The animal model of medical research: scientifically flawed

“A little learning is a dangerous thing. Drink deep, or taste not the Pierian spring.” - Alexander Pope, An Essay on Criticism

This short piece is intended firstly as a response to a piece by Granville Matheson (, itself written as a response to the World Day for Animals In Laboratories protest, and secondly as a general discussion around research using the animal model. Herein, I shall be relying solely on arguments from science / philosophy of science. A discussion from an ethical perspective is not required, especially as Granville implies agreement in his comments below his post as well as in statements like, ‘institutions killing animals unnecessarily deserve to be put down themselves’.

Here goes….

Firstly, I should define what the term ‘animal model’ in fact means.

There are two main ways in which animals are used by medicine, biology and biomedical research:

  • As models for human disease
  • As test subjects (e.g for drug and carcinogen testing)

There are also various other uses of animals in medical science that we will ignore as they are tangential to this discussion. These include, amongst others, use as spare parts (e.g heart valves), as factories (for, eg., insulin) and as tools for education.

So, why do we use animals as a model for humans in order to understand human diseases and test drugs intended to treat humans? In any field of science, a useful model or theory has to fit several criteria (including repeatability, falsifiability, etc.) The most important criterion, however, is predictability. Models are an application of axiomatic-deductive reasoning to predict certain outcomes. If models did not reliably predict that which they were meant to, they would be almost entirely useless. Using entrails to see into the future is an example of a model that is not predictive. Using complexity theory to map the climate is an example of a model that has proven predictive. We use the animal model because we think it offers predictive value. Technically, we call it a causal analogical model.

However, the animal model is not reliably predictive. While this might come as a surprise, considering the vast number of experiments we have performed on animals over the last 2 000 years and the degree to which we seem to rely on them for medical progress, there is a growing body of data to support this conclusion and, more importantly, some very simple logical arguments which remove some of the criteria necessary for causal analogy to be made (e.g, there should be no causally relevant dis-analogies between the model and the thing being modelled).

Why is the animal model not predictive? The primary reason is that the important differences between species are not simple quantitative ones (e.g genetically we’re 99% similar to chimpanzees), but contextual ones. Even if we share large parts of our genome with various animals (and plants – we’re around 50% banana!), where all the important (structural) genes sit relative to each other and relative to regulatory genes (genes that control pattern and growth by ‘switching’ structural genes on and off) makes all the difference in how or if they are expressed. The assumption that the animal model is predictive, then, relies on the erroneous assumption that differences in genetic makeup are measured quantitatively when, in fact, modern genetics tells us that small qualitative / contextual measures are what make a difference. For instance, a single amino acid difference is responsible for which cell receptors the HIV virus binds to in humans as distinct to which it binds to in other primates. A single amino acid difference is also responsible for cystic fibrosis and sickle-cell anemia.

But then why do we have drugs that are effective in humans that were tested on animals and had similar effects? Surely this indicates that the animal model is accurate SOME of the time?

No. What it indicates is that we can know which drugs had a similar effect on animals and humans (8-20% of the time, according to various metastudies) ONLY once we’ve tested them on humans. In other words, we know *retrospectively*. A model that only allows us to know something retrospectively is NOT a predictive model and thus has no value.

Substance A is tested on two different animals. It is ineffective on animal A and effective on animal B. Substance B is tested on two different animals. It is ineffective on animal B and effective on animal A. Which substance is going to be effective in humans? Well, we need to test both in humans to know and thus the use of the animal model is pointless as the humans become the first guinea pigs anyway.

By the way, this hypothetical example is in fact real: White New Zealand rabbits suffer from phocomelia (birth defects in their young) when they’re given Thalidomide. Mice do not. Mice, however, do react just like humans to penicillin. Rabbits do not.

Think about this. How is it different from throwing a dice to decide which of a set of drugs tested on animals to give to humans?

As for using animals as a model of human disease, well, here are some choice quotes from cancer researchers on just how well the animal model has served them:

“The history of cancer research has been a history of curing cancer in the mouse. We have cured mice of cancer for decades, and it simply didn’t work in humans.” – Dr Richard Klausner, Director, National Cancer Institute

“Animal carcinogenicity tests on new drugs are inaccurate, often insensitive and generally misleading.” - Dr John Griffin, Director of the Association of the British Pharmaceutical Industry

“It’s been well known for maybe two decades that many of these preclinical human cancer models have very little predictive power in terms of how actual human beings – actual human tumours inside patients – will respond…Preclinical models of human cancer, in large part, stink…Hundreds of millions of dollars are being wasted every year by drug companies using these animal models” – Prof. Robert Weinberg, Massachusetts Institute of Technology

“Let us look at some animal carcinogens – gold, DDT, clofibrate and bromocriptine. There is no doubt that all of these can rightly be regarded as carcinogenic for rodents, and yet there is really quite good evidence that they are not carcinogenic to man.” – Prof Andre McLean, Long Term Animal Studies – Their Predictive Value for Man

“As a cancer specialist engaged in clinical practice, I can’t agree with the researchers who believe that results obtained with laboratory animals are applicable to human beings.” – Dr Heinz Oeser


Let’s now look at some of the objections Granville’s piece raises (note that I have paraphrased Granville in some instances):

- The fact that the RSPCA and FYA actually allow animal testing, but with checks, is reason enough to query this position.

Response: This is an argument from authority; the RSPCA and FYA are likely advised by the medical research establishment and, as the animal model is still popular, it is unreasonable to think that non-specialised animal welfare organisations would be up to date with the debate. In fact, all they do is employ an argument from necessity.

- A whole intact body system is necessary for some research

Response: That’s true. However, logically, a whole intact HUMAN body system is what is required given the complexities of complex systems.

“All animals are examples of robust complex systems (on many levels) and hence have emergent, modular and nonlinear properties. Perturbations in complex system S1 that lead to affect A will not necessarily lead to effect A in complex system S2, regardless of how similar the two systems are or were.

Living complex systems also manifest different responses to the same stimuli due to: 1) differences with respect to genes present; 2) differences with respect to mutations in the same gene (where one species has an ortholog of a gene found in another); 3) differences with respect to proteins and protein activity; 4) differences with respect to gene regulation; 5) differences in gene expression; 6) differences in protein-protein interactions; 7) differences in genetic networks; 8) differences with respect to organismal organizations (humans and rats may be intact systems, but may be differently intact); 9) differences in environmental exposures; and last but no leads (10 differences with respect to evolutionary histories.

These are some of the important reasons why even two nearly identical complex systems (e.g., a chimpanzee and a human or even monozygotic twins) may respond differently to drugs and experience different diseases, and hence why one complex system/species cannot reliably predict responses for a different complex system/species.

Current biomedical research is studying disease and drug response at the level where the differences between complex systems (be they two different species or two different humans) become momentous, hence using animals (e.g., vertebrates) as predictive or causal analogical models (CAMs) for human diseases and drug testing is a scientifically invalid paradigm.” – Dr Ray Greek

- We study drugs and disease in lots of different animals which affords us more predictive value.

Response: Given the Thalidomide / penicillin example above, it’s clear that ‘more’ doesn’t mean ‘more accurate’, just ‘more choices’. In fact, the Thalidomide example is especially interesting because it took researchers a very long time and many, many species before they found one (the White New Zealand rabbit, and later a certain type of monkey) for which Thalidomide was teratogenic (although why this kind of retrospective ‘proof’ demonstrates anything more than defensive posturing is beyond me….Actually, there must have been a lot of opportunities for grant money ;-) If we had tested Thalidomide on lots and lots of species beforehand (until it became unfeasibly expensive), we probably still wouldn’t have tested it on this specific rabbit or monkey and thus would STILL not have discerned its teratogenicity.

- We either use animals or humans. Ethics preclude the latter, therefore the former must persist.

Response: This is a false dilemma. Firstly, because there are many, many alternatives (see the references at the bottom of this post) and secondly because if the animal model doesn’t work, it simply doesn’t work, no matter how much we may wish it did and regardless of any dearth of suitable ‘replacements’.

- More money is spent on welfare than on experiments – we care!

Response: Nobody really thinks that animal researchers are evil, unthinking psychopaths. In fact, most are probably loving, caring individuals with families and perhaps even well-loved pets. They are simply beholden to a specific (and, I assert, erroneous) discourse of how medical science progresses that for them axiomatically marginalises full ethical consideration of animals by appeal to notions like ‘the greater good’ or ‘necessary evil’.

If research wasn’t cruel, they would not have to resort to these notions. To try make it marginally less cruel whilst still inflicting the terrible cruelties of invoking cancers and other diseases as well as testing all sorts of drugs and documenting often horrific and deadly side effects, and also subjecting every single animal to a life of imprisonment, is a flippant, token notion of welfare at best. The Peter Singer analogy (mentioned in the comments below the original post) about welfarist rape involving a glass of wine and a soft carpet is relevant here.

Besides, this welfarist rhetoric simply does not stand up to the vast amount of filmed and otherwise documented (sometimes in the published papers of the researchers themselves) information available on actual lab conditions around the world.

- We use the 3 R’s.

Response: The 3 R’s based on the erroneous assumption that the animal model is working. If it is not, applying replace, reduce, refine is like applying this practice to reading the entrails of dead animals to predict the stock market….We would want to instantly abolish the latter as soon as we noted that it lacked reliable predictive value, not replace, reduce and refine it!

- Animal research is not only important for humans either: a great deal of progress is made into veterinary products from animal testing – both from human medicine testing, and in animal medicine testing. Many medications which work in humans, also work in animals.

Response: This is no more than a side effect of animal model research, which is primarily about finding cures for human ailments, and thus offers scant defence. Medical science progresses in times of war (WWII Nazi Germany, for example), but this is hardly an adequate defence for war.

Actually veterinary science, on the other hand, is another question entirely, and one which involves an ethical as opposed to scientific discussion.

- The animal testing industry is self-sustaining.

Response: Most animal testing money comes from university grants, private donors, large pharmaceuticals and the military.

- In 2006, 2.95 million animals were tested upon in England, Scotland and Wales ( and 1,012,713 animals were tested upon in the USA. I don’t know where this result comes from, but if the world powers are using only 0.4% of this total, then I feel obliged to be sceptical.

Response: Both this number and, unfortunately, the original number of ‘billions’ cited by a commenter to the original post, are incorrect.

Here’s some well researched data:

- Not even the RSPCA believes that animal testing can be completely replaced. These methods can replace elements, but not the whole of animal testing.

Response: Why are a mainstream welfarist organisation like the RSPCA the arbiters here? Surely what the RSPCA believes about modern medical science is about as relevant as what a person working in a CD store knows about a diminished seventh?

- 61% of birth defects are caused by drugs passed safe in animal tests. 68% of drug related deaths last year were as a result of legal drugs.

This would be significantly higher if animal testing were not used. Certainly these are not arguments for the cessation of animal medicinal testing.

Response: It would only be significantly higher if animal testing were not used because there would be many more new drugs available each year. In fact, in this case, you would most likely find a very direct correlation between number of new dugs and number of adverse reactions.

- At least 50 drugs on the market cause cancer in laboratory animals. They are allowed because it is admitted the animal tests are not relevant. Not relevant? I think not. The cancer which resulted would have been studied after humanely putting down the animals (Robinson, 2005), and found to result from an enzyme or some such which is not found in humans.

Response: There’s no strong case made by either the comment or the rebuttal here. The history of curing cancer in humans via recourse to animal models has been criticised best by actual cancer researchers:

- Less than 2% of human illnesses (1.16%) are ever seen in animals.

Encyclopaedia Britannica states:

“… it is likely that, for every known human disease, an identical or similar disease exists in at least one other species.”

Response: Note the use of the term ‘similar’. Cancer in mice (transgenic ones even) is ‘similar’ to cancer in humans. We can cure cancer in mice really well….but cancer in humans works / spreads in different ways, which makes sense when you consider differences in cell biology, etc.

- When asked if they agreed that animal experiments can be misleading ‘because of anatomical and physiological differences between animals and humans’, 88% of doctors agreed.

The group whose statement was used as a reference for this claim, have subsequently made a statement: “The AFMA poll did not ask doctors directly whether or not they believed animal experiments were relevant to human medicine. A more recent poll released on October 9 2006 demonstrates that the great majority of doctors believe that it is.”

Response: What is heartening is that there’s now a near consensus that animal experiments can be misleading. A fuller discussion around scientific credibility and criteria for predictive models now has to occur. Sure, a ‘great majority’ (according to Pro-Test, most likely) of doctors continue to believe that animal experiments were (note, ‘were’, not ‘are’) relevant to human medicine, but a majority of doctors probably also still believe that meat and dairy are essential components of a healthy human diet ;-)

- Animals are less intelligent, resulting in their being ruled by humans. I feel that humans’ higher intelligence renders suffering more salient in our lives as a result. I think humans are worse creatures, but ethically, I think that animals provide a subject matter whose suffering is probably less, and can be minimised with more ease.

Response: I will be writing another piece in response to Joe Roussos’s note ( in which I shall discuss various approaches to ethics and the affordance of rights and the criteria that should be brought to bear in this regard if we are to be consistent in applying the ethical practices we have constructed. My piece will approach ethics from a post-structuralist perspective, specifically a Foucaultian one which allows for ethics (as commonly understood, not entirely as Foucault uses the term) to be viewed as one practice amongst others in concrete relationships of feedback and contingency.

- Institutions killing animals unnecessarily deserve to be put down themselves :).

Response: I couldn’t agree more :-)

I must finally note that one of the sources in the original post, Speaking of Research (, is highly questionable:

It is unfortunate, albeit obvious, that there is a fair amount of PR and misinformation around a subject that, after all, involves a lot of money and entrenched power.

Further reading:

Specious Science – C Ray Greek, MD & Jean Swingle Greek, DVM
What Will We Do If We Don’t Experiment On Animals – C Ray Greek, MD & Jean Swingle Greek, DVM
Why Animal Experiments Must Stop – Vernon Coleman
From Guinea Pig to Computer Mouse (2nd edition) – Nick Jukes & Mihnea Chiuia (available free online!