When Using Animal Tests or Alternative Methods, We are Also Testing Ourselves

and application of the results of toxicity tests, which remains unresolved today. It is this. The correlation between the effects measured in a supposedly useful test with a model system (such as an animal or an in vitro system) and the likely effects in the object of interest (such as a humanbeing) must fall between 50% (the equivalent of tossing a coin, so a relationship below this would be useless) and 100% (which is unattainable, given the inevitably of significant differences between the test system and the object system, and variation in the responses in both of them). Bearing this in mind, how do we interpret and apply the result of an individual test? If, say, the correlation were 70%, which would be more than it is usually possible to obtain, how could we know, when we tested the next drug or chemical, whether the result was in the 70% agreement category or in the 30% disagreement category? This is an important question, as the answer to it has many implications of commercial, social and political concern, including: whether a chemical should be used and, if so, in what kinds of products and circumstances, and with what precautions, and whether a drug is likely to be sufficiently efficacious and sufficiently lacking in side-effects to be acceptable for use in treating patients. The central problem is that, while administrators, company lawyers, politicians, etc., and the general public, want clear yes-or-no answers on whether something is “safe” to use, science and testing cannot provide such certainty, but can only hope to make a reasonable judgement about the probability that it will be sufficiently safe. The conclusion that is eventually reached will usually involve consideration of the results of a number of different tests, along with many other kinds of available information, in a weight-of-evidence (WoE) approach. It also involves risk assessment and risk management, cost–benefit analysis, and, particularly in the case of medicines, postmarketing surveillance. This WoE approach is unavoidable, but it involves the temptation to believe that the combined consideration of uncertainties will lead to greater overall certainty. This is a somewhat biased description of meta-analysis, an approach used in evidence-based medicine for combining the findings from independent studies, in order, for example, to obtain a precise estimate of treatment effect, taking into account the outcomes of a series of clinical trials. However, as is emphasised by Crombie and Davies in a What is...? comment,1 “the validity of the meta-analysis depends on the quality of the systematic review on which it is based”. For example, the relevance and reliability of the individual studies must be considered, and a balanced assessment of their deficiencies must be made, as must decisions about the relative weights to be given to them in the meta-analysis. These are very important points, since, despite many grand attempts to make WoE reviews impersonal, objective and subject to rules, they inevitably involve strong subjective elements. I was reminded of this while watching an episode of Endeavour, a TV series about the young detective, Morse, who was the main character in a highlysuccessful earlier series about later stages in his career. Morse turns to his boss and says something like this ― “How you look at something determines what you see”. This reminded me of a famous quotation by Anaïs Nin, “We don’t see things as they are, we see things as we are”, also put like this in the Talmud: “We see the world, not as it is, but as we are”. In other words, the outcome of any evaluation is influenced by the attitudes, biases, characters, culture, experience, expertise, responsibilities, vested interests, etc., of the evaluators themselves. In the case of animal tests, one kind of evaluator will tend to concentrate on the benefits to humans, to their companies and to science, while another kind of evaluator will focus on the costs to the animals, their suffering, and their deaths. I mentioned this to one of my economist sons, who said that this reminded him of what is known as behavioural economics, which involves “the effects of social cognitive and emotional factors on the economic decisions of individuals and institutions and the consequences for market prices, returns and the resource allocations”.2 This can involve heuristics, when “people often make decisions based on approximate rules of thumb and not strict logic”, and framing, “the collection of anecdotes and stereotypes that make up the mental emotional filters individuals rely on to understand and respond to events”.2 ATLA 41, 141–143, 2013 141