Clinical Impact and Cost-effectiveness of Diagnosing HIV Infection During Early Infancy in South Africa: Test Timing and Frequency

Abstract

Background: Diagnosis of human immunodeficiency virus (HIV) infection during early infancy (commonly known as "early infant HIV diagnosis" [EID]) followed by prompt initiation of antiretroviral therapy dramatically reduces mortality. EID testing is recommended at 6 weeks of age, but many infant infections are missed.

Design/methods: We simulated 4 EID testing strategies for HIV-exposed infants in South Africa: no EID (diagnosis only after illness; hereafter, "no EID"), testing once (at birth alone or at 6 weeks of age alone; hereafter, "birth alone" and "6 weeks alone," respectively), and testing twice (at birth and 6 weeks of age; hereafter "birth and 6 weeks"). We calculated incremental cost-effectiveness ratios (ICERs), using discounted costs and life expectancies for all HIV-exposed (infected and uninfected) infants.

Results: In the base case (guideline-concordant care), the no EID strategy produced a life expectancy of 21.1 years (in the HIV-infected group) and 61.1 years (in the HIV-exposed group); lifetime cost averaged $1430/HIV-exposed infant. The birth and 6 weeks strategy maximized life expectancy (26.5 years in the HIV-infected group and 61.4 years in the HIV-exposed group), costing $1840/infant tested. The ICER of the 6 weeks alone strategy versus the no EID strategy was $1250/year of life saved (19% of South Africa's per capita gross domestic product); the ICER for the birth and 6 weeks strategy versus the 6 weeks alone strategy was $2900/year of life saved (45% of South Africa's per capita gross domestic product). Increasing the proportion of caregivers who receive test results and the linkage of HIV-positive infants to antiretroviral therapy with the 6 weeks alone strategy improved survival more than adding a second test.

Conclusions: EID at birth and 6 weeks improves outcomes and is cost-effective, compared with EID at 6 weeks alone. If scale-up costs are comparable, programs should add birth testing after strengthening 6-week testing programs.

Keywords: birth testing; cost-effectiveness; early infant HIV diagnosis; human immunodeficiency virus; nucleic acid test.

Figure 1.

One-year and lifetime per-patient healthcare costs of strategies for diagnosis of human immunodeficiency virus (HIV) infection during early infancy (commonly known as “early infant HIV diagnosis” [EID]) for HIV-exposed infants in South Africa. A, One-year total HIV care costs per HIV-exposed infant tested range from $50 for the no EID testing strategy to $100 for testing at birth and 6 weeks of age (hereafter, the “birth and 6 weeks” strategy). Costs of the EID program itself, including the costs of assays, personnel, specimen handling, and return of test results to the infant's caregiver, are shown in black and range from $30 (testing once) to $55 (testing twice). These costs exclude implementation costs associated with adding a new program if none existed previously. B, Total and component lifetime costs per HIV-exposed infant tested. Routine HIV care, CD4⁺ T-cell count and HIV load monitoring, opportunistic infections (OIs) and end-of-life care, and antiretroviral therapy (ART) comprise the majority of lifetime costs. EID program costs are shown in black and comprise 2%–3% of lifetime costs.

Figure 2.

Tornado diagram showing key parameters that change the cost-effectiveness of diagnostic testing for human immunodeficiency virus (HIV) infection during early infancy (commonly known as “early infant HIV diagnosis” [EID]) at birth and 6 weeks of age (hereafter, the “birth and 6 weeks” strategy), compared with testing at 6 weeks of age alone (hereafter, the “6 weeks alone” strategy). Key parameters varied in model sensitivity analyses are shown on the left. Values in parentheses indicate the range examined (from the value leading to the lowest incremental cost-effectiveness ratio [ICER] to the value leading to the highest ICER), with base-case values after the semicolon. ICERs for the comparison of birth and 6 weeks strategy versus the birth alone strategy are shown on the horizontal axis in 2013 dollars per year of life saved (YLS). The range of ICERs for each varied parameter is indicated by the blue bars. Longer bars indicate parameters to which the model results were more sensitive. The red line indicates the ICER of $2900/YLS for the birth and 6 weeks strategy versus EID at birth alone, using all base-case parameters. The dotted black line indicates 50% of South Africa's per capita gross domestic product (GDP; 0.5 × $6500 = $3250), the dashed black line indicates 100% of South Africa's per capita GDP ($6500), and the solid black vertical line indicates 300% of South Africa's per capita GDP ($19 500). The value for each parameter at which the ICER crosses the 50% GDP threshold is listed within each horizontal bar. This figure provides a framework for making decisions on cost-effectiveness grounds: the value within the horizontal bar indicates when one would favor the 6 weeks alone strategy over the birth and 6 weeks strategy on the basis of this criterion. Bars extending to the far-right axis indicate scenarios in which the birth and 6 weeks strategy results in an ICER of >$20 000/YLS, compared with the 6 weeks alone strategy, or becomes strongly dominated (more expensive and less effective) by the 6 weeks alone strategy. Abbreviations: ART, antiretroviral therapy; NAAT, nucleic acid amplification test; OI, opportunistic infection; PMTCT, prevention of mother-to-child transmission.

Figure 3.

Scaling up diagnostic testing for human immunodeficiency virus (HIV) infection during early infancy (commonly known as “early infant HIV diagnosis” [EID]) programs at 6 weeks of age alone (hereafter, “6 weeks alone”) or adding testing at birth. In scenario A, we projected the impact of scaling up existing 6 weeks alone EID programs and of adding birth testing to existing 6 weeks alone programs. For a 6 weeks alone program in which 50% of infants are tested (50% testing) and results are returned to caregivers of 50% of infants tested (50% result-return), 1-year survival among HIV-infected infants is projected at 63.4% (bar I). Adding birth testing to this program, also with 50% testing and 50% result-return, would improve outcomes minimally (to 64.4%; bar II). In contrast, scaling up the 6 weeks alone program with 75% testing and 75% result-return (1-year survival, 68.1%; bar III) or closer to 100% testing and 100% result-return (1-year survival, 74.9%; bar V) would improve outcomes to a much larger degree. Five-year and lifetime outcomes followed similar trends.