Drug studies on animals are often not reliable compared with human clinical trials, a review published Friday concludes.

Before drugs are tested on humans, researchers test if the candidate medication seems to work and is safe in animal models.

In Friday's online issue of the British Medical Journal, Ian Roberts of the London School of Hygiene and Tropical Medicine and his colleagues compared six interventions that were tested in human and animal trials.

The treatments were:

• Corticosteroids to treat head injuries and respiratory illnesses in babies.
• Antifibrinolytics to reduce bleeding.
• Thrombolysis and tirilazad for ischemic stroke.
• Bisphosphonates for osteoporosis.

In half the cases, the results did not match.

"Discordance between animal and human studies may be due to bias or to the failure of animal models to mimic clinical disease adequately," the team concluded.

Corticosteroids did not improve head injury in clinical trials, but in animal studies, they showed a beneficial effect. Antifibrinolytics reduced bleeding in humans, but the findings in animals were inconclusive. Tirilazad for stroke proved harmful in clinical trials, but beneficial on the model patients with tails. 

For the other drugs, the findings generally agreed. Both animals and humans had better outcomes from thrombolysis for ischemic stroke. Bisphosphonates consistently helped  bone mineral density. The corticosteroids improved respiratory distress in both types of studies, but survival only improved in humans.

Systemic reviews like this study may help reveal the limitations of animal models and translate findings between species, the researchers said.

For example, the animal model for stroke, which seemed to agree with the results from the clinical trials, seemed to be more representative of the condition in humans than the animal model for head injury.

Closer collaboration between the research communities would also help to improve the relevance of animal models to clinical trial design, the study's authors suggested.