To improve early infant HIV diagnosis (EID) programs, options include replacing lab-based tests with point-of-care (POC) assays or investing in strengthened systems for sample transport and return of results. We projected the clinical benefits and cost-effectiveness of these approaches.
We used the Cost-Effectiveness of Preventing AIDS Complications-Pediatric model, with programmatic and published data, to examine clinical benefits and costs of three strategies for EID in Zimbabwe for infants 6 weeks of age: 1) lab-based EID (LAB), 2) strengthened lab-based EID (S-LAB), defined as improved sample transport, two additional lab staff, and increased lab maintenance, and 3) POC EID. Assays differed in sensitivity (LAB and S-LAB 100%, POC 96.9%) and specificity (LAB and S-LAB 99.6%, POC 100%). LAB/S-LAB/POC algorithms also differed in: probability of result return (79/91/98%), time until result return (61/53/0 days), probability of linking to ART after confirmed positive result (52/71/86%), and total cost/test ($17.09/$29.80/$31.26), which included transport, salary, training, and maintenance costs derived from a resource utilization analysis in Zimbabwe. Monthly cost of HIV care and ART varied by age, CD4 count, regimen, and weight. We projected life expectancy (LE) and average lifetime per-person cost for all HIV-exposed infants, including those who did and did not acquire HIV. We calculated incremental cost-effectiveness ratios (ICERs) from discounted (3%/year) LE and cost results in $/year-of-life saved (YLS), defining cost-effective as an ICER <$1,330/YLS (Zimbabwe per-capita GDP). In multi-way sensitivity analyses, we varied differences between S-LAB and POC in: result return probability, result return time, and cost.
For infants who acquired HIV, LAB/S-LAB/POC led to projected one-year survival of 67/70/76% and undiscounted LE of 21.77/22.75/24.51 years. For all HIV-exposed infants, undiscounted LE was 63.34/63.38/63.42 years, at undiscounted costs of $330/$360/$390 per infant. S-LAB was dominated in cost-effectiveness analysis; the ICER of POC vs. LAB was $870/ YLS. In multi-way sensitivity analyses, S-LAB was only cost-effective if it cost $10 less than POC, had the same result return probability as POC, and had 10-day result return time (Figure).
Current EID programs will attain greater benefit for additional investments by integrating POC EID rather than strengthening lab-based systems; decreases in POC test cost will amplify the benefits of POC EID.