Cost-effectiveness and return on investment of 20-valent pneumococcal conjugate vaccine use among adults in England: analysis from the societal perspective

Background

Pneumococcal disease burden remains high in England despite use of 23-valent pneumococcal polysaccharide vaccine (PPV23) among all adults ≥65 years and those aged 18–64 years with risk conditions. We conducted cost-effectiveness (CE) and return on investment (ROI) analyses of adult vaccination with the available 20-valent pneumococcal conjugate vaccine (PCV20) to estimate societal net benefit associated use of PCV20 versus PPV23 in the current programme.

Methods

A probabilistic cohort model projected lifetime risks and costs of invasive pneumococcal disease (IPD) and all-cause community-acquired pneumonia (CAP) and expected impact of vaccination. Population included adults aged 18–64 years with risk conditions and all adults aged 65–99 years (N = 15,900,480). Inputs were based on publicly available data. Clinical outcomes included cases of IPD, CAP, and associated deaths. Economic costs included direct medical care and non-medical care costs. Analyses were conducted from the societal perspective (discounting, 3.5%/year). CE and ROI were estimated as cost per quality-adjusted life year (QALY) gained and averted total costs, respectively. An opportunity cost scenario and probabilistic sensitivity analyses were undertaken.

Results

Replacing PPV23 with PCV20 would prevent 10,196 IPD cases, 169,738 inpatient CAP cases, 74,707 outpatient CAP cases, and 27,259 pneumococcal-related deaths, increasing QALYs by 114,763. Total medical care costs decreased by £1,017·0 M and non-medical care costs by £144·0 M, and vaccination costs increased by £378·7 M. PCV20 yielded net savings of £782·3 M, and hence was dominant (vs. PPV23), with ROI of £3.1 for every £1 invested. Adjusting opportunity costs associated with suboptimal resource use in periods of excess demand on National Health Service increased total savings to £1,864·2 M and ROI to £5.9 per £1 invested.

Conclusions

Our findings suggest that use of PCV20—in lieu of PPV23—among all adults aged ≥65 years and those aged 18–64 years with underlying risk conditions in England will be cost-saving and have a substantial net benefit to society.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12962-025-00709-6.

Key summary points

• Why carry out this study?

• • ◦ Despite existing adult vaccination programmes, pneumococcal disease burden among adults in the UK remains high with pneumococcal infections being the main cause of pneumonia-related adult hospitalisations pre-COVID-19 pandemic.
    ◦ Novel, higher valency (≥20 serotypes) pneumococcal conjugate vaccines—which elicit a more durable immune response than 23-valent pneumococcal polysaccharide vaccine (PPV23) and provide protection against non-bacteraemic pneumonia—have the potential to address the substantial pneumococcal disease burden in the UK is used at coverage levels similar to currently recommended PPV23.
    ◦ We evaluated the cost-effectiveness and return on investment (ROI) of 20-valent pneumococcal conjugate vaccine (PCV20) to help inform decision makers in England and elsewhere who are considering use of novel pneumococcal vaccines to protect against pneumococcal disease.

• What was learned from the study?

• • ◦ Results suggest implementing PCV20 alone as a replacement for the current PPV23 programme among all adults aged ≥65 years and adults aged 18–64 years with underlying conditions in England would be cost-saving and likely to offer ROI of £3.1 per £1 invested by generating additional cost savings, from a societal perspective.
    ◦ Compared to previously published cost-effectiveness analyses from the healthcare system perspective, this societal perspective analysis suggests that cost savings from PCV20 would be 5 times higher.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12962-025-00709-6.

Introduction

Streptococcus pneumoniae is the main cause of invasive pneumococcal disease (IPD) and a leading cause of community-acquired pneumonia (CAP), both of which are associated with significant morbidity and mortality in children and adults in the United Kingdom (UK) [1, 2]. Routine vaccination against S. pneumoniae is currently recommended by the Joint Committee on Vaccination and Immunisation (JCVI) for infants, at-risk children and adults (i.e., aged ≥2 years), and all older adults (i.e., aged ≥65 years) in the UK to reduce the risk of pneumococcal infection [3]. The 23-valent pneumococcal polysaccharide vaccine (PPV23) is recommended for all people aged ≥2 years with underlying medical conditions who have high risk of serious disease. Although PPV23 coverage is high, surveillance data suggest that its use has provided only low to moderate short-term direct protection against IPD, and has not had a population-level impact on IPD burden given that serotypes unique to PPV23 continue to account for a large proportion of IPD among older adults [4, 5]. In children aged < 2 years, the 13-valent pneumococcal conjugate vaccine (PCV13) has been recommended since 2010 (the 7-valent PCV [PCV7] was in use between 2006 and 2009). Childhood PCV immunizations have prevented a significant number of cases of pneumococcal disease and associated deaths through the direct protection of young children and indirect protection of older children and adults [6, 7]. Nonetheless, the burden of both IPD and CAP remains substantial among adults [7–9].

A novel 20-valent PCV (PCV20) has been recently recommended by the JCVI for use in adults as an alternative to PPV23 [10]. PCV20 includes the serotypes in PCV13 (1, 3, 4, 5, 6A, 6B, 7F, 9 V, 14, 18C, 19A, 19F, and 23F) plus seven additional serotypes (8, 10A, 11A, 12F, 15B, 22F, and 33F). In a 2022 cost-effectiveness analysis considering only direct medical costs from hospitalised cases, we showed that vaccinating all adults aged ≥65 years and those aged 18–64 years with underlying comorbidities in England with PCV20 alone instead of PPV23 would result in estimated savings of £230 million to the National Health Service (NHS), as vaccination costs were entirely offset by cost savings from averted hospitalisations [11].

In addition to offering direct and indirect protection against pneumococcal infection and severe outcomes, and reducing the economic burden to the healthcare system, vaccinations are associated with broader societal benefits, such as the prevention of illness-related productivity losses [12–14]. These public health benefits, however, are often not reflected in standard cost-effectiveness evaluations. Therefore, return on investment (ROI) analyses—which exclusively focus on tangible direct and indirect costs to society at large—are used to estimate the cost savings to the healthcare system and society from use of novel immunisation strategies and support healthcare decision-making [15]. Moreover, research suggests that vaccine-preventable hospitalisations could be considered a suboptimal use of hospital beds, especially during periods when demand exceeds capacity, which in turn would increase the opportunity costs associated with the hospitalisation, a component NHS reference costs fail to account for [16, 17]. These higher costs may be especially relevant in the context of pneumonia, which tends to peak during winter when many hospitals are at capacity. A study by Neri et al. [18] suggests that opportunity costs due to suboptimal bed use are twice the cost of inpatient admissions due to pneumococcal disease, influenza, and respiratory syncytial virus. If accounted for, opportunity costs may reflect a broader consideration of economic burden of vaccine-preventable disease, which is relevant for policymakers. The aim of the current study is to estimate the cost-effectiveness and ROI of a PCV20 vaccination programme for adults in England aged ≥65 years and those aged 18–64 years with underlying conditions from the societal perspective.

Methods

Our methodology was similar to that employed in our previously published cost-effectiveness analysis [11], however, the present study included cases of outpatient CAP and associated costs, costs of treatment in the intensive care unit (ICU), and indirect costs associated with disease in all base case analyses.

Modelling framework

The lifetime risk of clinical outcomes and associated economic costs of IPD and all-cause CAP were projected using a probabilistic cohort model with a Markov-type process. Fig. 1 depicts the model structure adapted from Mendes et al. [11] The decision model considers a hypothetical population of all adults aged ≥65 years and adults aged 18–64 years with underlying conditions in the UK. The model population is stratified at model entry by age (in one-year increments) and risk profile (i.e., low, moderate- [immunocompetent with underlying medical conditions], and high-risk [immunocompromised]). Persons in the model population may receive PCV20 alone, PPV23 alone, or no vaccine at model entry, depending on the vaccination strategy and coverage.

Fig. 1.

Model schematic

Expected clinical outcomes and economic costs were projected annually for the model population, based on age, risk profile, disease/fatality rates, vaccination status, and medical care, non-medical and vaccination costs. Clinical outcomes and economic costs projected for each vaccination strategy include IPD and all-cause CAP cases and attributable deaths, life-years (unadjusted and quality-adjusted), vaccination costs, medical care costs, and non-medical costs. IPD included bacteraemia and meningitis, and all-cause CAP was stratified by care setting (inpatient and outpatient). Vaccinated persons were at lower risk of future IPD and all-cause CAP; the magnitude of vaccine-associated risk reduction depended on clinical presentation, vaccine type, age, time since vaccination, and risk profile. Risk of death from IPD, all-cause CAP, and other causes were dependent upon age and risk profile. Expected medical treatment costs for IPD and all-cause CAP were generated based on event rates and unit costs according to care setting; cost of intensive care unit (ICU) care was incorporated in inpatient costs. Vaccination costs were tallied in the year of vaccine administration (i.e., model year 1). Non-medical costs include paid and unpaid productivity loss due to illness and loss of future earnings.

Model estimation

A summary of methods employed to estimate key model parameters is presented below with estimates provided in Table 1.

Table 1.

Base case model input values

	Age and Risk Profile
	18–49 Years		50–64 Years		65–74 Years			75–84 Years			85–99 Years
	Moderate	High	Moderate	High	Low	Moderate	High	Low	Moderate	High	Low	Moderate	High
No. of adults (in thousands)	1,318.3	472.3	1,801.1	1,328.7	3,365.7	1,211.6	956.1	2,066.5	1,053.6	857.4	630.1	458.2	380.9
Annual disease incidence (per 100,000)
Bacteraemia	18.5	41.6	17.7	49.0	4.8	18.8	62.3	8.1	21.9	94.7	15.0	40.5	175.5
Meningitis	2.1	4.7	1.7	4.8	0.3	1.1	3.6	0.3	0.7	3.0	0.1	0.3	1.4
All-cause hospitalised CAP	247.7	452.4	700.1	1,278.5	769.6	2,301.1	4,202.0	939.9	2,810.3	5,131.9	1,499.5	4,483.4	8,187.1
All-cause outpatient CAP	173.2	316.4	398.3	727.4	290.8	869.4	1,587.7	471.2	1,409.0	2,573.0	624.4	1,867.0	3,409.3
Annual mortality/case-fatality (per 100)
General population	0.1	0.2	0.6	0.8	1.0	1.5	2.0	2.8	4.3	5.7	9.2	13.8	18.4
Bactaeremia	5.9	7.1	12.8	15.4	14.0	17.4	20.9	22.1	27.6	33.1	32.4	40.6	48.7
Meningitis	5.9	7.1	12.8	15.4	14.0	17.4	20.9	22.1	27.6	33.1	32.4	40.6	48.7
All-cause hospitalised CAP	3.4	4.1	6.7	8.0	7.9	9.9	11.8	10.3	12.9	15.5	14.5	18.1	21.7
All-cause outpatient CAP	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Medical care costs
ICU carea	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125	£16,125
Bacteraemia	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746	£6,746
Meningitis	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980	£7,980
All-cause hospitalised CAP	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903	£5,903
All-cause outpatient CAP	£43	£43	£43	£43	£43	£43	£43	£43	£43	£43	£43	£43	£43
Non-medical costs
Average work loss (days)
ICU careb	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0
General ward
Bacteraemia	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0
Meningitis	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0	11.0
All-cause inpatient CAP	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0	7.0
All-cause outpatient CAP	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5	3.5
Average annual earningsc	£37,682	£37,682	£27,935	£27,935	£3,873	£3,873	£3,873	£309	£309	£309	£175	£175	£175
Average daily unpaid work time (min)	201	201	181	181	174	174	174	174	174	174	174	174	174
General popuation health-state utilities	0.7745	0.7346	0.7234	0.6852	0.9283	0.7308	0.7082	0.8921	0.6799	0.6607	0.8191	0.6038	0.5643
Disutilities
IPD and all-cause hospitalised CAP	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300	0.1300
All-cause outpatient CAP	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040	0.0040
Vaccine uptake
PCV20	43.4%	37.5%	43.4%	37.5%	42.9%	79.0%	83.7%	73.3%	89.6%	95.5%	77.9%	85.9%	85.7%
PPV23	43.4%	37.5%	43.4%	37.5%	42.9%	79.0%	83.7%	73.3%	89.6%	95.5%	77.9%	85.9%	85.7%
VE PCV vs. VT-IPD
Year 1	81.5%	65.2%	79.2%	63.3%	75.0%	75.0%	60.0%	75.0%	75.0%	60.0%	75.0%	75.0%	60.0%
Year 5	81.5%	65.2%	79.2%	63.3%	75.0%	75.0%	60.0%	75.0%	75.0%	60.0%	75.0%	75.0%	60.0%
Year 10	63.1%	50.5%	61.2%	49.0%	58.0%	58.0%	46.4%	58.0%	58.0%	46.4%	58.0%	58.0%	46.4%
Year 15	37.2%	29.8%	36.2%	28.9%	34.3%	34.3%	27.4%	34.3%	34.3%	27.4%	34.3%	34.3%	27.4%
Year 16+	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%
VE PCV vs. VT-CAP
Year 1	55.6%	44.5%	51.3%	41.1%	45.0%	45.0%	36.0%	45.0%	45.0%	36.0%	45.0%	45.0%	36.0%
Year 5	55.6%	44.5%	51.3%	41.1%	45.0%	45.0%	36.0%	45.0%	45.0%	36.0%	45.0%	45.0%	36.0%
Year 10	43.0%	34.4%	39.7%	31.8%	34.8%	34.8%	27.9%	34.8%	34.8%	27.9%	34.8%	34.8%	27.9%
Year 15	25.4%	20.3%	23.5%	18.8%	20.6%	20.6%	16.4%	20.6%	20.6%	16.4%	20.6%	20.6%	16.4%
Year 16+	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%
VE PPV23 vs. VT-IPD
Year 1	32.7%	17.0%	32.4%	16.9%	55.9%	31.1%	16.2%	50.8%	28.2%	14.7%	37.6%	20.9%	10.9%
Year 5	24.9%	13.0%	24.7%	12.8%	42.6%	23.7%	12.3%	38.7%	21.5%	11.2%	28.7%	15.9%	8.3%
Year 10+	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%

Notes: CAP = community-acquired pneumonia; IPD = invasive pneumococcal disease; PCV20 = 20-valent pneumococcal conjugate vaccine; PPV23 = 23-valent pneumococcal polysaccharide vaccine; VE = vaccine efficacy; VT = vaccine type

^aAdditional cost of ICU care included for 7.4% of IPD (bacteremia and meningitis) cases and 7.4% of all-cause inpatient CAP cases among persons aged 18–64 years, and 3.4% of IPD and 3.4% of all-cause inpatient CAP cases among persons aged ≥65 years

^bAdditional ICU days included for 7.4% of IPD (bacteremia and meningitis) cases and all-cause inpatient CAP cases among persons aged 18–64 years, and 3.4% of IPD and all-cause inpatient CAP cases among persons aged ≥65 years

^cIncludes all persons irrespective of status of labor force participation

Population

The model population included adults aged 18–64 years with underlying conditions and all adults aged ≥65 years (N = 15.9 M). Population size and age distribution were based on the estimated population in England in 2024 [19]. Persons in each age group were allocated into low-, moderate, and high-risk subgroups based on risk factors defined in the Green Book [20] and published risk distributions [21]. The moderate-risk subgroup includes immunocompetent persons with conditions such as chronic respiratory disease, chronic heart disease, chronic liver disease, and diabetes. High-risk conditions include chronic kidney disease, immunosuppression, cochlear implants, and cerebrospinal fluid (CSF) leak.

Disease rates

Annual incidence rate of IPD was estimated by age (in one-year increments) and risk profile using age-specific disease rates from Ladhani et al. [6] and Djennad et al. [5], and age-specific distributions by risk profile and age-specific odds ratios for IPD by risk profile from van Hoek et al. [21]. IPD rates were apportioned between bacteraemia and meningitis based on a study by Amin-Chowdhury et al. [22].

Annual incidence rate of all-cause inpatient CAP was estimated by age and risk profile using age-specific rates of radiologically confirmed pneumonia presented in Hyams et al. [23], age-specific distributions by risk profile from van Hoek et al. [21], and age-specific fully adjusted ORs for all-cause inpatient CAP by risk profile from Campling et al. [24].

Annual rates of all-cause outpatient CAP were estimated by age (in one-year increments) and risk profile using age-specific rates of clinically diagnosed ambulatory pneumonia from Sun et al. [25], age-specific distributions by risk profile from van Hoek et al. [21], and age-specific fully adjusted OR for all-cause inpatient CAP by risk profile from Campling et al. [24].

As the published studies presented point-estimates for selected age groups, the reported point-estimates were anchored to selected ages within the age groups and methods of linear interpolation/extrapolation were employed to derive IPD rates across ages and risk profiles.

Mortality rates

IPD case fatality rates (CFRs) were estimated by age (in one-year increments) and risk profile using age-specific 30-day CFRs from Amin-Chowdhury et al. [22], age-specific distributions by risk profile from van Hoek et al. [21], and risk profile-specific relative rates of death from Lexau et al. [26] (1.25 for moderate-risk and 1.5 for high-risk persons [vs. low-risk persons]).

Inpatient CAP CFRs were estimated by age (in one-year increments) using age-specific 30-day CFRs from Shi et al. [27]. Because age-specific 30-day CFRs from Shi et al. [27] did not include estimates for persons aged < 65 years, estimates for the 18–49 and 50–64 year old age groups, respectively, were derived based on the absolute difference in the CFRs for persons aged 75–84 vs. 65–74 years. Age-specific estimates were then apportioned across risk groups based on van Hoek et al. [21] and relative rates of death from Lexau et al. [26]. Persons experiencing outpatient CAP were assumed not to be at elevated risk of mortality.

General population mortality rates were estimated using age-specific projected death rates for the 2022 in England [28], age-specific distributions by risk profile from van Hoek et al. [21], and assumed risk profile-specific relative rates of deaths [21]. General population mortality estimates accounted for deaths due to IPD and all-cause inpatient CAP.

Vaccine-type effectiveness

For PCV20, vaccine effectiveness (VE) against vaccine-type (VT) disease was assumed to be the same as that for PCV13 based on published results of an evaluation of PCV20 safety, tolerability, and immunogenicity [29]. For low- and moderate-risk persons aged ≥65 years, effectiveness of PCV20 against vaccine-type IPD (VT-IPD) and vaccine-type CAP (VT-CAP) in year 1 of the model (“initial VE”) was derived using data for adults aged ≥65 years in the Community-Acquired Pneumonia Immunization Trial in Adults (CAPiTA) per-protocol population [30]. For persons aged 50–64 years, initial VE was based on post-hoc analyses of CAPiTA data [31]. For persons aged 18–49 years, initial VE was assumed to be the same as that for persons aged 50 years. Initial VE against VT-IPD and VT-CAP for high-risk persons aged 18–99 years was assumed to be equal to 80% of corresponding values for low-/moderate-risk persons [32, 33]. Initial VE-PCV was assumed to persist for 5 years, consistent with CAPiTA [30, 34], and to wane thereafter, as follows: 5% annual decline during years 6–10, 10% annual decline during years 11–15, and no efficacy from year 16 through the end of the modelling horizon [31].

Risk-specific VE-PPV23 against VT-IPD for persons aged ≥18 years was derived from a real-world effectiveness study conducted among UK adults by Djennad et al. [5] VE-PPV23 against VT-CAP was assumed to be 0% based on published sources, and consistent with base case assumptions employed in a number of economic studies and previous JCVI statements [35–51]. Beyond year 1 of the modelling horizon, VE-PPV23 against VT-IPD was assumed to decline linearly across all age and risk groups, as follows: decline to 76.2% of initial VE by year 5, and decline to no efficacy by year 10 [5].

Serotype coverage

For PCV20-type disease, age-specific vaccine serotype coverage for IPD was based on 2016/2017 surveillance data collected by Ladhani et al. [6] for PPV23, serotype coverage reported in Ladhani et al. [6] was adjusted to reflect all 23 serotypes based on data from Djennad et al. [5] (Supplemental Table 1). Vaccine serotype coverage for all-cause inpatient and outpatient CAP was based on serotype coverage for pneumococcal CAP and the proportion of CAP due to pneumococcus in 2018–2020 as reported in Lansbury et al. [52]. In year 1, serotype coverage for all-cause CAP was assumed to not vary by age (Supplemental Table 2).

As a paediatric immunisation programme for PCV20 is assumed to be implemented in model year 3, herd effects for serotypes unique to PCV20 (vs. PCV13) were assumed to begin one year after the implementation of the immunisation programme (i.e., model year 4). Serotype coverage for PPV23 was also reduced to account for the impact of herd effects on serotypes common to PPV23 and PCV20. Reductions in serotype coverage due to herd effects from childhood vaccination with PCV20 were based on the age-specific observed impact of PCV13 on the 5 additional serotypes in PCV13 not in PCV7 (excluding serotype 3) [6]. Reduction in serotype coverage was assumed to peak by year 10 and to be sustained at that level thereafter. Serotype replacement and herd effects from adult PCV use were assumed to not occur [53–55].

Utility values and decrements

Age- and risk-specific health-state utility values were derived from data reported by Ara et al. [56]. Disutility due to IPD and all-cause inpatient CAP was assumed to be 0.13 in the year of occurrence, irrespective of age and risk profile, based on Mangen et al. [57]. For all-cause outpatient CAP, utility decrement was assumed to be 0.004 based on Melegaro et al. [58].

Direct costs

Medical care costs per episode in the general ward for IPD (bacteraemia, £6,746; meningitis, £7,980) and all-cause inpatient CAP (£5,903) were derived based on average costs per finished consultant episode (FCE) from the 2021/2022 NHS reference costs [59] and the average number of FCEs per admission from the Hospital Episode Statistics for England [9]. Per day cost of care in the ICU (£2,304) was derived from 2021/2022 NHS reference costs [59] and mean length of stay (LOS) for ICU cases was assumed to be 7 days based on Campling et al. [60]. Total medical care costs were estimated as a weighted average of general ward (18–64 years, 92.6%; 65–99, 96.6%) and ICU cases (18–64 years, 7.4%; 65–99, 3.4%) [60]. The cost of all-cause outpatient CAP (£43) was based on the 2022 unit cost per surgery consultation from the Personal Social Services Research Unit (PSSRU) [61]. All costs were inflated to 2023 GBP based on consumer price index (CPI) for health services and medical services [62, 63].

Prices of PPV23 (£16.80) and PCV20 (£56.50) were based on the list price from the British National Formulary (2023) [64]. Vaccine administration cost (£10.06) was based on the General Practice Contract (2019) [65].

Indirect costs

Non-medical costs (i.e., indirect costs) included productivity loss (paid and unpaid) due to illness and loss of future earnings due to disease-related mortality. To estimate expected productivity loss (i.e., work loss) due to illness, duration of illness was multiplied by age-specific average daily earnings for the UK population (derived by dividing average annual earnings reported by the National Transfer Accounts Project [66] by 365 days; earnings were inflated to 2023 GBP [67]). Duration of illness for inpatient cases was estimated based on mean LOS per episode [60, 68]; for all-cause outpatient CAP, duration of illness was assumed to be half that of an inpatient CAP episode. Loss of unpaid work due to illness was calculated using the duration of illness, average daily minutes of unpaid work time reported by the UK Office of National Statistics [69], and 2023 hourly wages from the Annual Survey of Hours and Earnings [70]. Lastly, loss of future earnings due to disease-related mortality was estimated using age-specific life expectancy and average annual earnings (i.e., paid productivity).

Analyses

Cost-effectiveness

Clinical outcomes and economic costs were projected for the model population considering PCV20 compared with PPV23. Vaccine coverage for each strategy was assumed on par with 2019–2020 PPV23 coverage levels [4]. Future costs, life-years (LYs), and QALYs were discounted at 3.5% annually; analyses were conducted from the societal perspective.

ROI

To calculate the ROI associated with PCV20 use, the incremental direct and indirect costs were summed, multiplied by −1, and divided by the additional vaccination costs incurred by vaccination with PCV20. An ROI of > 1 would indicate that the savings due to reduction in costs are greater than the additional costs of vaccination.

Scenario

Scenario analysis #1 considered medical care costs with suboptimal excess-demand opportunity costs (i.e., 2x base-case inpatient care costs based on Neri et al. [18]) for hospital admissions in periods where demand exceeds capacity to capture the true cost of hospital beds during periods of scarcity when vaccine-preventable admissions are suboptimal. Scenario analysis #2 assumed there were no herd effects from paediatric immunisation with PCV20 (i.e., serotype coverage remains the same throughout the modelling horizon with no reductions in overall rates of disease). Scenario analysis #3 assumed that all reductions in PCV20 non-PCV13-type pneumococcal disease would be countered by increases in pneumococcal disease caused by other serotypes; replacement was allocated across PPV23 non-PCV20 serotypes and non-vaccine serotypes (i.e., not included in either vaccine) proportionally to the distribution of serotypes in year 1. Scenario analysis #4 employed alternative VE-PCV20 estimates based on published real-world effectiveness studies for PCV13 [71, 72]; no adjustment was made for the increased serotype coverage in PCV20.

Sensitivity

Probabilistic sensitivity analyses (PSA) were conducted (1,000 replications) comparing PCV20 with PPV23 to account for uncertainty surrounding estimates of key model parameters. For each input parameter, the appropriate distribution was determined based on information available in the source material (Supplemental Table 3). The results from PSA were also used to estimate the incremental cost-effectiveness ratio (ICER) and ROI.

Ethics compliance

This study was conducted using publicly available data and previously conducted studies. It does not contain any new studies with human participants or animals performed by any of the authors, and hence was not subject to institutional review.

Results

Base case

Compared with use of PPV23, PCV20 alone would prevent an additional 10,196 IPD cases, 169,738 all-cause inpatient CAP cases, 74,707 all-cause outpatient CAP cases, and 27,259 pneumococcal-related deaths (Table 2). With PCV20 use, medical care costs for general ward cases, ICU cases, and outpatient cases were lower by £919.9 million, £94.4 million, and £2.7 million, respectively (vs. PPV23). With total vaccination costs higher by £378.7 million, use of PCV20 would save the NHS £638.3 million in healthcare costs. From the societal perspective, use of PCV20 would lower non-medical costs by £144.0 million. With total net costs (including medical care, non-medical and vaccination costs) lower by £782.3 million and discounted QALYs higher by 114,763, PCV20 was dominant versus PPV23 from the societal perspective. The ROI was estimated to be £3.1 per £1 spent.

Table 2.

CE and ROI for PCV20 versus PPV23 for eligible UK adults (N = 15,900,480)

I. Base Case		PPV23	PCV20	Difference
Clinical Outcomes
	No. of Cases
	IPD	120,922	110,726	−10,196
	All-Cause CAP
	Inpatient	7,704,640	7,534,902	−169,738
	Requiring Outpatient Care Only	3,371,158	3,296,451	−74,707
	No. of Deaths
	IPD	39,281	36,095	−3,186
	All-Cause CAP	1,166,576	1,142,503	−24,073
	Life-Years (discounted)	185,795,987	185,945,098	149,111
	Quality-Adjusted Life-Years (discounted)	136,785,246	136,900,009	114,763
Economic Outcomes
	Total Costs (millions)
	Medical Care (General Ward)	£28,732.6	£27,812.7	-£919.9
	Medical Care (ICU)	£2,992.6	£2,898.2	-£94.4
	Medical Care (Outpatient)	£89.5	£86.8	-£2.7
	Vaccination	£256.2	£634.9	£378.7
	Non-Medical	£4,545.5	£4,401.5	-£144.0
	Medical + Vaccination Costs	£32,071.0	£31,432.7	-£638.3
	Medical + Vaccination + Non-Medical Costs	£36,616.5	£35,834.2	-£782.3
Cost-Effectiveness (Societal)
	Cost per LY	—	—	Dominant
	Cost per QALY	—	—	Dominant
ROI for every £1 spent				£3.1
II. Scenario Analysis #1 (opportunity costs)		PPV23	PCV20	Difference
Economic Outcomes
	Total Costs (millions)
	Medical Care (General Ward)	£59,380.8	£57,479.7	-£1,901.1
	Medical Care (ICU)	£6,184.8	£5,989.7	-£195.0
	Medical Care (Outpatient)	£89.5	£86.8	-£2.7
	Vaccination	£256.2	£634.9	£378.7
	Non-Medical	£4,545.5	£4,401.5	-£144.0
	Medical + Vaccination Costs	£65,911.3	£64,191.1	-£1,720.2
	Medical + Vaccination + Non-Medical Costs	£70,456.8	£68,592.6	-£1,864.2
Cost-Effectiveness (Societal)
	Cost per LY	—	—	Dominant
	Cost per QALY	—	—	Dominant
ROI for every £1 spent				£5.9

Notes: CE = cost-effectiveness; ROI = return on investment; eligible population included moderate- and high-risk adults aged 18–64 years and all adults aged 65–99 years in the UK (N = 15,900,480)

Scenario

With suboptimal excess-demand opportunity costs included (Scenario #1), total medical care costs were lower by £1,901.1 million (including general ward, ICU, and outpatient costs) and thus yielded greater savings from the societal perspective (£1,864.2 M). The estimated ROI considering opportunity cost was £5.9 per £1 spent. Without consideration of herd effects from paediatric PCV20 immunisation (Scenario #2), PCV20 alone would prevent 13,007 IPD cases, 200,853 all-cause inpatient CAP cases, 88,378 all-cause outpatient CAP cases, and 32,658 pneumococcal-related deaths (Supplemental Table 4). With total net costs lower by £977.7 million and discounted QALYs higher by 134,016, PCV20 remained dominant versus PPV23 from the societal perspective; the estimated ROI increased to £3.6 per £1 spent. Results from Scenarios #3–4 can be found in Supplemental Tables 5 and 6.

Sensitivity

In PSA, PCV20 versus PPV23 was cost saving in 100% of simulations (WTP = £20,000/QALY) (Fig. 2). The mean ROI based on the PSA was estimated to be £3.1 (95% credible interval: £2.5 - £3.9) per £1 spent (Fig. 3).

Fig. 2.

PSA results: CE scatterplot for PCV20 versus PPV23 for eligible UK adults (N = 15,900,480). Notes: CE = cost-effectiveness; eligible population included moderate- and high-risk adults aged 18–64 years and all adults aged 65–99 years in the UK (N = 15,900,480); results are presented from the societal perspective

Fig. 3.

PSA results: ROI/£1 spent on PCV20 vaccination programme (versus PPV23) for eligible UK adults (N = 15,900,480). Notes: ROI = return on investment; PSA = probabilistic sensitivity analysis; eligible population included moderate- and high-risk adults aged 18–64 years and all adults aged 65–99 years in the UK (N = 15,900,480); results are presented from the societal perspective; the area graph depicts the density of data distribution from the probabilistic sensitivity analysis; grey bars present the 95% confidence interval intersecting with the error bars at the lower bound (LB) and upper bound (UB); box plot represents the interquartile range and median; scatterplot shows the spread of values

Discussion

The findings from our analysis of the cost-effectiveness of pneumococcal vaccination suggest replacing PPV23 with PCV20 for vaccination of all adults aged ≥65 years as well as adults aged 18 to 64 years with underlying conditions in England would reduce disease burden, save lives, and yield further cost savings from the societal perspective. These findings build on those previously published by Mendes et al. [11], which found PCV20 to be cost saving from the healthcare system perspective considering the impact of pneumococcal vaccination on hospitalised cases of IPD and all-cause CAP only. Our current analysis accounts for the broader disease burden (i.e., including outpatient CAP), more recent direct costs (including cost of ICU care), and includes societal costs, indicating that the cost savings are about 5 times those yielded in previously published analyses conducted from the healthcare system perspective.

In addition to the cost-effectiveness analysis, we estimated the potential savings associated with the implementation of PCV20 vaccination—relative to PPV23 alone (single dose)—using the ROI approach. Our results suggest that for every £1 invested to replace PPV23 with PCV20 in the adult pneumococcal vaccination programme in England, the ROI would be £3.1 from a societal perspective. Analysis incorporating opportunity costs of suboptimal vaccine-preventable hospitalisations in periods of excess demand would increase ROI to £5.9 per £1 invested, suggesting that the true value of vaccination which may be otherwise underestimated.

A similar cost-effectiveness analysis conducted in Belgium—which also accounted for the broader societal impact of vaccinations—suggests that use of PCV20 alone among adults aged ≥65 years and those aged < 65 years with risk factors would be cost-saving compared to the existing adult pneumococcal vaccines recommendation (i.e., 15-valent PCV + PPV23) [73]. Moreover, studies evaluated in a recent systematic literature review by Cho et al. also indicated that even from a healthcare system or a restricted societal perspective (i.e., excluding broader societal benefits) PCV20 was generally cost-effective or cost-saving compared to PPV23 [74]. Our finding that PCV20 is a cost saving replacement for PPV23 is based on analyses that account for the latest pre-COVID serotype coverage for pneumococcal disease among adults. Cost savings are largely attributable to the expected greater and longer-lasting protection provided by PCV20 against both IPD and pneumococcal CAP, offsetting both the increased cost of vaccination as well as the 7% IPD serotype coverage advantage of PPV23 (PPV23 non-PCV20 serotypes: 2, 9N, 17F, and 20; PCV20 non-PPV23 serotypes: 6A/6C) [5, 6, 8, 29, 30, 34, 51].

Additionally, studies have emphasized the need to move beyond cost-effectiveness analysis to better account for the broader benefits of vaccination programmes in terms of value of investment [12–15, 75–78]. One such evaluation of ROI from immunisation programmes in 94 low- and middle-income countries was conducted by Sim et al. [76] in an effort to inform funding replenishments and identify areas for resource allocation (e.g., procurement, data management, human resources training) to ensure the sustainability of immunisation programmes [76, 79, 80]. This ROI analysis employed methods similar to previously published assessments and adds to the growing literature on ROI for vaccination programmes [15, 78, 81, 82].

In the present study, conservative assumptions were employed for model parameters, where appropriate. For example, we assumed PCV20 would be added to the paediatric immunisation programme in 2026 and that herd effects would begin in 2027, despite uncertainty of its introduction. As shown in the scenario analysis without herd effects, in the event that PCV20 is not introduced in the national paediatric immunisation programme, the cost savings and ROI from adult PCV20 vaccination would be even greater [83]. Long-term clinical consequences and costs of pneumococcal disease also were not included though common following acute illness [84–86]. Despite evidence suggesting that risk of death may remain elevated for several years following an acute hospitalisation for IPD or CAP [85–87]—especially among the elderly and those with comorbidities (i.e., 65% of the English population) [21, 24, 28, 87]—we considered only in-hospital pneumonia deaths.

Additional limitations of our model include estimation of selected inputs and model assumptions. For incidence and serotype coverage estimates, we relied primarily on pre-COVID-19 data. Although more recent post-COVID-19 epidemiological data on pneumococcal disease are beginning to emerge [88, 89], it is not yet known whether these estimates are reflective of a truly stabilized post-pandemic environment. Given mounting evidence suggesting that incidence rates and serotype coverage distributions are returning to pre-pandemic levels/patterns [2, 88, 89], we assumed the pre-COVID-19 estimates to be more representative of pneumococcal disease during our modelling horizon [5, 6, 23, 52]. We therefore employed the same hospitalised IPD and CAP incidence rates as in the 2022 study by Mendes et al. [11] (i.e., data from Djennad et al. [5] and Ladhani et al. [6] for IPD and Hyams et al. [23] for CAP). While these estimates suggest that CAP incidence is 3 to 5 times higher than estimates employed in the 2016 JCVI analyses, they remain comparable to more recently published data by Sharp et al. [2], particularly for older age groups [8, 23, 90, 91]. Additionally, as noted in Mendes et al., PCV20 use remained cost effective in scenario analysis employing lower CAP incidence rates (cost/QALY = £6,473; willingness-to-pay threshold = £20,000) [11]. If lower incidence rates were employed in the present study, ROI from PCV20 use would be somewhat lower.

Similarly, we employed pre-COVID-19 IPD and CAP serotype coverage estimates. Although recent data on vaccine serotype coverage for IPD is available in the literature [89], the serotype distributions in pre- and post-COVID-19 data are largely similar. While there are more recent data on CAP serotype coverage and the proportion of CAP that is due to pneumococcus, these data were collected partly during the COVID-19 lockdowns (i.e., November 2021-November 2022), a period in which the epidemiology of respiratory infections varied considerably from historic trends [88]. Moreover, researchers believe that the proportion of CAP due to pneumococcus will rebound, similar to what has been observed for IPD [88]. Thus, lacking robust post-COVID-19 CAP serotype coverage data and in order to be consistent with methods employed for IPD serotype coverage, we employed CAP serotype coverage based on data spanning April 2018 – March 2020 [52]. Continued monitoring of the evolution of IPD and CAP serotype coverage as the effects of the pandemic wane over time will be necessary to inform future studies on the public health impact and cost-effectiveness of pneumococcal vaccines.

Additionally, we assumed that PPV23 conferred no benefit against CAP in the base case consistent with findings of multiple meta-analyses, economic evaluations, and the 2015 JCVI position [35–51, 92]. We note, however, that vaccination with PCV20 versus PPV23 remained dominant in the scenario analysis by Mendes et al. considering non-zero VE vs. VT-CAP [11]. Our analyses also do not account for possible serotype replacement following widespread paediatric uptake of PCV20, though the impact of serotype replacement on cost-effectiveness of PCV20 vs. PPV23 would be expected to be minimal, even if all serotype replacement occurred among PPV23 non-PCV20 serotypes. Our model also does not account for additional medical care costs incurred due to life-years gained with use of PCV20 (vs. PPV23); had these costs been included, they would have somewhat offset cost savings. Lastly, vaccine prices employed in the current analysis were NHS indicative prices and are subject to change as vaccines for national immunisation programme are centrally procured by the Department of Health and Social Care DHSC [93] and net prices agreed via tender are confidential. The estimated ROI may be smaller or larger depending on the difference between tender prices and assumed prices for both vaccines. Similarly, changes to medical care and indirect costs (e.g., increases due to inflation), could also impact the magnitude of cost savings and ROI with PCV20 use.

Conclusions

Our findings highlight the substantial public health impact and cost savings with use of PCV20—in lieu of PPV23—for prevention of pneumococcal disease among all adults aged ≥65 years and adults aged 18–64 years with underlying conditions in England. Inclusion of societal costs (i.e., current and future disease-related loss of earnings) and consideration of broader disease burden suggest that the cost savings from PCV20 adult vaccination would be approximately 5 times greater than previously published estimates which were limited to the healthcare perspective. Moreover, the estimated ROI suggests that, from a societal perspective, the use of PCV20 would yield a return of £3.1 per £1 invested. Future updates of PCV20 evaluation should incorporate post-pandemic epidemiological data when they become available.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CAP

Community-acquired pneumonia

CAPiTA

Community-Acquired Pneumonia Immunization Trial in Adults

CFR

Case-fatality rate

CPI

Consumer price index

DHSC

Department of Health and Social Care

FCE

Finished consultant episode

ICER

Incremental cost-effectiveness ratio

ICU

Intensive care unit

IPD

Invasive pneumococcal disease

JCVI

Joint Committee on Vaccination and Immunisation

LOS

Length of stay

LY

Life year

NHS

National Health Service

PCV7

7-valent pneumococcal conjugate vaccine

PCV13

13-valent pneumococcal conjugate vaccine

PCV15

15-valent pneumococcal conjugate vaccine

PCV20

20-valent pneumococcal conjugate vaccine

PPV23

23-valent pneumococcal polysaccharide vaccine

PSA

Probabilistic sensitivity analysis

PSSRU

Personal Social Servies Research Unit

QALY

Quality-adjusted life year

ROI

Return on investment

UK

United Kingdom

US

United States

VE

Vaccine effectiveness

VT

Vaccine-type

Authors contributions

Authorship was designated based on guidelines promulgated by the International Committee of Medical Journal Editors (2004). All persons who met criteria for authorship were listed as authors on the title page. The contribution of each of these persons to this study is as follows: conceptualization, all authors; acquisition of data, A.A., D.M., D.H.; formal analysis and interpretation of data, all authors; preparation of manuscript, D.H., A.A.; critical review of manuscript, all authors. All authors have read and approved the final version of the manuscript. The study sponsor, Pfizer Inc., reviewed the study research plan and study manuscript; data management, processing, and analyses were conducted by Avalere Health. All final analytic decisions and the decision to submit for publication were made solely by study investigators.

Funding

This study was sponsored by Pfizer.

Data availability

All data generated or analysed during this study are included in this published article/as supplementary files. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval

This study was conducted using publicly available data and previously conducted studies. It does not contain any new studies with human participants or animals performed by any of the authors, and hence was not subject to institutional review.

Consent for publication

Not applicable.

Prior presentation

Selected results from this study have been presented in a poster at the 17th Vaccine Congress held in Glasgow, Scotland, from September 24–27, 2023 [94].

Competing interests

Diana Mendes, Reiko Sato, Andrew Vyse, James Campling, Tendai Mugwagwa, and Jeffrey Vietri are employees and shareholders of Pfizer. Dhwani Hariharan was employed by Avalere Health during the design and conduct of this study. Mark Atwood, Ahuva Averin, and Derek Weycker are employees of Avalere Health, which received funding from Pfizer in connection with this study and the development of this manuscript.