Field evaluation of PIMA point-of-care CD4 testing in Rakai, Uganda

Objective: To assess the accuracy of PIMA Point-of-Care (POC) CD4 testing in rural Rakai, Uganda. Methods: 903 HIV positive persons attending field clinics provided a venous blood sample assessed on site using PIMA analyzers per manufacturer's specifications. The venous samples were then run on FACSCalibur flow cytometry at a central facility. The Bland-Altman method was used to estimate mean bias and 95% limits of agreement (LOA). Sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated for a CD4 threshold of <350 and <500 cells/uL for antiretroviral eligibility. Results: There was a high correlation between PIMA and FACSCalibur CD4 counts (r = 0.943, p<0.001). Relative to FACSCalibur, the PIMA POC CD4 had negative mean bias of -34.6 cells/uL (95% LOA: -219.8 to 150.6) overall. The dispersion at CD4<350 cells/uL was 5.1 cells/uL (95% LOA: -126.6 to 136.8). For a threshold of CD4<350 cells/uL, PIMA venous blood had a sensitivity of 88.6% (95%CI 84.8-92.4%), specificity of 87.5% (95%CI 84.9-90.1%), NPV of 94.9% (95%CI 93.1-96.7%), and PPV of 74.4% (95%CI 69.6-79.2%). PIMA sensitivity and PPV significantly increased to 96.1% and 88.3% respectively with increased threshold of 500 cells/uL. Conclusions: Overall, PIMA POC CD4 counts demonstrated negative bias compared to FACSCalibur. PIMA POC sensitivity improved significantly at a higher CD4 threshold of 500 than a 350 cells/uL threshold.