Flawed animal studies over several decades may have seriously hampered the development of new medicines, research conducted by two teams of scientists suggests.
One group from Canada found that a new kidney cancer drug's effectiveness had been over-estimated by as much as 45% because of the slack way pre-clinical tests had been conducted.
Lack of rigour may explain why only 5% of agents that show anti-cancer activity prior to patient trials are eventually licensed, said the researchers.
A second Scottish analysis of 147 randomly chosen studies published between 1941 and 2012 revealed that a basic step to avoid bias - ensuring researchers were "blind" to what experimental group animals had been assigned to - had only been taken in one in 20 cases.
The University of Edinburgh investigators also examined more than 1,000 animal studies from the UK's top five universities published between 2009 and 2010.
Less than a third of these research teams had recorded taking appropriate measures to reduce bias, they reported in the journal Public Library of Science Biology.
The kidney cancer drug, sunitinib, launched in 2006, was the focus of a critical in-depth study led by Dr Jonathan Kimmelman, from McGill University.
Sold under the brand name Sutent, the drug targets signalling pathways that cause certain cancers to grow. It is also used to treat rare stomach and pancreatic cancers.
The Canadian team claimed that studies reporting little or no anti-cancer effect from sunitinib were simply not published, and the drug's efficacy had been over-estimated by as much as 45%.
Dr Kimmelman said: "Only a fraction of drugs that show promise in animals end up proving safe and effective in humans.
"An important reason is because studies in animals are often not well designed, and because positive results have a higher chance of being published. They end up skewing what we think we know about the potential of a drug."
Few pre-clinical sunitinib studies were designed to avoid bias, said the scientists, writing in the journal eLife.
In addition, most of the experiments were conducted on one laboratory mouse strain that responded well to the drug - young females with compromised immune systems.
Tests involving other animal types, including mice that had spontaneously developed tumours, showed less of an effect.
In many cases it was not even clear how many animals had been tested because the sample size was not reported.
Sunitinib was said to show statistically significant anti-cancer activity against every different variety of tumour tested, a result which "strains credibility," said Dr Kimmelman.
The Scottish scientists led by Professor Malcolm Macleod, wrote: "It is sobering that of over 1,000 publications from leading UK institutions, over two-thirds did not report even one of four items considered critical to reducing the risk of bias, and only one publication reported all four measures."
Guidelines developed in the UK and aimed at improving standards of medical research conducted on animals were published in the journal Public Library of Science Biology in 2010.
The Canadian scientists called for new guidance on the design and reporting of pre-clinical cancer studies.
Dr Jarrod Bailey, senior scientist at the animal welfare group Cruelty Free International, formerly the British Union for the Abolition of Vivisection (BUAV), said: "We find the results and conclusions disturbing, yet unsurprising, adding as they do further support to existing evidence that shows many animal studies - as well as involving great suffering and by their very nature confounding our knowledge of human biology - to be very poorly designed and conducted, and often to provide results that are essentially meaningless."