There has been considerable debate over the management of HIV infected individuals admitted to the ICA. Mortality in HIV infected patients in the ICA correlates with the level of immune suppression. The majority of HIV infected individuals entering the ICA are antiretroviral naive. Despite the high mortality rates and the opportunity to intervene with antiretroviral therapy, physicians do not routinely administer highly active antiretroviral therapy (HAART) in the ICA. Early initiation of HAART, which improves immune function, could potentially reduce mortality. Numerous studies have shown that there is a dramatic drop in the HIV-1 RNA levels accompanied by an increase in the CD4 cell count within the first 2 to 4 weeks of therapy. Sufficient data now exist that antiretrovirals could be administered in the ICA with careful monitoring and attention to drug interactions. This study will evaluate the effect of HAART in patients admitted to the hospital with an AIDS-defining illness, pneumonia, or sepsis.

Upon entry into the study, patients are stratified according to a severity of illness score (SAPS I) and CD4 cell count. Patients then are assigned to 1 of 2 study arms: Arm A: HAART (lamivudine [3TC] and zidovudine [ZDV], or 3TC/ZDV, and nelfinavir [NFV] and efavirenz [EFV]); or an alternative HAART for 4 weeks. Arm B: No antiretroviral regimen. Evaluations of the following are performed: drug toxicity, immune status, viral load, arterial blood gas, ventilator parameters, and evolution of the presenting illness. Pharmacokinetic trough concentration analyses are performed on all patients in Arm A during 3 time points of their illness. Patients are followed for 24 weeks after entry. Patients in Arm A may elect to participate in two substudies. The first substudy will measure efavirenz and nelfinavir drug levels in the blood to determine how critical illness affects pharmacokinetics. The second substudy will evaluate the benefit of HAART in HIV infected patients being treated for pneumocystis carinii pneumonia.