University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Expanding antiretroviral options in resource-limited settings-a cost-effectiveness analysis
Global Health Sciences Literature Digest
Published April 21, 2010
Journal Article

Bendavid E, Wood R, Katzenstein DA. Expanding antiretroviral options in resource-limited settings-a cost-effectiveness analysis. Acquir Immune Defic Syndr. 2009 Sep;52(1):106-13.

In Context

The updated World Health Organization (WHO) guidelines (1) for highly active antiretroviral therapy (HAART) in low- and middle-income countries recommend starting treatment with two nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) such as AZT/3TC or TDF/(3TC or FTC), plus a non-nucleoside RTI (NNRTI), either NVP or EFV. Recommended second-line regimens include two different NRTIs and a boosted protease inhibitor (bPI), such as LPV/r. WHO has not given guidelines beyond second-line therapy, which limits treatment for those with drug resistance or drug-related toxicities.


To estimate the benefits, cost, and cost-effectiveness of expanding initial treatment guidelines to include three-drug regimens; to determine the influence of CD4 versus viral load (VL) monitoring on the outcomes of these different strategies


Mathematical simulation model


The model was based on evaluating the clinical course of HIV-infected people seeking treatment in South Africa, with input data taken from published cohort and clinical trials. Base-case information was obtained from studies in South Africa and, if these studies were not available, from other sub-Saharan studies and then from non-African trials.


Three HAART strategies were evaluated:

  1. the WHO strategy of a first-line regimen (two NRTIs plus an NNRTI) followed by second-line regimen (two NRTIs and LPV/r);
  2. a first-line regimen of triple NRTI (AZT/3TC/ABC) followed by the two-drug regimen of the WHO strategy; and
  3. the WHO sequence followed by a third-line treatment of a second-generation bPI (2bPI). Strategies were evaluated using CD4 counts alone or in combination with VL for monitoring drug response. Failure was defined as two VL measurements of >1000 copies/mL or a reduction in CD4 cell count by half after initial response. Rates of virologic failure and costs for VL were varied in the model.


Over the lifetime of the cohort, 25.6% were estimated to fail WHO first- and second-line regimens based on VL; however, the model estimates that only a quarter of these persons would be provided additional HAART if they were monitored with CD4 counts because of the delay in detection of failure. Adding a third-line regimen (with a 2bPI) could provide a 7-9 month gain of life expectancy per individual and a cost-effectiveness ratio of $2581 per year of life gained with CD4 monitoring and $6519 per year of life gained with VL monitoring. Using triple NRTIs as initial therapy increased mortality and opportunistic infections (OIs) and was not cost-effective.


Results of modeling suggest that the current recommended WHO regimens will provide adequate lifelong benefits for about three quarters of the HIV-infected population. Based on WHO and World Bank suggestions regarding acceptable costs, a third-line regimen with a second-generation bPI would be acceptable in South Africa, which has a per capita GDP of >U.S. $2000. It would be cost-saving elsewhere only if the price of second-generation bPIs dropped to <540/yr. Using a triple NRTI, a less efficacious regimen as first-line therapy, would worsen mortality and OIs regardless of monitoring strategy and would not be cost-effective.

Study quality

This appears to be a good quality modeling study. Results were hampered somewhat by limited input data on failure rates and on effectiveness of sequential regimens in sub-Saharan Africa.

Programmatic implications

The model from this study reinforces the relative efficacy and cost-effectiveness of current WHO recommendations for first- and second-line regimens in Africa. Its implications are somewhat limited, however, as it evaluated only one alternative initial strategy and a single third-line regimen. It did demonstrate that triple NRTI regimens should not be used as a first-line regimen. The model assumes that people are initiated and maintained on WHO regimens, whereas in reality, many have switched single or multiple intra-class drugs without VL monitoring. Therefore, with time, failure rates may be higher and more people will require a third-line regimen. At higher failure rates, a 2bPI will be "cost-effective," although the alternative is not to treat, with affected persons succumbing to OIs and progressive disease. Countries and the WHO should begin formulating cost-effective third-generation treatment recommendations. Viral load monitoring still is preferable for guiding treatment compared with CD4 counts, particularly with complicated regimens.


  1. WHO. Rapid Advice: Antiretroviral therapy for HIV infection in adults and adolescents. 30 November, 2009.