June 5, 2007
In head-to-head trials of two drugs, the one deemed better appears to depend largely on who is funding the study, according to an analysis of nearly 200 statin-drug comparisons carried out between 1999 and 2005.
UCSF researchers examined 192 published results of trials comparing one cholesterol-lowering statin drug to another, or to a non-statin drug.
Their findings found that two links stood out. If the reported results favored the test drug, the trial was about 20 times more likely to be funded by the maker of the statin rather than the comparison drug company. Even more striking, they say, if the conclusions or interpretation of the drug trial -- which reflect the impressions of the trial investigators -- favored the test drug, the trial was about 35 times more likely to be funded by the maker of that drug rather than the comparison drug.
The results of the new analysis are reported in the June 7 online edition of the journal "PLos Medicine."
"Many people are concerned about the growing proportion of drug trials funded by the drug's manufacturers," says Lisa Bero, PhD, UCSF professor of clinical pharmacy and health policy studies. "Results of drug trials affect what drugs are covered by medical plans, and so what drugs physicians will prescribe. If drug trial outcomes are largely determined by who pays for the trial, we don't really know what the best drug is." Bero is senior author on the PLoS paper.
The UCSF study examined the links between reported outcomes of the statin trials and many factors, including study design, sample size, thoroughness and type of analysis, as well as funding source. They examined only published randomized controlled trials. The trials involved seven different statins overall, all studied in head-to-head drug comparisons.
The analysis found that about half of the trials were funded by industry, and about a third did not disclose any funding source. Among those declaring industry funding, about one fifth explained the role of the sponsor, such as data analysis, or writing and preparing the manuscript. Trials with no disclosed funding sources were less likely to have conclusions favoring the test drug, compared to trials with industry funding, the researchers report.
The researchers note that a number of factors can result in the drug trial results favoring the trial drug's sponsor. Drug companies could selectively fund trials on drugs that are likely to produce a statistically significant result, the researchers explain. This can be accomplished, they say, by selecting non-equivalent doses of drugs for testing. Also, sponsors may choose not to report results that don't favor the drugs they sell. Or, they may report positive results in more than one journal, skewing the number of positive articles about their drug.
In addition, almost half of the trials lacked adequate blinding -- assuring that study scientists don't know which drug the patients were taking until the end of the trial. Blinding is considered of paramount importance in clinical trials. The researchers found that those studies with adequate blinding were less likely to report results favoring the test drug. This finding was independent of who funded the study -- in other words, funding source was a stronger predictor of outcome than blinding, but both had independent effects on outcome.
The most important weakness found in most of the trials was a lack of clinical measures of outcome, such as heart attacks or mortality -- considered better indicators in trial design than less direct measures such as lipid levels.
"The lack of true clinical outcome measures in these direct head-to-head comparisons of drugs is disappointing because the studies don't give us the best information we need to choose one statin over another," Bero says.
The analysis is one of the first large studies examining the influence of funding source on the outcomes of head-to-head drug comparisons, rather than comparing the effectiveness of one statin with no drug at all. The market for statins is competitive, so it is important to have valid information to choose one statin versus another, Bero explained. For policy makers, the relevant choice is not to select a statin versus a placebo, but to select one statin compared to another.
The study also differed from most other assessments of influences on drug trials by examining 11 different factors, and how they may interact to affect trial results. Most previous studies examined the link of results to one factor alone, such as funding source. But this analysis adjusted for "confounders" -- study aspects that can influence the results, such as study design, including randomization, blinding, sample size, even choice of comparison drug in the study.
Inclusion of the confounders still pointed to industry sponsorship as the most influential factor related to positive results and conclusions.
The study examined trials by all funding sources, as well as a subset of studies that were only industry funded. Favorable results and conclusions were associated not so much with industry sponsorship, but with the specific company that funded the study, Bero points out.
"The data available on choosing between statins based on head-to-head drug comparisons appears to be influenced by financial conflicts of interest," Bero concludes. "So decision makers -- those choosing drugs for a formulary or insurance plan -- should be skeptical about these kinds of trials. We need to know if a newer, more expensive drug is really better compared to older, less expensive drugs."
Co-authors of the study are Peter Bacchetti, PhD, professor of epidemiology, and Kirby Lee, PharmD, assistant professor of clinical pharmacy, both of UCSF; and Fieke Oostvogel, University of Leiden, Netherlands.
The research was funded by a California Tobacco Related Disease Research Program grant.
Note: This story has been adapted from a news release issued by University of California - San Francisco.