|Home | About | Contact | Vitamins for Schizophrenia||
December 21, 2004
Does funding source effect research outcomes?
An analysis of the effect of funding source in randomized clinical trials of second generation antipsychotics for the treatment of schizophrenia
John H. Montgomery, Matthew Byerly, Thomas Carmody, Baitao Li,
Controlled Clinical Trials 25 (2004) 598–612
This paper attempts to address the differences between studies of the second generation antipsychotics funded by pharmaceutical companies versus studies that are funded by alternate sources (government, nonprofit foundations, etc.) When these types of analyses have been performed for medications for other diseases, the studies funded by pharmaceutical companies have been nearly universally in favor of the new medication that the sponsoring company produces. There are many possible reasons for such a finding. Pharmaceutical companies have been known to pressure investigators not to publish negative studies or have maintained the data within the company and unpublished when a study does not show favorable response. The greater desire to publish “positive” results leading to increased publication of such studies is called publication bias. This can lead to misleading positive conclusions, especially when multiple trial results are put together in “meta-analysis.” There are other reasons that negative studies do not get published. Pharmaceutical companies will often end a study if it does not appear to be working as planned or seems to involve danger to the patients. While some of these results or trial terminations become public knowledge, often they are not published. One other issue that can lead to data that favors newer agents is when a new drug is compared to an older drug in such a way as to not allow the proper usage of the older drug. For example, in some comparisons with Haldol (haloperidol) investigators are not allowed to use medications to help with side effects, such as Cogentin (benztropine). This causes the Haldol group to have an increased rate of problematic side effects unfairly.
In this study, the authors systematically reviewed the medical literature for studies that involved schizophrenia and were organized as “double-blind, controlled, randomized trials.” A “double-blind” trial is one in which neither the investigators nor the patient knows which drug (or placebo) the patient is receiving. This helps to prevent the bias that one might unconsciously have towards favoring the patients on an active drug or assuming that they should be doing better than the subjects receiving an alternative treatment. “Controlled” means that the trial had an active group that received the experimental medication and the other group received either a placebo or an established treatment (like Haldol.). Randomized means that it was equally likely that a patient ended up in either group and that something like a computer or a coin toss was responsible for making the decision and not the investigators (who might be tempted to put particular patients in particular groups.)
The authors found that indeed in their analysis, industry sponsored trials did have a higher likelihood of showing that the experimental medication would be shown more positively. However, they also looked to see what might be responsible for that. They first looked at the methods of the studies to see if the pharmaceutical sponsored studies were not done as rigorously. They found that while there was some decrease in the quality of the pharmaceutical studies, it was not very significant and not likely responsible for the discrepancy. They also looked publication bias, but also conclude that journals do not have editorial policy to favor positive studies, nor did they appear to be more likely to be accepted for publication than negative studies. However, the issue of whether researchers are less likely to submit negative trials still exists. It is hard to know for sure how many studies are never submitted. However, due to recent political pressure, pharmaceutical companies are now going to be placing all trials online so that negative and positive results will all be available to the public.
However, this study is not without its limitations. First, it is hard to say that the groups of patients that are studied in industry sponsored versus non-industry sponsored research are the same groups. For example, the non-industry sponsored studies included a much higher proportion of studies using clozapine for treatment resistant patients. These patients have already shown that other medicines did not work for them and therefore are less likely to show response to clozapine perhaps. Also, this study was not done in a blinded fashion. The authors knew the funding sources of studies prior to rating them, leaving them open to the very same risk of bias that we demand to be absent from the trials they were reviewing.
Overall, this study provides further evidence that we need to use caution when interpreting data. Industry sponsored studies are the vast majority of studies in the literature, and while they provide useful data, they do not always provide complete answers. It will help to have the online data base for making it more easy to monitor for negative outcomes, but it will always be up to the informed consumer/physician to critically read studies and ascertain levels of bias in order to put together a complete understanding of the data.
The authors note no external source of funding for this project.Posted by Jacob at December 21, 2004 11:09 PM | TrackBack