Implementation and Operational Research: Implementation of Multiple Point-of-Care Testing in 2 HIV Antiretroviral Treatment Clinics in South Africa.

BACKGROUND: A plethora of point-of-care (POC) tests exist in the HIV and tuberculosis diagnostic pipeline which require rigorous evaluation to ensure performance in the field. The accuracy and feasibility of nurse-operated multidisciplinary-POC testing for HIV antiretroviral therapy (ART) initiation/monitoring was evaluated. METHODS: Random HIV-positive adult patients presenting at 2 treatment clinics in South Africa for ART initiation/monitoring were consented and enrolled. POCT was performed by a dedicated nurse on a venipuncture specimen; Pima (CD4), HemoCue (hemoglobin), Reflotron (alanine aminotransferase, creatinine), Accutrend (lactate) and compared with laboratory testing. External quality assessment, training, workflow, and errors were assessed. RESULTS: n = 324 enrolled at site1; n = 469 enrolled at site 2. Clinical data on n = 305 participants: 65% (n = 198) female with a mean age of 39.8 (21-61) years; mean age of males 43.2 (26-61) years; 70% of patients required 3 or more POC tests/visit. External quality assessment material was suitable for POCT. CD4, hemoglobin and alanine aminotransferase testing showed good agreement with predicate methodology; creatinine and lactate had increased variability. Pima CD4 misclassified up to 11.6% of patients at 500 cells per microliter and reported 4.3%-6% error rate. A dedicated nurse could perform POCT on 7 patients/day; inclusion of Pima CD4 increased time for testing from 6 to 110 minutes. Transcription error rate was 1%. CONCLUSIONS: Nurses can accurately perform multidisciplinary POCT for HIV ART initiation/monitoring. This will however, require a dedicated nurse as current duties will increase if POC is added to workflow. The use of Pima CD4 will increase patients initiated on ART. Connectivity will be central to ensure quality management of results, but overall impact will need to still be addressed.

Feasibility of performing multiple point of care testing for HIV anti-retroviral treatment initiation and monitoring from multiple or single fingersticks.

BACKGROUND: Point of Care testing (POCT) provides on-site, rapid, accessible results. With current South African anti-retroviral treatment guidelines, up to 4 fingersticks /patient/clinic visit could be required if utilizing POC. We determined the feasibility and accuracy of a nurse performing multiple POCT on multiple fingersticks followed by simplification of the process by performance of multiple POC on a single fingerstick. METHOD AND FINDINGS: Random HIV positive adult patients presenting at a HIV treatment clinic in South Africa, for ART initiation/ monitoring, were approached to participate in the study between April-June 2012. Phase I: n=150 patients approached for multiple POCT on multiple fingersticks. Phase II: n=150 patients approached for multiple POCT on a single fingerstick. The following POC tests were performed by a dedicated nurse: PIMA (CD4), HemoCue (hemoglobin), Reflotron (alanine aminotransferase, creatinine). A venepuncture specimen was taken for predicate laboratory methodology. Normal laboratory ranges and Royal College of Pathologists Australasia (RCPA) allowable differences were used as guidelines for comparison. In 67% of participants, ≥3 tests were requested per visit. All POCT were accurate but ranged in variability. Phase I: Hemoglobin was accurate (3.2%CV) while CD4, alanine aminotransferase and creatinine showed increased variability (16.3%CV; 9.3%CV; 12.9%CV respectively). PIMA generated a misclassification of 12.4%. Phase II: Hemoglobin, alanine aminotransferase and creatinine showed good accuracy (3.2%CV, 8.7%CV, 6.4%CV respectively) with increased variability on CD4 (12.4%CV) but low clinical misclassification (4.1%). No trends were observed for the sequence in which POC was performed on a single fingerstick. Overall, PIMA CD4 generated the highest error rate (16-19%). CONCLUSIONS: Multiple POCT for ART initiation and/or monitoring can be performed practically by a dedicated nurse on multiple fingersticks. The process is as accurate as predicate methodology and can be simplified using a single fingerstick.