Skip Navigation


Global Projects


Adherence-Based Viral Load Triage in Southern Africa

Gaborone, Botswana


Low cost methods for monitoring HIV-1-infected patients for virologic treatment failure are needed so that resource poor settings can identify failure early and thereby avoid development of resistance. In Botswana and South Africa, where an increasing number HIV-infected individuals are currently receiving antiretroviral therapy through one of Africa’s largest free National ARV Programs, over 87% of individuals that initiate HAART and follow-up for 6 month viral load testing have an HIV-1 RNA level that is <400 copies/mL plasma. These data document that the majority of patients initiating HAART in this setting become “undetectable” at this time-point and suggest that many of these viral loads could be avoided if non-viral load parameters were used to place patients initiating therapy into strata of probabilities of being undetectable. Broadly, we suggest that adherence and other clinical parameters besides HIV-1 RNA levels can be used to create a clinical prediction rule for an undetectable HIV-1 RNA level after initiating HAART, and that this rule can be used to guide patients into low- and high-intensity virologic monitoring pathways. Given resource constraints in Africa, use of current viral load assays cannot be considered for all patients routinely. Rather, virologic monitoring strategies should focus on those who are at high risk of virologic failure. This study aims to use cheap and clinically available parameters to develop a clinical prediction rule that successfully achieves this focus. The specific aim of this project is therefore to develop a clinical prediction rule for achievement of an undetectable plasma HIV-1 RNA level 6 months after initiation of ARV therapy in sub-Saharan Africa. The design of the rule is such that a positive “test” means that the plasma HIV-1 RNA level is undetectable (<400 copies/mL). We hypothesize that a prediction rule can be created such that the point estimate for the probability of being undetectable in patients with a positive test is at least 95% and the lower bound of the 95% confidence interval is > 90%. We feel that the probability of response should be at least 95% and no lower than 90% in order to be acceptable in ARV programs and to clinicians. Resources for virologic monitoring in such settings could therefore be logically focused on patients with negative tests who are at higher risk of virologic failure.


  • Start Date: 
  • End Date: