EUSAAT 2022 in Linz on 26 September

 Jan Lauwereyns will be speaking in Hall 3 from 14:00:

Tracking the Validity of Animal Models for Biomedical Research

In the more than six decades since Russell and Burch formulated the 3Rs, surprisingly little progress has been made in the scientific investigation of the validity of animal models for biomedical research. The issue of external validity (i.e., how relevant is the animal model?) is often ignored altogether, with researchers choosing animal models out of habit or on the basis of unchecked assumptions [1]. Worse, the issue of internal validity (i.e., how reliable or reproducible are the findings?) has placed animal-based biomedical research in a negative spotlight with the general public [2]. 

While most animal researchers may personally be committed to improving their research practices, the micro-motives of well-meaning individuals do not guarantee optimal macro-behavior at the aggregate level [3]. As has been described in behavioral economics, a minority of defectors can disrupt the development of cooperative dynamics. In the present context, defectors would be researchers who persist in substandard research practices, even if this is not out of malice or incompetence. All that is required for the “survival of the un-fittest,” or the continuation of substandard research practices, is a research culture with perverse incentives (e.g., short-term gains in publications).  

The issues can be vividly illustrated with the use of non-human primates during the COVID-19 pandemic. For instance, “[p]rimate researchers in the United States have banded together for an ambitious monkey study that would do head-to-head comparisons of the leading COVID-19 vaccine candidates” [4], while researchers in China reported that “primary exposure to SARS-CoV-2 protects against reinfection in rhesus macaques” [5], obtaining non-significant results from a complex study with a sample of seven, which proved to be entirely irrelevant, if not plain wrong. The banding together of primate researchers in the United States and the non-significant data from China were both reported in Science, arguably the pinnacle of scientific publishing, although anyone with a modicum of scientific reasoning could easily have shown that the primate researchers in the United States engaged in self-serving research bias that ignored the question of external validity, and that the sample of seven rhesus macaques in China did not approach anything like the sample size required for the given research design (as could have been assessed with readily available tools for power analysis, such as G*Power). 

The solutions are technically not difficult, but they require a sea change in research culture. In this respect, animal researchers can turn to the human behavioral sciences for a paradigm of a neighboring research field that is making great strides toward the required change in research culture. What is needed is a shift to open science, driven by leading journals and researchers, spreading the guidelines and facilitating the implementation of good practices. These rest on the twin pillars of the pre-registration of experiments and the use of power analysis to compute sample size. These twin pillars can effectively redress the issue of internal validity. 

For the external validity, the challenges for animal research are arguably unique. Previous efforts to compare the validity of animal models directly (e.g., [6]) have typically been limited to qualitative, categorical assessment. Moreover, the most important type of replacement – as according to Alexander Pope’s verse, “The proper study of mankind is man” – is all too easily forgotten. Here, I would propose that, before engaging in any animal research, researchers should establish the conditions under which it is impossible to work with human volunteers as the preferred animal model. Indeed, even for the most urgent issues or risky experiments, we may still find human volunteers. We just need proper guidelines and broad societal support (e.g., [7]). Then, to the extent that there are still types of research that require the use of animals, it is imperative that we track their validity – this means openly sharing and compiling all the data, and regulating the protocols, not just in academia, but also in industry. Sharing the data, and following academia- and industry-wide protocols, should be the entry point for a license to use animals for research. 


[1] Pound, P., Ritskes-Hoitinga, M. (2018). J. Transl. Med. 16, 304. doi:10.1186/s12967-018-1678-1 

[2] Dirnagl, U., Duda, G.N., Grainger, W.L. et al (2022). Adv. Drug Deliv. Rev. 182, 114118. doi:10.1016/j.addr.2022.114118 

[3] Lauwereyns, J. (2018). Rethinking the Three R’s in Animal Research. Palgrave: Pivot. 

[4] Cohen, J. (2020) Science 370, 154-155. doi:10.1126/science.370.6513.154 

[5] Deng, W., Bao, L., Liu, J. et al. (2020). Science 369, 818-823. doi:10.1126/science.abc5343 

[6] Lakdawala, S.S., Menachery, V.D. (2020). Science 368, 942-943. doi:10.1126/science.abc6141 

[7] Shah, S.K., Miller, F.G., Darton, T.C. et al. (2020) Science 368, 832-834. doi:10.1126/science.abc1076