Article information and data provided by NCBI

Quantitative analysis of sponsorship bias in economic studies of antidepressants.

by Baker CB, Johnsrud MT, Crismon ML, Rosenheck RA, Woods SW
The British journal of psychiatry : the journal of mental science.

Article Abstract:

BACKGROUND: Concern is widespread about potential sponsorship influence on research, especially in pharmacoeconomic studies. Quantitative analysis of possible bias in such studies is limited. AIMS: To determine whether there is an association between sponsorship and quantitative outcomes in pharmacoeconomic studies of antidepressants. METHOD: Using all identifiable articles with original comparative quantitative cost or cost-effectiveness outcomes for antidepressants, we performed contingency table analyses of study sponsorship and design v. study outcome. RESULTS: Studies sponsored by selective serotonin reuptake inhibitor (SSRI) manufacturers favoured SSRIs over tricyclic antidepressants more than non-industry-sponsored studies. Studies sponsored by manufacturers of newer antidepressants favoured these drugs more than did non-industry-sponsored studies. Among industry-sponsored studies, modelling studies favoured the sponsor's drug more than did administrative studies. Industry-sponsored modelling studies were more favourable to industry than were non-industry-sponsored ones. CONCLUSIONS: Pharmacoeconomic studies of antidepressants reveal clear associations of study sponsorship with quantitative outcome.

Eye opening paper - a must read!

By: Anonymous - Fri 11/25/2005 PM

There are so many things to say about this paper. First off, it is clear that $ speaks. What is not immediately clear is how insidious this fact is. For example: Did you ever think about the fact that PubMed does not index "Declaration of Interest" as a search term?? That is to say, nearly every high quality journal has a section for conflicts of interest. However, it is not currently possible to search PubMed for a study WITHOUT a reported conflict of interest (non-industry sponsored) This boggled my mind - and I wrote to the national library of medicine asking why this was so and received this reply: "Reply from the National Library of Medicine: PubMed/Medline does not contain the the information you are trying to search on. If the principle researcher's address or affiliation is a drug company than it will be listed and is searchable. Affiliation [AD] may include the institutional affiliation and address (including e-mail address) of the first author of the article as it appears in the journal. This field can be used to search for work done at specific institutions (e.g., Cleveland [ad] AND clinic [ad]). The only grant information capture on Pubmmed is Grant Number [GR] Research grant numbers, contract numbers, or both that designate financial support by any agency of the US PHS (Public Health Service). The three pieces of the grant number (LM05545 - number, LM - acronym, and NLM - institute mnemonic) are each individually searchable using the [gr] tag. They do not capture drug company grant information. PubMed/Medline does not capture conflict of interest informaiton either because it is not usually available in the literature. We do not know of any source that will provide you with this information. Customer Service National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 custserv@nlm.nih.gov"

Secondly, the questions asked were clever: "Is there an association between industry v. non-industry sponsorship of studies and quantitative conclusions?" and "Among industry sponsored studies and between industry sponsored v. non-industry-sponsored studies, is there an association between study design and quantitative conclusions?" However, there are LOTS more questions that could be asked -- for example: In a given journal how many advertisements are displayed for a given company relative to the same company's sponsored study with positive results in a given journal? Third, I would like to mention an important point not directly restated in the conclusion. "[Bias] was true whether industry sponsorship was defined as industry authorship, industry financial support alone, or both." That is to say, take all disclosures very seriously - even if it seems as if only a grant was offered to a researcher. The final point that I think is very important is related to the discussion on the causes of bias: "Industry motivated to enhance sales of its products, might only pursue studies on products and select comparators that would yield favorable results.... biased populations... influence analytical methods or models... veto submissions yielding unfavorable results... " Just some things to keep in mind with EVERY paper you read! JournalReview.org is the perfect forum to address these concerns.

[ Discuss Article ] [ Rate Article ]