Background
The
Streptococcus pneumoniae bacterium causes significant morbidity, mortality, and economic burden in older adults [
1]. Its most severe clinical picture is invasive pneumococcal disease (IPD), with a case fatality rate of 10–30% [
2]. Additionally,
S. pneumoniae is responsible for 9–30% of community acquired pneumonia (CAP) cases [
3], a leading cause of deaths among older adults. More than 100 pneumococcal serotypes exist. Two vaccines have been used in older adults: a 23-valent pneumococcal polysaccharide vaccine (PPV23) and a 13-valent pneumococcal conjugate vaccine (PCV13). PPV23 contains more serotypes but provides less effective and shorter-lasting protection against vaccine-type serotypes due to the absence of memory B cell activation [
4].
Childhood vaccination programs with PCVs have significantly altered the serotype distribution of IPD cases among older adults. In the Netherlands, no programmatic pneumococcal vaccination for older adults existed before 2020. Indirect protection from childhood vaccination with PCV7 since 2006 and PCV10 since 2011, both with over 90% vaccine uptake [
5], reduced the incidence of IPD from PCV10 serotypes by 96% among ≥ 60-year-olds between 2007 and 2019 [
6]. However, the burden of pneumococcal disease in older adults remains substantial due to an increase of non-PCV10 serotypes through serotype replacement. Indirect effects from childhood vaccination have also reduced the cost-effectiveness of pneumococcal vaccination for older adults, particularly for PCV13 [
7,
8], which saw a reduction in IPD incidence for 10 of its 13 serotypes. The cost-effectiveness of PPV23 was not substantially affected, as it covers additional serotypes. As a result, the Netherlands opted for the use of PPV23 for its program for older adults aged 60–79 years, starting in 2020. To prevent suboptimal protection due to waning immunity, revaccination is recommended every 5 years. Also, many other high-income countries recommend PPV23 for older adults, either alone or 1 year after an initial dose with PCV13.
Recognizing the limitations of PPV23 regarding its immunogenicity, higher-valency PCVs have emerged. In 2022, the European Medicine Agency licensed PCV15 for adults and children and PCV20 for adults with anticipated licensing for children. These vaccines expand on PCV13, adding two and seven serotypes, respectively. Additionally, PCV21, specifically formulated for adults, has entered phase 3 trials. PCV21 excludes nine PCV10 serotypes and introduces 11 new serotypes compared to PCV20, based on the serotypes most prevalent in adult IPD cases. Previous studies, except for one US study [
9], suggest that PCV20 is cost-effective or cost-saving for older adults compared to PPV23, PCV13, and PCV15 [
10‐
22], while PCV21 was estimated cost-saving compared to PCV20 [
23] (see Additional file
1: Table S1 for details). However, half of these studies did not account for the indirect protection resulting from a possible switch to PCV15 or PCV20 in the childhood vaccination program [
11,
13,
15‐
19,
21]. Two studies considering serotype replacement did not address the cost-effectiveness of PCV21 [
20,
22].
We evaluated the cost-effectiveness of different vaccination strategies with PPV23, PCV15, PCV20, and PCV21 in the Netherlands, taking into account indirect protection and serotype replacement from a switch to higher-valency PCVs in childhood vaccination. While focusing on the Netherlands, our findings hold wider policy implications for all countries considering higher-valency PCVs in older adults.
Discussion
We assessed the cost-effectiveness of new available PCV15 and PCV20 and the in-development PCV21 for older adults in the Netherlands from a societal perspective. We accounted for indirect protection and serotype replacement from a switch to higher-valency PCVs in childhood vaccination. We found that the impact of the indirect effects from childhood vaccination depended on the serotype overlap between the vaccines used for children and older adults. With PCV10, PCV13, and PCV15 in children, PCV20 was the cost-effective strategy for a 65-year-old cohort at a conventional Dutch threshold of €20,000/QALY gained. This strategy dominated PPV23 and PCV15 and costed approximately €10,000/QALY gained compared to no vaccination. However, with PCV20 in children, the ICER of PCV20 compared to no vaccination for the 65-year-old cohort increased to €22,250/QALY gained, which would no longer be cost-effective. As indirect effects progressed over time, the cost-effectiveness of PCV20 for older adults further diminished, and PPV23 dominated PCV20 for cohorts vaccinated 3 years after the switch to PCV20 in children. In-development PCV21 offered the highest QALY gains for older adults, and its cost-effectiveness was minimally impacted by indirect effects from PCV20 in children, thanks to its coverage of 11 different serotypes compared to PCV20.
Several countries, including the USA, decided to recommend PCV20 for older adults. While currently considered a cost-effective approach, the decision becomes less straightforward if the use of PCV20 in childhood vaccination is anticipated. Then, the long-term cost-effectiveness of pneumococcal vaccination for older adults relies on a price reduction of PCV20, or the adoption of a vaccine covering different serotypes compared to PCV20, like PCV21. Although PPV23 could eventually become cost-saving compared to PCV20 in older adults after a switch to PCV20 in children, its cost-effectiveness (and that of PCV15 + PPV23) compared to no vaccination would also be substantially diminished. Alternatively, PPV23 for older adults could be maintained until PCV21 becomes available. However, there is uncertainty around the timing and magnitudes of indirect effects, while introducing PCV20 for older adults would offer immediate health benefits. Clearly, these decisions also depend on the availability and pricing of PCV21, which are currently unknown, as well as on the flexibility in switching vaccines.
Projecting the timing and magnitudes of indirect effects proves challenging due to its unpredictable nature. We based our assumptions on observations from childhood PCV10 and PCV13 programs [
40,
41], where indirect protection on older adults began before serotype replacement. Indirect effects typically completed after 8 years, with no net effect on IPD incidence in non-vaccinated groups, though results varied between countries. Similar timescales were observed after PCV7 introduction [
54]. Our results were sensitive to the magnitude of indirect effects; PCV20 became cost-effective to the €20,000/QALY gained threshold again if indirect protection from PCV20 in childhood vaccination decreased from 80 to 40%. If only half of the IPD and NIPP cases prevented by indirect protection from PCV20 in children were replaced with non-PCV20 serotypes, the cost-effectiveness of PCV21 compared to no vaccination diminished but remained below the €20,000/QALY gained threshold if equally priced to PCV20. Note that we did not vary the distribution or invasive capacity of replacement serotypes.
Previous studies also concluded that adopting PCV15 and PCV20 in childhood vaccination reduced the cost-effectiveness of PCV15, PCV20, or PCV21 for older adults through indirect protection. However, their conclusions on the cost-effectiveness remained unchanged, often explained by the omission of a “no vaccination” strategy. The cost-effectiveness between two vaccines could hardly change if both the intervention strategy (e.g., PCV20) and reference strategy (e.g., PPV23) become less cost-effective compared to no vaccination. Two studies found that when considering serotype replacement from PCV20 in childhood vaccination, the cost-effectiveness of PCV20 compared to PPV23 was minimally affected [
20,
22]. This can be explained by the substantial overlap in serotypes between these vaccines. We demonstrated for PCV21 that serotype replacement could affect the cost-effectiveness of vaccinating older adults substantially if the vaccine offers broader protection against replacement serotypes.
Our study uses a static model that does not incorporate pneumococcal transmission dynamics. Dynamic modeling of pneumococcal carriage and transmission is complex, and competitive interactions between serotypes are not fully understood. Although dynamic transmission models are generally preferred for estimating indirect effects from childhood vaccination, the use of correction factors in a static model to adjust the incidence of infection in other age groups has often been applied successfully in economic evaluations of pneumococcal vaccines [
55]. A recent dynamic modeling exercise for Germany estimated that the continued use of PCV13 in childhood vaccination could result in a 20% increase in overall IPD incidence among ≥ 60-year-olds between 2022 and 2031 [
56]. This increase is explained by a continued rise in non-PCV13 serotypes. However, for some serotypes categories, the short-term predictions did not align well with observed IPD trends in Germany, illustrating the challenge of modeling serotype interactions. In our study, we assumed the IPD incidence in older adults to remain at the level observed in the period 2017–2019. If replacement leads to higher disease incidence in older adults than currently observed, our cost-effectiveness estimates could be considered conservative.
Our analysis has limitations. Firstly, uncertainties exist regarding the incidence and serotype distribution of hospitalized NIPP cases, the VE against NIPP, and vaccine protection duration. We addressed these with sensitivity analyses, yielding consistent outcomes across most scenarios. Secondly, we extrapolated the vaccine-type specific VE of PCV13 to PCV15, PCV20, and PCV21, despite lower immune responses in PCV15 and PCV20 for most serotypes shared with PCV13 [
57,
58]. However, both PCV15 and PCV20 met the chosen non-inferiority criteria, and the clinical relevance of these lower immune responses is unclear. Thirdly, we focused on mortality within 30 days post-diagnosis, neglecting non-hospitalized deaths. Also, literature suggests that the risk of dying from IPD may persist for at least 1 year post-diagnosis [
28], although with uncertain comorbidity factors. If included, the cost-effectiveness of pneumococcal vaccination would improve. Fourthly, we did not include the prevented burden of pneumococcal disease in primary care. However, a previous cost-effectiveness study of PCV13 for Dutch older adults in 2015 estimated that including vaccine effectiveness against NIPP in primary care changed the ICER of vaccinating adults aged 65–74 years from €8890/QALY gained to €8647/QALY gained [
46]. Therefore, we do not anticipate that this omission will significantly impact our outcomes.
Strong points of our analysis include the utilization of high-quality data from one country for many model parameters. The VE estimation for PPV23, primarily derived from UK data, aligns well with early impact estimates of the PPV23 vaccination program for older adults in the Netherlands [
59]. Also, we included a no vaccination strategy without requiring counterfactual calculations, since the Netherlands had no pneumococcal vaccination program for older adults until 2020. By taking indirect protection and serotype replacement from childhood vaccination into account, our study enhances the knowledge about the impact and cost-effectiveness of new higher-valency PCVs for older adults. This insight is valuable for all countries facing similar decisions in addressing pneumococcal diseases in older adults.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.