Background: There is concern that antiretroviral therapy (ART) use with only clinical monitoring for failure will result in high rates of transmission of virus with resistance to drugs currently in use.
Methods: A stochastic simulation model of transmission of HIV, natural history and the effect of ART, was developed and used to predict the proportion of new infections with resistance according to whether and when viral load monitoring is introduced.
Results: In our base model, there was predicted to be 12.4% of new HIV infections with primary antiretroviral resistance in 2020 if clinical monitoring is used throughout, compared with 5.4 and 6.1% if viral load-guided switching (based on viral load measured every 6 months, with switch determined by a value >500 copies/ml) was introduced in 2010 or 2015, respectively. The death rate for those on ART was lowest when viral load monitoring was used, but the overall death rate in all infected people was higher if viral load monitoring was introduced at the expense of scale-up in HIV diagnosis and ART initiation beyond their 2010 coverage levels (4.7 compared with 3.1 per 100 person-years).
Interpretation: To preserve current first-line drugs for the long term there is an eventual need for some form of cheap and practical viral load monitoring in resource-limited settings. However, a delay in introduction of 5 years has limited consequences for resistance transmission so the current priority for countries' ART programmes is to increase HIV testing and provide treatment for all those in need of ART.