Use of adjuvanted trivalent influenza vaccine in older-age adults: a systematic review of economic evidence

ABSTRACT

Seasonal influenza is a very common disease. Yearly vaccination of at-risk population groups is a well-recognized cost-effective/cost-saving preventive measure. It is, however, unclear which available alternative has the most favorable economic profile. Some available options are: trivalent (TIV) and quadrivalent (QIV) inactivated vaccines, adjuvanted TIV (aTIV). Because of immunosenescence, aTIV has been specifically developed for elderly. The present study aimed at assessing the available evidence of aTIV use in elderly from the economic perspective. A systematic literature review targeting aTIV economic evaluations in adults aged ≥65 years was performed using Medline via Ovid, Embase, DARE and NHS/EED. Of a total of 3,654 papers screened, 18 studies (13 full papers, 5 conference abstracts) were included. It emerged that compared with both non-vaccination or non-adjuvanted vaccines, aTIV was cost-effective or cost-saving. The vaccinations strategies incorporating aTIV based on age and/or risk profile are associated with the most favorable economic outcomes.

KEYWORDS:

• Influenza
• vaccine
• trivalent
• adjuvated
• MF59
• older age
• systematic review
• economic evidence

Introduction

Seasonal influenza is a common disease affecting every year up to 10% of adults and up to 30% of children.¹ Among other communicable diseases, influenza ranks at the top in terms of its burden. Indeed, it has been estimated that in Europe the average annual burden of influenza amounts to 81.8/100,000 disability-adjusted life years (DALYs) (cf. tuberculosis, which ranked second, estimated 53.5/100,000 DALYs, i.e., 1.53 times less than influenza).² The enormous population burden of influenza is, however, highly dependent on age. While the infection attack rate is generally the highest in the pediatric population,³ about 90% of influenza-related deaths occur in the elderly.⁴

Annual influenza vaccination is a cornerstone Public Health measure that is able to significantly reduce the burden of disease.^1,5 In the last 70–80 years, several types of influenza vaccine types have been developed, passing from live attenuated to inactivated, whole-virion to subunit, monovalent to quadrivalent, unadjuvanted to adjuvanted, egg-based to recombinant and cell-based, standard-dose to high-dose.^1,6,7 The current worldwide armamentarium of available options includes: trivalent (TIV) and quadrivalent (QIV) inactivated vaccines, adjuvanted TIV (aTIV), high-dose TIV (hdTIV), intradermal TIV (idTIV), cell culture-derived QIV (ccQIV), recombinant QIV (rQIV), and some others.⁷ However, the number of available alternatives is highly different among single countries, being the highest choice in North America, while in some countries only one vaccine type is available. This is likely a result of the different commercial attractiveness of single markets.

It is well-known that compared with younger age groups, standard influenza vaccines like TIV have suboptimal immunogenicity and efficacy in the elderly^8,9 owing to the phenomenon of immunosenescence and a relatively high prevalence of chronic conditions and multi-morbidity among seniors. In order to address these issues, three types of vaccines have been specifically developed for the elderly, namely aTIV, idTIV (15mcg) and hdTIV. It is worth noting that all of these are trivalent. While idTIV enhances immunogenicity thanks to the particular immunological environment of the derma,¹⁰ hdTIV elicits a more robust immune response through the quadruple antigen content.¹¹ By contrast, aTIV takes advantage of the MF59® adjuvant that induces strong and long-lasting T- and B-cell responses against both vaccine-like and drifted/heterotypic strains.¹²

It is now widely acknowledged that annual influenza vaccination is a good value for money; this has been demonstrated in several systematic reviews of economic studies.^13–15 It is particularly cost-effective/cost-saving in the elderly;¹³ indeed, seniors are the main internationally recognized target group for annual immunization campaigns.^1,16

Systematic reviews of economic evaluations are becoming increasingly common. It has been argued¹⁷ that, differently from systematic reviews/meta-analyses of clinical efficacy data, systematic reviews of economic studies should not produce a single result, rather they should inform decision-makers (in a robust and transparent manner) on the potential impact of different resource allocations and how these may vary from setting to setting. For instance, a systematic review of cost-effectiveness studies on replacing TIV with QIV¹⁸ has established that the incremental cost-utility ratio (ICUR) of QIV vs TIV ranged from likely not cost-effective [≥$140,000 per quality-adjusted life year (QALY) gained] to cost-saving.

Considering the above, the present study aimed at systematically assessing the available pharmacoeconomic evidence behind the use of aTIV in the elderly population; indeed, among the three aforementioned elderly-specific vaccines, aTIV was licensed first in 1997,¹² but no comprehensive reviews have been conducted so far. The published systematic evidence^13–15 suggests that influenza vaccination is cost-effective/cost-saving in the elderly but it is unclear which of the different available alternatives⁷ has the most favorable economic profile. Available systematic reviews on economic assessments of influenza vaccination programs have included a few studies^13,15 that specifically addressed aTIV or other elderly-specific vaccines. We, however, hypothesized a larger pool of available evidence on the cost-effectiveness of aTIV, given that it has been on the market over the last 20 years. Moreover, our study was prompted by the recent decision of Public Health England¹⁹ on the preferential use of aTIV in the over-65 age group. Similarly, the very recent normative document issued by the Italian Ministry of Health²⁰ advocated for the preferential use of aTIV in the oldest age group. In summary, the objective of this study was to assess the economic profile of using aTIV in the elderly population as compared to both “doing nothing” and available alternatives and to figure out how this depends on the study setting. The present study should therefore aid decision-makers of different jurisdictions (aTIV is available in more than 30 countries²¹) in establishing an optimal influenza immunization strategy in the elderly.

Methods

Compliance with international reporting standards

The study was carried out by consulting the Joanna Briggs Institute (JBI) guidelines on conducting systematic reviews of economic evaluation evidence.²² Our findings were reported in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement.²³ The internal protocol (in Italian) is available upon request.

Eligibility criteria

The primary studies of interest were model-based economic evaluations including both full [e.g., cost-effectiveness (CEA), cost-utility (CUA), budget impact (BIA), cost-benefit (CBA) analyses] and partial [e.g., cost-consequences analysis (CCA)] economic assessments. However, study-based economic evaluations were also potentially eligible candidates for inclusion.

According to the current indication of aTIV,²¹ the study population included adults aged ≥65 years independently of age subgroup (e.g., 65–74 or ≥75 years), risk profile (e.g., at low or high risk of developing influenza-related complications), institutionalization (e.g., community-dwelling or nursing home residents). aTIV comparators could include both non-vaccination or either competitor.

Papers on economic evaluations of (i) the pediatric formulation of aTIV, (ii) previously available adjuvanted vaccines (e.g., virosomal), (iii) off-label (i.e., age <65 years) use of aTIV, (iv) highly selected population groups non-representative to the general population and (v) pandemic/pre-pandemic adjuvanted vaccines were excluded. Moreover, when a study was first presented at a conference or published in a “grey literature” modality but then published in a peer-reviewed journal, we considered only the latter record.

Search strategy

The automatic search strategy was adapted from previous systematic reviews on aTIV^24,25 and those on economic evaluations of influenza vaccines^13,18 and is reported in Supplementary file 1. Medline via Ovid, Embase, Database of Abstracts of Reviews of Effects (DARE) and the NHS Economic Evaluation Database (EED) were the primary information sources. Considering a relatively small amount of output produced by the DARE/NHS EED databases and for the purpose of increasing sensitivity, only terms relative to aTIV (“MF59”, “adjuvanted influenza vaccine”) were searched. No time, language or other restrictions were applied.

We then proceeded with searching for conference abstracts and grey literature (at www.abstract-archive.org and www.opengrey.eu) using the same simplified search strategy. The subsequent manual search included the following modalities: (i) standard reference checking; (ii) screening of the most pertinent journals (Vaccine, Human Vaccines and Immunotheraputics, Expert Review of Vaccines, Journal of Vaccines and Vaccination, Influenza and Other Respiratory Viruses, The European Journal of Health Economics, Expert Review of Pharmacoeconomics and Outcomes Research, Health Affairs, Health Economics, Journal of Health Economics, PharmacoEconomics, Quality of Life Research, Value in Health); and (iii) searches for articles in Google Scholar that cited the included papers. Finally, in a post-hoc modality, we cross-checked the reference lists of all known systematic reviews on economic evaluations of influenza vaccination in the elderly.^13,18

Potentially eligible records from all information sources were pooled into a spreadsheet and screened for duplicates. As a standard procedure, titles and abstracts were then evaluated to remove clearly irrelevant records; subsequently full texts of potentially eligible studies were assessed for inclusion and exclusion criteria. The search was performed on 12–16/07/2018 independently by IL and cross-checked by IL and AS; any disagreement was solved by consensus.

Data extraction and abstraction

In order to extract potentially useful information, the JBI data extraction form for economic evaluations²² was first consulted. It was then adapted to stay in line with the study objectives. In particular, the following key data were extracted and abstracted: authors, publication year, publication type (journal article or conference abstract), country/territory, currency and reference year, study population and setting, type of economic evaluation and key features of model structure/key assumptions, strategies evaluated and/or comparators, perspective, time horizon and discount rate, methods for principal data extraction/imputation, primary outcome measures, funding source, main results (both the base case and uncertainty analyses) and authors’ conclusions.

We would like to point out the following notes on our extraction/abstraction strategy. First, while the inclusion of conference abstracts that were not published in full may sometimes mitigate the publication bias,²⁶ the information available in the abstracts is provided in limited detail; moreover, the peer-review process of contributions submitted to conferences may be less rigorous.²⁷ Indeed, a recent systematic review by de Boer et al.¹⁸ excluded conference abstracts. We, however, decided to retain the abstracts with the aim of (i) providing the reader the broadest horizon possible and (ii) checking whether the unpublished abstracts differ somehow from the published full texts (i.e., non-positive results are often less likely to be published in full). In any case, in order to circumscribe the intrinsic shortcomings of the conference abstracts we reported them separately from ordinary journal articles.

Second, a note should be made for what concerns the study intervention/strategy. We decided to dichotomize the included studies into “comparison of aTIV with non-vaccination and/or competitors” and “comparison of Public Health strategies”. Indeed, a Public Health vaccination strategy could compare different scenarios, in which different vaccine types are administered to different population groups (e.g., QIV to subjects aged 18–64 years and aTIV to subjects aged ≥65 years) or in different proportions to account for real world market splits (a usual case of BIA). Indeed, aTIV is a vaccine specifically designed for the over 65, while unadjuvanted TIVs/QIVs may be administered to everyone aged of 6 months and above. However, from our personal experience, dynamic models usually include all age-cohorts to see the indirect effect of influenza vaccination (e.g., reduction of influenza cases in the elderly by vaccinating children). By contrast, in a head-to-head comparison of aTIV with “doing nothing” or other competitors, it is intrinsically assumed that a given intervention (i.e., either vaccine or non-vaccination) is given to 100% of a given population group. Therefore, we decided that the variable “study intervention/strategy” should be split into the aforementioned categories as the between-category comparison would be difficult to interpret or even misleading.

Third, all costs were extracted and reported in their original currency. However, in order to make results more comparable, the costs were first trended to 2017 and then exchanged to 2017 US dollars (₂₀₁₇$) by using data of the Organization for Economic Co-operation and Development (OECD),^28,29 as described earlier.¹⁸ If the exact costing year was not explicitly reported, it was imputed as a difference between the year of publication and 3 years.¹⁸

Data were extracted/abstracted by IL and cross-checked by IL and AS; any disagreement was solved by discussion.

Quality appraisal

The quality assessment was performed by using the JBI appraisal checklist for economic evaluations.²² This tool evaluates the quality of economic evaluations through 11 items with four possible response options, namely “Yes”, “No”, “Unclear” and “Not applicable”. The JBI appraisal tool covers different domains, including the available comparators, (e.g., Is there comprehensive description of alternatives?), effectiveness (e.g., Has clinical effectiveness been established?), costs and outcomes (e.g., Are costs and outcomes valued credibly?), etc. The tool also provides an explicit guideline on voting procedure.

The quality appraisal of the included studies was independently conducted by IL and MT; the inter-rater agreement was measured by Cohen’s κ. The identified disagreements were solved by consensus. The quality of conference abstracts was not assessed.

It is worth noting that no study was removed on the basis of a low score obtained by applying the JBI checklist, but pertinent remarks were made in the text.

As suggested by a peer-reviewer, we performed a post-hoc quality reporting assessment of the included full papers by using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement.³⁰

Overall synthesis of results

All studies were summarized qualitatively. In order to summarize the key findings of included studies, whenever possible, we followed the 3-by-3 dominance ranking matrix proposed by JBI.²² Briefly, this scheme envisages three categories (always aTIV vs comparator), namely “Strong dominance” [i.e., (i) aTIV is more effective and less costly; or (ii) aTIV is effective and less costly; or (iii) aTIV has equal cost and is more effective], “Weak dominance” [i.e., (iv) aTIV is equally costly and effective; or (v) aTIV is more effective and more costly; or (vi) aTIV is less effective and less costly] and “Non-dominance” [(vii) aTIV more costly and less effective; or (viii) aTIV is equally as costly and less effective; or (ix) aTIV is more costly and as effective]. Considering that the 3-by-3 dominance ranking matrix takes into account only incremental cost-effectiveness measures, the results of BIAs were not summarized using this tool.

Results

Study selection process

The initial automatic search produced a total of 3,817 results. After duplicate removal (N = 163), a total of 3,654 titles and abstracts were screened and clearly irrelevant records (N = 3,634) were removed. The full text assessment allowed including 11 papers. The list of excluded studies is provided in Supplementary file 2. Furthermore, two potentially eligible studies and five conference abstracts were identified through manual search. A total of 18 studies, corresponding to 13 full papers^31-43 and 5 conference abstracts^44-48 were included in the qualitative assessment. The whole process of study selection is illustrated in Figure 1.

Figure 1. Study selection process.

Characteristics of the included studies: full papers

The main characteristics of the included studies are reported in Table 1. Most studies (N = 10) were written in English,^{31,32,34–37,39–41,43} while the remaining two in Spanish^38,42 and one in Italian,³³ respectively. All studies were conducted in high-income countries, namely Italy (N = 5),^{32,33,36,41,43} the US (N = 3),^34,37,40 Spain (N = 2),^38,42 Canada (N = 2)^35,39 and France (N = 1).³¹ Most studies (N = 8)^{31–33,35,37,41–43} were funded/supported by pharmaceutical companies and/or included industry employees as authors; three studies^34,39,40 were funded/supported by a public body, while the funding source of other two studies^36,38 was not clearly stated. The study population was set to individuals aged 65 years and older in 10 studies,^{32–34,36,38–43} while in the earliest study³¹ it consisted of the elderly at high risk. The over 65 population, in fact, can be considered at high risk of developing complications due to influenza because it has at least one serious chronic disease (diabetes, heart attack, angina pectoris, other heart disease, stroke, chronic bronchitis, chronic obstructive pulmonary disease, liver cirrhosis, malignancy, Alzheimer’s, senile dementia, parkinsonism, renal failure). The remaining two studies^35,37 included all age cohorts (people aged <65 years could receive TIV and/or QIV).

Table 1. Main characteristics of the included studies.

Authors, year [Ref]	Country, currency (year)	Population	Evaluation type	Model type	Prospective	Time horizon	Strategies	Uncertainty analysis
Piercy et al., 2004^Citation31	France, € (Unclear)	Elderly^a at high risk (with heart or lung disease)	CEA	Static	Payer	1 year	TIV; aTIV	One-way deterministic
Baio et al., 2006^Citation32	Italy, € (2003–2004)	Elderly^a	CEA	Static	Payer	1 year	TIV; aTIV; non-vaccination	Probabilistic
Iannazzo et al., 2009^Citation33	Italy, € (Unclear^b)	Elderly^a	CEA	Static	Payer	1 year	TIV; aTIV; non-vaccination	One-way deterministic; scenario
Lee et al., 2009^Citation34	US, $ (2007)	Elderly^a	CUA	Static	Societal; payer	1 year	TIV; aTIV	Probabilistic
Fisman et al., 2011^Citation35	Canada, C$ (2009)	All ages	CUA	Dynamic	Payer	10 years	TIV to all age-classes; TIV to children (<6 years), while aTIV to the elderly; aTIV to both children and the elderly. Each strategy could include or not TIV for people aged 6–64 years.	One-way deterministic; best-worst case scenario
Iannazzo, 2011^Citation36	Italy, € (Unclear^c)	Elderly^a	BIA	Static	Payer	1 year	TIV; aTIV; non-vaccination	One-way deterministic; scenario
Mullikin et al., 2015^Citation37	US, $ (Unclear)	All ages	CUA	Dynamic	Societal; payer	1 year	QIV to people aged <65 years, while aTIV to the elderly; QIV to all age-classes; TIV to all age-classes	One-way deterministic; scenario
Ruiz-Aragón et al., 2015^Citation38	Spain, € (Unclear^d)	Elderly^a	BIA	Static	Payer	1 year	TIV; aTIV	One-way deterministic
Garlapati, 2016^Citation39	Canada, C$ (Unclear)	Elderly^a	CCA	Static	Unclear^e	Unclear^f	TIV; aTIV; non-vaccination	None
Raviotta et al., 2016^Citation40	US, $ (2014)	Elderly^a	CUA	Static	Societal	Unclear^f	Non-vaccination; TIV; QIV, hdTIV; aTIV^g	One-way deterministic; probabilistic
Barbieri et al., 2017^Citation41	Italy, € (2017)	Elderly^a	BIA	Static	Payer	1 year	TIV, aTIV, idTIV, QIV at actual and hypothetical market shares	Scenario
Perez-Rubio et al., 2018^Citation42	Spain, € (Different^d)	Elderly^a	BIA	Static	Payer	1 year	TIV; aTIV	One-way deterministic
Capri et al., 2018^Citation43	Italy, € (2017)	Elderly^a	CUA	Static	Payer	1 year	TIV; aTIV; idTIV; QIV; non-vaccination	One-way deterministic; probabilistic

Notes

a: Defined here as ≥65 years old; b: The vaccine price was a 3-year (2006–2008) average; c: The vaccine price was a 4-year (2006–2009) average; d: It is likely that different reference years were used; adjustment methods were not reported; e: Probably, payer; f: Probably 1 year; g: aTIV was considered only in the secondary analysis

TIV: trivalent inactivated vaccines;

aTIV: adjuvanted trivalent inactivated vaccines;

QIV: quadrivalent inactivated vaccines;

hd-TIV: high-dose trivalent inactivated vaccines;

idTIV: intradermal trivalent inactivated vaccines;

CEA: cost-effectiveness analysis;

CUA: cost-utility analysis;

BIA: budget impact analysis;

CCA: cost-consequences analysis.

From the point of view of modelling characteristics, most studies could be classified as CEAs/CUAs (N = 8)^{31–35,37,40,43} and BIAs (N = 4).^36,38,41,42 One study³⁹ was declared by the author as CEA; however, it seemed to be a CCA with some elements of CBA. All ICURs in CUAs were expressed as cost/QALY gained. Among CEAs the following outcomes were used: cost/life year gained,³¹ cost/death avoided,^31,33 cost per episode of influenza-like illness (ILI) avoided³³ and incremental net benefit (INB).³² A note on BIAs should be also made. Only the most recent BIA by Barbieri et al.⁴¹ considered a “realistic” scenario where each of the alternatives available to the elderly was “weighted” by its market share, while the remaining three BIAs^36,38,42 assumed a universal use aTIV vs TIV and/or non-vaccination. Most models were static, conducted from the third payer’s perspective in a time horizon of one year (one influenza season). At least one type of uncertainty analysis was reported in all studies but one.³⁹ Three studies were classified as “comparison of Public Health strategies”, while the remaining studies compared aTIV with other competitors and/or non-vaccination (Table 1).

Characteristics of the included studies: conference abstracts

All the five abstracts^44-48 were CUAs. Four studies^44-47 included an industry employee as an author, while the funding source was unclear in the fifth.⁴⁸ All the models were likely conducted from the third payer’s perspective of the National Healthcare System (NHS) of Germany,⁴⁴ Spain,⁴⁵ Italy,⁴⁶ the UK⁴⁷ and Canada.⁴⁸ Two models^47,48 were dynamic, included all-age cohorts and compared different Public Health strategies, while all abstracts by Petri et al.^44–46 were static, included only the elderly population and compared single vaccines (TIV, aTIV, idTIV and QIV) and non-vaccination.

Comparison between aTIV and non-vaccination

ativ was explicitly compared to “doing nothing” in five full papers,^{32,33,36,39,43} mostly in the context of Italy^32,33,36,43 (Table 2). Compared with non-vaccination, aTIV seems to give good value for money at least from the third payer perspective being cost-saving^32,33 or cost-effective.⁴³ The main ICUR drivers were natural influenza attack rate, aTIV effectiveness, aTIV acquisition price, vaccine administration cost, cost of hospitalization due to respiratory complications and QALY loss due to an episode of influenza. Likewise, its universal use in the elderly was associated with net budget savings³⁶ (Table 2).

Table 2. Key findings of the included studies (full papers): adjuvanted trivalent inactivated vaccine (aTIV) vs non-vaccination.

Authors, year [Ref]	Key assumptions	Key findings: base case	Key findings: uncertainty analysis
Baio et al., 2006^Citation32	ILI attack rate: 16.6%. Vaccination coverage: 52.6%. Reduction in cases generated by aTIV: pharmacological treatments (62.6%); GP visits (50.9%); hospitalization for influenza/pneumonia (44.2%), chronic heart failure (31.7%), respiratory disease (35.4%); mortality (54.8%). aTIV price (with administration): €14.12 (2017$19.81).	Results were reported graphically with no explicitly stated base case. For example, at the WTP threshold of €500 (2017$702) per death avoided, the probability aTIV produces a positive incremental net benefit was approximately 80%.	The probability of having a positive incremental net benefit was >50% at WTP thresholds of up to €6,000 (2017$8,418) per death avoided.
Iannazzo et al., 2009^Citation33	ILI attack rate: 16.8%. Vaccination coverage: 64.9%. Reduction in cases generated by aTIV in the elderly: H1N1 (11.8%); H3N2 (23.8%); B (19.6%). Reduction in cases generated by aTIV in the elderly at high risk: H1N1 (10.4%); H3N2 (26.6%); B (23.8%). aTIV price: €5.59 (2017$7.25).	In terms of both €/ILI avoided and €/death avoided aTIV was dominant. aTIV would produce €83.7 (2017$108.5) million (−7.9%) of savings in terms of total direct costs.	In the scenario analysis (complete mismatch) aTIV remained the dominant strategy: it would produce a net saving of €35.6 (2017$46.1) million (−3.4%) in front of health gains. In the one-way sensitivity analysis ICER aTIV vs non-vaccination was constantly negative (aTIV dominant strategy).
Iannazzo, 2011^Citation36	ILI attack rate: 16.8%. Vaccination coverage: 64.9%. aTIV effectiveness against ILI in the low- and high-risk elderly of 53% and 55%, respectively. aTIV price: €5.58 (2017$6.99).	aTIV would produce €74.0 (2017$92.7) million (−6.8%) of savings in terms of total direct costs.	In the scenario analysis (complete mismatch) aTIV would still produce a net saving of €14.6 (2017$18.3) million (−1.3%). In the one-way sensitivity analysis aTIV was constantly associated with net savings in terms of total costs.
Garlapati, 2016^Citation39	Unclear attack rate. aTIV effectiveness: 40%. aTIV price: C$12.59 (2017$10.30).	The benefit-cost ratio of aTIV vs non-vaccination was 49.8.	–
Capri et al, 2018^Citation43	Influenza attack rate: 5.4%. Vaccination coverage in the low- and high-risk elderly of 55.1% and 66.0%, respectively. aTIV effectiveness against laboratory-confirmed influenza: 72.5%. aTIV price: €6.99 (2017$7.88).	ICUR: €12,305 (2017$13,866)/QALY	In the one-way analysis, the major ICUR drivers were: influenza attack rate, aTIV effectiveness, aTIV price, aTIV administration cost and QALY loss due to influenza; the min-max variation [from €5,104 (2017$5,752)/QALY to €19,358 (2017$21,814)/QALY] of these parameters were below the adopted ICUR threshold of €30,000 (2017$33,807)/QALY. In the probabilistic sensitivity analysis (10,000 Monte Carlo rounds) aTIV was cost-effective (ICUR<€30,000/QALY) in 80.9% of simulations, dominant in 1.8%, likely not cost-effective (ICUR>€30,000/QALY) in 17.3% of simulations.

aTIV: adjuvanted trivalent inactivated vaccines;

ICUR: incremental cost-utility ratio;

ICER: incremental cost-effectiveness ratio;

QALY: Quality Adjusted Life Years

ILI: Influenza-like illness;

Results of the comparison with non-vaccination were also reported in three conference abstracts: aTIV has been found to be a cost-effective strategy in Germany [ICUR: €925/QALY (₂₀₁₇$1,094/QALY)]⁴⁴ and Italy [ICUR: €12,305 (₂₀₁₇$14,061)/QALY]⁴⁶ and dominant in Spain [incremental costs: -€0.48 (2017$0.55)/person, incremental QALYs: 0.0016/person).⁴⁵

Comparison between aTIV and other vaccine types

TIV was the most common active comparator used in all included full papers (Table 3). Among the included CEAs/CUAs aTIV was deemed cost-saving^32-34 or cost-effective.^31,43 The earliest paper by Piercy et al.³¹ is the only to report an incremental cost-effectiveness ratio (ICER) separately for each virus strain: it was the lowest for A(H3N2) and the highest for A(H1N1). In the sensitivity analysis the following variables drove significantly the results: influenza attack rate, aTIV and TIV effectiveness, and acquisition price of both vaccines. From the point of view of budget impact, the use of aTIV was associated with net savings in Italy^33,36 and Spain.^38,42 Three conference abstracts by Petri et al.^44–46 reported consistent results. In Germany,⁴⁴ it was stated that aTIV was dominant over TIV at a relative price increase of up to 31% [€9.20 (₂₀₁₇$10.88) and €12.06 (₂₀₁₇$14.26), respectively]. It was analogously cost-effective [at the willingness-to-pay (WTP) threshold of €30,000/QALY] in Spain⁴⁵ and Italy⁴⁶ with an incremental price of 352% [€3.10 (₂₀₁₇$3.54) vs €14.00 (₂₀₁₇$15.96)] and 156% [€6.15 (₂₀₁₇$7.03) vs €15.75 (₂₀₁₇$16.30)], respectively.

Table 3. Key findings of the included studies (full papers): adjuvanted trivalent inactivated vaccine (aTIV) vs non-adjuvanted trivalent inactivated vaccine (TIV).

Authors, year [Ref]	Key assumptions	Key findings: base case	Key findings: uncertainty analysis
Piercy et al., 2004^Citation31	Relative vaccine effectiveness of aTIV vs TIV was 6.0%, 15.0% and 18.2% against H1N1, H3N2 and B, respectively. Relative price increase of aTIV vs TIV: 50%.	ICERH1N1: €17,496 (2017$24,752)/LYG or €87,040 (2017$122,243)/DA; ICERH3N2: €3,759 (2017$5,279)/LYG or €18,701 (2017$26,265)/DA; ICERB: €4,821 (2017$6,671)/LYG or €23,985 (2017$33,686)/DA.	In the sensitivity analysis aTIV was constantly cost-effective or cost-saving. In particular, the influenza attack rate was a strong driver of ICER: at the attack rate of 10% ICERB became negative in terms of both CPLYG and CPDA (aTIV was dominant), while at the attack rate, while it was €19,435 (2017$27,296)/LYG or €96,686 (2017$135,791)/DA at the attack rate of 2%. Variations in life expectancy (from 1 to 7 years) produced a ranger of ICERB of €3,426 (2017$4,812) to €26,383 (2017$37,054) per LYG, while different discount rates (0–10%) did not change ICER significantly.
Baio et al., 2006^Citation32	aTIV was associated with a greater relative reduction than TIV: pharmacological treatments (18.0%); GP visits (13.4%); hospitalization for influenza/pneumonia (20.1%), chronic heart failure (32.6%), respiratory disease (22.0%); mortality (10.0%). Relative price increase of aTIV vs TIV: 9.7%.	Results were reported graphically with no explicitly stated base case. For example, at the WTP threshold of €0 per death avoided, the probability aTIV produces a positive incremental net benefit was >90%.	The probability of having a positive incremental net benefit was >50% at WTP thresholds of up to €6,000 (2017$8,418) per death avoided.
Iannazzo et al., 2009^Citation33	Relative vaccine effectiveness of aTIV vs TIV against ILI: 29.3% and 34.1% in the low- and high-risk elderly, respectively. Relative price increase of aTIV vs TIV: 46.5%.	In terms of both €/ILI avoided and €/death avoided aTIV was dominant. A total replacement of TIV with aTIV would produce €129.5 (2017$167.8) million (−11.7%) of savings in terms of total direct costs.	In the scenario analysis (complete mismatch) aTIV remained the dominant strategy: it would produce a net saving of €123.3 (2017$159.8) million (−10.7%) in front of health gains. In the one-way sensitivity analysis ICER aTIV vs TIV was constantly negative (aTIV dominant strategy). In the breakeven analysis aTIV remained dominant with a relative price increase of up to 473.5%.
Lee et al., 2009^Citation34	Use of the adjuvant overcomes the imunnosenescence, i.e. the elderly had the same vaccine effectiveness as non-elderly adults. Cost of adjuvant varied from $15.75 (2017$18.61), which is the average TIV price, to $100.00 (2017$118.17).	From both the societal and payer perspective, aTIV was a dominant strategy with a price increase of up to $30 (2017$35). From the societal perspective, replacement of TIV with aTIV would save $824 (2017$974) million, $595 (2017$703) million, $369 (2017$436) million and $167 (2017$197) million at an extra price of aTIV of $0, $10 (2017$12), $20 (2017$24) and $30 (2017$35), respectively. From the payer perspective, replacement of TIV with aTIV would save $842 (2017$995) million, $614 (2017$726) million, $394 (2017$466) million and $167 (2017$197) million at an extra price of aTIV of $0, $10, $20 and $30, respectively.	aTIV was always the dominant strategy as long as the extra price over TIV was >$20 (2017$24). When aTIV had an extra price of $40 (2017$47), it ceased to be a favourable strategy when the WTP was >$750 (2017$886)/QALY.
Iannazzo, 2011^Citation36	Relative vaccine effectiveness of aTIV vs TIV against ILI: 29.3% and 34.1% in the low- and high-risk elderly, respectively. Relative price increase of aTIV vs TIV: 43.7%.	A total replacement of TIV with aTIV would produce €124.3 (2017$155.8) million (−10.9%) in savings in terms of total direct costs.	In the scenario analysis (complete mismatch) aTIV would still produce a net saving of €124.6 (2017$156.2) million (−10.4%). In the one-way sensitivity analysis ICER aTIV vs TIV was constantly negative (aTIV dominant strategy).
Ruiz-Aragón et al., 2015^Citation38	Relative vaccine effectiveness of aTIV vs TIV against ILI: 34.1%. Relative price increase of aTIV vs TIV: 14.7%.	A total replacement of TIV with aTIV would produce €76.1 (2017$87.8) million in savings in terms of total direct costs.	Replacing TIV with aTIV was a cost-saving strategy when varying vaccination coverage (from 32.9% to 70.9%, base case: 57%) with a range of €43.9–78.2 (2017$50.7–90.3) million of net savings. Diminishing the aTIV price by 10% did not produce significant changes [€78.2 (2017$90.3) million in net savings].
Garlapati, 2016^Citation39	Relative vaccine effectiveness of aTIV vs TIV: 100%. Relative price increase of aTIV vs TIV: 67.9%.	The benefit-cost ratio of aTIV vs TIV was 87.6.	–
Perez-Rubio et al., 2018^Citation42	Relative vaccine effectiveness of aTIV vs TIV against ILI: 34.1%. Relative price increase of aTIV vs TIV: 48.3%.	A total replacement of TIV with aTIV would produce €82.5 (2017$94.5) million in savings in terms of total direct costs with a benefit-cost ratio of 12.83.	In the one-way analysis the benefit-cost ratio was constantly >9.
Capri et al, 2018^Citation43	Relative vaccine effectiveness of aTIV vs TIV against laboratory-confirmed influenza: 25.0%. Relative price increase of aTIV vs TIV: 30.7%.	ICUR: €4,527 (2017$4,797)/QALY	In the one-way analysis, the major ICUR drivers were: TIV (from aTIV being dominant to dominated) and aTIV [from €222 (2017$250)/QALY to €119,477 (2017$134,638)/QALY] effectiveness and TIV price [from aTIV being dominant to €24,479 (2017$27,785)/QALY]. In the probabilistic sensitivity analysis (10,000 Monte Carlo rounds) aTIV was cost-effective [ICUR<€30,000 (2017$33,807)/QALY] in 58.3% of simulations, dominant in 18.1%, likely not cost-effective (ICUR>€30,000/QALY) in 13.1% and was dominated by TIV in 10.5% of simulations.

aTIV: adjuvanted trivalent inactivated vaccines;

ICUR: incremental cost-utility ratio;

ICER: incremental cost-effectiveness ratio;

QALY: Quality Adjusted Life Years

ILI: Influenza-like illness;

A head-to-head comparison aTIV vs idTIV was reported in a full paper by Capri et al..⁴³ From the Italian NHS perspective, aTIV dominated idTIV in the base case scenario. In the probabilistic sensitivity analysis, the acceptability curves of four vaccines (TIV, aTIV, idTIV and QIV) showed that at the threshold of €30,000/QALY idTIV had the probability of 25.5% to be the best choice, while the same probability for aTIV was 58.8%.

A full paper by Capri et al.⁴³ reported results on a direct comparison aTIV vs QIV. Considering that there was no effectiveness estimate for QIV in the elderly, the latter was assumed to be more effective than TIV by 35% in case of B lineage mismatch (relative to B lineage strain included in trivalent formulations); the frequency of B lineage mismatch was set to 61.8%. In the base case scenario, QIV was clearly dominated by aTIV from the Italian NHS perspective [incremental costs: -€5.2 (₂₀₁₇$5.86)/person, incremental QALYs: 2x10⁻⁷/person]. In the probabilistic sensitivity analysis, QIV remained constantly dominated by aTIV independently from the WTP threshold set.

Finally, Raviotta et al.⁴⁰ compared different vaccines available to the US elderly; however, aTIV was used as a comparator only in a secondary analysis (and therefore several relevant data were not described) since at the time the study was conducted, aTIV was not available in the US. In the primary analysis, hdTIV was found to be cost-effective in comparison with non-vaccination, TIV and QIV. The authors then conducted a secondary two-way sensitivity analysis by varying the relative effectiveness of aTIV vs TIV in a range of 0–50% and its price in the range of $10–50. At the WTP threshold of $100,000/QALY, aTIV was not favored if its relative effectiveness increase was <15% but was favored if its relative effectiveness was ≥32%. Moreover, if it is assumed that the relative effectiveness vs TIV was the same as hdTIV vs TIV (24.2%), it would be favored if its acquisition price was less than $31.20 (₂₀₁₇$32.29).

Comparison between different public health strategies

Fisman et al.³⁵ compared the universal use of TIV in Canada with two alternative strategies, namely (i) use of aTIV in the elderly and TIV in children aged <6 years and (ii) use of aTIV in both children and the elderly. Moreover, each scenario could also include the immunization with TIV of subjects aged 6–64 years. Compared with the universal offering of TIV to the entire population aged >0.5 years, the strategy in which aTIV was used in the elderly while TIV in persons aged 0.5–64 years produced an ICUR of C$2,111 (₂₀₁₇$1,853)/QALY gained. When aTIV offer was extended to children, the ICUR dropped to <C$500 (₂₀₁₇$439)/QALY gained. The latter strategy was also highly cost-effective [C$1,612 (₂₀₁₇$1,415)/QALY gained] when compared with that of TIV to people aged 0.5–64 years and aTIV to the elderly. A similar picture was seen in the comparison of scenarios when people aged 6–64 years were excluded from the vaccination offer [TIV to children and the elderly vs TIV to children and aTIV to the elderly: C$1,424 (₂₀₁₇$1,250)/QALY; TIV to children and the elderly vs aTIV to both children and the elderly: <C$300 (₂₀₁₇$263)/QALY; aTIV to both children and the elderly vs TIV to children and aTIV to the elderly: C$1,190 (₂₀₁₇$1,045)/QALY].

The US study by Mullikin et al.³⁷ compared the differential use of QIV and aTIV (administered to subjects aged <65 and ≥65 years, respectively) to the universal all-age use of QIV and TIV. This comparison was made using nine scenarios according to the intensity of seasonal influenza activity (low, average and high) and matching (low, average and high correspondence between circulating and vaccine strains) levels. From the societal perspective, the differential use of QIV/aTIV was dominant over QIV independently from influenza intensity and match level. QIV/aTIV versus TIV was the dominant strategy in all but three scenarios where the former was deemed cost-effective [ICURs of $9,980 (₂₀₁₇$10,649)/QALY, $28,761 (₂₀₁₇$30,690)/QALY and $1,955 (₂₀₁₇$2,086)/QALY in the scenarios low intensity + average match, low intensity + high match and average intensity + high match, respectively]. A similar picture was seen when only direct costs were considered. The differential use of QIV/aTIV was constantly dominant over QIV. Compared with the universal use of TIV, the strategy QIV/aTIV was dominant in three scenarios (average intensity + low match, high intensity + low match, high intensity + average match), while it was cost-effective in other six scenarios with the ICUR range of $1,619–34,404 (₂₀₁₇$1,728–36,712)/QALY gained.

A recent BIA by Barbieri et al.⁴¹ compared the real-world strategy with the observed market share of all available vaccines for the elderly in Italy (TIV, aTIV, idTIV and QIV) with a hypothetical strategy in which each vaccine is administered according to the age (65–74 and ≥75 years) and risk profile (low and high risks). Switching to the alternative strategy was associated with a significant reduction in terms of occurrence of influenza-related events and slightly higher total vaccination campaign costs (+1.5%) that, however, were offset by a reduction of costs associated with lower number of influenza events and therefore generating net budget savings. In the scenario analyses where QIV took the whole quotum of TIV in low-risk people aged 65–74 years (with or without QIV price reduction of 20%), only a slight increase in total costs was seen [€0.07–0.24 (₂₀₁₇$0.08–0.27)/person].

Finally, two conference abstracts compared different Public Health strategies in the UK⁴⁷ and Canada,⁴⁸ by using dynamic modelling approach. Nguyen et al.⁴⁷ compared the current (not well-defined) UK strategy with a hypothetical one where aTIV is preferentially used in people aged ≥65 years. At a WTP threshold of £20,000 (₂₀₁₇$26,667)/QALY and aTIV price of £16 (₂₀₁₇$21.3), the strategy would be highly cost-effective with an ICUR of £3,540 (₂₀₁₇$4,720)/QALY. In the probabilistic sensitivity analysis, the ICUR was <£19,048 (₂₀₁₇$25,397)/QALY in 97% of simulations. It was also estimated that the use of aTIV would be cost-neutral to the UK NHS when aTIV price is within a range of £10–13 (₂₀₁₇$13.3–17.3). A recent conference abstract by Fisman et al.⁴⁸ has reported that in Canada, over a 10-year period, a differential vaccination strategy (aTIV for both infants and the elderly, QIV for children and TIV for adults) would produce 325,000 QALYs gained and C$1.33 (₂₀₁₇$1.04) billion in savings, as compared with non-vaccination.

Quality assessment of the included full papers

Although the absolute agreement in judging single quality attributes was relatively high (79.2%), Cohen’s κ was deemed only moderate (0.46). Following the reconciliation process, it was observed that the overall quality of the included studies could be judged moderate to good; each item was met by at least 60% of studies (Figure 2). However, only two studies^31,43 were deemed to meet all quality criteria. A notable exception was the study by Garlapati³⁹ that did not meet 8 out of 11 quality attributes (Supplementary file 3). That study was identified through manual search and was published in a journal not currently indexed in either Medline or Embase. There were no obvious patterns relating the study quality to the year and language of publication, funding source and overall study findings.

Figure 2. Quality appraisal of the included studies (Joanna Briggs Institute checklist).

Notes: Item 1, Is there a well-defined question?; Item 2, Is there comprehensive description of alternatives?; Item 3, Are all important and relevant costs and outcomes for each alternative identified?; Item 4, Has clinical effectiveness been established? Item 5, Are costs and outcomes measured accurately? Item 6, Are costs and outcomes valued credibly? Item 7, Are costs and outcomes adjusted for differential timing? Item 8, Is there an incremental analysis of costs and consequences? Item 9, Were sensitivity analyses conducted to investigate uncertainty in estimates of cost or consequences? Item 10, Do study results include all issues of concern to users? Item 11, Are the results generalizable to the setting of interest in the review?

The CHEERS checklist produced almost similar results (Supplementary file 4). The most frequent regarded reporting of currency, price date and conversion (item 9); conflict of interest (item 24); study setting and location (item 5) and analytical methods used (item 17). Indeed, these items were not reported/clearly reported by 69%, 54%, 38% and 38% of included studies.

Overall synthesis of results

The JBI 3-by-3 dominance ranking matrix showed that all pairwise comparisons could be categorized as strong or weak dominance of aTIV (Supplementary file 5). In particular, compared with non-vaccination, aTIV was more effective and less costly in 4 comparisons and more effective and more costly in 2 comparisons. When compared with TIVs, aTIV established a strong dominance in 6 comparisons and a weak dominance in 2 comparisons. aTIV was directly compared to idTIV and QIV only in the study by Capri et al.;⁴³ aTIV was strongly dominant over both comparisons (Supplementary file 5).

By analyzing the studies that compared different Public Health vaccination strategies, it seems that the strategies incorporating aTIV on the basis of age and/or risk profile are associated with the most favorable economic outcomes.

Discussion

The present paper is the first to comprehensively assess the economic profile of the use of aTIV in the elderly population. Our main conclusion is that aTIV is “good value for money” independently from the comparator evaluated. The concept of “value for money” is central to health policy (especially among the universal healthcare systems) and the main reason for that is related to accountability. In particular, payers should be reassured that their money is being spent wisely, while patients should be reassured that their claims on the NHS are being treated fairly and consistently.⁴⁹

In line with this concept, a comparison with non-vaccination revealed that aTIV was highly cost-effective or cost-saving in all included studies; this finding is consistent with the previously published systematic evidence on the overall economic value of influenza vaccination in the elderly.¹³ Similarly, aTIV has been found to be cost-effective/dominant in comparison with TIV and dominant over QIV and idTIV. From the point of view of “real-world” Public Health strategies, the available evidence suggests that there is a need to establish an optimal combination of existing vaccines that is based on age and/or risk profile of the target population groups. The latter finding is in line with recently formulated guidelines issued by health authorities in the UK,¹⁹ Italy²⁰ and Australia.⁵⁰ In particular, in the UK general practitioners and community pharmacists were advised to preferentially use aTIV in all people aged ≥65 years and QIV in people at risk aged 18–64 years.¹⁹ The most recent Italian recommendations issued by the Ministry of Health²⁰ suggested that both aTIV and QIV may be used in people aged ≥65 years. However, given that the A(H3N2) subtype has the highest burden in the oldest age group (≥75 years) and the evidence of better effectiveness of aTIV in this particular age group, aTIV should be preferred over both QIV and TIV among over-75 elderly people. In other younger target populations, QIV should be preferred over TIV (unless the former is unavailable).²⁰ Similarly, in Australia⁵⁰ both aTIV and hdTIV are preferentially recommended over QIV in ≥65-year-olds, while QIVs are to be used in all target groups <64 years. In summary, our review advocates the recent policy changes from the health economics point of view. Furthermore, a somewhat similar theoretical concept of the appropriateness of the use of different influenza vaccines has been proposed by Bonanni et al..⁵¹ It is, however, interesting to note that in the US, although hdTIV, aTIV and QIV are available, the Recommendations of the Advisory Committee on Immunization Practices (ACIP), 2018–19 Influenza Season for the elderly population does not express any preference for the type of vaccine to be used. For persons aged ≥65 years, any TIV age-appropriate formulation (standard-dose or high-dose, trivalent or quadrivalent, unadjuvanted or adjuvanted) or QIV are acceptable options.⁵² Additional data concerning these vaccines from studies examining immunogenicity and non-laboratory-confirmed influenza outcomes are needed. The main problem in the US is the lack of robust studies so ACIP will continue to review data concerning the efficacy and effectiveness of these vaccines as more information emerges and will state new recommendations.

The present systematic review has both strengths and limitations. The main strength is that we were able to analyze a significantly higher number of original studies on the economic profile of aTIV than any previously published systematic review focusing on influenza vaccines of any type.^13-15 This was likely achieved thanks to a sufficiently high sensitivity and specificity of the adopted search strategy. Another strength concerns the fact that the included studies had similar conclusions and therefore the overall review summary and interpretation were straightforward. The main drawback was that the inter-rater agreement of the quality appraisal was not high (even if it was judged moderate). The most probable explanation is that both raters studied the checklist independently and did not agree before how to vote. However, to make comparisons, we were not able to identify any assessment on the reliability and other psychometric properties of the JBI checklist. However, our post-hoc application of the CHEERS checklist produced roughly comparable to the JBI tool results; indeed, the latter is partially based on the former.

We then would like to discuss the principal limitations of the included papers. First, most comparisons were made with TIV and non-vaccination, while aTIV was compared to idTIV, hdTIV and QIV only in two studies. This is not surprising since about half of studies were conducted in the period when only TIV and aTIV were available. However, given that newer vaccine formulations have recently entered the market of some jurisdictions (e.g. ccQIV and rQIV), novel models should be constructed. Indeed, the recent observational study by the Food and Drug Administration showed that ccQIV and hdTIV were more effective in preventing hospitalizations for influenza-related events than standard egg-based QIV by 11% and 8% in the US elderly.⁵³ Second, the reporting quality should be improved. One may observe that several studies were published about 10–15 years ago when economic evaluations of the vaccines had just commenced. Third, most studies were conducted from the third payer perspective only. This is likely due to the fact that most of the studies included were carried out in countries with National and/or Regional tender policies, like Italy and Spain. However, the recent recommendation issued by the World Health Organization (WHO) on the economic evaluation of influenza vaccination⁵⁴ has stated that the societal perspective should be ideally adopted, while (where possible) other perspectives should be also reported separately. Indeed, the third payer perspective underestimates indirect costs even in the case of elderly since the latter may contribute to the economy by looking after grandchildren during influenza epidemics.⁴³ Forth, most included studies somehow (directly or indirectly) involved the industry. The same pattern has been established by the recently published systematic review of cost-effectiveness studies on QIV.¹⁸ This so-called “funding” or “sponsorship” bias⁵⁵ may skew study outcomes in favor of the sponsoring company’s product. However, the exact mechanisms and level of influence of the funding bias are still largely unknown and inconclusive.⁵⁶ In any case, we believe that despite a limited budget available to governmental and Regional Public Health authorities (especially on the condition of healthcare spending review seen in recent years) independent economic evaluations of the available influenza vaccine formulations should be performed and made readily available to all relevant stakeholders given that annual influenza vaccination in the elderly is a well-recognized, cost-effective and targetable goal.⁴³ Indeed, independently funded economic studies would further reassure both taxpayers and patients. The fifth limitation identified by the authors is conditional and regards the fact that most of the included studies were static and not dynamic. While the static models are relatively easy to construct and interpret, they do not account for herd protection effects;⁵⁴ this, indeed, underestimates the total vaccine-induced savings. However, most models specifically targeted the elderly, where herd protection is less important, while all age-class CUAs models were dynamic. Moreover, static models usually produce more conservative estimates.¹⁸ Therefore, this limitation has a limited impact on our findings.

Apart from the above-mentioned suggestions for future research, the authors would like to acknowledge that most economic evaluations of QIV (identified through the systematic review by de Boer et al.¹⁸) compared the latter to a generic category of “TIVs”. It is therefore not clear whether the “TIV” category was composed of only standard unadjuvanted TIV or included all trivalent formulations together (i.e., TIV plus enhanced elderly specific aTIV, idTIV and hdTIV). Indeed, the second hypothesis is unacceptable given that enhanced trivalent elderly-specific vaccine formulations have been found more immunogenic and effective than both TIVs and QIVs.^25,57-60 Compared with trivalent vaccine formulations QIV is expected to provide a higher effectiveness by covering both B lineages. Type B lineage mismatch is a frequent event and the degree of cross-lineage protection provided by TIV is still controversial.^61,62 On the other hand, a recent systematic review and meta-regression analysis by Beyer et al.⁶³ has established that the cross-lineage protection highly depends on age of vaccinees; QIV is therefore expected to provide significant benefits in younger populations, while its advantage over TIV is rather limited in the elderly.⁶³

To conclude, aTIV use in people aged 65 years and over seems to be a good value for money and it may be cost-saving in some jurisdictions. Future economic evaluations should be up-to-date and include either available alternatives. Health authorities should give general practitioners the opportunity to choose the most appropriate vaccines tailored to specific patients. In this regard, an availability of clinical guidelines on which specific vaccine type is to be administered to a specific population group (such as those recently issued in the UK,¹⁹ Italy²⁰ and Australia⁵⁰) may be a core component. Indeed, formal guideline development expertise among physicians may represent a bottom-up approach by training end-users of such guidelines in order to develop their own explicit evidence-based guidelines which meet nationally recognized criteria.⁶⁴

Disclosure of potential conflicts of interest

IL has been serving as consultant for Seqirus srl since January 2019 for an unrelated project. AS, AN, MT declare no conflict of interest.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

This review received no specific funding.