PHILADELPHIA (AP) - A combination drug therapy that has been extremely successful in suppressing HIV levels during closely monitored clinical trials does not work nearly as well in real-world practice, according to a study.

The study published in this week's Annals of Internal Medicine looked at 273 patients infected with HIV-1, the most common type of the virus, who received a powerful multidrug regimen as outpatients at the Johns Hopkins HIV Clinic in Baltimore between March 1996 and February 1998.

The researchers compared the progress of those patients with the results for patients in clinical trials.

Both groups were treated with highly active antiretroviral therapy, or HAART — more commonly known as combination or "cocktail" therapy — which has produced dramatic reductions in mortality rates associated with HIV infection in the United States since its introduction in 1996.

According to the study, 37 percent of the Johns Hopkins patients getting the cocktail treatment had undetectable HIV levels one year after starting therapy. Only 23 percent suppressed the virus in all three time periods studied — 1-90 days, 3-7 months and 7-14 months.

Clinical trials using similar drugs show suppression rates twice as high as those numbers.

The cocktail, a mixture of protease inhibitors and other anti-HIV drugs, can inhibit HIV's ability to replicate, causing the virus to plummet to undetectable levels.

But the therapy must be precisely followed to successfully combat HIV, the virus that causes AIDS. That's not always easy. Patients have to take lots of pills and adhere to complicated dosing schedules. The drugs also may cause side effects.

According to the study, missed clinic visits were the strongest independent risk factor that prevented patients from reducing their body's HIV levels as low as those in more strictly monitored clinical trials. Poorer outcomes were also found among non-Caucasian groups and people above age 40.

One physician said the findings mirror what some doctors have been seeing in their own experiences but questioned whether study's comparison of two very different groups was valid.

"Patients who tend to participate in clinical studies tend to be more self-motivated to begin with," said Dr. Ian Frank, director of antiretroviral clinical research at the University of Pennsylvania. "And patients less likely to adhere to a complex pill-taking regimen are generally not admitted to clinical studies."

The study was conducted by three doctors at the Johns Hopkins University School of Medicine. They were not available for comment Monday.

Copyright 1999 The Associated Press. All rights reserved.