Poor investigator brochures: Resarchers call for standardisation

Poor animal study design and reporting thwarts the ethical review of proposed human drug trials, according to the study led by researchers under Daniel Strech at Hannover Medical School, Germany, in cooperation with researchers from McGill University, Canada. They analysed the descriptions of animal studies found in "investigator brochures" (IBs) – the documents used by regulatory authorities and ethics committees to assess the potential efficacy of drugs that are being tested in patients for the first time.

Independent assessments of animal evidence are key to ensuring that patients are not exposed to undue risk when volunteering in trials. Based on documents obtained from three prominent German medical research centers, the study authors recommend that regulators need to develop standards to ensure the rigorous design and reporting of preclinical animal studies when trials of new drugs are launched.

Strikingly, less than one-fifth of investigator brochures referenced animal studies that had been through a peer-reviewed publication process. Less than 20% of animal studies that tested the efficacy of the new drug described the use of simple techniques, like randomization blinding or sample size calculation, that can reduce the effects of bias. And worryingly, of the more than 700 animal studies that the authors found in the investigator brochures, only 4% did not show positive effects of treatment.

"Our analysis shows that the vast majority of these documents lack the information needed to systematically appraise the strength of evidence supporting trials," said senior author Daniel Strech. "We were also struck by the rarity of negative outcomes of animal studies in investigator brochures", said co-author Jonathan Kimmelman, from McGill University. "With a median group size of 8 animals, these studies had limited ability to measure treatment effects precisely. Chance alone should have resulted in more studies being negative- the imbalance strongly suggests publication bias" said Susanne Wieschowski, a postdoctoral fellow in Strech’s team.