Reflection and Reaction http://infection.thelancet.com Vol 6 July 2006 385 1 Colebunders R, Moses KR, Laurence J, et al. A new model to monitor the virological efficacy of antiretroviral treatment in resource-poor countries. Lancet Infect Dis 2006; 6: 53–59. 2 Weidle PJ, Downing R, Sozi C, et al. Development of phenotypic and genotypic resistance to antiretroviral therapy in the UNAIDS HIV Drug Access Initiative—Uganda. AIDS 2003; 17 (suppl 3): S39–48. 3 Grant RM, Hecht FM, Warmerdam M, et al. Time trends in primary HIV-1 drug resistance among recently infected persons. JAMA 2002; 288: 181–88. 4 Sutthent R, Arworn D, Kaoriangudom S, et al. HIV-1 drug resistance in Thailand: before and after National Access to Antiretroviral Program. J Clin Virol 2005; 34: 272–76. In June 2005 it was estimated that 6·5 million HIV- infected people living in resource-limited countries were in urgent need of antiretroviral treatment and, through a variety of international funding initiatives, access to treatment in these regions is expanding. 1 However, an important obstacle to the effective use of these life-saving drugs is the issue of how to monitor patients following initiation of treatment and, more specifically, how to detect treatment failure. Early detection of virological failure is important in the effective management of patients with antiretroviral therapy. If patients continue to receive a failing regimen, multiple antiretroviral drug resistance mutations may accrue. As a result, the likely efficacy of second-line regimens is compromised and such patients may also transmit drug-resistant virus, leading to primary drug resistance within the community. 2 Furthermore, failure to diagnose virological failure early subsequently leads to immunological failure with the associated risk of morbidity and mortality. Plasma viral load measurement, the gold standard used for monitoring in high-income countries, is considered prohibitively expensive or logistically unfeasible in many settings. Robert Colebunders and colleagues 3 provide a very useful review of these issues and discuss alternative strategies to viral load measurement. They suggest that simple clinical and immunological monitoring (advocated by WHO in the absence of viral load monitoring 4 ) is too insensitive. They instead propose a more sophisticated model, which incorporates data relating to prior treatment history, adherence data, clinical events, and measurements of haemoglobin, total lymphocyte count, and CD4 cell count. 3 To provide a substitute for viral load monitoring, such a model would have to be both sensitive and specific. Alternatively, use of a high sensitivity/low specificity model might be used as a screening tool to identify patients at high risk of virological failure, allowing targeted viral load testing in settings with limited availability of such tests. The proposed model, derived from clinical experience in Uganda, has not been formally evaluated and requires operational assessment. 3 Since 2002 we have run a large antiretroviral treatment service in Gugulethu, Cape Town, South Africa, which has previously been described in detail. 5–7 Over 1500 patients are currently receiving antiretroviral therapy at this site. All patients are carefully characterised at baseline and are monitored by clinical assessment and laboratory investigations every 16 weeks during long-term follow-up. Laboratory monitoring includes CD4 cell counts, plasma viral load, full blood counts, and biochemistry. Adherence data is routinely assessed by pill counts. All data are entered prospectively into an electronic database. Patients with a viral load over 1000 copies per mL at any follow-up visit are reviewed and receive a targeted adherence intervention. Viral load measurement is then repeated 6–8 weeks later; those with a second viral load over 1000 copies per mL (a minority) are diagnosed as having virological failure and treatment is switched to a second-line regimen. To evaluate the sensitivity of the model proposed by Colebunders and colleagues, we identified 24 patients (including one child) in this cohort treated between September 2002 and August 2005 who had confirmed virological failure diagnosed at least 6 months after initiation of antiretroviral therapy. Their median age was 33 years, 19 were women, and all but one were antiretroviral therapy-naive on programme entry. Virological failure was confirmed after a median of 37 weeks (interquartile range 29–61) of treatment with a median viral load of 3·9 log 10 copies per mL (range 3·2–5·5). Application of existing WHO criteria for treatment failure 3 would not have identified any of Evaluating a model for monitoring the virological efficacy of antiretroviral treatment in resource-limited settings