]
• Diazepam:
  • What the lead says: The wholesale cost in the developing world is about US$0.01 per dose as of 2014.[4]
  • What the source says about:
    • Suppliers: The cited database entry has no information about supplier prices.
    • Buyers: The cited database entry lists two governments that reported buying 10 mg tablets of diazepam in 2014. Both governments report a price of US$0.0100 per pill, so the lowest, highest, and median price are identical, and the High/Low ratio is 1.
    • The drug: Each pill contains 10 mg of diazepam. The defined daily dose is 10 mg. There are two other entries in the MSH database for this drug in 2014,[5] and three for 2015.[6]

Background

Drug price availability

Drug prices vary according to a number of factors, including strength (e.g., how many milligrams per pill), formulation (e.g., pill or liquid), manufacturer, country, wholesale vs. retail, what the patient pays vs. what the health program pays, prices before rebates and subsidies, etc. Even within the category of wholesale prices, there are multiple measurements. Many prices are kept confidential, so there are few sources available. Additional information about drug prices can be found at Category:Drug pricing.

Example articles

You can see several hundred examples of how drug prices are being presented at User:Colin/ExistingPrices.

Related discussions

There was a previous RFC related to drug prices in 2016. Recent discussions on what advice we should give in guidelines include a long thread at Wikipedia talk:Manual of Style/Medicine-related articles#Product pricing and another at Wikipedia talk:WikiProject Medicine/Archive 130#MEDPRICE. This RFC was written at the direction of the community in this discussion and mostly developed at WT:MEDMOS.

This is a marathon, not a sprint

The next goal is to update the Wikipedia:Manual of Style/Medicine-related articles to include advice on when and how to present information about prices and pricing for drugs. It is likely that there will be at least one more RFC related to this subject area. Everyone's ultimate goal is to build great encyclopedia articles about pharmaceutical drugs.

Discussion

Main question:

Do you think that the content in the examples above complies with Wikipedia's standards for verifiability, due weight, no original research, what Wikipedia is not, and how to write a lead section?

This is a request for your comments, not just a vote! You don't need to "support" or "oppose" anything. Please feel free to ask questions and to respond to this RFC broadly. If you think the text in our articles could be improved, please tell us how you would improve it. If you think an example shows a good feature, please tell us why it's good. Understanding your thought process and what matters most to you will directly help us develop a solid proposal for updating WP:MEDMOS and will ultimately help us improve articles about drugs.

We're going to try to keep comments about the same general subject area together. Don't worry: you don't have to stick tightly to the subject! We know that these areas overlap significantly, so just put your comments in whichever section you think is closest. You can put all of your ideas in one section, or you can put different comments in separate sections.

Comments about verifiability and original research

• The thing I really like...
• The problem I see...

• Unless there's some background I'm missing, the ethosuximide example appears grossly inappropriate. Is a single price from one supplier to one country really being given as the cost for the entire developing world? Is there some reason to believe that price is typical or representative? And why the four-figure precision? Even before we think about over-extrapolation, fluctuations in the US exchange rate alone make $27.77 misleading on account of overprecision. Adrian J. Hunter^{(talk•contribs)} 07:33, 24 January 2020 (UTC)
• I agree with Adrian. Unless the drug is specific to one or a small group of countries, or there is a negotiated price for a drug (eg a charity, producer or grouping has negotiated a wholesale price, which in itself is relevant to mention in my opinion), I think it's misleading and inappropriate to give a single number as the cost of a drug. It's misleading, because this is the wholesale price, which is not relevant to readers who will be expecting the consumer price. It's highly likely to vary between countries, and regardless of what a source may say I am extremely doubtful that there is a meaningful single stable price for drugs that should be included in the lead, particularly one that readers, who are likely from English speaking and therefore developing to developed countries. --Tom (LT) (talk) 21:56, 24 January 2020 (UTC)

• What calculation leads to the example for carbamazepine? Since the database appears to be a series of examples, unless it’s exhaustive I don’t think you can infer a range from that without original research (the extremes of a sample are not the extremes of the entire population). Also, "the developing world" is an ambiguous category, and even if a specific classification can be decided on, it will change as development status generally improves over time. Sunrise (talk) 00:40, 25 January 2020 (UTC)

• I am confused why the first two entries in this section are nothing but blank starting statements. Are the originators of this RFC so insecure that they feel they must control the inputs of everyone who comments? Verifiability and original research have little in common as it pertains to the supposed examples. Are you asking whether we can verify original research or verify content or whether verifiable information is being used for original research. Also is there a particular reason you kept the one users's, who bludgeoned the discussion to start this RFC's, original research in the opening prose. Anyone thinking this RFC is a neutral representation of current content should look at the discussion that lead to this RFC. Be careful it will take hours.AlmostFrancis (talk) 04:47, 25 January 2020 (UTC)
  • @AlmostFrancis: your comment is a confusing mix of substantive comments and process complains. I suggest you separate the two. I am confused why the first two entries in this section are nothing but blank starting statements. Obviously, they are an attempt to encourage people to engage in discussion behavior, rather than in voting behavior. --JBL (talk) 02:18, 26 January 2020 (UTC)

• About 25% of our articles give a developing world cost per dose. The source does not give a cost per dose, nor say what is a typical dose (the amount of a drug one takes at one time), so any such cost is unverifiable. Most drugs treat several conditions and there are lots of common doses, which vary with indication and patient characteristics (age, weight, health).

For example, with diazepam we give the cost for an untypical 10mg tablet dose, a tablet strength that isn't available from any suppliers to the developing world. The correct typical dose for diazepam for anxiety is 2mg three times a day taken long term. In the UK this costs the NHS £0.027 per dose, which is meaningless to our readers if you don't know the "2mg three times a day" treatment plan (forbidden by WP:MEDMOS). For emergency treatment of a prolonged seizure then the typical adult dose of 10mg is give and then repeated after 10 minutes if required. This is not a pill, but given either by injection or per rectum, costing £0.55 or £1.50 per dose. There is no one cost per dose. -- Colin°^Talk 11:41, 26 January 2020 (UTC)

• About 30% of our articles give a developing world treatment cost per day or month.. The source does not give the cost per day or month, nor does it suggest a treatment plan ("take two 50mg tablets twice a day for seven days", or "take one 100mg extended-release tablet a day and come back to see me in six months") so we are unable to verify which tablet strength to cite, how many to take, how often, and for how long.

These costs are derived by an editor using the defined daily dose (DDD) combined with an arbitrary choice of tablet strength and assumptions about dose frequency and treatment duration. The DDD is a calculated metric designed for drug usage statistics not treatment cost. It isn't guaranteed to divide by any available pill size, or correspond to a dose anyone is typically or ever prescribed. The WHO say that using this for detailed cost analysis is an abuse of the metric, and the only valid use they suggest is to compare the cost of two different formulations of the same drug. Our source does not use the DDD to give the treatment cost of a drug, so neither should we. -- Colin°^Talk 11:41, 26 January 2020 (UTC)

• We give a "wholesale cost in the developing world", an International Reference Price. The source Guide says "buyer" prices should not be used as reference prices. Instead, the WHO recommend taking the median of many "supplier" prices to make a reliable reference price. A third of our drug articles have no suppliers, and another third only have one or two suppliers: for these drugs, the majority, the Guide is not a reliable source of "wholesale cost in the developing world". -- Colin°^Talk 11:41, 26 January 2020 (UTC)

• See PriceMistakes for examples of drugs where original research, unsourced data and misuse of WHO metrics has led to clear errors. 20% of our prices are simply wrong, even ignoring all the other issues. -- Colin°^Talk 11:41, 26 January 2020 (UTC)

Comments about neutrality and due weight

• I don't think this belongs in the lead, per WP:LEDE ("introduction to the article and a summary of its most important contents") and WP:UNDUE ("depth of detail...prominence of placement"). It's just one loose fact, not generally a defining characteristic of the drug or a summary of substantial article content (no secondary ref to help place this idea in context, etc.). If the topic of pricing of a certain drug is covered in substantial detail in the article (for example, highlighted as an example by secondary refs for differences of US vs foreign, or sudden markup by manufacturer, or in comparison to other treatments for a medical condition), I could envision mentioning that topic there (but again, probably not the single value). DMacks (talk) 09:43, 24 January 2020 (UTC)

• For anything but the most basic information, using only a database as a source isn’t much better than having no source at all. Even passing mentions in stronger sources are not generally sufficient to support inclusion, because they give no indication of the information’s importance; the price itself has to be the subject of discussion. A database contains no context or analysis, and is only usable in the first place if there’s no doubt about accuracy. However, to actually ensure that, the examples above would have to be much more specific, because these are prices for specific times and places, and even specific transactions ("in year A, according to the records in database B, company C sold D doses of drugs to group E for a price of F in currency G, which includes insurance and transportation charges but not handling fees or taxes"), and it rapidly becomes clear that the weight is insufficient. Considering only questions of weight, I suppose one could argue that a price is in fact basic information (similar to molecular weights or chemical formulas), meaning a database citation would be acceptable, although the argument is significantly weakened by the issues with inaccuracy and the fact that the value isn't constant. If it's accepted regardless, perhaps it might not be so out of place to include prices in an infobox entry attached to an appropriate caveat, but even then a sentence in the lead (or arguably even a sentence in the body) is far more weight than it should be given without support from stronger sources. Sunrise (talk) 00:40, 25 January 2020 (UTC)

What does neutrality have to do with due weight in these examples? Are you claiming that the entries are not neutral in some way, and if so could that be pointed out? Due weight is a better question. Not the question that was asked but a better question. I think an RFC based on what would be due weight would be useful but based off of these three examples is pointlessAlmostFrancis (talk) 04:57, 25 January 2020 (UTC)

Weight violations are a type of neutrality violation; WP:WEIGHT is a subsection of WP:NPOV. Sunrise (talk) 12:24, 25 January 2020 (UTC)

Comments about what Wikipedia should include

• I don't think WP:NOT per se should be used as an a reason to exclude pharmaceutical prices. To me, the approximate price of pharmaceuticals is self-evidently encyclopedic and a matter of public interest, as are the price of oil, the price of carbon, and even the price of a Big Mac. WP:NOTCATALOGUE has a vital maintenance role, preventing Wikipedia getting spammed with ephemeral information about which vendor sells what product for 10% less than the competition, or who's got a sale on right now!!!. Such practical concerns need not stop us citing globally relevant information from non-commercial secondary sources.

Obviously this is moot in cases where we cannot convey pricing information without contravening WP:VERIFIABILITY or otherwise misleading readers. Adrian J. Hunter^{(talk•contribs)} 07:36, 24 January 2020 (UTC)

• I do think the drug price is a notable topic deserving inclusion in an article. Additionally as I mention above, things related to drug prices can be notable and deserve mention, such as: notable agreements to lower or fix prices, notable criticism about price variation of a specific drug between countries or produces, notable coverage of extortionate increases or extortionate prices, or prices in relation to the prices of inferior or fake versions of the drug. Normal procedures should apply to this, ie. notability determined by coverage of these issues in reliable sources, and normal procedures should apply to coverage in the lead - viz. likely not in most cases --Tom (LT) (talk) 22:01, 24 January 2020 (UTC)

• A price can be the direct subject of analysis, as described in Tom’s comment, in which case it is encyclopedic. However, the implication from the examples seems to be that the price should be included automatically, cited only to a database. Including these prices in such a manner, including the examples given above, makes Wikipedia seem more like a commercial catalogue and thus undermines our status as an authoritative reference work. One of the purposes of WP:NOT is to prevent this; in such cases the information should be included in Wikidata instead. Overcoming NOTPRICES in this sort of situation requires a strong overriding argument based in another policy, or otherwise WP:IAR.

As a separate point, in one part of the long pre-RfC discussion, I saw one argument regarding deaths caused by medical misinformation, which would certainly fit the criteria for a strong IAR argument. However, by the same analysis, damage to the encyclopedia’s reputation itself causes deaths because people have to be willing to believe the information in the first place (and that would affect all the rest of the information in the same articles, as well as everything else that has important implications anywhere on Wikipedia). So there is a potential "more harm than good" issue that needs to be resolved. Furthermore, while medical misinformation is certainly a major problem, I refer to the comments in the other sections: I'm not convinced that using prices in the manner described above wouldn’t just cause even more harm by contributing to that misinformation instead. Sunrise (talk) 00:40, 25 January 2020 (UTC)

Comments about the Manual of Style

• The thing I really like...
• The problem I see...

• I think the complexity of drug pricing means, if it should be included, it should be as a section or subsection, certainly not in the lead. For comparison, coverage in articles about stable foods is inconsistent (eg milk - subsection, bread - no section; rice - section) --Tom (LT) (talk) 22:01, 24 January 2020 (UTC)
  • I agree that drug prices (unless exceptionally notable) should not be in the lead (if they appear at all). Kaldari (talk) 03:52, 26 January 2020 (UTC)

Comments about drug prices or statistics

This is the section for comments that are more focused on the subject matter than on Wikipedia's policies and guidelines.

• The prices of pharmaceuticals vary over several orders of magnitude. Just in the US, Kalydeco apparently costs $25,000/month, while aspirin costs <$25/month. Perhaps much conflict could be avoided if the proponents of including pricing could focus on conveying the order of magnitude of a drug's cost. Is it in the same ballpark as Kalydeco or as aspirin? Trying to give exact pricing runs into issues with overprecision and whether particular known prices are representative. From this perspective, I think the example given for diazepam is fine, being an order-of-magnitude estimate to one significant figure. Adrian J. Hunter^{(talk•contribs)} 07:38, 24 January 2020 (UTC)
  • Also drug prices vary even in the same country on the same drug Dq209 (talk) 13:43, 24 January 2020 (UTC)
    • Agree with your point; a single number is almost meaningless. That is because what is discussed above is the wholesale price which is the price of the drug sold to businesses. As I mention above it is confusing and misleading to use the wholesale price without any context, for the very reason you mention. There is usually a lot of variation between this and the prices charged by businesses or end distributors to consumers.--Tom (LT) (talk) 22:08, 24 January 2020 (UTC)

• As far as I can tell, the prices are more likely than not to be misleading. Adding specific qualifiers or putting together numbers from several different places or times doesn’t materially change this, because there are at least hundreds of possibilities. Per Adrian, a "which order of magnitude" estimate is reasonable, but nothing more precise than that. Also, I'll repeat that all of this has the same exception that sources that directly analyze the price are a separate issue. However, if e.g. certain medicines are important specifically because of their low (or high) price, which I can certainly believe, then it should be easy to find a source that simply says that and use it directly. A generic statement about medicines from a certain category would probably work for that and could be copied across multiple articles. Sunrise (talk) 00:40, 25 January 2020 (UTC)
• would be ok with just reflecting the "order of magnitude"...the price has been stable over the last 20 years, the one supplier IDA Foundation sells to more than 130 LMIC at the stated price...IMO--Ozzie10aaaa (talk) 02:29, 26 January 2020 (UTC)

• The most obvious cost to offer readers is treatment cost for a named indication. Either for complete treatment or an idea of ongoing costs for lengthy treatment. Giving the cost of an arbitrary pill or dose isn't meaningful if you don't know how many pills to take, how many times a day to take that dose, and for how many days. The problem is, our drug database sources simply do not contain the information needed to work this out, and our medical guideline forbid giving dose or treatment plan information to our readers. And that's assuming it even makes sense to talk of one treatment cost, for a drug with multiple indications and multiple patient types.

Even if we were able to source a single treatment cost for a drug, some have suggested offering a ballpark figure (Adrian J. Hunter). The $0.01 for diazepam is an illusion: it isn't a ballpark figure but just the result of the two buyer prices being the same "$0.0100" price. Rounding to 1 significant figure, for example, still gives the illusion of precision to a tenth of the price: $0.06 appears twice as expensive as $0.03 when in fact both prices may have huge error bars and fluctuate by that amount in a few years. Newspapers can collapse multiple prices into one by saying "cost as much as $25,000" or "cost as little as $0.20" or cherry pick with "can cost $100" and do so to make a case about how expensive or cheap a drug is. Unless citing such a source, we can't do that, so how could Wikipedia give a ballpark figure that the reader knows is a ballpark and understands how imprecise it is? -- Colin°^Talk 11:50, 26 January 2020 (UTC)

Other concerns

• The thing I really like...
• The problem I see...

• The problem I see with this is WP:NOTPRICES. Neutral view or not, it cannot override policy.

"An article should not include product pricing or availability information unless there is an independent source and a justified reason for the mention. Encyclopedic significance may be indicated if mainstream media sources (not just product reviews) provide commentary on these details instead of just passing mention. Prices and product availability can vary widely from place to place and over time. Wikipedia is not a price comparison service to compare the prices of competing products, or the prices and availability of a single product from different vendors or retailers."

Accordingly, I don't see a reason to override policy here. If we're going to include prices (which is an inherently political issue when it comes to medicine with WIDELY varying standards), it's just plain too unclear to be accurate. I can't see a single instance where such a quote wouldn't have to have an in-depth explanation. The only instance I can see its inclusion would be where the price was in the news for some reason (very high or very low). Without addressing this issue at Wikipedia Talk:What Wikipedia is not, neutrally phrased or not, such a change to this guideline cannot override a policy. Buffs (talk) 16:54, 24 January 2020 (UTC)

• Comment. [COI statement: I used to do consultancy work for the pharma industry in the area of HIV & antivirals.] Commenting generally here as I find the attempt to channel comments into areas unnecessarily limiting. I have not read any of the previous discussions. Some drug prices are, in my experience, of huge interest to patients and in particular patient advocate groups. They are often the subject of considerable discussion in news-type sources. I don't see why they should be excluded because they are not sourced to meet the medical project's idiosyncratic definition of reliable sources for medical articles, as they are not themselves medical information. There again, there is a problem with the amount of detail that is necessary to convey any real understanding of the real price of a drug. Point information in the infobox is likely to be misleading, and will date rapidly. A couple of sentences in the lead, as in say nevirapine, whilst not precisely undue weight, feels less critical than much of the other information in the lead. In general, a section in the body on pricing might be a better way of conveying the issue, unless there is a cogent reason for including it in the lead. Espresso Addict (talk) 23:20, 25 January 2020 (UTC)

Espresso Addict, just to note that nobody has suggested politico/economic data like drug costs should meet WP:MEDRS. Meeting basic core policy WP:V, WP:OR and WP:WEIGHT and not misleading our readers would be a start. The notability requirements of WP:NOTPRICES are very much "mainstream media sources" and not specialist journal papers, drug database records, or documents in the WHO archives, for example. -- Colin°^Talk 11:57, 26 January 2020 (UTC)

• I agree with Buffs: I think that WP:NOTPRICES is the relevant policy and that it applies cleanly to the particular case of drug pricing. There may be individual drugs whose pricing in encyclopedically notable (insulin is clearly an example), but for those drugs one should expect to find many sources addressed to a broader audience, allowing for meaningful coverage on WP; these exceptions are already allowed for in the relevant policy. --JBL (talk) 02:26, 26 January 2020 (UTC)

Notices

As of RfC launch Wugapodes has agreed to act as a closer. Depending on the nature of the conversation it might or might not make sense to do a three editor panel. Best, Barkeep49 (talk) 23:02, 23 January 2020 (UTC)

It looks like Ymblanter should also be available as a closer, based on their most recent comment at AN. Advertising the discussion early at ANRFC might help with finding additional closers as well. Sunrise (talk) 00:49, 25 January 2020 (UTC)

I had missed that. Thanks. Best, Barkeep49 (talk) 01:15, 25 January 2020 (UTC)

So far, I only volunteered to close this RfC. This might or might not make me involved, but I am not yet prepared to make statements now on the RfC at which talk page we are now in.--Ymblanter (talk) 07:52, 25 January 2020 (UTC)

My mistake, I hadn't realized your comment was about the other RfC. Sunrise (talk) 12:34, 25 January 2020 (UTC)

No problem. Let us do it step by step. First I need to close that one.--Ymblanter (talk) 18:07, 25 January 2020 (UTC)