Articles www.thelancet.com Vol 371 April 26, 2008 1443 Outcomes from monitoring of patients on antiretroviral therapy in resource-limited settings with viral load, CD4 cell count, or clinical observation alone: a computer simulation model Andrew N Phillips, Deenan Pillay, Alec H Miners, Diane E Bennett, Charles F Gilks, Jens D Lundgren Summary Background In lower-income countries, WHO recommends a population-based approach to antiretroviral treatment with standardised regimens and clinical decision making based on clinical status and, where available CD4 cell count, rather than viral load. Our aim was to study the potential consequences of such monitoring strategies, especially in terms of survival and resistance development. Methods A validated computer simulation model of HIV infection and the effect of antiretroviral therapy was used to compare survival, use of second-line regimens, and development of resistance that result from different strategies— based on viral load, CD4 cell count, or clinical observation alone—for determining when to switch people starting antiretroviral treatment with the WHO-recommended first-line regimen of stavudine, lamivudine, and nevirapine to second-line antiretroviral treatment. Findings Over 5 years, the predicted proportion of potential life-years survived was 83% with viral load monitoring (switch when viral load >500 copies per mL), 82% with CD4 cell count monitoring (switch at 50% drop from peak), and 82% with clinical monitoring (switch when two new WHO stage 3 events or a WHO stage 4 event occur). Corresponding values over 20 years were 67%, 64%, and 64%. Findings were robust to variations in model specification in extensive univariable and multivariable sensitivity analyses. Although survival was slightly longer with viral load monitoring, this strategy was not the most cost effective. Interpretation For patients on the first-line regimen of stavudine, lamivudine, and nevirapine the benefits of viral load or CD4 cell count monitoring over clinical monitoring alone are modest. Development of cheap and robust versions of these assays is important, but widening access to antiretrovirals—with or without laboratory monitoring—is currently the highest priority. Funding None. Introduction For antiretroviral treatment to be introduced as widely and rapidly as possible in lower-income settings, initial treatment will inevitably be provided without monitoring of viral load and, in many cases, without monitoring of CD4 cell counts; furthermore, the role of resistance testing is likely to be in monitoring emergence of resistance patterns at a population level rather than for individualised care. If second-line regimens are to be available in these lower-income settings the decision of when to switch a patient from a first to a second-line regimen will have to be based on incidence of new HIV-related clinical disease and—if available—CD4 cell count changes, rather than on the basis of viral load (ie, observed virological failure), as is standard in high-income countries. 1–10 Since clinical monitoring is being used for most of the 2 million patients currently receiving antiretrovirals in lower-income countries 11 and will continue to be used for many years as the scale-up of antiretroviral therapy continues, it is important to fully consider the potential long-term consequences of its use, especially in terms of survival and resistance development. Here, we use a model of HIV progression and the effect of antiretroviral therapy that has been extensively tested against clinical data to predict outcomes from the use of various clinical and CD4 cell count-based monitoring strategies for determining when to switch to a second-line regimen, and compare these with strategies based on viral load monitoring. Methods The model of HIV infection and the effect of antiretroviral therapy (HIV Synthesis) that we used was originally developed for well-resourced settings and has been described in detail elsewhere. 12 In brief, this stochastic computer simulation model generates data on the progression of HIV infection and the effect of antiretroviral therapy on simulated patients. At the time of infection, information generated includes calendar date, age, viral load, and CD4 cell count. Each individual’s data are then updated every 3 months. Presence of resistance is Lancet 2008; 371: 1443–51 See Comment page 1396 HIV Epidemiology and Biostatistics Group, Department of Primary Care and Population Sciences, and Royal Free Centre for HIV Medicine (Prof A N Phillips PhD), Centre for Virology, Department of Infection (Prof D Pillay FRCPath), Royal Free and University College Medical School, University College London, London, UK; Centre for Infections, Health Protection Agency, London, UK (D Pillay); London School of Hygiene and Tropical Medicine, London, UK (A H Miners PhD); Department of HIV/AIDS, WHO, Geneva, Switzerland (Prof C F Gilks FRCP, D E Bennett MD); and Centre for Viral Diseases/KMA, Rigshospitalet and Faculty of Health Sciences, University of Copenhagen, Denmark (Prof J D Lundgren DMSc) Correspondence to: Prof Andrew N Phillips, Department of Primary Care and Population Sciences, Royal Free and University College Medical School, University College London, London, UK a.phillips@pcps.ucl.ac.uk