Pharmaceutical companies are overstating the effectiveness of their drugs, and may be placing patients at greater risk, because animal laboratory studies they fund are biased, it was claimed yesterday.

A survey of nearly 300 animal-test studies involving six different experimental drugs suggested that such flawed methodology is rampant in the drug-testing industry.

About two-thirds of the studies, which were all aimed at testing drugs with the potential to treat stroke patients, did not use a proper "randomised blind" methodology, the British Association's Science Festival in York was told.

A similar proportion did not conform to the standard methodology where the experimenters were deliberately left "blind" as to which animals have been given the drug until the end of the experiment. "We show that animal experiments modelling human stroke often overstate how good drugs are at treating stroke," said Malcolm Macleod, a consultant neurologist at Stirling Royal Infirmary, who led the stroke study.

"This goes some way to explaining why stroke drugs tested in animals do not appear to work nearly so well in humans.

"It is certainly the case that human health in clinical trials would be better served with higher quality animal data," he added.

Dr Macleod said that most research using animals to test the effectiveness of drugs suffers from poor-quality controls which lead to subjective assessments and result in an inherent bias that makes the drug seem more powerful than it really is.

"It is highly likely that these flaws are also present in other fields of science. As scientists we need to improve how we conduct experiments using animals if we are to retain the confidence of the public," he said.

Animal studies looking at the effectiveness of a stroke drug called NXY-059, made by AstraZeneca, found that it improved outcome by more than 50 per cent.

However, when animal studies involving the drug were properly randomised and blinded, they showed that the improvement was actually between 25 and 30 per cent.

"We have reported similar findings for other interventions, but what is disturbing about the data for NXY-059 is that – for a drug where most of the published work was funded by the drug manufacturers – the impact of poor study quality was much more pronounced," Dr Macleod said.

"Another concern with the data for NXY-059 is that the number of animals used... was generally too small to allow a precise estimate of outcome," he added.

Professor Michael Bracken, an epidemiologist at Yale University, said that the serious flaws and inadequacies in animal research raised questions about the way animal studies are evaluated for their relevance to medicine. "This lack of advanced scientific methods leaves many questions about the value of animal research unanswered, and exposes patients and research volunteers to clinical trials that could be based on flawed animal studies," Professor Bracken said.

"The general public takes the view that it will only tolerate animal experiments if the results improve human health, but how much animal experiments improve human health is a scientific question. The key question is whether animal studies translate to human medicine." .

Derek Fry, a Home Office inspector of animal experiments, said that the authorities are concerned about the conduct of many studies looking at the effectiveness of drugs. "Scientists are expected to be objective but they are only just realising how subjective they can be," Dr Fry said."When you look at the published papers, it's often quite difficult to see whether they are randomised," he said.

"However, this only applies for a very limited subject. Most animal trials involving drugs are done for safety reasons,rather than efficacy."

Animal experimenting

* The number of experimental procedures on animals increased by 4 per cent last year to just over 3 million

* 83 per cent of experiments involved mice, rats and other rodents. Fish accounted for 9 per cent and birds 4 per cent.

* Dogs, cats, horses and monkeys – which are given special legal status – were involved in less than 0.5 per cent of experiments

* In 2005, 957,500 procedures – one in three of the total – involved genetically modified animals. This compared with just 8 per cent in 1995

* Rodents accounted for 98 per cent of the genetically modified animals that were used for experiments