You would hope that a fake company, proposing to test a risky medical procedure, would be turned down flat. But that's not what happened in an elaborate "sting" operation set up to probe the US system for protecting volunteers in clinical research.

Trials of new drugs or medical devices can only begin if approved by an Institutional Review Board (IRB). Often these are attached to the hospitals or universities where the research will take place. But the task is increasingly being performed for profit by commercial IRBs, prompting fears that some may be rubber-stamping risky trials without proper scrutiny.

Now it appears these fears may be justified. At a congressional hearing on 26 March, the Government Accountability Office revealed the results of an investigation commissioned by the US House of Representatives Committee on Energy and Commerce. To test the responses of commercial IRBs, the GAO created a proposal from a fictitious company called Device Med-Systems that wanted to test the ability of a gel, poured into the abdomen, to reduce the growth of scar tissue after surgery. The protocol for the trial matched multiple examples described as posing "significant risk" by the Food and Drug Administration.

The GAO submitted the fake proposal to three commercial IRBs, two of which rejected it. But Coast IRB of Colorado Springs approved the proposed trial by a unanimous vote, describing it as "probably very safe". The Committee on Energy and Commerce also found that Coast approved all 356 proposals it received in the past five years, and earned $9.3 million for its services in 2008 alone.

Committee chairman Henry Waxman has vowed to "push for reforms that will protect the health and safety of the American people".