Reader Comments

Post a new comment on this article

It pays industry to get it right

Posted by plosmedicine on 30 Mar 2009 at 23:43 GMT

Author: Hywel Williams
Position: Professor of Dermato-Epidemiology
Institution: University of Nottingham
Submitted Date: June 06, 2005
Published Date: June 6, 2005
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Vintage Smith this one - but provocation aside, there are some distrubing issues for debate.

The box listing the methods that Pharmaceutical companies use for getting the results they want from clinical trials is very similar to one that I have developed in a lecture that I am currently writing up on "how to cheat in clinical trials" based on game theory (design games, analysis games, write-up games and publication games).

Design games include: avoiding active comparators, choosing an obscure comparator, using the comparator at the wrong dose, and selecting outcome measures that inflate the treatment effect.

Analysis games include torturing that data with lots of statistical tests (only some of which are reported), torturing the data again with unspecified subgroup analyses, using relative as opposed to absolute changes to sound impressive, only testing change in score for each treatment separately but not against each other, and not performing an intention-to-treat analysis.

Write-up games include misinterpreting inconclusiveness with therapeutic equivalence, avoid mentioning adverse effects in abstract, ignoring patient-generated outcomes, inflating the scales on figures and emphasising clinically meaningless composite scales.

The publishing games are fairly straightforward ie don't publish "negative" studies, delay publishing negative studies, publish such studies in obscure journals, and duplicate and salami publishing of "positive" studies.

And I am sure the list goes on. I would suggest that it applies to academics as well as the pharmaceutical industry.

Yet one of Smith's central points (that studies sponsored by indutsry are more likely to show favourable results) may have another explanation ie in addition to publication bias. Phase 3 industry trials of a new drug, are often designed on the basis of a lot of prior knowledge in the form of previous case series and Phase II studies. The prior probabilities of their drug "working" is therefore quite high, especially against placebo - I suggest that clinical equipoise is such cases is seldom truly fulfilled.

Because their future salaries depend on picking winners that are likely to show positive results, colleagues in the pharmaceutical industry clearly have a vested interest in getting it right.

Competing interests declared: None