Background:  Timely diagnosis of pediatric HIV infection is critical for effective antiretroviral treatment to reduce HIV-related morbidity and mortality. The WHO clinical criteria for presumptive diagnosis of severe HIV disease in children under 18 months requiring ART when HIV virologic testing is not available were formulated on “expert opinion” basis. Not yet scientifically validated, this study aimed to determine utility of these criteria (clinical algorithm) in predicting ART need in settings where there is no ready access to HIV virologic confirmatory tests.

Methods:  Symptomatic HIV-exposed children <18 months were prospectively enrolled over a 17-month period at 3 hospitals in Malawi, Tanzania, and Uganda and evaluated for clinical features of severe HIV disease. All children were tested for HIV using dried blood spot DNA-PCR and evaluated for immunological disease stage using CD4%. The algorithm’s utility and discriminating accuracy in predicting ART need was determined against a gold standard for ART eligibility (2008 revised WHO guidelines for initiating ART in children aged <12 months and 12 to 18 months).

Results:  In this study, 301 symptomatic HIV-exposed children were evaluated. The median age was 6 months, 248 (82.4%) were <12 months. 141 (46.8%) were HIV-DNA PCR positive; 118 (39.2%) met the criteria for presumptive diagnosis of severe HIV disease; 122 children met the gold-standard criteria for ART (2008 revised WHO guidelines). The 2006 WHO criteria for presumptive diagnosis correctly predicted ART requirement with 68.9% sensitivity (95%CI 59.8 to 76.9); 81.0 % specificity (95%CI 76.57 to 85.44); 71.2% PPV, 79.2% NPV; and 74.5% accuracy of distinguishing between those who do and do not need ART.

Conclusions:  Results confirm the utility of a simple clinical algorithm as a tool to support early recognition of HIV-exposed infants who need ART when there is no ready access to confirmatory virologic diagnosis. This is viewed from the urgency of making a decision to initiate ART in clinical settings where virologic testing is not readily available and the potential morbidity and mortality averted. With this tool, 68.9% of infants urgently requiring ART will be correctly identified and given potentially lifesaving treatment pending confirmatory tests. These results will strengthen guidelines for clinical ART initiation decision making in resource poor settings where there is no ready access to virologic diagnosis.