Health officials in Brazil have launched an investigation after claims that at least 10 impoverished Brazilians from an Amazon village may have contracted malaria while being used as human "guinea pigs" during a study by an American university.

The $1m (£570,000) research project, funded by the National Institutes of Health and conducted by the University of Florida, was being carried out in three villages on the Matapi river in the northern state of Amapa. It intended to study feeding patterns among mosquitoes over a four-year period in order to help control malaria outbreaks.

But critics say villagers were manipulated into taking part in the project and that those who allegedly contracted malaria as a result were offered no medical treatment. Brazil's Medical Council suspended the project last Wednesday following the allegations.

Villagers in Sao Raimundo da Pirativa, Sao Joao and Santo Antonio were paid between R$12 and R$20 (£3-£5) to collect mosquitoes on their bodies, while some were required to expose themselves to mosquito bites for periods of up to six hours.

"These are people with little means of feeding themselves, they are poor people who live on the river," said Haroldo Franco, the local prosecutor. "They were used as human bait and after that they became guinea pigs ... It is a question of whether these researchers have the right to endanger someone's life."

The study's chief researcher, Robert Zimmerman from the University of Florida, said it was "completely false" to say people had caught malaria because of his project. "There is no way you could show that," he told the Guardian. "The reason they have malaria is that they live in a place where malaria is endemic."

He said that humans had only been used to feed mosquitoes on one occasion, in the Sao Joao settlement, before the practice was stopped.

"We didn't go there and say you have got to do this," he added. "All they had to do is ask and we'd have left."

Gysélle Saddi Tannous, a counsellor from the Brazilian Medical Council, said no project which involved deliberately exposing villagers to mosquito bites or paying them would have been authorised by the council's ethics board (Conep).

She confirmed that a version of the project was approved on October 19 2001, but said it did not "foresee the use of human bait or the payment of the participants".

"This was definitely not authorised," she said. "It represents exploitation."

The Guardian had access to the five-page consent form that participants were required to sign. One paragraph explains that they will be "asked as volunteers to feed 100 mosquitoes on your arm or leg", while another says they will receive a "normal salary" for taking part.

Dr Zimmerman admitted that errors had been made when the project was submitted to Conep for approval in 2001. He said his presentation had not mentioned that participants would be paid - a prohibited practice for foreign organisations working in Brazil - but that researchers who worked in the same region had done the same in the past.

"I know this is a problem," he said. "I thought this was the way it was done. No one from the [Brazilian] government told me it was wrong."

Guardian Unlimited © Guardian Newspapers Limited 2005

###