1.    Introduction

Pneumococcal disease is an infection caused by the Streptococcus pneumoniae (S. pneumoniae) bacterium. There are currently over 90 serotypes (STs) recognized worldwide, 15 of which cause the majority of disease [1]. The burden of disease is important, as infection is a leading cause of life-threatening invasive pneumococcal disease (IPD) which mainly includes meningitis and bacteraemia, as well as non-invasive illness such as pneumonia and acute otitis media (AOM) [2]. In Poland, between 2010 and 2016, the National Reference Center for Diagnosis of Central Nervous System Infections (KOROUN) detected 3,447 IPD cases. There were significant differences between provinces and between the analyzed periods in the same provinces. Data reported suggested underreporting and the differences among provinces most likely reflect inconsistence medical practices i.e., low number of blood cultures and serotyping test performed. In 2016, the highest incidence rates of 4.76 and 5.43 per 100,000 people in Poland were observed in those aged over 65 years and under two years, respectively. The IPD case fatality rate was 49.3% (over 65-year-olds), 38.4% (45-64-year-olds), 33.3% (25-44 and 5-9-year-olds), and 6.7% in children under two years [3]. AOM is a very common childhood disease typically treated with antibiotics, with 60-70% of clinical cases caused by bacteria [4]. A meta-analysis reports that on average 35.9% of AOM episodes are caused by S. pneumoniae and 32.3% by non-typeable Haemophilus influenzae (NTHi) [5].

Vaccination in infants is an effective means of preventing diseases, not only in vaccinated but even in unvaccinated infants and in older age groups due to herd immunity [6-10]. Moreover, it plays an important role given increased antimicrobial resistance among some pneumococci [11-13]. A recent Polish study found antibiotic resistance in children with AOM was an important cause of treatment failure [14]. Pneumococcal conjugate vaccines are recommended by the World Health Organization (WHO) [15] for routine immunisation of infants and in many countries introduced in their National Immunization Programme (NIP). There are two licensed conjugate vaccines available in Poland: the pneumococcal non-typeable Haemophilus influenzae protein D conjugate vaccine (PHiD-CV; Synflorix, GSK), and the 13-valent pneumococcal conjugate vaccine (PCV-13; Prevenar13, Pfizer). Since 2017, PHiD-CV is administered as part of the NIP for all healthy infants, in a 2+1 schedule given at two, four and 13 months of age. The vaccine is intended to protect against all spectrum of pneumococcal disease i.e., IPD, pneumonia and AOM. Both vaccines can be used for risk groups in a 3+1 schedule [16]. While the vaccine STs contained in the two vaccines differ to some extent i.e., number of STs, carrier protein, there is evidence of protection exerted from PHiD-CV against cross-related STs. Recently, WHO’s global systematic review [10,17] on the impact of PHiD-CV and PCV-13, and 2 other industry-independent studies, concluded that at this stage, there is no evidence of difference on the net impact on pneumococcal diseases despite the difference of composition between the two vaccines [18,19]. Regardless of the vaccine used and of local epidemiology, countries which have implemented a childhood pneumococcal immunization program with a high coverage, observed a significant reduction in the burden of IPD in children [20].

The objective of this study was to assess the cost-effectiveness of introducing a routine infant vaccination programme against pneumococcal disease in Poland. Two analyses were performed: comparing PHiD-CV to no vaccination, and, comparing PHiD-CV to PCV-13.

2.    Materials and Methods

2.1.   Model and population

A previously published Markov cohort model [21] was adapted for Poland to assess the cost-effectiveness of introducing routine infant vaccination against pneumococcal disease, from the public payer perspective. Vaccination with PHiD-CV was compared to no vaccination and to vaccination with PCV-13.

The model is an age-compartmental, deterministic, static cohort model that assesses the health and economic impact of IPD caused by S. pneumoniae (i.e., meningitis and bacteraemia), all-cause pneumonia, and AOM caused by S. pneumoniae and NTHi, in a birth cohort over lifetime, using monthly cycles. Fig. 1 shows the flow diagram of the model. [22]

Fig. 1 Model flow diagram

Circle boxes represent mutually-exclusive health states. Dashed rectangles (sequelae and deaths) and natural deaths in susceptible individuals show the populations removed from the model. Age-specific incidence was applied monthly to the susceptible population. Costs and benefits were calculated monthly and aggregated over the cohort’s lifetime. Non-consulting AOM were included in the quality of life impact calculation. Redrawn from Delgleize et al.[22]. AOM: acute otitis media; Sp: Streptococcus pneumoniae; TPP: tympanostomy tube placement.

 A Polish birth cohort of 373,527 babies [23] was modelled; during each model cycle, the probability of an individual entering a specific health state was governed by age-specific incidence rates and applicable vaccine efficacy (VE) levels. Infants received, in the vaccine arms, two primary doses of one of the vaccines, at two and four months old, with a booster dose at 13 months old [4,24]. Age-specific overall monthly mortality rates for the general population were obtained from Polish life expectancy tables [25]. Age-specific disease incidence and disease progression were represented by health states (i.e., disease with or without general practitioner (GP) visits, hospitalisation, complications, long-term sequelae or death), linked to resource utilisation (e.g., hospitalisation, GP visits, and specific interventions such as myringotomies for AOM), resulting in direct costs and quality-adjusted life years (QALYs) over lifetime.

In the model, lifetime direct costs and QALYs associated with morbidity and mortality due to IPD, pneumonia and AOM were compared for PHiD-CV versus no vaccination and PHiD-CV versus PCV-13. The incremental cost per QALY gained or incremental cost-effectiveness ratio (ICER), expressed as Polish złoty (PLN)/QALY, was calculated for both comparisons.

Table 1. Vaccine effectiveness: model inputs and assumptions

2.2.   Comparators and vaccine effectiveness assumptions

The model compared vaccination with PHiD-CV versus no vaccination and versus vaccination with PCV-13. A 2+1 regimen as previously described was used for both vaccines.

The vaccines have never been compared directly in trials. As they contain different STs, vaccine effectiveness against IPD, pneumonia and AOM were based on ST-specific efficacy from clinical trials (Table 1). Both vaccines directly protect against disease caused by 10 common STs 1, 4, 5, 6B, 7F, 9V, 14, 18C, 19F and 23F. In addition, PCV-13 directly protects against STs 6A, 19A and 3 [4,24], while PHiD-CV, which includes STs 6B and 19F, has shown evidence of cross-protection against 19A [8] and a decrease of 6A in real-world evaluations after its introduction [26]. Cross-protection occurs when a vaccine demonstrates effectiveness against STs not included in the vaccine.  and 19A [8]. Cross-protection occurs when a vaccine demonstrates effectiveness against STs not included in the vaccine.

Eight of ten STs of PHiD-CV are conjugated to a protein D from NTHi, eliciting robust immune responses [27]. Higher serum antibody levels to protein D have been found to be associated with reduced risk of future AOM [28]. Efficacy against NTHi caused IPD was omitted in this analysis for both vaccines.  Vaccine efficacy was assumed to increase with the number of doses (i.e., 50% efficacy with dose one, 90% with dose two and 100% with dose three) [29,30], and to wane exponentially from three until ten years of age.

2.3.   Effectiveness against IPD

PHiD-CV and PCV-13 were assumed to have the same VE (94.7%, [95% confidence interval (CI): 87.0; 99.9]) for the ten common STs, calculated as the average efficacy of STs contained in the predecessor of PCV-13, i.e. the 7- valent pneumococcal conjugate vaccine (PCV-7) [30]. PHiD-CV was assumed to have no efficacy against ST3, while PCV-13 was assumed to have 26% VE (95%CI: -69; 68) [32]. Even not being significant, this assumption was made as this value was previously used in health technology assessment and tender discussions with the government. Many post-marketing surveillance studies [32-34] have shown no consistent impact of PCV-13 on ST 3 IPD, with most studies showing no impact or lack of effectiveness against ST 3 IPD [8]. As diverse vaccine effectiveness can be found in literature, the impact of 0% and 79.5% (95%CI: 30.3; 94.8) VE against ST 3 IPD (i.e., the highest published [35]) was tested in scenario analyses.

Since PHiD-CV was shown to elicit a similar antibody response to PCV-7 against ST 6A, a cross-protection effectiveness of 76% (95%CI: 39.0; 90.0) was assumed in the model [36]. PHiD-CV was shown to elicit a stronger response against ST 19A than PCV-7 [36-39]. Recent large effectiveness studies observed a clear cross-protective effect against ST 19A ranging from 62% (95%CI: 20; 85) to 82% (95%CI: 10.7; 96.4) [37-39]. This prompted many countries to update the Summary of Product Characteristics (SmPC) to include protection against 19A [37-40]. Therefore, the base case analysis assumed a conservative cross-protection effectiveness against ST 19A (62%) [39] which was left out (0%) and increased to 82% in the scenario analyses [38].

VE against STs 6A and 19A IPD was assumed to equal the other STs in PCV-13 (94.7%, 95%CI: 87.0; 99.9).

Indirect or herd protection resulting from continual vaccination of sequential birth cohorts was taken into account for IPD only (we conservatively assumed no herd protection for pneumonia and AOM). Serotype replacement offsets the incremental effect of indirect protection resulting in a net indirect effect. In the model, indirect protection adjusted for the opposing impact of serotype replacement was applied as a fixed effect to the residual disease incidence. This net indirect effect was estimated at 30%, removing the necessity to account separately for the effect of serotype replacement [41,42]. All efficacy estimates and the net indirect effect applied in the model are in line with Tregnaghi et al. [43] and were validated by a panel of experts (GSK PHiD-CV Health Economics Advisory Board. Leuven, Belgium, September, 2013) [22].

2.4.   Effectiveness against pneumonia

Both vaccines were assumed to have the same effectiveness in reducing GP visits (by 8.7%) and hospitalisation (by 21.8%) due to S. pneumoniae [43].

2.5.   Effectiveness against AOM

Efficacy against AOM was modelled against S. pneumoniae vaccine STs (VT) and against non-vaccine STs (NVTs), as efficacy data by ST or by dose were not available. Vaccine effectiveness against the ten common pneumococcal vaccine STs was assumed to be the same for both vaccines, based on the COMPAS study for PHiD-CV; i.e., 69.9% (95%CI: 29.8; 87.1) [43]. As STs 6A and 19A are included in PCV-13, VE of PCV-13 against STs 6A and 19A AOM were also assumed to be 69.9%, based on the COMPAS study findings for vaccine types included in PHiD-CV.  Effectiveness against NVTs was also assumed to be the same for both vaccines, based on PCV-7 data (FinOM study); i.e., -33% (95%CI: -80.0; 1.0) [44]. Negative numbers indicate ST or pathogen replacement, despite no replacement being observed for PHiD-CV in the COMPAS study [43]. Effectiveness of PHiD-CV was 63.7% (95%CI: -13.9; 88.4) [45] against ST 6A, 0% against ST 6C, and, 45.8% against ST 19A (estimated by taking the ratio of the vaccines’ efficacy against ST 19A IPD). Effectiveness of PCV-13 against ST 6C was assumed to be 63.7% (based on same effectiveness used for PHiD-CV against ST 6A).

Efficacy was also modelled against NTHi. Conservatively, PHiD-CV was assumed to have an efficacy against NTHi caused AOM of 21.5% (95%CI: -43.5; 57.0) [43,46] versus -11.0% (95%CI: -34.0; 8.0) for PCV-13, based on FinOM study for PCV-7 [43,44].  NTHi efficacy was included in the overall reduction of AOM-related GP visits and myringotomy procedures. The latest was used as proxy for inpatient AOM estimated to be 21.2% with PHiD-CV and 12.7% with PCV-13. With PCV-7, Black and colleagues (2000) [47] observed a reduction in ventilatory tube placement (tympanostomy tube placement, TTP) of 20.1% (95%CI: 1.5; 35.2%) and a reduction in AOM episodes of 7.0% (95%CI: 4.1; 9.7%). The VE of PCV-7 against TTP was found, based on the FinOM [44] and Kaiser Permanente [48] studies, to depend on the incidence of TTP, assuming an inverted exponential relationship. According to the Polish incidence of TTP, VE for PCV-7 was estimated using this exponential relationship, and extrapolated to PHiD-CV a ratio of the modelled overall VE against AOM of PHiD-CV (23.8%) over PCV-13 (14.3%), i.e 1.7. The FinIP [29] and POET [45] study estimates are in line with this extrapolation.

2.6.   Health outcomes, resources, costs and utilities

Health and economic benefits that could be achieved through vaccination with either PHiD-CV or PCV-13 were a reduction in cases, sequelae, resource use, healthcare costs, and, a gain in QALYs.

The age-specific population data, incidence and deaths related to IPD, pneumonia and AOM, as well as disease management (e.g. rates of GP visits and hospitalisations) were specific to Poland from available evidence and expert opinion (Table 2) [3,51-56]. (See Online Resource 1 for epidemiology inputs). The ST distribution in IPD in Poland from 2006 to 2016 was estimated from recent national laboratory surveillance reports [3,51,52,55]. (See Online Resource 2).

Data from the National Health Fund [57] and expert opinion were used to estimate the average cost of an acute episode of meningitis, bacteraemia, pneumonia and AOM, and annual costs for the management of sequelae (e.g., hearing loss or neurological) for the reference year 2016. The base case analyses used the published vaccine prices [49,58] of PHiD-CV (PLN 100) and PCV-13 (PLN 150), while in a scenario analysis, price parity between the vaccines was assumed (i.e., PLN 125: average of PHiD-CV and PCV-13 published prices). The cost-effectiveness threshold of PLN 134,514/QALY was defined as 3x Gross Domestic Product (GDP)/capita according to the Polish Agency for Health Technology Assessment and Tariff System [59] (based on PLN 44,838 per capita for 2013–2015 [60] ).

Table 2. Resource use, costs and utilities: inputs and assumptions

Normative population utilities by age group were based on data from Poland [61]. Disease specific disutility data from Poland were not available. Therefore, values from the United Kingdom (UK) were used and applied for an inpatient or outpatient episode of meningitis, bacteraemia, pneumonia or AOM, and, for their long-term sequelae assumed to persist over the individual's remaining lifetime.

2.7.   Currency, price date, and discounting

All costs are in PLN and were updated to 2016 values. Costs and QALYs were discounted at 5.0% and 3.5% per annum, respectively, as per Polish health technology assessment (HTA) guidelines [67] .

2.8.   Sensitivity and scenario analyses

2.8.1. One-way sensitivity analyses

The impact of any important model input on the outcome was evaluated in a one-way sensitivity analysis, whereby key model inputs (e.g., epidemiology, resource use, costs and disutility inputs) were varied one by one, using the lower and upper limits of their 95% confidence interval or in some cases, a range based on plus or minus 20%. A tornado diagram presents the most important impacts on the ICER.

2.8.2. Probabilistic sensitivity analysis

A probabilistic sensitivity analysis (PSA) was conducted to assess the robustness of the ICER when uncertainty in key model inputs was considered simultaneously. In each PSA simulation, the model randomly draws from a probability distribution describing each parameter, thus varying each input simultaneously with every PSA run. The result from 2,000 simulations is presented in a cost-effectiveness plane and cost-effectiveness acceptability curve.

2.8.3. Scenario analyses

Several scenario analyses were also conducted. The first scenario analysis compared PHiD-CV to PCV-13, assuming price parity; with both vaccines costing PLN 125, the average of their published prices. The next scenario analysis (of PHiD-CV versus PCV-13) assessed the impact of parametrising vaccine effectiveness of PHiD-CV against ST19A IPD from the conservative base case value of 62% to a no protection (0%) or to a more optimistic 82.2%, based on recent studies [38] . A final scenario analysis assessed the impact of changing vaccine effectiveness of PCV-13 against ST3 IPD from the base case value of 26% to both 0% and 79.5%, based on post-marketing studies [32-35].

3.    Results

3.1.   Base case results for PHiD-CV versus no vaccination

Based on the model, the introduction of PHiD-CV vaccination versus no vaccination reduced the number of undiscounted IPD cases from 269 to 164 (by 39.2%), pneumonia cases from 261,609 to 252,468 (by 3.5%) and AOM cases from 1,213,534 to 1,113,705 (by 8.2%). This resulted in an increase of 978 QALYs (undiscounted) and a decrease in healthcare costs of PLN 39.6 million(M) (undiscounted), with vaccination costs of PLN 108.5M (undiscounted). The overall incremental cost of the vaccination programme was PLN 71.4M (5% discounted) with 752 QALYs gained (3.5% discounted), resulting in a cost per QALY gained of PLN 94,933 for PHiD-CV versus no vaccination. Therefore, under these assumptions, PHiD-CV was a cost-effective option versus no vaccination, with the cost per QALY gained well below the threshold for cost-effectiveness of PLN 134,514 (Tables 3&4).

Table 3. Undiscounted health and costs outcomes (PLN) (per birth cohort)

Table 4. Base case incremental health and cost outcomes and cost-effectiveness (5% discount on costs and 3.5% discount on effects) 

3.2.   Base case results for PHiD-CV versus PCV-13

When comparing the two vaccines, PHiD-CV had comparable outcomes to PCV-13 regarding IPD and pneumonia health outcomes. PHiD-CV, however, achieved better health outcomes against AOM with fewer inpatient and outpatient cases (38,462), and lower discounted direct costs (PLN 8.3M). Overall, PHiD-CV was associated with a gain in 170 QALYs (discounted at 3.5%) and a reduction in direct disease management costs (- PLN 61.1M, discounted at 5%) compared with PCV-13. Thus, PHiD-CV was the dominant strategy for Poland (Tables 3&4).

Fig. 2 One-way sensitivity analysis results (PHiD-CV versus no vaccination)

The 12 model inputs that had the biggest impact on the ICER (comparing PHiD-CV to no vaccination) when varied in the one-way sensitivity analysis. The central line indicates the base case cost per QALY. Blue: using a lower input, Red: using a higher input for the variable at the left side. AOM: acute otitis media; CAP: community acquired pneumonia; H. influenza: Haemophilus influenzae; NTHi: non-typeable H. influenza; nVT: non-vaccine type; QALY: quality-adjusted life-years; Sp: Streptococcus pneumoniae; VT: vaccine type.

3.3.   One-way sensitivity analyses

Using the one-way sensitivity analysis, the tornado diagrams in Fig. 2 and Fig. 3 show the 12 most impactful model inputs on the ICER when comparing PHiD-CV to no vaccination (Fig. 2) and to PCV-13 (Fig. 3). 

Fig. 3 One-way sensitivity analysis results (PHiD-CV versus PCV-13) The 12 model inputs that had the biggest impact on the ICER (comparing PHiD-CV to PCV-13) when varied in the one-way sensitivity analysis. The central line indicates the base case cost per QALY. Blue: using a lower input, Red: using a higher input for the variable at the left side. AOM: Acute otitis media; IPD: Invasive pneumococcal disease; GP: General Practitioner; H. influenza: Haemophilus influenzae; PCV-13: 13-valent pneumococcal conjugate vaccine; PHiD-CV: Pneumococcal non-typeable Haemophilus influenzae protein D conjugate vaccine; NTHi: Non-typeable Haemophilus influenzae; Sp VT: Streptococcus pneumoniae vaccine type.

When comparing PHiD-CV to no vaccination, the most influential factors were variations in the percent reduction in tube placement, hospitalisations for community-acquired pneumonia (CAP) and percent reduction in CAP hospitalisation. Only a very unrealistic 10% increase in tube placement after vaccination would result in an ICER above the threshold (i.e., PLN 142,939). In all analyses, PHiD-CV resulted in a gain in QALYs versus no vaccination and remains below the threshold of PLN 134,514. Therefore PHiD-CV remained cost-effective (Fig. 2).

When comparing PHiD-CV to PCV-13, assumptions around AOM were the most influential factors, as present in 10 of the top 12 variables. Variations in the value for ‘GP visits for AOM’ had the largest impact on the ICER; with the cost per QALY ranging from -PLN 512,565 to -PLN267,675 per QALY gained. In all analyses, PHiD-CV resulted in a gain in QALYs versus PCV-13 and was cost-saving. Therefore, despite variations in the ICER, PHiD-CV remained the dominant choice over PCV-13 in all cases (Fig. 3).

3.4.   Probabilistic sensitivity analysis

The PSA results showed that the model outcomes are robust to simultaneous probabilistic variation of key model inputs. When comparing PHiD-CV to no vaccination, PHiD-CV was cost-effective in 79.6% of simulations (Fig. 4). When comparing PHiD-CV to PCV-13, PHiD-CV was the dominant strategy in 94.1% of the simulations (Fig. 5).

Fig. 4 Cost-effectiveness plane (A) and acceptability curve (B) for PHiD-CV vs no vaccination. Panel A: Cost-effectiveness plane. Each PSA run is represented by a dot in terms of difference in direct costs against difference in QALYs gained for PHiD-CV compared to no vaccination. Results below the cost-effectiveness threshold (3xGDP/capita = PLN 134,514/QALY) are considered cost-effective. The red box represents the base case result. Panel B: The cost-effectiveness acceptability curve shows the probability (i.e., proportion of PSA runs) that PHiD-CV is cost-effective compared to no vaccination for increasing cost-effectiveness threshold values. The red circle highlights the proportion of PSA runs when the cost per QALY gained remains below the threshold value (3xGDP/capita = PLN 134,514/QALY). GDP: Gross Domestic Product; cap: capita; QALY: quality-adjusted life-year.

3.5.   Results of scenario analyses

Results of all scenarios are summarised in Table 5.

Table 5. Incremental health and cost outcomes as well as cost-effectiveness (5% discount on costs and 3.5% discount on effects) for different scenarios 

3.5.1. Scenario analysis: PHiD-CV versus PCV-13, assuming vaccine price parity (PLN 125)

As in the base case, PHiD-CV resulted in more QALYs gained versus PCV-13, and when assuming price parity between the vaccines, PHiD-CV costs remained lower (by PLN 8.2M) than in the PCV-13 arm, making PHiD-CV the dominant strategy over PCV-13. 

3.5.2. Scenario analysis: PHiD-CV versus PCV-13, assuming 0% and 82.2% VE for PHiD-CV against ST 19A IPD

Omitting any cross protection from PHiD-CV against 19A reduced the incremental QALYs gained from 170 to 147, and the total cost saving from PLN 61.1M to 60.0M, however PHiD-CV will still be dominant in this scenario.

Increasing the VE of PHiD-CV against ST 19A IPD from the base case value of 62% to 82.2% resulted in a further gain of QALYs versus PCV-13 (from 170 to 178 QALYs) as well as an extra cost savings of PLN 310,000 versus base case, due to better health outcomes achieved. As a result, PHiD-CV remained the dominant strategy versus PCV-13 in both scenarios.  

Fig. 5 Cost-effectiveness plane (PCV-13 vs PHiD-CV). Each PSA run is represented by a dot in terms of difference in direct cost against difference in QALYs gained for PHiD-CV compared to PCV-13. Results in the South-East quadrant represent dominance of PHiD-CV over PCV-13 (i.e. PHiD-CV provides more QALY gain at a lower cost than PCV-13). The red box represents the base case result.GDP: Gross Domestic Product; PCV-13: 13-valent pneumococcal conjugate vaccine; PHiD-CV: Pneumococcal non-typeable Haemophilus influenzae protein D conjugate vaccine; QALY: quality-adjusted life-year.

3.5.3. Scenario analysis: PHiD-CV versus PCV-13, assuming 0% and 79.5% VE for PCV-13 against ST 3 IPD

Changing the VE of PCV-13 against ST 3 from the base case value of 26% to 0% or to 79.5% had a limited impact on the overall health and economic outcomes for PCV-13 versus PHiD-CV; i.e., loss or gain of 1 QALY respectively versus PHiD-CV, and, increasing the cost difference by PLN 3,000 or decreasing the cost difference by PLN 5,000, respectively. PHiD-CV remained dominant versus PCV-13 in both scenarios.

4.    Discussion

Based on a previously published Markov model [21], this analysis estimated that adding PHiD-CV to the NIP in Poland will improve health outcomes in the population by reducing cases, and associated long-term sequelae and deaths, of IPD, pneumonia and AOM. Despite the additional cost of vaccination, routine infant vaccination is a cost-effective strategy for healthcare payers (ICER of PLN 94,933 per QALY gained) versus no vaccination.

When comparing the two pneumococcal conjugate vaccines, PHiD-CV and PCV-13, despite some differences between them (e.g., potentially higher effectiveness of PCV-13 against STs 6A and 19A), the overall effectiveness of PHiD-CV was significantly higher. The model predicted very similar health outcomes with both vaccines for IPD and pneumonia which is in line with industry-independent systematic reviews, showing no significant differences in the two vaccines impact on IPD and pneumonia [17-19]. However, the model projected a stronger reduction in AOM with PHiD-CV versus PCV-13, and the direct costs associated with PCV-13 were much higher than with PHiD-CV.

The impact of PCV programs on the incidence of overall IPD and the distribution of VTs and NVTs in children <5 years was analysed from post-marketing studies and from high-quality surveillance data (available for at least two years before and three years after implementation of programmes with PHiD-CV or PCV-13. The study shows that there was a substantial reduction in the burden of overall IPD following introduction of either vaccines in countries with high coverage, regardless of the PCV used and of differences in pneumococcal epidemiology. IPD was mainly due to VTs before PCV introduction, whereas NVTs are the major contributor today in children and adults [20]. As AOM is less severe in nature than IPD, it might not be the primary focus of vaccination programs. However, it is a frequent cause of physician visits and it leads to a significant use of antibiotic use, both of which can result in a high impact on the public health budget. The prevention of a large number of AOM cases through vaccination may be an important contributor to minimising the growing problem of antibiotic resistance [9], as well as reducing indirect costs due to parent absenteeism from work. Based on the clinical FinIP trial setting, a recent study showed that vaccinating 5 infants with PHiD-CV prevented one antimicrobial purchase for uncomplicated AOM during 24 months after administration of the first dose in 2+1 and 3+1 vaccination schedules combined [68].

Overall, PHiD-CV was the dominant strategy compared with PCV-13, providing more health gains (+170 QALYs) at a cost-saving (-PLN 61.1M). PHiD-CV has a lower public price than PCV-13 in Poland, however, the scenario analysis comparing the two vaccines at the same price also found PHiD-CV to be the dominant strategy. Similarly, in scenario analyses assuming lower vaccine effectiveness for PHiD-CV against ST19A IPD or higher effectiveness for PCV-13 against ST3, PHiD-CV remained the dominant option. This can, on one hand, be argued with observations of recent systematic reviews confirming no consistent impact of PCV-13 on ST3 diseases, with most studies showing no impact or a lack of VE for PCV-13 against ST3 IPD [8,18]. On the other hand, considering the relatively low proportion of 19A caused illnesses, the favoring differences for PCV13 over PHiD-CV to protect against 19A diseases [8] can be out-weighted by the benefit of PHiD-CV to prevent other PCV related diseases, such as AOM preventable diseases [68] and/or potentially to reduce the replacement by PCV-13 non-vaccine type IPD [18,69,70]. Moreover, varying the vaccine effectiveness against ST3 and ST19A of both vaccines according published 95% confidence limits was found to have almost no effect on health and economic outcomes regarding IPD and pneumonia for the UK and Canada [68]. Finally, these findings were supported by a recent health economic evaluation from an industry-independent organisation in Quebec (INSPQ, Canada), estimating that for most of the realistic increases of 19A over time when using PHiD-CV, the cost-effectiveness ratio will remain in favour of PHiD-CV (at least 39 and 94 more cases of 19A-IPD per year must be averted in children under 5 years of age by PCV-13 compared to PHiD-CV in order to outweigh the extra cost difference per dose (25$) and to become cost-effective against PHiD-CV at 1x and 3x GDP/capita - for a birth-cohort of about 83,000 new-borns) [71,72]. Therefore, from a public health perspective, it is more important to consider the overall impact, i.e. the overall effectiveness, of both vaccines rather than to link protection with the number or ST coverage of each vaccine.

In one-way sensitivity analyses versus no vaccination, PHiD-CV remained cost-effective in all analyses except when assuming a lower percent reduction in tube placement. However, the estimated assumption used for the lower reduction is lower than 0, reflecting an increase in TTP when children are vaccinated, which is very unlikely based on results of clinical trials, even for both vaccines [29,50]. One-way sensitivity analyses versus PCV-13 found that AOM inputs and assumptions had the greatest influence on outcomes, however PHiD-CV remained dominant in all analyses. Uncertainty in model inputs was tested in probabilistic sensitivity analyses, and the outcomes for PHiD-CV were found to be robust, with PHiD-CV remaining cost-effective versus no vaccination in 79.6% of simulations and remaining dominant versus PCV-13 in 94.1% of simulations.

The findings of this study in Poland are comparable to other cost-effectiveness studies comparing PHiD-CV to PCV-13, where the comparable benefits of both vaccines in preventing IPD and pneumonia were complemented by the larger benefits of PHiD-CV regarding AOM prevention. For example, for studies in Sweden [73], Norway [74], and the UK [22], the outcome was a gain in QALYs with PHiD-CV at a cost-saving. A recent systematic review of 46 pneumococcal vaccine studies concluded that PCV-13 and PHiD-CV are relatively comparable, with PHiD-CV tending to be more cost-effective due to its additional effect on prevention of AOM, which although less severe than IPD, is more prevalent [75]. A study in Germany, by contrast, found PCV-13 to be more cost-effective than PHiD-CV, if the assumptions made around additional indirect effects with PCV-13 were significant [76]. Finally, a recent comparison of the two vaccines’ effectiveness in reducing the incidence of IPD in Sweden, where PCV-13 is used in some counties and PHiD-CV in the others, found no significant differences between the vaccines’ overall effect on IPD, despite serotype differences [18].

This analysis had some limitations, principally due to the use of a static model to evaluate the impact of herd effect or serotype replacement. Serotype replacement was represented in the model by reducing VE, and herd effect was included as a fixed effect at equilibrium. A dynamic model would be better able to model indirect effects such as serotype replacement and herd effect. Moreover, some inputs were from other countries when Polish data were lacking (e.g., disutility data). A conservative approach was taken when assumptions were made.

The analysis also did not consider the burden or costs due to adverse events from vaccine administration in either vaccination arm. However, based on clinical trial and post-marketing safety data, this is likely to be marginal.  Indirect costs due to work absenteeism of patients or parents of patients were out of scope for this study as well as health benefits and cost savings of reducing antibiotic resistance. The latest is hard to be estimated due to lack of data, however this is likely to make both vaccines somewhat more cost-effective that what has been estimated here.

5.    Conclusion

Introducing mass vaccination of infants in Poland with PHiD-CV is likely to reduce the disease burden due to IPD, pneumonia and AOM, while being a cost-effective strategy (versus no vaccination) and a cost-saving strategy (versus PCV-13) for the healthcare payer.

End 2016 the Ministry of Health selected PHiD-CV for their NIP [77,78].

6. Acknowledgements

The authors would like to thank Business & Decision Life Sciences platform for editorial assistance and manuscript coordination, on behalf of GSK. Stephanie Garcia and Lyes Derouiche (publications managers, Business & Decision Life Sciences on behalf of GSK) coordinated manuscript development and editorial support. The authors also thank Kavi Littlewood (Littlewood Writing Solutions, on behalf of GSK) for providing medical writing support.

7. Compliance with Ethical Standards

GlaxoSmithKline Biologicals SA funded this analysis and was involved in all stages of research conduct, including analysis of the data. GlaxoSmithKline Biologicals SA also took in charge all costs associated with the development and publication of this manuscript.

JO, AK and RR are employees of the GSK group of companies. BS, KW and IK were employed by GSK at the time of the study. MMG is a partner of Pracownia HTA Company which received remuneration from the GSK group of companies for the development of a HTA report on Synflorix (analysis disclosed in the present manuscript) and other vaccines. AR discloses no conflict of interest.

8. Trademark

Synflorix is a trademark of the GSK group of companies. Prevenar13 is a trademark owned or licensed by Wyeth LLC.

Supplementary Materials

Sup. 1. Plain Language Summary