Field Evaluation of a Point-of-Care CD4 Analyzer for Monitoring HIV Patients in the Interior of the Amazon Region, Brazil

OBJECTIVE To evaluate the accuracy of the PIMA point-of-care CD4 analyzer (PIMA) under field conditions in comparison to the current CD4 count system (FACSCalibur), and to evaluate the operational suitability and acceptability of health professionals (HP) and HIV-patients in using the PIMA in health clinics in the Amazon Region. METHODS CD4 counts were measured onsite by the PIMA using fingerprick blood and in the reference laboratory by both the PIMA and FACSCalibur using venous blood. We used the Bland-Altman method to estimate the mean bias, and calculated the sensitivity and specificity at <200 and <500 cell/μL thresholds. Patients (n = 404) and HP (n = 7) were interviewed on the acceptability and operational suitability of the PIMA. RESULTS Using fingerprick blood (n = 337), the PIMA showed a concordance correlation coefficient (Rc) of 0.81, mean difference of -111.9 cell/μL, 93.1%/98.5% sensitivity, and 89.2%/56.7% specificity at <200 and <500 cell/μL thresholds, respectively. Venous blood (n = 340) showed an Rc of 0.89, mean difference of -83.4 cell/μL, 98.3%/97.5% sensitivity, and 93.9%/66.0% specificity at <200 and <500 cell/μL thresholds, respectively. The capillary PIMA was well accepted and found operationally appropriate by patients and HP. CONCLUSIONS The agreement between both instruments was poor and the PIMA underestimated CD4 cell counts, which was more pronounced at CD4 counts ≥500 cell/μl. The PIMA's performance with fingerprick blood was less reliable than its performance with venous blood. In Brazil, where antiretroviral treatment is initiated regardless of CD4 counts, the PIMA's systematic bias towards CD4 underestimation may limit its role for monitoring HIV-patients.