6

A 2016 editorial in the BMJ makes the following claim:

The top 10 prescription drugs in sales have a cumulative clinical response rate of less than 20%.

The claim is not the central theme of the article, which is an argument for doing clinical trials differently.

The author references this claim with a link to a Nature Comment piece. But it is unclear whether the claim is a fair summary of the argument from Nature or the consensus view of the literature.

Hence the question: do the most prescribed drugs have a cumulative clinical response rate of less than 20%? What does the claim itself mean in layman's terms and is it a fair summary of scientific opinion on the effectiveness of prescription pharmaceuticals?

matt_black
  • 56,186
  • 16
  • 175
  • 373
  • 1
    The BMJ is a highly-respected medical journal. Do you have a reason to think it is publishing incorrect information, or to believe that random people on the internet are better suited to judge the claim than the BMJ? – DJClayworth May 25 '16 at 16:45
  • @DJClayworth well, the claim itself was not *central* to the topic of the article but was more of a throwaway remark. This context was (I think overzealously) edited out by Dawn and I haven't decided whether to rollback yet. – matt_black May 25 '16 at 16:48
  • There is also a reference backing up the claim in the editorial. – DJClayworth May 25 '16 at 16:51
  • @DJClayworth True, and the reference is also fairly respectable. But still worth questioning and checking against broader sources. If the evidence is strong and we know what it means it is still worth recording here. – matt_black May 25 '16 at 16:54
  • 1
    The fact that this claim is (in your view) a throwaway remark doesn't affect the claim or how we should evaluate it. Also, @DJClayworth just because the claim comes from a reputable source doesn't at all affect how we handle it here... we still just present the best evidence concerning the topic. If that evidence is already given as a reference in the BMJ article, so be it. –  May 25 '16 at 17:48
  • 1
    @DJClayworth We take questioners at face value that they're simply looking for the best evidence. They don't need to assert *why* or even *that* they doubt a claim. –  May 25 '16 at 17:50
  • I changed the title because cumulative response rate does not translate to specific response rate ([fallacy of division](https://en.wikipedia.org/wiki/Fallacy_of_division)). – Sklivvz May 25 '16 at 23:59
  • @Sklivvz I'm not entirely sure you are right and, more importantly, I am sure you are introducing technical language into the title that will confuse people. I'd have thought that any answer would deal with this sort of technicality, but I'm sure it obfuscates the question. I'm not going to rollback any edits for not, though. – matt_black May 26 '16 at 00:10
  • The language was already in the quote and in the body of the question, is it not confusing there? – Sklivvz May 26 '16 at 00:44
  • I had a brief look at what was involved in answering this question. [Humira is the top selling drug by sales](http://www.medscape.com/viewarticle/849457), but its [clinical response rate was different](http://www.medscape.com/viewarticle/849457) at Week 8 versus Week 52 (as was the placebos), so I am not sure if the claim is adequately specified. (Maybe there is a well-known result that they are implicitly referring to?) – Oddthinking May 26 '16 at 00:52
  • yes, it's linked in the article: http://www.nature.com/news/520609a-i2-jpg-7.25660?article=1.17411 – Sklivvz May 26 '16 at 00:53
  • They all have different response rates, 1:5 is only the average... – Sklivvz May 26 '16 at 00:53
  • :-( I just saw it was linked in the article. That should be mentioned in the question. Edited. – Oddthinking May 26 '16 at 00:55
  • @matt_black: This claim appears to have been derived from [this infographic](http://www.nature.com/news/personalized-medicine-time-for-one-person-trials-1.17411#/imprecision). If your skepticism is actually about the numbers in this infographic in particular, (as opposed to whatever transmogrification was used to work out "cumulative clinical response"), I think this question would be easier to answer. – Oddthinking May 26 '16 at 01:14
  • @Sklivvz I think titles should be simple. There is room in the body to deal with the complexity and detail. – matt_black May 26 '16 at 08:25
  • @Oddthinking I think the claim is OK but a meaningful answer would need to explain exactly what it means as there is more than one interpretation. The claim as it stands has been made. I suspect a good answer would explore how the numbers were derived (perhaps even going behind the infographic), what they mean and whether they are credible. – matt_black May 26 '16 at 08:29
  • The claim has a link to an actual peer reviewed article that supports it. It is pointless to ask a question that needs to be found by a Google search, when the answer can be found by clicking a link literally next to the claim. If one didn't have the patience to check the given sources, they won't certainly look here. Closing as non notable. – Sklivvz May 26 '16 at 09:23
  • @Dawn If someone asked this question without the BMJ link, and an answer was provided linking to the BMJ and the article it references, everyone would say "what a great answer" and vote it up. Why would exactly that answer be bad just because the OP links to the article up front? – DJClayworth May 26 '16 at 15:26
  • 1
    @DJClayworth http://meta.skeptics.stackexchange.com/a/2528/30557 Also, it is only our *presumption* that this is the best evidence available. Maybe there is other, conflicting, or better evidence. To close the question because we *think* it already contains the best evidence is not very skeptical. FYI, I would not upvote an answer that contained a single paper as reference without some reason why I can rule out that paper having been cherrypicked. –  May 26 '16 at 15:30
  • If the OP is skeptical about the evidence linked in the article, they are free to reword the question accordingly, and it's likely to be reopened. As it stands, there's no value on posting a reference *here* which is already available *there right next to the claim*, and that fully, and conclusively, explains what the claim is about. – Sklivvz May 26 '16 at 15:40
  • In other words, if the OP has read the linked article and is still skeptical, simply posting it here won't help anyone, in that case the question needs to be reworded so it refers to the article. On the other hand, if the OP has not read the linked article, then they should read it first and then elaborate a better question based on that if they are still skeptical. – Sklivvz May 26 '16 at 15:49
  • @Sklivvz I am astounded at your logic here. The claim is *clearly* notable by any normal standards here (and being easy to answer, even if true which is arguable in this case, has not ruled out many other questions here on many other topics). And, if having a single reference to the literature is good enough to count as "not notable", there would either be few questions here at all or a large number of completely contradictory answers. – matt_black May 26 '16 at 15:58
  • @matt_black it hasnt been closed for non-notability. –  May 26 '16 at 16:20
  • @dawn see the comment by sklivvz about 6hr ago "closing as non notable" which seems to imply it was. – matt_black May 26 '16 at 16:22
  • @matt_black He has since corrected that. Note the close reason, and this [meta comment](http://meta.skeptics.stackexchange.com/questions/3633/does-the-non-notabile-close-reason-apply-to-claims-that-already-ostensibly-provi#comment11323_3635). –  May 26 '16 at 16:23
  • 1
    @Dawn I missed much of the action today and the meta discussion. I'm still feeling perturbed at the moderator's logic which doesn't appear consistent with other judgements here. I think the problem isn't the *question* but is inherent in the claim quoted in the question. The issue, I think, would be better resolved by a good *answer* than by putting the question itself on hold. This may require a lot more discussion on meta. – matt_black May 26 '16 at 16:37
  • 2
    @matt_black You might find support for your position in [this meta q/a](http://meta.skeptics.stackexchange.com/questions/3230/how-much-we-need-to-define-in-a-question). I think that stands for the principle that if the claim *itself* is ambiguous, that doesn't necessarily warrant closure. –  May 26 '16 at 16:45

0 Answers0