A South African economic model found that giving PCV20 to older adults living with HIV could be cost-effective, while offering it to all adults aged 65 or older was not cost-effective at the modeled price. The analysis compared PCV20 with PPSV23 from the public-sector payer perspective.
The model estimated a substantial annual disease burden in the public-sector cohort: 94 invasive pneumococcal disease cases and 50 deaths, 3,581 hospitalized pCAP cases and 430 deaths, and 13,563 outpatient pCAP cases. It also estimated 5,874 quality-adjusted life years, a measure combining length and quality of life, lost and USD 12.30 million in medical costs.
The price of wider coverage
At 20% coverage, PCV20 produced 317.1 more modeled QALYs than PPSV23, but added USD 5,492,699 in cost. Its incremental cost-effectiveness ratio, or ICER, was USD 17,323 per QALY, compared with decision benchmarks of USD 3,322 and USD 6,267 per QALY. The ICER measures the extra cost for each additional QALY.
At the model's base-case price of USD 20 per dose, the maximum acceptable price for PCV20 was USD 12.75 per dose at the primary USD 3,322 threshold, a 36.2% reduction. The modeled maximum was USD 12.65 at a USD 3,134 threshold and USD 14.28 at USD 6,267.
Price was the strongest one-way driver of the ICER. The next most influential inputs were PCV20 effectiveness against community-acquired pneumonia, or CAP, hospitalized pCAP incidence and PPSV23 effectiveness against CAP.
The uncertainty analysis kept the broader rollout's modeled prospects low. The researchers ran 1,000 simulations and retained 994 valid ones; the median ICER was USD 22,477 per QALY, with a 95% uncertainty interval from USD 9,579 to USD 102,779. The modeled probability of cost-effectiveness was 1.90% at each of the three stated thresholds.
A budget question as well as a health one
Over five years, the base-case rollout was projected to require USD 88.43 million in vaccination costs. Medical cost offsets were USD 1.54 million, leaving a net budget impact of USD 86.90 million. The model projected that 4,654 cases and 129 deaths would be averted.
Why HIV targeting changes the result
To examine HIV-associated risk, the model separated HIV-positive and HIV-negative older adults. It used annual IPD incidence of 237.2 per 100,000 person-years in the HIV-positive group and 4.1 per 100,000 person-years in the HIV-negative group, rather than applying a relative-risk multiplier.
With national HIV-stratified incidence, the general-population ICER was USD 10,700 per QALY. A scenario without a vaccine-effectiveness discount for people living with HIV produced an ICER of USD 9,301, but both figures remained above the study's decision benchmarks.
When vaccination was targeted to HIV-positive older adults, the ICER was USD 2,104 per QALY under national incidence and USD 96 per QALY in the lower-bound Soweto scenario. In the upper-bound Soweto scenario, the result was dominant, meaning the strategy was modeled to cost less while delivering more health benefit. Every CD4-distribution scenario tested kept the targeted strategy below the USD 3,322 primary benchmark.
The result depends on the model's boundaries
The work used a static decision-tree model for the South African public-sector payer, with 2024 as its base year and a five-year budget-impact horizon. Its cost-effectiveness and budget calculations covered a modeled cohort of 3,062,092 public-sector adults aged 65 or older, not a recruited patient sample.
That design leaves important questions open. The model did not include transmission dynamics, herd protection or serotype replacement, and its HIV-specific vaccine-effectiveness adjustments were assumption-based rather than trial-derived. The findings are projections from that model, not direct clinical measurements.
Paper data and sources
Original title: Optimizing pneumococcal prevention in South African older adults: Cost-effectiveness of PCV20 versus PPSV23 in the context of HIV-associated risk
Authors: Wang W, Tan Y, Bellua Sam N et al.
Journal/Repository: Human vaccines & immunotherapeutics
Status: Peer-reviewed
First online: 2026-10-10
DOI: Not available
Original paper · Full text