Washington ? Weeks before President Bush announced a plan to protect African babies from AIDS, top U.S. health officials were warned that research on the key drug was flawed and may have underreported severe reactions, including deaths, government documents show.

During PResident Bush's tour of Africa, protesters display a sign to Bush's passing motorcade criticizing the high cost of HIV medications, in Entebbe, Uganda, in July 11, 2003, file photo. Weeks before President Bush announced a plan to protect African babies from AIDS, top U.S. health officials were warned that research on the key drug was flawed and may have underreported severe reactions, including deaths, government documents show.

The 2002 warnings about the drug, nevirapine, were serious enough to suspend testing for more than a year, let Uganda’s government know of the dangers and prompt the drug’s maker to pull its request for permission to use the medicine to protect newborns in the United States.

But the National Institutes of Health, the government’s premier health research agency, chose not to inform the White House as it scrambled to keep its experts’ concerns from scuttling the use of nevirapine in Africa as a cheap solution, according to documents obtained by The Associated Press.

“Everyone recognized the enormity that this decision could have on the worldwide use of nevirapine to interrupt mother-baby transmission,” NIH’s AIDS research chief, Dr. Edmund Tramont, reported March 14, 2002, to his boss, Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases.

The documents show Tramont and other NIH officials dismissed the problems with the nevirapine research in Uganda as overblown and were slow to report safety concerns to the FDA.

An unidentified baby is fed at a home for HIV/AIDS on World Aids Day in this file photo in Soweto, South Africa, Dec. 1, 1998. The United States sent doses of nevirapine to Africa to treat pregnant mothers with HIV.

NIH’s nevirapine research in Uganda was so riddled with sloppy record keeping that NIH investigators couldn’t be sure from patient records which mothers got the drug. Instead, they had to use blood samples to confirm doses, the documents show.

Lane said NIH officials were aware in spring 2002 about the impending White House announcement on nevirapine but did not tell presidential aides of the problems because they were confident that the underlying science was solid.

Two years later, after hundreds of thousands of doses of nevirapine have been distributed to African mothers and children, the FDA has recommended NIH stop using the drug with certain patients.