testing must be of the highest quality
experiments have an important role, but much research is biased and substandard.
schoolboy, I hated biology for one reason: dissection. I found the smells
and sights so repulsive that I vowed to give the subject up, which I did.
like so many people, my head rules my heart when it comes to vivisection.
It disgusts me, yet I am convinced that animal experiments have an important
role in medical research – unlike the extremists who have threatened
children, disinterred relatives and resorted to fire bombs and death threats
to bring them to an end.
I had always assumed that any scientist prepared to endure such opprobrium,
and to wade through the red tape that accompanies animal experiments, would
make sure that the quality of the work was high enough to make it all worthwhile.
But it is with a mounting sense of disbelief that in recent months I have
seen evidence that much animal research is biased and substandard.
days ago, in New Scientist, we reported a study by Dr Malcolm Macleod, a
neurologist at the University of Edinburgh, who analysed animal studies
in stroke medicine. Having reviewed more than 1,300 experiments in 525 papers,
his team estimated that one in six animal research studies is never published,
most likely because they described boring or neutral phenomena.
analysis suggested that this total did not include another 200 unpublished
trials, which failed to show that a new treatment had any beneficial effect:
a bias that could mean that we exaggerate the benefits of new drugs by up
to 30 per cent.
work by Dr Macleod's group has suggested that these factors may be even
more pronounced for drugs to treat other diseases, such as multiple sclerosis.
This may explain why a remarkably high number – 99 per cent in the
case of drugs for stroke – perform poorly when given to humans instead
of animals. But
it's not just this bias that is worrying, but the methodology that is being
used. Last year, the National Centre for the Replacement, Refinement and
Reduction of Animals in Research (known as the NC3Rs) did a systematic survey
of animal research. Only one in eight of the studies reported used randomisation.
But if you are comparing a drug with a placebo, for example, and you don't
randomly assign animals to each group, then it's possible you might unconsciously
use the weaker, easy-to-catch animals in the placebo group and stronger,
harder to catch animals in the drug group.
only one in seven was a "blind" test, with the experimenters not
knowing which animals had been given the drug and which had a placebo. Four
per cent of the papers didn't mention how many animals were used in an experiment,
and not one of the 48 studies that were examined more closely explained
why researchers had chosen that particular number of animals, which is important
to demonstrate that a study has sufficient statistical power – that
is, can deliver a result that is beyond what is possible with chance.
the chief executive of the NC3Rs, Vicky Robinson, told me: "There is
no excuse for poor reporting, particularly when the ramifications for human
health and animal welfare are so significant."
come so much animal research is second-rate? The reason so much poor quality
research has endured for so long is down to politics.
sad truth is that in the understandable scramble to present a unified front
on animal research, the scientific establishment has been reluctant to be
critical of its own troops in the face of vicious opposition.
Home Office regulates animal research tightly, and welfare standards here
are far better than in many other countries. But insisting on best practice
in the research – including a statistically meaningful number of animals
– is beyond its remit. Bringing in such a measure would improve the
science, but could cause adverse publicity due to a short-term surge in
the number of animals used.
about this issue is simple: hardline antivivisectionists want a ban, pure
and simple, while sophisticated scientists will say my complaints are naive
and unhelpful. Others will warn that by raising the bar in Britain, more
animal research will end up being carried out in countries where regulations
if we are to defend this work, we have to be honest about its shortcomings.
Scientists must ensure that it is of the highest quality, and be seen to
commit to this. This is not just a matter of animals suffering needlessly
or dying in vain. The more animal experiments that pass muster, the more
human trials that will succeed – and the more money and lives will
Highfield is the Editor of 'New Scientist'.