One of the hopes of this blog is to create a repository of maneuvers through which clinical trials can be gamed to get results. The series of posts laying out some of the less well known tricks are filed under the Hiding the Bodies blog category. To be more generally useful, this repository needs others to contribute further maneuvers to make it comprehensive and to contribute examples from areas of medicine other than mental health to help people more generally to spot what is going on. I will add to the lists below as examples come in.
After this post, (and after a pharma-myth and personal account), the next series of posts will tackle the notion that randomized controlled trials (RCTs) offer a Gold Standard when it comes to determining what drugs cause. The argument in brief will be that RCTs provide something close to an ultimate, and at present unrecognized, bias behind which companies can hide bodies. These posts will be filed under the Spin & Data blog category.
Clinical trial maneuvers typically listed in the literature include:
- Switching endpoints after the results come in
- Surrogate rather than substantive outcome measures
- Inappropriate use of Number Needed to Treat (NNT)
- Last Observations Carried Forward (LOCF)
- Relative Risks: results expressed in relative rather than absolute risks
- Short term trials, when long term administration is contemplated
- Co-administration of agents that will minimize side effects
- Publication management: non-publication, ghostwriting of publications, etc.
Recent posts have demonstrated a series of less commonly listed strategies:
- Moving bodies to boost apparent risks in the placebo group
- Creating doubt – using statistical significance to hide risks
- Splitting hazards so they appear under different coding groups
- Inappropriate use of duration of treatment to hide risks
- Running trials with high baselines of adverse events to drown out hazards
- Actively withholding trials from regulators (this can be done quite legally – the regulator has to ask for exactly the right thing, and regulators encouraged to partner industry are perhaps less likely to get the questions right)
- Companies commonly use placebo washout periods to get rid of placebo responders.
- Companies also use washout weeks to eliminate anyone having an adverse response to the active treatment.
Here is a variant on maneuver 1 outlined in Drug companies use studies the way a drunk uses a lamppost where the placebo washout was used by Lilly, GSK, and Pfizer to hide suicidal acts. When reviewing the issue of antidepressants and suicide in 2003, the MHRA asked for, but didn’t get, all trials and asked companies not to repeat this placebo washout maneuver.
Unable to move bodies from the washout phase, GSK instead took three suicides from the post-trial period and filed them under placebo – see Figure 1.
- After his trial was over one of the patients had been put on clotiapine, oxazepam, and fluoxetine. He went on to commit suicide, and this Prozac-treated patient was coded as a placebo suicide.
- A second killed himself on day 33 after the trial was over. The post-trial period at the end of these GSK trials was a 30-day period. This patient shouldn’t have been included.
- The third patient might have committed suicide 19 days after completing the study, but all that is filed is a brief mention that the patient’s brother called to report the death with no confirmation of cause of death. We don’t know if this patient was also put on an SSRI and then committed suicide.
A good case can be made for excluding one, or two, or all three of these suicides. But part of the mystery is this: when there was only one suicide on Paxil and none on placebo in these trials, one suicide in about 34,000 months of study, why should three turn up in this withdrawal month, all from the placebo group? It is a statistical freak.
In the face of this statistical freak why…
In the face of this freak why did the regulator (MHRA) include these suicides in the mix, helping in the process to hide the risks not just on Paxil but on all SSRIs?
Did the MHRA think that just because they had told companies not to import suicides from the washout period that this would bring an end to all efforts to find bodies elsewhere?
Why did David Gunnell, Julia Saperia, and Deborah Ashby, who produced an article for BMJ (19 February 2005) on the data from antidepressant trials and suicide, include these three placebo suicides?
When details of the suicides were drawn to their attention, why did DG/JS/DA do nothing about it?
Why did no one at BMJ spot the issue?