CD4 criteria improves the sensitivity of a clinical algorithm developed to identify viral failure in HIV-positive patients on first-line antiretroviral therapy

By  Dr. Denise Evans  Dr Mhairi Maskew  Lynne McNamara  Patrick MacPhail,  Christopher Mathews  Professor Ian Sanne  Dr. Matthew Fox  |  | 


Background: Routine viral load monitoring of patients on antiretroviral therapy (ART) is neither affordable nor available in most resource-limited settings. We used data from an electronic patient management system to develop an algorithm to identify patients at risk of viral failure using a combination of accessible and inexpensive markers. Methods: We analyzed data from HIV-positive adults initiated on ART at Themba Lethu Clinic, South Africa between April 2004-February 2010. Viral failure was defined as 2 or more consecutive HIV-RNA viral loads >400copies/ml following suppression below ≤400copies/ml. We used Cox-proportional hazards models to calculate hazard ratios (HR) and 95% confidence intervals (CI). Weights for each predictor associated with viral failure were created as the natural logarithm of the adjusted HR and categorized into low, medium and high risk groups. We assessed the diagnostic accuracy of predictor scores and risk categories, with and without CD4 criteria (CD4< 100cells/mm3; CD4< baseline CD4; drop in CD4 >30%), using sensitivity, specificity, positive and negative predictive value. Results: Of 7369 patients, 922 (12.5%) experienced viral failure at a rate of 39.7/100 person-years. In our model for predictor scores, the following each received a weight of 1: baseline or current CD4< 100cells/mm3, baseline WHO stage III/IV, albumin< 25g/L, diastolic blood pressure< 70mmHg, MCV< 100fL, worsening WHO stage, suboptimal adherence and new condition/symptom. Odds ratio estimates and Kaplan Meier curves showed discrimination of viral failure by risk score or risk category. However, the sensitivity/specificity of the risk score (≥4 vs. < 4) or risk category (medium vs low/no risk) with CD4 criteria was 24.4%/88.6% and 21.1%/89.4%, respectively. The sensitivity/specificity without CD4 criteria was 12.3%/95.8% and 11.8%/95.9%, respectively. Conclusions: The algorithm may be a feasible and useful screening tool for resource-limited settings to identify patients at risk of treatment failure allowing more frequent monitoring or targeting for laboratory testing.

Conference: AIDS 2012, Washington DC, USA

Publication details