Background:  Most Africans are not being monitored for viral response to ART because of the cost of viral load testing. We assessed a combination of clinical and laboratory parameters to predict viral status in Kampala, Uganda.

Methods:  This was a cross-sectional study of 496 Ugandans established on non-nucleoside reverse transcriptase inhibitor (NNRTI) -based ART. Treatment, adherence, and clinical and laboratory parameters were evaluated for association with virological failure (VF; >400 copies/mL) and suppression (<400 copies/mL) using logistical regression. Parameters found to be independently associated with virological status were combined to produce a failure score. The failure score was then used to create a monitoring algorithm in which patients are categorized according to “likely” virological status.

Results:  Of the total, 65% were female, median age 38 years, and median duration of ART was 13 months; 49 experienced virological failure (9.9%). In univariate analysis, significantly associated with virological failure were:  a history of non-HAART, ever paying for treatment, time on ART, any treatment interruption >2 days, any missed dose in last 30 days, weight loss below baseline, new or poorly responsive HIV-related symptoms (eg, itchy skin), CD4 fall <baseline and >30% from peak achieved on ART (CD4³⁰), and fall in hemoglobin <baseline. Only 4 patients (0.8%) suffered a WHO stage 4 event after 6 months of ART. Multivariate analysis showed treatment interruption (OR 3.6, p = 0.001) and CD4³⁰ (OR 3.7, p = 0.007) as independently associated. No parameters significantly associated with virological suppression. ROC characteristics of virological failure generated from these 2 significant parameters gave AUC of 0.70 (0.61 to 0.78). The monitoring algorithm categorized patients into “failure unlikely” (according to negative predictive value [NPV] 0 virological failure) and “status unknown”; 384 were categorized as “failure unlikely” (NPV = 95%, 91 to 96). Viral load therefore only required for 112 (ie, “status unknown”), but 21 with virological failure missed (43%). For viral load cut-off of 1000 copies/mL, NPV = 97% and 13 of 39 virological failure missed (see the table). If WHO failure criteria were applied to this population, 37 (76%) patients with virological failure (>400 copies/mL) would be left on ART and 60 with virological suppression would be switched unnecessarily to second line.

Conclusions:  Clinical and laboratory parameters may allow rationing of viral load testing in resource-limited settings and improve upon currently available guidelines.