Loading the dice?

Loading the dice?

Pharmaceutical companies sponsor trials of already approved products to demonstrate their superiority to other therapies. Somehow the sponsor’s product tends to look good.

From the Washington Post (Comparison of Schizophrenia Drugs Often Favors Firm Sponsoring Study)

Pharmaceutical giant Eli Lilly and Co. recently funded five studies that compared its antipsychotic drug Zyprexa with Risperdal, a competing drug made by Janssen. All five showed Zyprexa was superior in treating schizophrenia.

But when Janssen sponsored its own studies comparing the two drugs, Risperdal came out ahead in two out of three.

In fact, when psychiatrist John Davis analyzed every publicly available trial funded by the pharmaceutical industry pitting six new antipsychotic drugs against one another, nine in 10 showed that the best drug was the one made by the company funding the study.

Why might this be? One possibility is that the companies only publicize studies that come out in their favor. If so we may soon gain a better perspective now that the industry has agreed to make public almost all trials. (Read the press releases here.) Another possibility is that sponsors introduce just enough bias into their study designs –e.g., by selecting certain patient sub-populations or comparing a more effective dose of their own product against a weaker dose of a competitor’s– in order to influence the results.

Probably both things are occurring. If so, the impact of greater transparency might be to increase sponsors’ use of “exploratory studies.” These studies, which help establish hypotheses for broader testing, are excluded from the industry’s commitment to fuller transparency. That means increased scrutiny of study design and the funding of comparison trials by neutral parties remain vital.

April 14, 2006

2 thoughts on “Loading the dice?”

  1. I did some work for some of the reported studies — you’ve left off another reason that a sponsor might show more efficacy for its trials. As the WaPo article (nicely done, IMHO) indicates, many of these trials do not study the broad spectrum of schizophrenia, and very few study effectiveness “in the wild.” Instead, they are designed to study a narrow range of disease symptoms or adverse effects — that range that the specific drug is designed to treat, or those effects where there is differentiation. No surprise, then, that the studies generally turn out well for the sponsor.

    As the article mentions, in the field of CNS drugs, neurologists have almost too many choices. For schizophrenia, there are 4 major atypical neuroleptics (5 counting clozapine), along with a large set of older drugs; for epilepsy, there are an almost uncountable set of drugs. There is an even larger set of off-label uses…

    Many neurologists have one or two “favorite” drugs, which differ from neurologist to neurologist. Where are the studies establishing the evidence-based approach for pharmacotherapy?

    The same thing holds for the more everyday drugs — only Kaiser took the effort to compare generic formulations of statins to brand-name statins.

  2. As a person who has been a study coordinator for over a decade in schizophrenia, the previous commenter is correct about the rigid inclusion criteria for the studies as a way of stacking the decks that a study will produce the favorable results. This can be as specific as a person must have at least moderate severity of one symptom domain and mild or absent in others. The effect is that you are increasing the chance you are gathering a specific group of patients with very limited co-morbitidities. One argument for the utility of these restricted types of studies is that it helps to identify/predict who will respond to a certain medications.

    However in all reality, individuals with schizophrenia, one of the most severe forms of mental illness, get seen by prescribers once every three months at best in the public mental health system and that is for a 15 minute med check. The prescribers simply do not have the time to conduct extensive, objective symptom assessments similar to those in clinical trials.

    However, I must admit, these studies do make pretty slides to cover at the CME presentations where conversations are limited and slides that show the inclusion/exclusion criteria and group demographic are rarely shown.

    Advice to prescribers: always ask about the inclusion and exclusion criteria and the differences in the groups. This will give you a good indication if the information is applicable to your practice.

Leave a Reply

Your email address will not be published. Required fields are marked *