When most people learn about the results of clinical trials for new medicines, it's either because something very good or very bad happened, according to a study reported in the Pittsburgh Post-Gazette.

"There seems to be two narratives people can handle: Experiments are dangerous and you shouldn't take part, or experimentation is the engine for the next great wonder drug," said Alex John London, an associate professor of philosophy at Carnegie Mellon University who is the co-author of an article in the current issue of the journal PLoS Medicine that looks at this and related issues with clinical trials.

That's important, London said, because if people believe there are only two possible outcomes at either extreme, they might not be willing to participate in clinical studies.

London and his colleague, Jonathan Kimmelman, an associate professor in the biomedical ethics unit at McGill University in Montreal, co-authored the article titled, "Predicting Harms and Benefits in Translational Trials: Ethics, Evidence and Uncertainty."

While arguing that expectations of clinical trials need to be more realistic and informed, the authors also believe trials need to be proposed, and run, more rigorously.

The article deals with problems in both pre-clinical animal-based trials, and the transitional, first-in-human clinical trials in which so many proposed drugs fall apart.

Kimmelman said, "only rarely do the effects seen in animal studies translate into human studies."  Even though researchers regularly find success with a proposed drug during research on animals, they infrequently achieve similar success with human test subjects.

"What we're really trying to do here is check the expectations associated with major clinical findings," said Kimmelman. That has important implications not only for possible human test subjects—who need to better understand potential outcomes—but for organizations that provide research funding, London said.

Still, they believe clinical trial outcomes might be improved were it not for two main problems they have found. The first is that most animal studies don't use the same generally accepted methods used in human trials—such as randomization and blinded outcome assessment—to prevent researcher bias.

One recent analysis of animal studies found that only 12% used random allocation and only 14% used blinded outcome assessment.  These are important because "researchers have an inherent bias in wanting to see a positive outcome,” Kimmelman said.

The second problem is that they believe researchers don't look widely enough at other related research in making their predictions about the possible outcome once they begin human trials. Most researchers, they said, when predicting the outcome of their study, only consider other studies involving the particular agent, or drug, they are testing. They don't look at agents that might be different but that work on the same pathway. London and Kimmelman refer to this process as "evidential conservatism."

Howard Mann, a program associate in the Division of Medical Ethics at the University of Utah School of Medicine, said this point is troubling. "What should be alarming is the persistent conduct of pre-clinical research that will not yield clinically relevant information because researchers are not aware of, or have not made sufficient attempts to identify, the negative results of prior completed relevant research," Mann said.

Ultimately, London said, "We want to make sure we design our trials as well as possible so we can learn from them—even if it is a failure in outcome. It's not that we learn from every failure, but it's that when they are well designed, we can learn much more from them."