Results produced by the toxicologist who investigated the death of the government weapons expert, Dr David Kelly, have been questioned by a group of experts.

Alexander Allan, who examined the body of Dr Kelly after it was found in woods near his home on 18 July last year, concluded that the Government scientist had swallowed at least 20 co-proxamol tablets, which contributed to his death.

But writing in the British Medical Journal, the International Toxicology Advisory Group say that the science of measuring drug levels in the blood after death is based on flawed evidence.

In the Kelly case, Dr Allan reported that he had found blood level concentrations of 97 micrograms of paracetamol and 1.0 micrograms of dextropropoxyphene per millilitre. He calculated that it was equivalent to approximately 20 tablets.

Dr Kelly had been found with three 10-tablet blister packs of co-proxamol in his pockets. Only one tablet remained. The scientists also had a cut to his wrist, which had severed the ulnar artery.

Robert Forrest, a professor of forensic toxicology at the University of Sheffield and one of the authors of the BMJ article, said yesterday: "Alexander fell into the trap of trying to estimate the number of tablets from the blood concentrations.A lot of toxicologists felt that was an extrapolation from the data that was inappropriate."

In the BMJ the group writes: "Drug concentrations are likely to have changed after death. For many drugs, including those found in David Kelly, concentrations may increase by as much as tenfold."

The Hutton inquiry concluded that Dr Kelly committed suicide after accepting evidence from Nicholas Hunt, the forensic pathologist, that the scientist died primarily of a self-inflicted wound causing haemorrhage, with co-proxamol ingestion as a secondary cause.

Professor Forrest said he accepted the suicide verdict, but the case highlighted a wider problem with the interpretation of toxicology results that had led to miscarriages of justice in America and Europe.

Measuring drug concentration levels in living patients is straightforward, involving factors such as how the drugs were administered and the number of doses. For dead subjects, however, this information is almost never available, meaning that conclusions about drug levels are incomplete.

The toxicology group intended to examine the problem and propose reforms, Professor Forrest said.