Many emergency treatments for severely injured patients such as car crash victims have never been properly evaluated and could be worse than no treatment at all, a trauma medicine researcher says.

He said public funding was essential to evaluate such medicines because there was little incentive for companies to conduct the research. "If you have an injury you will be exposed to treatments that we really don't know whether they will do more harm than good," Ian Roberts of the London School of Hygiene and Tropical Medicine told the British Association festival of science yesterday.

After decades of use there was little or no hard evidence for the treatments' effectiveness. And there was a postcode lottery over which treatments accident and emergency departments use. "There's huge variation," he said, "That's usually a sign that the evidence base is lousy and it absolutely is in this situation."

In 2004 Prof Roberts published a study in the Lancet exposing the dangers of corticosteroids, used to reduce brain inflammation. Rather than improving the patients' condition it actually increased their chances of dying. The Lancet estimated 10,000 people had been killed worldwide by the treatment in the 1980s and before. It is now rarely used in the UK.

He is also worried about other treatments, including hyperventilation to lower blood pressure in the brains of patients with head injuries.