It’s an interesting petition, though I have grave doubts about whether the No. 10 petition site is of any value, after some early headlines it now seems to be ignored.
Looking at the points raised:
1. that animal researchers are trained in experimental design and statistics.
Fair enough, though most already have such training, but why just animal researchers? Surely all aspiring scientists, biologists, chemists, physicists, should get a good grounding in statistics and experimental design during their undergraduate studies, with perhaps more specific training later. A decade ago when I was an undergraduate I felt that not enough stress was placed on statistics, with units being optional and in my view not detailed enough to be really useful. Most colleagues studied the statistics in much more depth later when they started doing their own research. Perhaps things have improved since then but I suspect more could be done. Indeed given the evident confusion among politicians, media commentators and the general public where statistics are concerned perhaps this is an issue that should be addressed in part at school.
It might also be worth examining whether the pressure to limit the numbers of animals in studies is in any way responsible for some underpowered studies.
p.s. my name is not Ben Goldacre
2. that each licence application includes references to systematic reviews of existing relevant studies.
As Georgina pointed out for the vast majority of proposed animal studies there will be no systematic reviews to refer to. There may be possible exceptions in the case of some more immediately pre-clinical studies where there are a significant number of earlier animal studies, and even some preliminary human data, but in most cases animal research is done at a stage in the research process that is too early for data required for systematic reviews to be available. It is also worth noting that project licence applications must contain an account of the research that underpins the proposed study, which will of course include references to published research and unpublished data.
Currently summaries of project licences are made public, and provide enough information to avoid unnecessary duplication. It has to be remembered that some duplication is almost always a necessary to independently verify the results of a study, in most instances a scientific discovery is not considered valid until it is independently replicated.
3. that approved applications are made available to researchers and others with relevant interest
This is an idea that is fine in theory but problematic in practice. Forcing publication of detailed account of planned research on granting the licence would put UK scientists at a serious competitive disadvantage with respect to scientists in other countries (such as the USA) who have access to greater resources and have less bureaucratic regulatory systems to slow them down. Of course there are also security problems associated with making information that would allow the identification of individual researchers freely available.
4. that only the highest quality experimental designs of animal studies and laboratory practice are permitted.
Hardly a groundbreaking comment!
Several signatories to the petition have participated in some recent systematic reviews of animal research; these include Perel, Sandercock and Pound. Although critical of the standard of methodology in many cases, these reviews did not suggest that the animal studies which they looked at were irrelevant.
Systematic reviews of clinical trials are normally undertaken by clinical experts in the subject being reviewed. In the case of systematic reviews of animal research it would make sense for experts in animal research to undertake the reviews.
In systematic reviews making direct comparisons between human and animal medical research, it would also be sensible to have a balance of experts from both fields, to protect against any possible bias of one field against the other.
I’d certainly agree that systematic reviews are very useful when evaluating and designing clinical trials, and would benefit some pre-clinical evaluations of new treatments in animals. In these cases there should be a sufficient number of studies whose methodology is similar enough so that it is possible to formulate the right literature queries and design exclusion criteria to discard studies that are of insufficient quality or would be expected to bias the review.
I don’t see how they can be successfully undertaken prior to most animal studies where evidence from a wide variety of sources, for example in vitro expression, protein interaction, clinical data and animal studies of related proteins or pathways, needs to be considered when reviewing literature for grant proposals and project licence applications. Applying some of the principles of systematic review to these literature searches might be of benefit, but care would need to be taken to avoid narrowing the search too much and thereby missing relevant information. Trying to come up with consistent exclusion criteria for data would be very difficult, and any meaningful meta-analysis next to impossible.
