2.1 Model Structure

The cost-effectiveness of the 4CMenB vaccine was evaluated by adapting a validated, published, dynamic transmission-based cost-effectiveness (DyCE) model to the French context [24, 25]. In this model, the number of IMD cases over time with and without vaccination is described using a ‘susceptible–infected–susceptible’ model. The model distinguishes between three categories of serogroups, that is, serogroup B, serogroups ACWY (considered as one group), and ‘other’ (Fig. 1). Furthermore, the model allows the impact of vaccination on cross-protection and carriage to be evaluated. The contact matrix was derived from French-specific data [26]. Carriage prevalence across age groups for different serogroups was derived from a meta-analysis, which included data from 89 cross-sectional and longitudinal studies reporting on nasopharyngeal meningococcal carriage by age group [5]. The model was calibrated against French-specific IMD incidence by serogroup between 2009 and 2019, by adjusting the force of infection and the probability of infection for individuals being carriers of a specific serogroup. Model assumptions were validated by independent experts, as recommended by the French health technology assessment body [27] (Online Resource 1, electronic supplementary material [ESM] Table 1).

Fig. 1

Adapted from Beck et al., 2020 [25]. Ψvi Proportion of people effectively protected, calculated by multiplying vaccination coverage per vaccine efficacy, λi, x infection force at age i for IMD strain x (x = B, ACWY, other); Ri, x clearance rate at age I for IMD strain x (x = B, ACWY, other). IMD invasive meningococcal disease, IMD ACWY serogroup A, C, W or Y invasive meningococcal disease, IMD B serogroup B invasive meningococcal disease

Schematic of the dynamic transmission model/decision-tree model.

Output from the dynamic transmission model served as input for a decision-tree analysis, which evaluated the health and cost outcomes for each vaccination strategy [24]. Additional input parameters for the decision tree included the proportion of patients with IMD who developed sequelae, case-fatality rate, quality of life (QoL), and cost data (Table 1).

Table 1 Model input parameters—epidemiological and vaccine characteristics

The base case analysis adopted a collective perspective, while one scenario analysis was carried out from a societal perspective (including productivity losses and other types of informal care costs). The collective perspective covers all people or institutions affected by the production of an intervention in terms of health effects or costs. This covers informal care, home healthcare, inpatient and outpatient care, as well as personal care services associated with the health issue. Health and cost outcomes were calculated over a life-time horizon (i.e., 100-year time horizon), in line with other cost-effectiveness modeling studies [28,29,30]. The first year considered was 2020; for each subsequent year, a new birth cohort entered the model and was followed for 100−x years, where x was the number of years after 2020. A discount rate of 2.5% for costs and health outcomes was applied for the first 30 years, followed by a discount rate of 1.5% thereafter, as recommended in the 2020 health economics guidelines issued by HAS [27].

The statistical software R was used for the dynamic transition model [31]. The model was implemented in Excel 365 via BERT, which allows for seamless linkage between the two software packages [32].

2.2 Population

The model considered the entire French population stratified by age as of January 2020 [33]. The population eligible for vaccination was the birth cohort of each year, with infants receiving three doses of 4CMenB, at 3, 5, and 13 months.

2.3 Intervention and Comparator

The 4CMenB vaccination strategy was compared with no vaccination strategy, as there is no alternative serotype B-specific vaccine currently approved for use in France.

2.4 Outcomes

MenB IMD incidence and related deaths, MenB survivors who developed sequelae, costs, and quality-adjusted life-years (QALYs) were evaluated for each vaccination strategy; incremental cost-effectiveness ratios (ICERs) were calculated for the base case and for each scenario analysis, considering benefits and costs in the overall population over a 100-year time horizon.

2.5 Model Input Parameters2.5.1 Demographics and IMD Epidemiology

The French population count by age group as of January 2020 was extracted from the National Institute of Statistics and Economic Studies (INSEE) database [33]. Mortality rates for 2019 were obtained from the National Institute for Demographic Studies (INED) [34]. The birth cohort for 2020 (infants aged < 1 year) was estimated to be 686,900 in 2020, based on 2018 data from INSEE and annual growth projections [35].

IMD is a notifiable disease in France, and incidence rates are published each year by Public Health France (Santé Publique France–SPF) [10]. Age-specific IMD incidence rates were calculated by serogroup, based on cases notified between 2009 and 2019 and corrected for incomplete notification (Table 1). A notification rate of 92% was reported in a French study evaluating completeness of IMD notification in France; this rate was confirmed by SPF in 2017 [36, 37]. IMD incidence varies in an unpredictable fashion over time; therefore, a smoothing function was applied to the incidence rate, in line with a previously developed health economic model for 4CMenB [1]. In 2019, IMD incidence (all serotypes) was estimated to be 0.76 cases per 100,000 inhabitants, while IMD B incidence was estimated to be 0.36 per 100,000 individuals [37]. Age-adjusted incidence and the relative proportion of MenB IMD to all IMD cases are provided in Table 1 and Online Resource 1, ESM Table 2). Serogroup-specific mortality rates for IMD cases were derived from data from SPF (Table 1). The proportion of patients who developed sequelae was obtained through a retrospective analysis of the national healthcare insurance database (Système National d’Information Interrégimes de l’Assurance Maladie–SNIIRAM) (Table 1) (Online Resource 1, ESM Table 3) [38]. The types of sequelae were validated through a systematic literature search of observational and medico-economic studies and independent assessment of their clinical relevance by three independent experts [3].

Table 2 Health economic model input parametersTable 3 Public health impact and health economic outcomes for the base case analysis: 100-year time horizon2.5.2 4CMenB Efficacy and Waning

A vaccination coverage of 95% for the first and second doses and 90% for the third dose of 4CMenB was assumed. Vaccination coverage was assumed to be high and similar to that reported for other infant vaccines recommended by the French Health Authority [22, 43]. Vaccine effectiveness was derived from real-world data, including the latest updated report from Public Health England; in England, 4CMenB vaccination was included in the national immunization schedule in 2015 (Table 1) [41]. The immunogenicity conferred by 4CMenB vaccination decreases over time. The model assumed an exponential decrease in vaccine efficacy, in line with other meningitis vaccination models [29, 44,45,46], and the duration of protection was estimated to be 33 months after the first and second doses and 38 months after the third dose (Table 1).

2.5.3 Utility Data

French utility values by age group in the general population were used as reference value; utility values were obtained via the EQ-5D-3L, a validated tool for QoL measures [47]. Utility decrements were applied to these reference values, distinguishing between the acute phase with and without sequelae and the long-term phase governed by sequelae (Table 2). Utility decrements for the acute phase were sourced from a study reporting on utility values in an English IMD cohort and decrements due to sequelae were applied as reported in a previous model for England [28, 48]. Utility values for specific sequelae were identified through a literature search, giving preference to studies using the EQ-5D tool, as recommended by the HAS health economic guidelines (Online Resource 1, ESM Table 3) [27]. A reduction in QoL was also assumed for family networks of IMD survivors, corresponding to 48% of the utility decrement experienced by survivors (range: 17–79%), both during the acute and the long-term phases [49]. For parents losing a child to IMD, the model assumed a QALY loss proportional to the QALY loss induced by the premature death of the child, corresponding to 9% of QALYs lost by the patient (range: 0–9.6%) [50].

2.5.4 Cost Data

The following cost items were considered in the model: vaccination costs; costs for fever management following vaccination; cost of acute IMD care; cost of sequelae (acute and long-term); outbreak management costs, including logistics for the identification of contact persons and deployment of preventive treatment; and costs associated with disability, such as allowance, special education, and institutionalization (Table 2, Online Resource 1, ESM Table 4). The most recent costs available (2016–2018 costs, except for the cost of public education, which was available for 2014 only) were used in the model. According to The Directorate of Research, Studies, Evaluation, and Statistics (DREES), costs related to goods and services have undergone minimal changes between 2014 and 2018 (< 2%) [59]. Furthermore, the index might not systematically include items related to allowances perceived by people needing special education and institutionalization. Therefore, the expected error introduced due to different reference years is considered marginal.

Table 4 Base case and scenario analyses

It was assumed that 4CMenB would be administered during routine infant health check-ups, hence no additional costs for a medical consultation were considered. The proportions of infants requiring medical attention due to fever after the first (0.96%), second (1.68%), and third dose (1.65%) were derived from clinical studies [60,61,62,64]. Costs for outpatient visits were taken from official French tariffs [64]. The proportion of individuals requiring hospitalization (0.17%) due to 4CMenB-related fever was derived from a study conducted in the UK [65].

Costs associated with IMD during the acute and long-term phases (including costs related to sequelae) were estimated by means of a retrospective database analysis, extracting data from the SNIIRAM database for the period January 2012 to December 2017. IMD cases were identified using the tenth revision of the International Classification of Diseases (ICD) codes (v.10), and the index date was set to the first diagnosis with a meningitis-related ICD-10 code. For the acute phase, transfers occurring immediately prior and/or subsequent to the index hospitalization were grouped together to yield a combined overall length of hospitalization and costs related to the acute phase. The mean length of stay was 14.8 days (95% confidence interval [CI] 14.0–15.6), and the mean cost was €11,269 (95% CI 10,869–11,643) [12]; costs by age group are provided in Table 2.

IMD sequelae were identified by searching for several indicators that occurred within a given timeframe from the index date. ICD-10 codes for primary, associated, and secondary diagnosis; reimbursement codes for prescription drugs and medical equipment specific for the treatment of sequelae (e.g., implants for hearing loss); and medical procedures or hospitalizations related to sequelae were identified by a pre-specified algorithm that has been validated by independent experts (Online Resource 1, ESM Table 3) [12]. Costs were estimated separately for the first year after IMD diagnosis and the following years. Costs due to sequelae were adjusted using a multivariable regression model to estimate incremental costs due to sequelae compared with observed costs in subjects without sequelae (Online Resource 1 ESM Table 3).

Families of children with IMD sequelae receive allocations to cover educational expenses and care (Allocation d’Education de l’Enfant - Handicapé – AEEH) [66]. These allocations are proportional to a child’s disability and are financed by the national social security system [67]; therefore, these costs form part of the collective perspective.

People with a disability are eligible for compensation services due to their disability (Prestation Compensation du Handicap—PCH), which comprises financial aid personalized for each individual who has limited autonomy due to a disability [68, 69]. For both AEEH and PCH, mean cost was calculated, considering the total number of beneficiaries in France and total aid received annually. The mean annual costs estimated for AEEH and PCH were €3667 and €6433, respectively (Online Resource 1 ESM Table 4).

Parents who stop working to care for their child during the acute phase of IMD are eligible for an additional allowance (Allocation Journalière de Présence Parentale – AJPP). An average time of 6 months off work was assumed, based on expert opinion which considered results from a survey of affected families and published literature [13]. A single cost of €5905 was estimated for AJPP.

Children with IMD sequelae can be schooled in specialized or public facilities, provided that appropriate support and logistics are available. The incremental costs related to specialized facilities, human resources, and specific materials are detailed in Online Resource 1 ESM Table 3. Similarly, adults with a disability are also eligible for various types of financial and material support (Online Resource 1 ESM Table 4).

For local outbreaks comprising two or more IMD cases in the same place and at the same time, a limited vaccination campaign is recommended, to provide protection to any exposed groups (e.g., high-school students if cases occurred in this setting) [70]. In France, two local MenB IMD outbreaks were recently reported [71, 72]. Costs for outbreak management in the present modeling study were estimated based on information from these cluster outbreaks, as well as another outbreak that involved MenW [73]. For isolated cases, contagion management is limited to antibiotic prophylaxis and immunization of contacts. Mean costs due to local outbreaks and isolated cases were estimated at €1,078,688 and €132, respectively, leading to a weighted mean of €21,703.55 for outbreak management (Online Resource 1 ESM Table 4).

2.6 Sensitivity and Scenario Analyses

Deterministic and probabilistic sensitivity analyses (DSA and PSA) were carried out for predefined parameter ranges (Tables 1 and 2). For PSA, 1000 simulations were carried out; at each iteration, input variables were sampled within the pre-specified limits of uncertainty and using the distribution shown in Online Resource 1 ESM Table 5. Twelve scenario analyses were conducted to investigate the impact of model structure parameters, vaccine effectiveness, and epidemiological parameters on ICERs. Parameter assumptions that were modified included discount rates, the QoL adjustment factor, residual cross-protection of 4CMenB against serotypes W and Y, carriage, the number and proportion of sequelae considered, and the impact of MenB IMD on life expectancy. Further information is provided in Online Resource 1 ESM Tables 3 and 6.