Treatment for human immunodeficiency virus (HIV) infection has progressed rapidly since the introduction of zidovudine monotherapy in 1987. Although combination treatment with two or more drugs is now widely accepted, the advantages and disadvantages of this strategy have been difficult to delineate because of differences in study design and type of patient enrolled. Jordan and colleagues reviewed the evidence for combination therapy as the initial treatment of newly diagnosed patients with HIV infection.

They searched electronic databases for studies of initial treatment of HIV infection in patients 12 years or older using drugs approved for HIV treatment in Great Britain or the United States. The studies had to be controlled trials of good quality lasting at least 12 weeks. Data were collected on disease progression, mortality, viral load, CD4 count, and all other relevant outcomes, including dropout rate. Statistical tests were used to assess publication bias.

From more than 700 papers initially identified, the study finally included 90 papers containing data from 54 trials involving 20,404 patients. More than 80 percent of patients were men, averaging 27 to 40 years of age. Mean baseline CD4 counts were 83 to 660 cells per mm³ (83 to 660 × 10⁶ per L), and mean viral load ranged from 2.35 to 7.35 log copies per mL. The patients were followed from 12 weeks to 4.8 years.

The only monotherapy trials used zidovudine compared with placebo or no treatment. Dual therapies usually consisted of two nucleosides that were compared with zidovudine or didanosine monotherapy. Triple therapy usually incorporated a protease inhibitor or a non-nucleoside into therapy with two nucleosides.

Single therapy with zidovudine significantly reduced disease progression and mortality compared with no treatment or placebo, but the effect waned over time and was not apparent beyond three years of therapy. Overall, double therapy produced significantly better results than monotherapy, and combinations including protease inhibitors appeared to be more effective than other combinations. Disease progression and death outcomes were significantly better with triple therapy than with double therapy, but there was substantial variation among trials in combinations of medications. Dropout rates were similar for monotherapy and double therapy but significantly higher in trials containing protease inhibitors. Quality-of-life indicators were incorporated into only four trials, and the results were inconsistent.

The authors conclude that despite the limitations of meta-analysis and the variations between studies, the effectiveness of combination therapies increases as the number of drugs is increased, up to at least three. They call for further studies to investigate the most effective and practical combination of drugs in triple therapy and suggest that quadruple therapies be explored.