Diagnostic accuracy of the WHO clinical staging system for defining eligibility for ART in sub-Saharan Africa: a systematic review and meta-analysis.

Introduction: The World Health Organization (WHO) recommends that HIV-positive adults with CD4 count ≤500 cells/mm(3) initiate antiretroviral therapy (ART). In many countries of sub-Saharan Africa, CD4 count is not widely available or consistently used and instead the WHO clinical staging system is used to determine ART eligibility. However, concerns have been raised regarding its discriminatory ability to identify patients eligible to start ART. We therefore reviewed the accuracy of WHO stage 3 or 4 assessment in identifying ART eligibility according to CD4 count thresholds for ART initiation.

Methods: We systematically searched PubMed and Global Health databases and conference abstracts using a comprehensive strategy for studies that compared the results of WHO clinical staging with CD4 count thresholds. Studies performed in sub-Saharan Africa and published in English between 1998 and 2013 were eligible for inclusion according to our predefined study protocol. Two authors independently extracted data and assessed methodological quality and risk of bias using the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2) tool. Summary estimates of sensitivity and specificity were derived for each CD4 count threshold and hierarchical summary receiver operator characteristic curves were plotted.

Results: Fifteen studies met the inclusion criteria, including 25,032 participants from 14 countries. Most studies assessed individuals attending ART clinics prior to treatment initiation. WHO clinical stage 3 or 4 disease had a sensitivity of 60% (95% CI: 45-73%, Q=914.26, p<0.001) and specificity of 73% (95% CI: 60-83%, Q=1439.43, p<0.001) for a CD4 threshold of ≤200 cells/mm(3) (11 studies); sensitivity and specificity for a threshold of CD4 count ≤350 cells/mm(3) were 45% (95% CI: 26-66%, Q=1607.31, p<0.001) and 85% (95% CI: 69-93%, Q=896.70, p<0.001), respectively (six studies). For the threshold of CD4 count ≤500 cells/mm(3) sensitivity was 14% (95% CI: 13-15%) and specificity was 95% (95% CI: 94-96%) (one study).

Conclusions: When used for individual treatment decisions, WHO clinical staging misses a high proportion of individuals who are ART eligible by CD4 count, with sensitivity falling as CD4 count criteria rises. Access to accurate, accessible, robust and affordable CD4 count testing methods will be a pressing need for as long as ART initiation decisions are based on criteria other than seropositivity.