More myths busted
At UAR, we do a lot of myth-busting which is a bit of a shame, because it limits the opportunity to have a more nuanced conversation about the strengths and weaknesses of animal research. Worse still, organisations which are potentially well-placed to contribute to this can choose instead to generate myths which we’re then obliged to unpick.
The website of the Safer Medicines campaign is a case in point. It states that it is
“an independent patient safety organisation of scientists and doctors with extensive expertise in drug development. Our aim is to change the way medicines are tested, so they are safer for humans: moving from a system based mainly on tests on animals to one focused firmly on human biology.”
By and large this might appear to be a laudable aim, and has the potential to contribute to an honest conversation about the strengths and limitations of animal models. Unfortunately, Safer Medicines goes a bit far in making some ropey claims which don’t really stand up to scrutiny, not least that animal models make medicines less safe for the general public. Their FAQ page, widely shared by activists on comment boards and on social media, states
“Actually, drugs would be safer than they are now if the animal testing phase was eliminated.”
This is of course patently untrue. Each drug, following preclinical testing using cell cultures, computer models and animal trials, will go on to three stages of human clinical trials, in healthy volunteers, small numbers of patients with the condition in question and finally a much larger trial in patients, before being introduced to the wider public via hospital consultants, pharmacists and GPs. By the time an individual comes to take a new drug, it will typically have been taken by thousands of people. All removing the animal stage would do, would be to make it more likely that a human volunteer would be harmed in a stage 1 clinical trial.
Safer Medicines also claims that:
“Many studies have shown that animals predict correctly for humans less than 50% of the time: worse than tossing a coin!(31)”
The references given for this statement are:
· Perel, P and colleagues. British Medical Journal (2007) 27: 197–200;
· Hackam, DG and Redelmeier DA. Journal of the American Medical Association (2006) 296: 1731–1732;
· Bailey, J. Biogenic Amines, vol.19, N° 2, pp 97-146, May 2005.
Let us take a closer look at these references. The Perel paper can be found here http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1781970/ and in it he says that:
“It would be inappropriate to make general statements about the utility of animal research on the basis of only six interventions.”
Yet this is exactly what Safer Medicines has done.
The second reference (Hackam et al) is a review looking at highly cited papers in leading journals and comparing data, where available, to human outcomes. The authors discovered just 79 papers which they considered to be relevant and methodologically sound, so are working with an artificially constrained sample size, but do not give a list these papers so it is impossible to weigh the quality of their conclusions.
The last reference in the Journal of Biogenic Amines is written by three anti-research campaigners and regards the narrow field of birth defects induced by maternal exposure to exogenous agents during pregnancy. Looking at the paper, the data does not seem to support the abstract’s claim of a 50% correlation. Table 3, for instance, shows a mean 28.7% of cases where animal effects didn’t correlate to humans is recorded. 100% minus 28.7% is 71.3%, not 50%.
Interestingly, the data given by the same author (Bailey) in subsequent studies (An Analysis of the Use of Dogs in Predicting Human Toxicology and Drug Safety, ATLA 41, 335–350, 2013 and An Analysis of the Use of Animal Models in Predicting Human Toxicology and Drug Safety ATLA 42, 181–199, 2014), when run through standard Bayesian modelling give the following positive (PPV) and negative (NPV) prediction values:
· Dogs: PPV median = 70%, NPV median = 94%.
· Rats: PPV median = 90.5%, NPV median = 97.9%.
· Mice: PPV median = 92.9%, NPV median = 95.6%.
· Rabbits: PPV median = 98.8%, NPV median = 99.2%.
This is before combining second-species data and further statistical modelling pushes confidence beyond the 90% mark. In this context it seems unreasonable to cite an animal to human translation rate of 50% or less.
Next is one of the most insidious claims:
“More than 10,000 people are killed every year in the UK by side effects of prescription medicines (32)"
The implication being that death by side-effect is due to these side-effects not being revealed during animal testing. The claim cites this paper http://www.bmj.com/content/329/7456/15.long looking at hospital admissions.
However the authors of this paper write that “Most reactions were either definitely or possibly avoidable. Drugs most commonly implicated in causing these admissions included low dose aspirin, diuretics, warfarin, and non-steroidal anti-inflammatory drugs other than aspirin, the most common reaction being gastrointestinal bleeding.” So, in most cases these were known effects. Aspirin accounted for 61% of fatalities.
The authors further note that “Nevertheless, it is impossible to be absolutely certain of a causal link between a drug and an ADR (adverse reaction). For example, with low dose aspirin, up to half of the cases of bleeding may have occurred anyway, irrespective of aspirin use.”
Most importantly, the Safer Medicines quote implies that the ADRs were an unknown side-effect, experienced in humans following general release, when in fact these are known effects which are listed as possibilities in the warning booklet that comes with every medicine. Again, the drug would have undergone human trials following animal trials and be released with known side-effects.
Ironically, Aspirin is one of the few drugs that was not tested in animals because it was already in use – it fulfills Safer Medicines’ dream of an animal-free drug, but causes the majority of the fatal ADRs Safer Medicines warns about on its website!
Elsewhere, Safer Medicines raises an old and misleading statistic:
92% of new drugs fail in clinical trials, after they have passed all the safety tests in animals (36)
The statistic is from the FDA (Food and Drug Administration), used to illustrate inefficiencies in drug development and has been thoroughly debunked by a Fellow of the Royal Society here. The FDA analysis essentially refers to all preclinical testing methods including human tissue samples, DNA chips and computer models, not just animal testing.
There are a number of reasons for this rate of failure, including drugs that have been proved safe using animal models then proving to be ineffective at a safe dose in humans, trials being halted because competitors have already brought a product to market or, rarely, a side-effect manifesting that was not predicted by the animal models. Many drugs will have some potential adverse effects- consequently the decision to use them is a balance of the costs and benefits.
It is worth noting that around 90% of drugs are removed at every stage of safety tests, i.e. 90% are removed at non-animal pre-clinical safety tests, 90% at animal stages, and 90% during human clinical trials. In fact, working backwards, if we have 1000 drugs entering animal safety tests, 900 of them fail, of which 20 might be safe in humans. Of the remaining 100, 92 fail human tests, therefore: 90.5% of dangerous drugs have been kept out of clinical trials thanks to animal safety tests.
Finally, Safer Medicines claims that:
Computers enable scientists to design and test new medicines. Though they may never be able to replicate a human patient completely, they will always reap more accurate findings than an entirely different species.
This claim is not referenced, but is completely wrong. It cannot be said that a computer will always reap more accurate findings, not least because a computer requires pre-existing knowledge in order to model accurately. In other words, you cannot model what you do not yet understand. Limitations in the understanding of molecular dynamics and cell biology as well as the absence of available computer processing power force large simplifying assumptions that constrain the usefulness of present computer (in silico) models.
So Safer Medicines have published a number of false statements demonstrating a position that tries to undermine the validity of animal testing, rather than entering into a critique of animal testing.
Where I’d love the public debate to be on this issue is having a nuanced discussion about the strengths and limitations of animal models, and the difficulties facing ethics committees which have to decide whether or not experiments with animals get a licence.
Furthermore, we need to be focused on creating and validating alternatives, and encouraging their uptake internationally.
Instead we are spending too much time unpicking false or misleading claims from activists. I am sure there are some who think it’s a good idea to do ‘whatever it takes’ to force this issue, but it’s an issue which cannot be forced.
Regulators are not going to move away from requiring safety testing in animal models on the basis of misrepresented scientific papers, dodgy statistics and made-up technology, even if it is enough to fool activists and mislead the public into repeating these false claims on social media.