Map of life expectancy at birth from Global Education Project.

Friday, June 06, 2014


You'll often see people who I agree with about a whole lot of important subjects defending the importance of animal research to the advancement of medicine and the relief of human suffering and disability.

The picture most people, including scientists have of the typical course of biomedical research is that you first create an animal "model" of human disease; you study the animals to try to understand the disease better; then you test treatments in the animals. Ones that seem promising then get clinical trials in humans. According to this putative reality, If PETA gets its way, and we can't do this any more, we will never find cures for multiple sclerosis and Alzheimer's disease and pancreatic cancer and all those other horrors. It's too bad for the animals, but human welfare and survival are more important.

According to Pandora Pound and Michael Bracken, writing in BMJ, not so. In order to make the best judgments about what treatments work in humans, we do systematic reviews and meta-analyses to pull together information from all the available studies, assess its quality, and combine the highest quality information to generate conclusions and assess  how confident in them we can be. Lately, investigators have worked hard to overcome the problems of publication bias, cherry picking of results and other flaws in the scientific enterprise that often lead to overuse of treatments without good evidence. We have a long way to go, but we're getting better at it.

In animal studies, however, there is far less accountability for quality and synthesis of results. Pound and Bracken cite one survey that found only 12% of animal studies used random allocation and only 14% had blind assessment. Selective analysis and publication bias result in distortion of "entire bodies of research," according to another reviews. Meta-analyses are rare.

When treatments enter clinical trials in humans based on animal research, they are overwhelmingly likely to fail. P&B also note that decades of research have failed to yield a single treatment for stroke, ALS, and other conditions.

Not only are there problems with research quality, reporting and synthesis, but it is questionable how useful animal models are for understanding human health or predicting human responses to treatment in the first place. There are just too many differences between humans and rodents, or even other primates, for treatments to translate in most cases.

There is an enormous vested interest in animal research, and you know these claims will generate massive yelling and screaming. Note, of course, that this is unrelated to the conduct of basic research intended to understand the animals themselves. We're talking about biomedical research purported to ultimately benefit humans. That relationship is far more tenuous than most people believe, and the effort generates enormous waste and massive failure much more often than it generates anything useful. That's what Pound and Bracken are arguing. This is a debate that people need to approach with an open mind.

No comments: