Strengthening Existing Laboratory-Based Systems vs. Investing in Point-of-Care Assays for Early Infant Diagnosis of HIV: A Model-Based Cost-Effectiveness Analysis

Abstract

Background: To improve early infant HIV diagnosis (EID) programs, options include replacing laboratory-based tests with point-of-care (POC) assays or investing in strengthened systems for sample transport and result return.

Setting: We used the CEPAC-Pediatric model to examine clinical benefits and costs of 3 EID strategies in Zimbabwe for infants 6 weeks of age.

Methods: We examined (1) laboratory-based EID (LAB), (2) strengthened laboratory-based EID (S-LAB), and (3) POC EID (POC). LAB/S-LAB and POC assays differed in sensitivity (LAB/S-LAB 100%, POC 96.9%) and specificity (LAB/S-LAB 99.6%, POC 99.9%). LAB/S-LAB/POC algorithms also differed in: probability of result return (79%/91%/98%), time until result return (61/53/1 days), probability of initiating antiretroviral therapy (ART) after positive result (52%/71%/86%), and total cost/test ($18.10/$30.47/$30.71). We projected life expectancy (LE) and average lifetime per-person cost for all HIV-exposed infants. We calculated incremental cost-effectiveness ratios (ICERs) from discounted (3%/year) LE and costs in $/year-of-life saved (YLS), defining cost effective as an ICER <$580/YLS (reflecting programs providing 2 vs. 1 ART regimens). In sensitivity analyses, we varied differences between S-LAB and POC in result return probability, result return time, ART initiation probability, and cost.

Results: For infants who acquired HIV, LAB/S-LAB/POC led to projected one-year survival of 67.3%/69.9%/75.6% and undiscounted LE of 21.74/22.71/24.49 years. For all HIV-exposed infants, undiscounted LE was 63.35/63.38/63.43 years, at discounted lifetime costs of $200/220/240 per infant. In cost-effectiveness analysis, S-LAB was an inefficient use of resources; the ICER of POC vs. LAB was $830/YLS.

Conclusions: Current EID programs will attain greater benefit from investing in POC EID rather than strengthening laboratory-based systems.

FIGURE 1.

Early survival of CWH survival for CWH through the first 6 months of life, with survival percentage along the vertical axis and time, in months, along the horizontal axis. Survival curves for CWH receiving EID at 6 weeks of age are shown for laboratory-based EID (LAB, blue), strengthened laboratory-based EID (S-LAB, red), and POC EID (POC, green). The point at which infants receive results and initiate ART is marked with arrows for each strategy. The absolute difference in survival between LAB, S-LAB, and POC is shown as a percent at 6 months.

FIGURE 2.

Cost-effectiveness of POC as a function of S-LAB per test cost, result return time, result return probability, and ART initiation probability. Cost effectiveness of POC at varied cost, result return time, result return probability, and ART initiation probability of S-LAB. Varied S-LAB per test cost is shown in each colored box, from $0.40 less than POC to $10 less than POC. Along the horizontal axis of each box, result return probability of S-LAB is shown, from 15% less than to the same as POC. Along the vertical axis of each box, result return time of S-LAB is shown, from 60 days longer to 10 days longer than POC. Each figure panel shows probability of ART initiation of S-LAB, from 20% (A) less than POC to the same as POC (D). For each combination of parameters, the cost effectiveness of POC is shown. Blue: The ICER of POC is ≤50% of the per capita GDP; S-LAB is weakly dominated. Grey: The ICER of POC is >50% of the per capita GDP and ≤100% of the per capita GDP. Black: The ICER of POC is >100% of the per capita GDP and S-LAB is ≥50% the per capita GDP and ≤100% the per capita GDP. At base-case costs, the ICER of POC remained ≤50% of the per capita GDP and the preferred strategy over S-LAB unless S-LAB result return and ART initiation probabilities were the same as POC and result return time was reduced to 10 days. At lower S-LAB costs, POC was the preferred strategy over S-LAB unless S-LAB result return time, result return probability, and ART initiation probability were close to those of POC, representing marked improvements compared with base-case values. S-LAB, strengthened laboratory-base early infant HIV diagnosis.