Luckily, the trial is actually a rather neat spoof paper, written for educational purposes to highlight bad practices in the design and writing up of clinical trials. It's accompanied by a serious piece which analyzes these problems. They're both open access so you can take a look.
The sham study is ostensibly a trial of a new antidepressant, "placeboxetine", compared to an older drug for depression - but it was really written by one of the Editors of the journal via "Common shortcomings in manuscripts submitted to the Indian Journal of Psychiatry were collated into a single manuscript". These shortcomings are certainly not limited to Indian papers.
The problems included:
- No placebo group. This is extremely common in trials comparing two drugs, so it's "accepted practice", but it's still a bad thing.
- The "placeboxetine" was given at a higher dose, relative to its dose range, than the comparison drug but they don't say why.
- Side effects are reported but they don't explain how these were assessed. If you specifically ask about them you find a lot more than if you rely on patients to spontaneously complain.
- Subtle, but important, issues with the statistics, such as reliance on t-tests over more appropriate methods.
- The trial had some unusual features for a depression trial - with no explanation. Most patients were males in their 20s, while the norm is for about 65% females and an average age in the 40s; very few people dropped out; very few people who were screened were excluded, whereas most trials exclude loads of people for all kinds of reasons.
- The effectiveness of both drugs was remarkably high (75% cure rate over 6 weeks - better than any treatment, drug or therapy, would be expected to show.) Yet they don't mention this.
- It was badly written. The title in particular was far too long and clumsy.
- It turns out that the trial was sponsored by the fictional pharmaceutical company, and was probably conducted to help get placeboxetine sold in India - but we only find this out in the small print at the end.
- Hot pink and white is not a good color scheme for your graphs, or for anything except marshmallows. (I may have added this myself.)
Andrade C (2011). A 6-week, multicentre, randomized controlled clinical trial to evaluate the safety and efficacy of placeboxetine hydrochloride in the treatment of major depressive disorder in an Indian setting. Indian journal of psychiatry, 53 (1), 69-72 PMID: 21431013Andrade C (2011). Placeboxetine for major depressive disorder: Researcher, author, reader, and reviewer perspectives on randomized controlled trials. Indian journal of psychiatry, 53 (1), 73-6 PMID: 21431014
7 comments:
Good to see a publication like this, but I have to disagree with you on one point - pink and white is a fantastic graph colour scheme. I myself am currently embracing the use of magenta and black. It makes your graphs look like graphics in a gay nightclub, i.e. awesome.
"No placebo control."
Might this be due to ethical concerns? Withholding a proven treatment for a disease seems wrong. Comparing to treatment as usual might not be ideal from a study design perspective, but for (say) someone with depression to go 6 weeks with a placebo seems cruel and dangerous.
Loved the journal for both articles, thanks for sharing!
The complaint of no placebo arm is completely bogus and would have been unethical. The Declaration of Helsinki declares that:
32. The benefits, risks, burdens and effectiveness of a new intervention must be tested against those of the best current proven intervention, except in the following circumstances:
The use of placebo, or no treatment, is acceptable in studies where no current proven intervention exists; or Where for compelling and scientifically sound methodological reasons the use of placebo is necessary to determine the efficacy or safety of an intervention and the patients who receive placebo or no treatment will not be subject to any risk of serious or irreversible harm. Extreme care must be taken to avoid abuse of this option.
http://www.wma.net/en/30publications/10policies/b3/index.html
In my opinion using a placebo leg would have been unethical. There are no compelling scientific or methodological reasons to use a placebo. The drug status of the trial participants should have been blinded to the evaluators, not necessarily to the physician adjusting dosage. The non-equal dosages is irrelevant.
daedalus: The problem is this; a new treatment ought to be tested against placebo and it also ought to be tested against the current standard, however, as it stands these are often does as two separate kinds of trials.
This allows trial sponsors to stack the odds in their favour. Make the placebo-controlled trial as "difference finding" as possible, and make the comparison trial as "conservative" as possible.
Hey presto : the drug is better than placebo and no worse than the alternatives!
Whereas what you ought to do is to compare it to placebo and comparator in one trial (which would incidentally provide new data on the effectiveness of the competitor).
Since symptoms of depression often rise and fall, a placebo arm would help distinguish between treatment effects and background changes that normally occur over time. As long as the participants of such a trial clearly understand the protocol, are judged to be stable enough to not be unreasonably harmed by a lack of treatment if they are in the placebo arm (while of course taking into account that one or both of the drugs in this kind of trial might be more harmful than no treatment), and there is monitoring to catch warning signs of worsening symptoms (especially suicidal thoughts) in all three arms; then it seems an ethical case can be made for a placebo arm.
Another problem: the trial doesn't look like it's been registered in a clinical trial registry (like clinicaltrials.gov), a must for publication in a number of journals.
Well spotted.
Maybe it should have been registered on the ROFL (Registry Of Fake triaLs)
Post a Comment