Cost-effectiveness analysis of pediatric immunization program with 15-valent pneumococcal conjugate vaccine in Japan

Abstract

Background

The 15-valent pneumococcal conjugate vaccine (PCV15 or V114) has recently been approved for pediatric vaccination against pneumococcal diseases (PDs) in Japan. The study aims to evaluate the cost-effectiveness of pediatric vaccination with V114 versus 13-valent PCV (PCV13) in Japan.

Methods

The study used a decision analytical Markov model to estimate the cost and effectiveness outcomes for a birth cohort in Japan over a 10-year time horizon. The model tracked the occurrences of acute PD events, including invasive PD (IPD), non-bacteremic pneumococcal pneumonia (NBPP) and pneumococcal acute otitis media (AOM) and the long-term impact of post-meningitis sequalae. Vaccine effectiveness was estimated based on literature and assumptions, and accounted for indirect effects and vaccine waning. The base case took the societal perspective, including both direct and indirect costs, while a healthcare payer perspective was modeled in a scenario analysis. Additional scenario analyses and sensitivity analyses were conducted.

Results

In the base case, V114 was associated with an incremental gain of 24 quality-adjusted life years and a reduction of ¥365,610,955 in total costs compared to PCV13. It was expected to reduce the number of pneumococcal AOM, NBPP, and IPD cases by 1,832, 1,333 and 25, respectively. All scenario analyses and most sensitivity analyses showed that V114 was a dominant strategy compared to PCV13.

Conclusions

Pediatric vaccination with V114 is expected to lead to cost savings and more health benefits compared to PCV13 in Japan from both societal and healthcare payer perspectives. The findings are robust under plausible assumptions and inputs.

Keywords:

• Cost-effectiveness
• V114
• pneumococcal diseases
• pediatric vaccination
• Japan

Introduction

Pneumococcal disease (PD) is a major cause of morbidity and mortality globally, especially in young children and older adults.¹ PD includes a group of diseases caused by Streptococcus pneumoniae (S. pneumoniae, pneumococcus).^1–3 The most common types of PD are non-bacteremic pneumococcal pneumonia (NBPP), which can affect individuals of all ages, and pneumococcal acute otitis media (AOM), which primarily affects children.^3,4 Invasive pneumococcal disease (IPD), including meningitis, bacteremia without focus, and bacteremic pneumonia, is a less common but more severe form of PD.^2–4 Meningitis, in particular, has a high mortality rate and can lead to long-term disabling neurological sequelae, such as hearing loss, seizures, and cognitive impairment.^5,6

The introduction of pneumococcal conjugate vaccines (PCVs) was a critical milestone in reducing the burden of PD worldwide.^1,7 In Japan, the 7-valent pneumococcal conjugate vaccine (PCV7) was approved in 2009 and included in the National Immunization Program (NIP) in 2013.⁸ However, it was soon replaced by PCV13, which provides protection against six more serotypes, and has been included in the NIP for the routine pediatric vaccination against PD since 2013.⁸ Currently, the Japan Pediatric Society recommends 3 + 1 schedule for PCV13, consisting of three doses at 2, 3, and 4 months in the primary series and one booster dose at 12–15 months.⁹

Routine immunization with PCV7 and PCV13 has drastically reduced incidence of PDs caused by PCV13 serotypes among children in Japan.^10–14 Additionally, vaccine-serotype PDs have markedly declined in the adult population in Japan following the introductions of PCV7 and PCV13, which suggests that vaccination with PCVs may reduce the PD incidences in unvaccinated populations—a phenomenon known as the ‘indirect effects’ of pediatric PCV use.^13,15–18 For example, the percentage of IPD attributable to PCV13 serotypes decreased by more than half in adults from 2010 to 2016.¹³ Similar declines were observed in PCV13 serotypes in community acquired pneumonia in Japan adult population.^16,17 Combining the direct and indirect effects, pediatric vaccination with PCVs has substantially reduced burden of PD in Japan.^10–18

However, despite the success in the pediatric PCV vaccination program in Japan, serotype replacement, which refers to an increase in the incidence rates of PD caused by non-vaccine serotypes (NVTs) after introduction of PCVs, poses new challenges in PD prevention.^10,13,14 Between 2014 and 2016, NVTs accounted for nearly half of the pneumococcal serotypes in children, a substantial increase compared to their contributions to the pneumococcal serotypes in the pre-PCV7 era.¹³ In addition, PCV13 may have limited effectiveness against certain vaccine serotypes, such as serotype 3.^13,19–22 Consequently, the epidemiological, economic and humanistic burden of PD in Japan remains high during the post-PCV era.³ Therefore, it is necessary to develop more effective vaccines with broader serotype coverage to further reduce the burden of PD in Japan.

A 15-valent PCV (PCV15 or V114) is approved and recommended for prevention of PDs in children and adults in many countries and regions, including the United States, Canada, and European Unions.^23–26 It has recently been approved for pediatric and adult vaccination against PD in Japan.²⁷ V114 provides protection against two additional serotypes, 22 F and 33 F, which contribute to 9.3–10.9% of IPD cases in Japan among children <5 years in the post-PCV13 period.^10,28 Pediatric clinical trials have demonstrated that V114 is well-tolerated with a safety profile that is consistent with PCV13, and has a non-inferior immune response for the 13 shared serotypes^29–31 and statistically superior immune responses for serotypes 3, 22 F and 33 F as compared to PCV13.³⁰ This study evaluated the cost-effectiveness of routine infant vaccination with V114 versus PCV13 in Japan from a societal perspective.

Method

Model overview

A previously published decision analytical Markov model³² was utilized to assess the cost-effectiveness of vaccination strategies with V114 versus PCV13 in the pediatric population in Japan (Figure 1). Both strategies adopted the 3 + 1 dosing schedule as recommended by the guidelines of the Japan Pediatric Society.⁹ The model assumed that the primary series and the booster dose were administered in the first- and second-year of life, respectively, with all doses from the same vaccine type.

Figure 1. Decision analytic model of pediatric pneumococcal vaccination.

Abbreviations: V114: 15-valent pneumococcal conjugate vaccine; PCV13: 13-valent pneumococcal conjugate vaccine.

The Markov model consisted of three health states, namely ‘no PD,’ ‘post-meningitis sequelae (PMS),’ and ‘death,’ and was designed to track acute PD events, including IPD, NBPP, and pneumococcal AOM (Figure 1). Individuals entered in the ‘no PD’ state at cycle 0 and were at risk of developing PD events in each model cycle (i.e. one year). All PD events were modeled as acute events that would resolve within a year. These events would incur short-term resource utilization, costs, and a decrement in quality-adjusted life years (QALYs) during the cycle in which the events occurred.

Individuals with meningitis were at risk of developing PMS, including neurologic deficits and hearing loss,³³ and would remain in the ‘PMS’ state until death. They were also at risk of developing non-meningitis IPD, pneumonia, and AOM in later model cycles. Moreover, the general background mortality was applied to all health states, while individuals with IPD or inpatient NBPP would experience additional disease-related death.

In the base case, the model followed a single vaccinated cohort for a 10-year time horizon. The 10-year time horizon was selected considering the duration of a public health program launched by municipalities and the time horizon needed to incorporate the vaccine waning effect, which was assumed to start in Year 6 from the last dose. A societal perspective was applied in the base case, including direct medical costs and indirect costs. The outcomes included incremental cost-effectiveness ratio (ICER), expressed as cost per life year (LY) gained and cost per QALY gained, as well as clinical outcomes, reported as the number of IPD, NBPP, pneumococcal AOM, and PMS cases prevented, and the number of IPD- and pneumonia-related deaths avoided. An annual discount rate of 2% was applied to costs, QALYs and LYs.³⁴ Clinical outcomes such as numbers of PD cases or deaths were not discounted.

Model inputs

Epidemiological inputs

In the base case, the model followed a single birth cohort of 781,093 infants in 2023, the year when V114 was assumed to be available to the pediatric population in Japan. The birth rate and background mortality were obtained from the government data.^35,36 While the vaccines were given to the birth cohort, the model considered the direct effects of the vaccines on the birth cohort as well as the indirect effects on other age groups.

Annual PD incidence rates and case fatality rates are presented in Table 1. For IPD, incidence rates for children <5 years were estimated using the overall IPD incidence rates and the distribution of different IPD types published by the National Institute of Infectious Disease (NIID) in Japan.^37,38 For the age groups between 5 and 65 years, the incidence rates were sourced from Chang et al.³⁹, Fukusumi et al.⁴⁰ and Tamura et al.¹⁵ For the age groups ≥65, the rates were obtained from the NIID as well.³⁸ The proportions developing PMS among patients with meningitis were only considered for the age groups <18 years¹⁰ due to lack of data in the adult population. Annual incidence rates for NBPP by age group were estimated by multiplying all-cause pneumonia incidence rate^11,41 with the percentage of pneumonia incidence attributable to S. pneumoniae in the corresponding age group.^11,41,42 The inpatient and outpatient NBPP incidence rates were then estimated by the proportions reported in the government data.³⁵ Annual incidence rates for simple and recurrent pneumococcal AOM were estimated using all-cause AOM rates based on the Sado Otitis Media Study,⁴³ the proportions of simple and recurrent AOM^43,44 and the percentage attributable to S. pneumoniae.⁴⁵ The case fatality ratios for IPD were obtained from the studies by Iwata et al.¹⁰ Nakano et al.⁴⁶ and Chang et al.³⁹ and those for inpatient NBPP were derived from the studies by Tashiro et al.⁴⁷ and Morimoto et al.⁴¹

Table 1. Model inputs on pneumococcal disease incidence and case fatality rates, health utilities and costs.

	Input value
	Age group (years)
Parameters	0–1	1–2	2–5	5–18	18–35		35–50	50–65	65–75	75–85	85–100
Epidemiological inputs
PD annual incidence rates (per 100, 000)
IPD
IPD that is meningitis^a^,^b	0.5	0.5	0.5	0.2	0.2		0.2	0.2	0.3	0.3	0.3
IPD that is bacteremia without focus^a^,^c	5.8	5.8	5.8	0.1	0.1		0.1	0.1	1.5	1.5	1.5
IPD that is bacteremic pneumonia^a,c	1.3	1.3	1.3	0.5	0.5		0.5	0.5	2.0	2.0	2.0
PMS^d
% meningitis cases leading to neurological deficits	11.7	11.7	11.7	11.7	–		–	–	–	–	–
% meningitis cases leading to hearing loss	3.8	3.8	3.8	3.8	–		–	–	–	–	–
Inpatient NBPPe	539.3	561.6	464.6	5.7	11.8		11.8	89.6	436.1	1,217.4	1,414.8
Outpatient NBPP^e	280.8	280.8	280.8	20.0	53.7		53.7	69.9	73.6	37.0	4.0
Simple pneumococcal AOM^f	827.9	1,886.6	915.3	79.6	–		–	–	–	–	–
Recurrent pneumococcal AOM^f	72.1	176.9	75.3	6.6	–		–	–	–	–	–
Case fatality ratios
Meningitis^g	0.05	0.05	0.05	0.07	0.07		0.07	0.07	0.12	0.12	0.12
Bacteremia without focus/ bacteremic pneumonia^h	0.04	0.04	0.04	0.14	0.14		0.14	0.14	0.19	0.19	0.19
Inpatient NBPP^i	–	–	–	–	0.003		0.009*	0.023*	0.800	2.200	6.800
Utility inputs
	Age group (years)
	0–19	20–29	30–39	40–49		50–59	60–69	70–79	80–100
Health state utility values
Baseline (general population without PD)^j	0.9724	0.9520	0.9485	0.9460		0.9278	0.9285	0.8826	0.7980
PMS, neurological deficits^k	0.68
PMS, hearing loss^k	0.73
	Age group (years)
	0–18	18–100
QALY decrements related to each episode of PD events^l
Meningitis	0.0230	0.0709
Bacteremia without focus	0.0080	0.0709
Bacteremic pneumonia	0.0080	0.1300
Inpatient NBPP	0.0060	0.1300
Outpatient NBPP	0.0040	0.1300
Simple AOM	0.0050	–
Recurrent AOM	0.0050	–
Cost inputs
Vaccine Costs (in 2022 ¥)
Vaccine acquisition costs (per dose)^m
PCV13		7,200
V114		7,200
Vaccine administration costs (per dose)^n
PCV13		4,400
V114		4,400
	Age group (years)
	0–1	1–2	2–5	5–18		18–35	35–50	50–65	65–75	75–85	85–100
Costs per episode of PD events (in 2022 ¥)
Direct medical costs^o
Meningitis	819,243	819,243	813,915	741,301		539,339	539,339	539,339	499,997	499,997	499,997
Bacteremia without focus	402,734	402,734	385,855	329,291		801,931	801,931	801,931	875,430	875,430	875,430
Bacteremic pneumonia	143,878	142,151	142,151	142,151		376,903	376,903	376,903	552,340	552,340	552,340
Inpatient NBPP	143,878	142,151	142,151	142,151		312,202	312,202	312,202	478,538	478,538	478,538
Outpatient NBPP	26,919	26,919	27,629	16,419		33,923	33,923	33,923	33,923	33,923	33,923
Simple AOM	30,737	33,930	78,509	44,291		–	–	–	–	–	–
Recurrent AOM	30,737	33,930	78,509	44,291		–	–	–	–	–	–
Indirect costs^p
Meningitis	187,016	187,016	187,016	187,016		299,728	459,965	450,756	344,486	333,188	333,188
Bacteremia without focus	101,193	101,193	101,193	101,193		277,819	426,343	417,807	319,305	308,833	308,833
Bacteremic pneumonia	90,946	90,946	90,946	90,946		211,156	324,043	317,555	235,600	227,874	227,874
Inpatient NBPP	90,946	90,946	90,946	90,946		165,063	253,307	248,235	235,600	227,874	227,874
Outpatient NBPP	12,809	12,809	12,809	12,809		–	–	–	–	–	–
Simple AOM	28,180	28,180	28,180	28,180		–	–	–	–	–	–
Recurrent AOM	28,180	28,180	28,180	28,180		–	–	–	–	–	–
	Age group (years)
	0–1	2–15	16–19	20–24		25–34	35–44	45–54	55–64	65–74	75–100
Annual costs of PMS and premature death (in 2022 ¥)
Direct medical costs of PMS^q
PMS, neurological deficits	505,374	435,911*
PMS, hearing loss	76,311	74,999*
Costs of premature death
Annual earnings (in 2022 ¥)^r	–	–	424,107	1,892,936		2,693,717	3,250,851	3,670,283	3,044,312	1,221,045	304,845

Abbreviations: IPD: invasive pneumococcal disease; NBPP: Non-bacteremic pneumococcal pneumonia; AOM: acute otitis media; PMS: post meningitis sequelae; QALY: Quality-adjusted life year; PCV13: 13-valent pneumococcal conjugate vaccine; V114: 15-valent pneumococcal conjugate vaccine; NIID: National Institute of Infectious Disease; MHLW: Ministry of Health, Labour and Welfare; NHI: National Health Insurance; JPY: Japanese Yen; DPC: diagnostic procedure code.

Notes:

^a For individuals aged <5 years, annual IPD incidence rates were estimated using the overall IPD incidence rate and the distribution of different IPD types published by the NIID.^37,38 For the age groups >65, the rates were obtained from the Infectious Agents Surveillance Report (2023).³⁸

^b For the age groups between 5 and 65 years, annual meningitis incidence rates were sourced from Chang et al.³⁹

^c For the age groups between 5 and 65 years, annual incidence rates for bacteremia without focus and bacteremic pneumonia were estimated using the number of non-immonocomprised patients from Fukusumi et al.⁴⁰ and Tamura et al.¹⁵.

^d The percentages of individuals with meningitis who developed PMS for the age groups less than 18 years were obtained from Iwata et al.¹⁰

^e Annual incidence rates for NBPP by age group were estimated multiplying all-cause pneumonia incidence rate with the percentage of pneumonia incidence attributable to S. pneumoniae in the corresponding age group.^11,41 The incidence rates of all-cause pneumonia were sourced to from Takeuchi et al.¹¹ for age < 18 years and from Morimoto et al.⁴¹ for age ≥ 18 years. The proportions of inpatient and outpatient NBPP were derived from e-stat³⁵. For age groups <5 years, the percentage of inpatient pneumonia attributable to S. pneumoniae was obtained from Okada et al.⁴²; for outpatient pneumonia, the percentage was assumed to be 14.5%. For age groups ≥18 years and 5–18 years, the percentages of pneumonia incidence attributable to S. pneumoniae were obtained from Morimoto et al.⁴¹ and Takeuchi et al.¹¹, respectively.

^f All-cause AOM rates were obtained from the Sado Otitis Media Study.⁴³ The proportions of recurrent AOM (including tube placement) were obtained from Otsuka et al.⁴³ and Yamanaka et al.⁴⁴ and were applied to estimate the incident rates for simple and recurrent AOM. The percentage attributable to S. pneumoniae was 10.1% for simple AOM and 0.7% for recurrent AOM⁴⁵ and were multiplied by the all-cause simple and recurrent AOM incidence rates, respectively, to estimate simple and recurrent pneumococcal AOM incidence rates.

^g The case fatality ratios of meningitis for children <5 years were obtained from Iwata et al.¹⁰; those for the age groups ≥5 years were obtained from Chang et al.³⁹

^h The case fatality ratios of bacteremia without focus and bacteremic pneumonia for children <5 years were obtained from Nakano et al.⁴⁶; those for the age groups ≥5 years were obtained from Chang et al.³⁹

ⁱ The case fatality ratios of inpatient NBPP were obtained from Tashiro et al.⁴⁷ for the age groups 18–64 years, and from Morimoto et al.⁴¹ for the age groups ≥65 years. *0.009 was the case fatality rate for the age group of 35–55 years and 0.023 was the case fatality rate for the age group of 55–65 years. For patients ≤18 years, the model assumed no case fatality associated with inpatient NBPP compared to the background mortality rates.

^j The baseline utility values for the individuals without PD in Japan were reported by Shiroiwa et al.⁵⁷. Weighted average utilities of male and female populations were estimated for each age group.

^k Health state utilities for PMS were obtained from the study by Rubin et al.⁵⁸.

^l For children under 18 years old, the QALY decrements of IPD, pneumonia, and AOM were obtained from the study by Rubin et al.⁵⁸ For individuals 18 years and older, the QALY decrements were obtained from the studies by Mangen et al.⁵⁹ and Glick et al.⁶⁰

^m The vaccine acquisition cost for PCV13 was obtained from the MHLW data.⁶² Price parity was assumed for V114.

ⁿ Vaccine administration costs were estimated based on the MHLW guidance report and the NHI treatment fee schedule^{9, 63}, and included outpatient pediatric visit fee, injection fee, vaccine premium and consumer tax.

^o For the age groups <5 years, the costs for meningitis, bacteremia without focus, simple AOM and recurrent AOM were obtained from Hoshi et al.⁶⁴ For the age group 5–18 years, the costs for meningitis, bacteremia without focus, outpatient pneumonia and AOM were derived from Shiragami et al.⁴⁹ The diagnostic procedure code (DPC) statistics and DPC NHI scores were used to estimate the costs of meningitis and bacteremia without focus for the age groups 18–65 years, costs of bacteremic pneumonia for all age groups and costs for inpatient pneumonia for age groups <65 years.^65,66 For the age groups of ≥65 years, the costs of meningitis and bacteremia without focus were obtained from Jiang et al.⁶⁷; the costs of inpatient pneumonia were estimated using the weighted average of number of patients and costs of general ward and intensive care unit, which were obtained from Konomura et al. 2017 and converted into JPY using 101.5 as currency converter.⁶⁸ For outpatient pneumonia, the age groups <5 years were obtained from Ishiwada et al.⁶⁹ and the costs for age groups ≥18 years were obtained from Konomura et al.⁶⁸

^p Indirect costs refer to productivity loss related to PD treatment, which was incurred by adult patients and caregivers of pediatric patients. The indirect costs for the age groups <18 years were obtained from the study by Shiragami et al.⁴⁹ For the age groups ≥18 years, productivity loss related to hospitalization due to IPD and inpatient NBPP was considered and were calculated by multiplying the length of stay, which was obtained from the DPC statistics, and the median annual earnings for each age group.^65,66,71

^q Annual medical costs for PMS were estimated separately for the age groups of <2 years and 2–18 years, based on the discounted lifetime costs.⁷⁰

* These costs were estimated for the age groups 2–18 years.

^r Productivity loss due to premature death for the age groups ≥16 years was calculated by multiplying the labor force participation rate with the median annual earnings for each age group.^35,71

Vaccine effectiveness

The vaccine coverage rate (VCR) was assumed to be 100% for primary series and the booster dose for both vaccines based on the data published by the Ministry of Health, Labour and Welfare (MHLW) and a previous cost-effectiveness analysis (CEA) study.^48,49 Vaccine effectiveness (VE) was estimated based on the literature and assumptions, and was reported as a percentage reduction in baseline PD incidence rates resulting from pneumococcal vaccine administration. Serotype-specific VEs were applied to age-specific serotype distributions for all PDs in the model (Table 2).

Table 2. Vaccine effectiveness of PCV13 and V114 in the base case.

	Vaccine type	Serotype-specific VE (%)
Disease	Serotype	1	4	5	6B	7F	9V	14	18C	19F	23F	3	6A	19A	22F	33F
IPD^a	PCV13	87	93	87	94	97	100	94	97	87	98	80	86	86	-	-
	V114	87	93	87	94	97	100	94	97	87	98	80	86	86	86	86
AOM^b (pneumococcal)	PCV13	86	57	86	57	86	57	57	57	57	57	15	100	91	-	-
	V114	86	57	86	57	86	57	57	57	57	57	15	100	91	57	57
NBPP^c	PCV13	77	77	77	77	77	77	77	77	77	77	77	77	77	-	-
	V114	77	77	77	77	77	77	77	77	77	77	77	77	77	77	77

Abbreviations: PCV13: 13-valent pneumococcal conjugate vaccine; V114: 15-valent pneumococcal conjugate vaccine; VE: vaccine effectiveness; IPD: invasive pneumococcal disease; AOM: acute otitis media; NBPP: Non-bacteremic pneumococcal pneumonia.

Notes:

^a The vaccine effectiveness (VEs) against IPD for PCV13 were estimated based on the VEs for PCV7⁵⁰ and the study by Moore et al.⁵¹ The VEs for serotypes 22 F and 33 F were assumed to be equal to the average VE among the shared 13 serotypes.

^b The VEs against pneumococcal AOM were obtained from the PCV7 clinical trial⁵² and the study by Pichichero et al.⁵³ The VEs for serotypes 22 F and 33 F were assumed to the same as the VE in the PCV7 clinical trial.⁵²

^c The VEs against inpatient NBPP were obtained from Lewnard et al.,⁵⁴ which were different between the primary series and the booster dose.

The VEs against IPD and pneumococcal AOM were estimated using similar methods. Specifically, the serotype-specific VEs for PCV13 were estimated first. V114 was assumed to have the same VEs as PCV13 for the shared 13 serotypes. For IPD, VEs for PCV13 were estimated based on the VEs for PCV7⁵⁰ and the study by Moore et al.⁵¹ The VEs for serotypes 22 F and 33 F in V114 were assumed to be equal to the average VE among the shared 13 serotypes. The VEs against pneumococcal AOM were obtained from the PCV7 clinical trial⁵² and the study by Pichichero et al.⁵³ The VEs for serotypes 22 F and 33 F were assumed to the same as the overall VE in the PCV7 clinical trial.⁵² The same VEs were assumed for simple and recurrent AOM in the base case. The VEs against NBPP were obtained from Lewnard et al.⁵⁴ and applied to both inpatient and outpatient NBPP.

In addition, the model considered vaccine waning and indirect effects when estimating the VEs. The base case assumed that vaccine waning started in Year 6 after the last dose and decreased linearly to 0% over the next 10 years.^55,56 The indirect effects were applied to serotype 22 F and 33 F IPD, assuming an annual incidence rate reduction of 7.8%⁵⁵ for the first five years. Based on this, a relative reduction in 22 F- and 33 F-related IPD incidence rate of 7.8%, 15.5%, 21.6%, 27.7% and 33.4% compared to the baseline was applied in the first, second, third, fourth, and fifth or more years upon V114 launch, respectively.

Utility weights and QALY decrements

The utilities for ‘no PD’ were assumed to be the same as those in the general population in Japan, as reported by Shiroiwa et al.⁵⁷ Weighted average utilities of male and female populations were estimated for each age group (Table 1). For PMS, the utility values of 0.68 and 0.73 from the study by Rubin et al.⁵⁸ were applied to neurological deficits and hearing loss, respectively, across all age groups. The QALY decrements associated with PD events were sourced from the study by Rubin et al.⁵⁸ for children <18 years, and the studies by Mangen et al.⁵⁹ and Glick et al.⁶⁰ for adults (Table 1).

Cost inputs

The societal perspective was utilized in the base case, which included both direct medical costs and indirect costs (Table 1). All costs were adjusted to 2022 Japanese Yen (JPY) using the National Health Insurance (NHI) conversion rates.⁶¹

Direct medical costs

The vaccine acquisition cost for PCV13 was ¥7,200, which was obtained from the MHLW data.⁶² Price parity was assumed for V114. Vaccine administration costs were estimated at ¥4,400 for both vaccines based on the MHLW guidance report and the NHI treatment fee schedule,⁶³ and included outpatient pediatric visit fee, injection fee, vaccine premium and consumer tax.

Treatment costs per episode for IPD, NBPP, AOM were derived from the literature and varied by age group.^49,64–69 Annual medical costs for PMS were estimated separately for the age groups of <2 years and 2–18 years, using the discounted lifetime costs.⁷⁰

Indirect costs

The model considered productivity loss related to the treatment of PDs, which was incurred by caregivers of pediatric patients and adult patients. The productivity loss was estimated using the human capital approach, which accounted for the wage loss of caregivers or adults patients for each PD episode. The indirect costs for the age groups <18 years were obtained from the study by Shiragami et al.⁴⁹ For the age groups ≥18 years, productivity loss related to hospitalization due to IPD and inpatient NBPP was considered. Specifically, the length of stay, which was obtained from the Diagnostic Procedure Code (DPC) statistics, were multiplied by the median annual earnings for each age group.^65,66,71 In addition, the model accounted for productivity loss due to premature death, which was derived based on the average annual earnings for the age groups ≥16 years, calculated by multiplying the labor force participation rate with the median annual earnings for each age group.^35,71

Sensitivity analyses

A series of scenario analyses were carried out to assess the uncertainties associated with specific assumptions and inputs in the model. These analyses involved alternative assumptions regarding the target population, the time horizon, the model perspective, vaccine coverage rate, and the direct and indirect vaccine effects. Specifically, Scenario 1 included the entire Japan population as the target population and used 100-year (i.e. lifetime horizon); Scenario 2 took a healthcare payer perspective instead of the societal perspective; Scenario 3 used VEs against all-cause pneumonia and all-cause AOM to estimate pneumonia- and AOM-related outcomes; Scenario 4 assumed no indirect effects; and lastly, Scenario 5 and Scenario 6 assumed a vaccine coverage rate of 95% and 90%, respectively. Furthermore, one-way sensitivity analysis (SA) was conducted by varying one model input at a time to assess the sensitivity of the model results to various model input values. Additionally, a probabilistic sensitivity analysis (PSA) was conducted using a Monte Carlo simulation with 1,000 iterations generated by simultaneously varying key model parameters based on pre-specified distribution assumptions (Supplementary Table 1). Except for the population inputs and vaccine acquisition costs, the key parameter variations were defined based on the 95% confidence intervals or the standard errors (SEs), as reported in the respective source studies. If the variation of an input parameter was not presented in the source study, an SE of 20% of the base-case value was applied. Lastly, given the uncertainty of the V114 price, a threshold analysis was conducted to estimate the maximum price of V114 for it to be cost saving compared to PCV13.

Table 3. Base-case results.

Outcomes	V114	PCV13	Incremental outcomes (V114 versus PCV13)
Clinical outcomes (undiscounted)
IPD cases	262	287	−25
NBPP cases	22,168	23,701	−1,333
Pneumococcal AOM cases	42,042	43,875	−1,832
PMS cases	3	4	−1
IPD deaths	13	15	−2
Pneumonia deaths^a	0	0	0
LYs and QALYs (discounted)
Total LYs	7,071,214	7,071,207	7
Total QALYs	6,875,443	6,875,418	24
Cost outcomes (in 2022 ¥, discounted)
Vaccine acquisition costs	22,375,381,278	22,375,379,359	1,919
Vaccine administration costs	13,673,844,114	13,673,842,942	1,172
IPD treatment costs	91,575,473	100,714,034	−9,138,561
Pneumonia treatment costs	2, 164,407,772	2,290,603,743	−126,195,971
AOM treatment costs	2,260,355,950	2,359,259,161	−98,903,212
PMS treatment costs	6,104,753	7,005,897	−901,144
Indirect costs	2,504,470,095	2,634,945,255	−130,475,159
Costs of premature death^b	–	–	–
Total costs	43,076,139,476	43,441,750,390	−365,610,955
ICER			V114 dominant

Abbreviations: PCV13: 13-valent pneumococcal conjugate vaccine; V114: 15-valent pneumococcal conjugate vaccine; IPD: invasive pneumococcal disease; AOM: acute otitis media; PMS: post meningitis sequelae; LY: life year; QALY: quality-adjusted life year; ICER: incremental cost-effectiveness ratio.

Notes:

^a The model assumed no case fatality associated with inpatient NBPP compared to the background mortality rates for patients ≤18 years.

^b Costs of premature death were only estimated for individuals aged ≥16 years and thus were not applicable for the base case, which only followed up a single birth cohort for 10 years.

Results

Base case

The base-case results (Table 3) showed that by vaccination of a birth cohort of 781,093 infants, V114 was expected to reduce the number of IPD, NBPP and pneumococcal AOM cases by 25, 1,333 and 1,832, respectively, over the 10-year time horizon, compared to PCV13. Furthermore, V114 was projected to prevent two IPD-related deaths and to save seven LYs with an incremental QALY gain of 24, relative to PCV13. V114 was associated with slightly higher vaccine acquisition and administration costs (¥1,919 and ¥1,172, respectively), due to the reduction in the IPD-related deaths associated with V114 during the first year. However, it was predicted to save ¥9,138,561 in IPD treatment costs, ¥126,195,971 in pneumonia treatment costs, ¥98,903,212 in AOM treatment costs and ¥901,144 in PMS treatment costs. Additionally, V114 was predicted to reduce the indirect costs by ¥130,475,159. The total cost savings associated with V114 versus PCV13 amounted to ¥365,610,955. Because V114 was associated with extended LYs and QALYs and lower costs, it was a dominant strategy compared to PCV13 in the base case.

Table 4. Incremental costs and effectiveness outcomes comparing V114 vs. PCV13 in the scenario analyses.

	Incremental outcomes (V114 vs. PCV13)
	Scenario 1	Scenario 2	Scenario 3	Scenario 4	Scenario 5	Scenario 6
Description of scenario analyses
Key assumptions/inputs in the scenario	Target population: Entire Japan population^c Time horizon: 100 years (i.e. lifetime)	Perspective: Healthcare payer	Pneumonia- and AOM-related outcomes were modeled based on VEs against all-cause pneumonia and all-cause AOM^d	No indirect effects	Vaccine coverage rate for V114 and PCV13: 95%	Vaccine coverage rate for V114 and PCV13: 90%
Corresponding assumptions/inputs in the base case	Target population: A single birth cohort Time horizon: 10 years	Perspective: Societal	Pneumonia- and AOM-related outcomes were modeled based on VEs against NBPP and pneumococcal AOM	Indirect effects on IPD were applied	Vaccine coverage rate for V114 and PCV13: 100%	Vaccine coverage rate for V114 and PCV13: 100%
Outcomes in the scenario analyses
Clinical outcomes, cases/deaths prevented (undiscounted)
IPD cases	2,664	25	25	24	24	23
Pneumonia cases	44,955	1,333	1,500	1,333	1,267	1,200
AOM cases	60,529	1,832	1,832	1,832	1,741	1,649
PMS cases	18	0	0	0	0	0
IPD deaths	346	1	1	1	1	1
Pneumonia deaths^a	138	0	0	0	0	0
LYs and QALYs (discounted)
Total LYs	3,011	7	7	7	7	7
Total QALYs	3,221	24	24	23	22	21
Cost outcomes (in 2022 ¥, discounted)
Vaccine acquisition costs	36,843	1,919	1,919	1,895	1,740	1,570
Vaccine administration costs	22,515	1,172	1,172	1,158	1,063	959
IPD treatment costs	−789,347,837	−9,138,561	−9,138,561	−8,767,865	−8,801,307	−8,646,054
Pneumonia treatment costs	−2,617,065,661	−126,195,971	−147,954,319	−126,196,000	−119,886,147	−113,576,325
AOM treatment costs	−1,937,158,151	−98,903,212	−98,870,853	−98,903,250	−93,958,029	−89,012,847
PMS treatment costs	−38,036,527	−901,144	−901,144	−870,309	−866,550	−831,957
Indirect costs	−2,966,418,822	–	−144,381,859	−130,368,258	−123,985,344	−117,495,530
Costs of premature death^b	−4,719,511,186	–	–	–	–	–
Total costs	−13,067,478,826	−235,135,796	−401,243,644	−365,102,630	−347,494,574	−329,378,183
ICER	V114 dominant	V114 dominant	V114 dominant	V114 dominant	V114 dominant	V114 dominant

Abbreviations: PCV13: 13-valent pneumococcal conjugate vaccine; V114: 15-valent pneumococcal conjugate vaccine; VE: vaccine effectiveness; IPD: invasive pneumococcal disease; AOM: acute otitis media; PMS: post meningitis sequelae; LY: life year; QALY: quality-adjusted life year; ICER: incremental cost-effectiveness ratio.

Notes:

^a The model assumed no case fatality associated with inpatient NBPP compared to the background mortality rates for patients ≤18 years.

^b Costs of premature death were only estimated for individuals aged ≥16 years and thus were not applicable for scenario analyses 2–4, which only followed up a single birth cohort for 10 years.

^c In this scenario, all age groups in the entire Japan population were included at the beginning of the model, while in the base case, only newborns were included at the beginning of model.

^d In this scenario analysis, the VE against all-cause inpatient pneumonia was 7.9% and 8.2% for the primary and booster series of PCV13, and 10.5% and 11.0% for the primary and booster series of PCV15. The VE against all-cause outpatient pneumonia was 2.1% for both primary and booster series of PCV13 and 2.7% for PCV15. The VE against all-cause simple and recurrent AOM was assumed to be the same for the primary and booster series. The VE against all-cause simple AOM was 0.6% and 1.0%, respectively, for PCV13 and PCV15. The VE against all-cause recurrent AOM was 0.04% and 0.1%, respectively, for PCV13 and PCV15.

Sensitivity analysis

All scenario analyses conducted were consistent with the base-case results, indicating that V114 was the dominant vaccination strategy in comparison to PCV13 (Table 4). In scenario 1, following the entire Japan population over lifetime (i.e. 100 years) resulted in the highest cost savings of ¥13,067,478,826 for V114, with 3,011 extended LYs and 3,221 total QALY gain. When the healthcare payer perspective was applied, the total cost savings associated with V114 relative to PCV13 became ¥235,135,796. In scenario 3, where VEs against all-cause pneumonia and all-cause AOM were applied, the incremental LYs and QALYs were similar to the base case but the total cost savings were greater. Moreover, the scenario analysis assuming no indirect effects resulted in little changes compared to the base case, mainly because the indirect effects were only applied to IPD.

The one-way SA and PSA further supported the robustness of the model results. All ICERs in the one-way SA indicated that V114 was cost saving and yielded higher QALYs (Figure 2). The ICER was most sensitive to inputs related to inpatient pneumonia, including serotype-specific VE for V114 and PCV13, direct and indirect cost per episode, baseline incidence rate, percentage attributable to S. pneumoniae, serotype distribution, and QALY decrement. Certain inputs related to simple AOM and bacteremia without focus were also among the most impactful ones. In addition, the model inputs for the age group of 2–5 years generally had a greater impact on the results than other age groups (Figure 2). The threshold analysis showed that the maximum price per dose for V114 was ¥7,318 for it to remain cost saving compared to PCV13, i.e. ¥118 higher than the PCV13 price. Results from the PSA showed that V114 was the dominant strategy in 98.7% of the simulations (Figure 3).

Figure 2. One-way sensitivity analysis results (showing the most impactful parameters).

Abbreviations: ICER: incremental cost-effectiveness ratio; PCV13: 13-valent pneumococcal conjugate vaccine; V114: 15-valent pneumococcal conjugate vaccine; QALY: quality-adjusted life year; VE: vaccine effectiveness; IPD: invasive pneumococcal disease; AOM: acute otitis media; ST: serotype.

Notes: Tornado diagram showing the results of the one-way sensitivity analysis. The purple bars show the change in ICER from the base case when the higher value of an input was used whereas the light blue bars show the change in ICER from the base case when the lower value of the selected input was used while all other inputs remain constant.

Figure 3. Scatter plot of probabilistic sensitivity analysis.

Abbreviations: QALY: quality-adjusted live year; PSA: probabilistic sensitivity analysis.

Discussion

The emergence of PDs caused by NVT and antibiotic-resistant serotypes present new challenges in PD prevention and treatment in Japan,^10,13,46 which could be potentially addressed by newer vaccines. V114 includes the 13 serotypes in PCV13 plus two additional serotypes, 22 F and 33 F, which are associated with high mortality, antibiotic resistance, and meningitis.¹ Furthermore, V114 provides superior immune response against the shared serotype 3 as compared with PCV13,³⁰ potentially offering better protection against this serotype. Adding to the previously published literature on V114’s immunogenicity and safety, the current study provides economic evidence on the use of V114. The current study shows that V114 is a cost-saving strategy in Japan from both the societal and healthcare payer perspectives compared to PCV13. To our knowledge, this is the first study to evaluate the cost-effectiveness of V114 in Japan. The base case predicts that V114 is associated with total cost savings of ¥365,610,955 and 24 more QALYs. It had the greatest impact in reducing the number of AOM cases, followed by pneumonia cases, while the cost savings were highest for pneumonia treatment costs, followed by treatment costs for AOM and IPD. Indirect cost savings accounted for approximately one-third of the total cost savings in the base case. Despite the variation in the incremental costs and QALYs comparing V114 versus PCV13 in the sensitivity analysis, V114 was consistently the dominant strategy under all plausible alternative assumptions and input values in the scenario analyses and deterministic sensitivity analyses and 98.7% of the simulations in the PSA, demonstrating the robustness of the model results.

The conclusions from the current study align with the recently published cost-effectiveness analyses of pediatric use of V114 versus PCV13 in the United States.^32,72 Both US models applied a similar model structure as the one in the current model. Huang et al.³² tracked PD incidence in the entire US population over a lifetime, while Prasad et al.⁷² tracked PD incidence in a single birth cohort up to 17 years (15 years after the last PCV dose). Despite the differences, both studies showed that that V114 was the dominant strategy compared to PCV13 in the base case and all the sensitivity analyses, including the scenario analysis of no indirect effects. Based on these CEA studies as well as the demonstrated immunogenicity and safety of V114 in the clinical trials, the Centers for Disease Control and Prevention Advisory Committee for Immunization Practices recommends V114 for pediatric immunization in the US.²⁶ The current study adds to the previous CEAs comparing V114 with PCV13 in pediatric populations. Together, they demonstrate that V114 is expected to prevent more PD cases, extend LYs and QALYs, and lead to cost savings in both direct and indirect costs as compared to PCV13. Moreover, consistent across all studies, the conclusions were sustained in almost all scenario analyses and sensitivity analyses, suggesting that the cost saving of V114 vs. PCV13 is robust under different settings, assumptions, and input values.

While the results from the current study cannot be directly compared with other CEAs of pneumococcal vaccines in Japan due to different comparators, the model structure is similar to that of previously published models.^49,64,73 Notably, the current study incorporated distinctive improvements upon earlier models. For example, the current study modeled bacteremia without focus and bacteremic pneumonia separately, and treated simple AOM and recurrent AOM as separate PD events with different baseline incidence rates. Additionally, the current study estimated the number of NBPP cases, which is specific to S. pneumoniae vs. all-cause pneumonia. And lastly, the model incorporated indirect effects, which are included in many CEA models in other countries^{32,58,74–76} and is supported by the evidence in Japan.^13,15–18 The study shows that indirect effects had a negligible effect on the cost savings and QALY gains because the model conservatively applied the indirect effects to IPD caused by 22 F and 33 F only and did not consider the indirect effects on other PD types.

The model should be reviewed in the context of certain limitations. As with all models, limited data availability is a major constraint in the current model. While thorough effort was made to obtain the most relevant model inputs specific to the Japan populations, certain inputs are not available in the literature, such as disutilities associated with PD events in Japan, case fatality ratios related to inpatient pneumonia for children, and VEs for certain serotypes, including 22 F and 33 F. With such limitations, reasonable and conservative assumptions were made with extensive sensitivity analyses to evaluate the uncertainty. One example is indirect effects, which were conservatively assumed to be applicable to IPD in the base case and were removed in the scenario analysis. Another example is the conservative assumption of the same VE between V114 and PCV13 for PDs caused by serotype 3 despite the demonstrated superior immunogenicity of V114 for this serotype.³⁰ Additionally, due to lack of information in Japan, certain inputs were not considered in the current model, e.g. direct non-medical costs and indirect costs associated with PMS. However, based on studies conducted in the US, direct non-medical costs were generally insignificant when compared to direct medical costs and indirect costs due to work loss.^77–81 Therefore, the exclusion of direct non-medical costs is unlikely to affect the conclusions of the current study. Similarly, the omission of indirect costs associated with PMS reduces the magnitude of cost saving by V114, resulting in more conservative results, and thus would not impact the study conclusions. Another key model input with uncertainty is the price for V114. V114 price in the pediatric indication was unknown at the time of the analysis. To address this limitation, a threshold analysis was conducted to estimate the maximum price at which V114 remains cost saving compared to PCV13. Furthermore, the current study did not account for certain trends in recent epidemiology of PD. One of them is antibiotic resistance, which may increase the PD treatment costs over time. Notably, serotypes 22 F and 33 F are associated with antibiotic resistance;¹ by providing protection against these two serotypes, V114 is expected to further reduce treatment costs compared to the current estimate in the base case. Moreover, the model did not explicitly consider time-varying serotype distribution. However, the study applied vaccine waning to account for reducing VEs over time due to serotype replacement.

Despite the limitations of the study, the robust findings suggest that V114 has the potential to address the emerging challenges posed by PD and reduce the burden of communicable diseases in Japan. The study provides timely economic evidence that enables healthcare decision-makers in Japan to make informed decisions about adopting V114 in the NIP. In addition, the study highlights the importance of continuing to monitor the disease and update the model to accurately assess the potential impact of V114 as new information becomes available.

Conclusions

The study demonstrates that using the 3 + 1 dosing schedule of V114 for infant vaccination is a dominant strategy compared to PCV13 in Japan, resulting in both direct and indirect cost savings and improved QALYs. V114 has the potential to prevent more cases of PD, particularly NBPP and AOM. The findings are robust under a wide range of model input values and plausible alternative assumptions, providing valuable economic evidence to healthcare decision-makers in Japan for the adoption of V114.

Transparency

Author contributions

All authors were involved in the model conception, design, and analysis, as well as the interpretation of the results. All authors contributed to the first draft of the paper and subsequent revisions, and the final approval of the version to be published. All authors agree to be accountable for all aspects of the work.

Acknowledgements

We would like to acknowledge CHEORS for assistance in the model evaluation and Analysis Group, Inc. for assistance in model input collection. We would also like to acknowledge Jipan Xie, MD, PhD, at XL Source, Inc. for medical writing assistance.

Declaration of funding

This work was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA.

Declaration of financial/other relationships

Atsushi Tajima and Machiko Abe are employees of MSD K.K., Tokyo, Japan. Min Huang and Jessica Weaver are employees of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA. A reviewer on this manuscript disclosed that they have been part of a health technology assessment project on PCV15. Peer reviewers on this manuscript have received an honorarium from JME for their review work but have no other relevant financial relationships to disclose.