Transparency In The Research Community Saves Lives
The aim of biomedical research is to create treatment options for patients. Scientists use animal models if they can’t find another way to answer a question. When a project requires animals to be use, government agencies, universities, or funders generally ask for documentation of the need for the animals, their planned care, and how the scientists will protect their welfare. Yet these protections don’t mean animals won’t suffer — and this suffering will be in vain if the scientific community doesn’t learn about the experiments and their outcomes.
This learning takes place through myriad journals that inform the research community about new developments. Typically, once scientists complete a research project, they write up the results and send it to a journal for publication. They may need to re-submit their article to various journals until they find one that agrees to print it. However, not all studies are published, and multiple submissions are no guarantee. What seems to matter more is a statistically significant outcome. This affects the likelihood that it will be accepted by a journal and ultimately published. This is known as “publication bias” and has been a problem for several decades.
Even if a study is never published, or isn’t yet published, there is another way other researchers can find out about it. Experiments can be preregistered. Before a project begins, details about the planned protocols are posted to an online platform. However, while clinical trials are more standardized and often required preregistration, this is not the case for preclinical trials.
It’s not known what proportion of preclinical biomedical research projects involving animals never find a publisher. To calculate this failure rate, researchers gathered data using a set of 67 animal study protocols from the University Medical Center Utrecht in the Netherlands, to quantify the number of experiments which led to a publication. They allowed for a follow up period of seven years since it takes time after a study is complete to place it in a journal.
Analysis of the data showed that 30 full text papers and 40 conference abstracts resulted from the 67 applications. This equates to a publication rate of 46% for full articles and 60% with abstracts included. These studies used a total of 5,590 animals. Publications mentioned animal lives only 26% of the time. Smaller animals such as mice and rats were less likely to be mentioned than larger animals. The average time from project approval to publication was 30.7 months. Failure to publish resulted most often for three reasons. Projects lacked a statistically significant finding, they were pilots for another study, or there were technical problems with the animal model.
The authors plead with the scientific community to preregister planned preclinical research projects. To that end, they have created a new, international website for such registrations, with the belief that preregistration should reduce reporting and publication bias in preclinical research. Also, researchers may be less likely to duplicate experiments if they can see what’s already been done, even if the results were never published. They can find others working on similar topics or who have experience with similar animal models. Prospective registration can improve study design and create transparency between the planned experiment and the results. As animal advocates, we would prefer animals not be used in research, but anything that can reduce animal suffering is a step forward.
https://openscience.bmj.com/content/4/1/e100051