Cost-effectiveness analysis of quadrivalent influenza vaccination in at-risk adults and the elderly: an updated analysis in the UK

Abstract

Objective:

To update an earlier evaluation estimating the cost-effectiveness of quadrivalent influenza vaccination (QIV) compared with trivalent influenza vaccination (TIV) in the adult population currently recommended for influenza vaccination in the UK (all people aged ≥65 years and people aged 18–64 years with clinical risk conditions).

Methods:

This analysis takes into account updated vaccine prices, reference costs, influenza strain circulation, and burden of illness data. A lifetime, multi-cohort, static Markov model was constructed with seven age groups. The model was run in 1-year cycles for a lifetime, i.e., until the youngest patients at entry reached the age of 100 years. The base-case analysis was from the perspective of the UK National Health Service, with a secondary analysis from the societal perspective. Costs and benefits were discounted at 3.5%. Herd effects were not included. Inputs were derived from systematic reviews, peer-reviewed articles, and government publications and databases. One-way and probabilistic sensitivity analyses were performed.

Results:

In the base-case, QIV would be expected to avoid 1,413,392 influenza cases, 41,780 hospitalizations, and 19,906 deaths over the lifetime horizon, compared with TIV. The estimated incremental cost-effectiveness ratio (ICER) was £14,645 per quality-adjusted life-year (QALY) gained. From the societal perspective, the estimated ICER was £13,497/QALY. A strategy of vaccinating only people aged ≥65 years had an estimated ICER of £11,998/QALY. Sensitivity analysis indicated that only two parameters, seasonal variation in influenza B matching and influenza A circulation, had a substantial effect on the ICER. QIV would be likely to be cost-effective compared with TIV in 68% of simulations with a willingness-to-pay threshold of <£20,000/QALY and 87% with a willingness-to-pay threshold of <£30,000/QALY.

Conclusions:

In this updated analysis, QIV was estimated to be cost-effective compared with TIV in the UK.

Keywords::

• Cost-effectiveness
• Quadrivalent
• Seasonal influenza
• Trivalent
• UK
• Vaccination

Introduction

Influenza is an acute viral disease, which is self-limiting in most people but can cause serious complications or death. Population groups at elevated risk of influenza complications include infants, elderly people (aged 65 years or more), and people with chronic illnesses such as respiratory or heart disease1. In temperate climates such as the UK, influenza is generally a seasonal disease, with most cases occurring in the winter1. Mainly two types of influenza virus, influenza A and influenza B, can cause significant clinical disease in humans1. Influenza A undergoes antigenic drift and is responsible for seasonal epidemics and occasional pandemics. Influenza B has two lineages, Yamagata and Victoria, which have co-circulated annually at varying levels since 2001–20022^,3.

Annual vaccination to protect against seasonal influenza is recommended by the World Health Organization (WHO) for high-risk groups4. The UK has recently extended its influenza vaccination program to include administration of a live attenuated intra-nasal influenza vaccine to children aged 2 to <17 years1^,5. Continuous surveillance of circulating strains is required to identify the influenza strains that should be included in each season’s influenza vaccine. Until recently, most seasonal influenza vaccines were trivalent, with two influenza A sub-types and one influenza B lineage. However, co-circulation of the two influenza B lineages could limit the effectiveness of trivalent vaccines, especially in years with a substantial mismatch between the circulating influenza B lineage(s) and the influenza B lineage selected for that year’s seasonal vaccine. There is considerable variation between seasons in the circulating influenza virus strains and lineages, and in the degree of matching between the seasonal influenza vaccine and the circulating viruses. In the UK, the vaccine influenza B lineage and the circulating influenza B lineage were at least partially mismatched in 6 of the 10 influenza seasons from 2000/2001 to 2009/20106. Quadrivalent vaccines that include both influenza B lineages could improve protection against influenza B. A quadrivalent inactivated influenza vaccine (Fluarix Tetra™) is now available7, and was introduced in the UK in time for the 2013–2014 influenza season8.

Economic data on the cost of influenza in the UK are scarce. In review of economic studies on influenza published in 2013, no UK studies reporting national cost-of-illness estimates were identified9. Since the publication of this review, new evidence on the burden and cost of influenza in the UK has been published10–13. An economic evaluation of quadrivalent influenza vaccination compared with trivalent influenza vaccination in elderly people and clinical risk groups was conducted in 2013 by one of the authors of the present study (G. Meier) and colleagues14. Since this economic evaluation was conducted, the new burden and cost of illness studies and the price of the quadrivalent vaccine in the UK have become available.

The objective of the present study was to update the earlier economic evaluation of quadrivalent influenza vaccination, taking into account current vaccine prices, updated reference costs, the most recent influenza strain circulation data, and results from new database studies on the burden and cost of influenza in the UK. This analysis estimates the cost-effectiveness of quadrivalent influenza vaccination compared with trivalent influenza vaccination in the adult population currently recommended for influenza vaccination in the UK, i.e. all people aged 65 years or more and people aged 18–64 years with clinical risk conditions (defined according to National Institute of Health and Care Excellence [NICE] criteria)15.

Patients and methods

Model design and structure

Full details of the model structure and the rationale for the model design have been published previously14. Briefly, this is a static multi-cohort Markov model with a 1-year cycle time, in which several cohorts in different age groups enter the simulation and are followed over a lifetime of consecutive influenza seasons, i.e., until the youngest cohorts at entry reach the age of 100 years. This modeling approach is better able than a 1-year model to capture long-term consequences and to reflect real-world influenza management, in which individuals are vaccinated in each successive influenza season.

In each annual cycle, individuals could be vaccinated against influenza, become infected with influenza, receive medical attention from a general practitioner (GP) or accident and emergency (A&E) department, receive antiviral treatment, develop influenza-related complications (respiratory or non-respiratory) that may in turn lead to hospitalization or outpatient treatment, or die from influenza or other causes. Surviving individuals moved to the next annual cycle14. The criteria for receiving post-exposure prophylaxis (PEP) or treatment with antiviral medications were based on UK guidelines15^,16.

The base-case analysis compared quadrivalent influenza vaccination with trivalent influenza vaccination over a lifetime time horizon. An additional analysis considered a 10-year time horizon. The main outcome measure was the number of quality-adjusted life-years (QALYs) gained and the incremental cost-effectiveness ratio (ICER) per QALY gained. Costs and outcomes were discounted at 3.5% per year in the base-case, in line with the UK guidelines17. The effects of different discount rates were explored in sensitivity analysis. The base-case analysis was conducted from the perspective of the UK National Health Service (NHS), the national healthcare provider, and, thus, included only the direct medical costs. A secondary analysis was conducted from the societal perspective. Herd protection effects were not included in the model. The model was developed using Microsoft Excel 2010, and has been previously validated14.

Population

The population considered in the present analysis differs from the one in the previous publication. In the present analysis, the base-case population vaccinated included all people aged 65 years or more and people aged 18–64 years with clinical risk conditions, i.e. adults and elderly individuals only. This is consistent with the guidance from the Joint Committee on Vaccination and Immunization (JCVI), as a live attenuated intra-nasal vaccine was recommended for the program extension to children and adolescents5, rather than the inactivated quadrivalent and inactivated trivalent influenza vaccines compared in the present analysis. Thus, the two age groups covering children and adolescents in the earlier analysis were excluded from this one. The other seven age groups covered adults and elderly people (18–49, 50–64, 65–69, 70–74, 75–79, 80–84, ≥85 years)14. The at-risk group was further sub-divided into people with clinical risk conditions and people living in residential care homes. Clinical risk conditions included patients with one or more of the following characteristics: chronic respiratory disease; chronic heart disease; chronic renal disease; chronic liver disease; chronic neurological conditions; diabetes mellitus; or immunosuppressed (including transplanted patients)15. The UK guidelines issued by JCVI and Public Health England also recommend vaccination of pregnant women and workers in health and social care1^,5. However, these groups were excluded from the model.

A scenario analysis considered vaccination of people aged 65 years or more only.

Input data

Influenza circulation data

The present analysis differed from the previous evaluation as it excluded influenza circulation data for the influenza pandemic year of 2009–2010, when the seasonal influenza vaccination considered in this analysis would not have been appropriate. It also used the most recent surveillance data on circulation of influenza A and influenza B reported by the UK Health Protection Agency (HPA)18. The average proportions of influenza A and influenza B over 10 seasons from 2002–2003 to 2012–2013, excluding the pandemic year of 2009–2010, were 74.3% and 25.7%, respectively.

Population data

Table 1 shows the population distribution by age and risk status. UK population and age distribution data were updated to the most recent available (2012 and 2011, respectively) using data published by the Office for National Statistics (ONS)19. All-cause mortality data for the general UK population were updated to the most recent available20, and distributed between the at-risk and healthy populations by assuming that all-cause mortality in the at-risk population was 10-times the all-cause mortality in the healthy population14^,21.

Table 1. UK population distributed by age group and risk status.

Age groups	Population distribution19	Distribution within age-group
		Healthy	At-risk
			Clinical at-risk12^,13^,22*	Residential care27†
18–49 years	43.99%	88.63%	11.38%	0.00%
50–64 years	18.23%	82.38%	17.63%	0.00%
65–69 years	4.86%	46.25%	48.75%	5.00%
70–74 years	3.88%	46.25%	48.75%	5.00%
75–79 years	3.19%	46.25%	48.75%	5.00%
80–84 years	2.38%	46.25%	48.75%	5.00%
>85 years	2.22%	46.25%	48.75%	5.00%

*Defined according to National Institute for Health and Care Excellence criteria15.

†Defined as people living in residential care facilities.

The proportion of each age group categorized as clinically at-risk was obtained from the percentage of patients at-risk in the General Practice Research Database (GPRD), now known as the Clinical Practice Research Datalink (CPRD), in a recent study12^,13^,22. The proportion of people in residential care was unchanged from the earlier analysis14.

Probability and vaccine data

The age-dependent probabilities of moving from healthy to at-risk in each annual cycle were assumed to be independent of vaccination status and influenza exposure, and were calculated from all-cause mortality data and the age distribution of the at-risk population19–21. These data are shown in Table 2.

Table 2. Rate of movement from healthy status to at-risk status in each annual cycle by age*.

Age (years)	Rate	Age (years)	Rate	Age (years)	Rate	Age (years)	Rate
18	0.000,15
19	0.000,18
20	0.000,18
21	0.000,20	41	0.000,68	61	0.004,47	81	0.045,54
22	0.000,21	42	0.000,75	62	0.004,92	82	0.052,02
23	0.000,19	43	0.000,78	63	0.005,35	83	0.058,78
24	0.000,19	44	0.000,84	64	0.440,12	84	0.065,25
25	0.000,23	45	0.000,92	65	0.008,69	85	0.073,02
26	0.000,22	46	0.001,01	66	0.009,71	86	0.081,74
27	0.000,24	47	0.000,99	67	0.010,43	87	0.091,38
28	0.000,26	48	0.001,15	68	0.011,32	88	0.102,95
29	0.000,27	49	0.071,59	69	0.012,64	89	0.114,71
30	0.000,29	50	0.001,63	70	0.014,93	90	0.129,41
31	0.000,31	51	0.001,84	71	0.015,91	91	0.148,88
32	0.000,32	52	0.002,05	72	0.017,57	92	0.143,53
33	0.000,35	53	0.002,24	73	0.018,61	93	0.168,72
34	0.000,39	54	0.002,39	74	0.021,44	94	0.192,62
35	0.000,43	55	0.002,66	75	0.023,04	95	0.215,51
36	0.000,46	56	0.002,94	76	0.025,96	96	0.231,47
37	0.000,49	57	0.003,19	77	0.029,02	97	0.256,23
38	0.000,52	58	0.003,44	78	0.032,32	98	0.276,12
39	0.000,56	59	0.003,81	79	0.036,23	99	0.300,32
40	0.000,62	60	0.004,18	80	0.040,95	100	0.352,44

*Calculated from all-cause mortality data and the age distribution of the at-risk population19–21.

Vaccine efficacy data used in the analysis are shown in Table 3. The average efficacy of the trivalent vaccine against influenza A was estimated from two Cochrane reviews in healthy adults23 and elderly people24, and its efficacy against vaccine-matched and mismatched influenza B lineages was estimated from a meta-analysis in adults25. The average matching with the circulating influenza B lineages over 10 seasons from 2002–2003 to 2012–2013, excluding the pandemic year of 2009–2010, was calculated from HPA surveillance data18. The average of these 10 seasons, allowing for cross-protection, was used for the estimation of the average trivalent vaccine efficacy against influenza B in the base-case. The quadrivalent vaccine was assumed to have the same efficacy as the trivalent vaccine against influenza A, and the same efficacy against influenza B as the trivalent vaccine with optimal matching14. It is assumed that there was no difference in vaccine efficacy between healthy and at-risk groups. No adverse effects of vaccination were considered, in line with a previous systematic review and economic evaluation26.

Table 3. Vaccine efficacy and coverage rates.

	Age group (years)
	18–49	50–64	65–69	70–74	75–79	80–84	85+
Vaccine efficacy against influenza A
Trivalent and quadrivalent vaccine	61.00%	61.00%	58.00%	58.00%	58.00%	58.00%	58.00%
Vaccine efficacy against influenza B
Trivalent vaccine*	63.69%	60.22%	57.28%	57.28%	54.28%	54.28%	54.28%
Quadrivalent vaccine†	77.00%	73.00%	69.00%	69.00%	66.00%	66.00%	66.00%
Vaccine coverage rates
Healthy	0%	0%	71.13%	71.13%	71.13%	71.13%	71.13%
At-risk	34.07%	100%	71.13%	71.13%	71.13%	71.13%	71.13%

*Allowing for cross-protection, and with vaccine matching to circulating strains of influenza B set at the average of the 10 seasons from 2002–2003 to 2012–2013, excluding the pandemic year of 2009–2010.

†Equivalent to the efficacy of the trivalent vaccine with optimal matching to circulating strains.

Vaccine coverage was derived from data reported in the GPRD over 8 seasons from 2000–2001 to 2007–20081 2^,13^,22. The vaccine coverage rates are shown in Table 3.

Age-dependent probabilities for events in the model are shown in Table 4. Probabilities for receiving PEP, influenza infection, and receiving antiviral treatment were derived from previous publications26–28 and were unchanged from the previous model14. Probabilities for seeking medical advice, developing complications, hospitalization, and death due to influenza were calculated from two recent UK database studies10–13^,22^,29^,30. Although the model can differentiate between respiratory and non-respiratory complications and between different types of complication within each of these categories, the input values for parameters associated with complications (e.g. hospitalization rate, death rate, cost, duration, and disutility) used in this analysis were the same for all. Table 4, thus, does not show the probabilities for the different types of complication, since in the present analysis the distribution between different complication types did not influence the results. It was assumed that the probability of death from influenza without complications was zero in all age and risk groups.

Table 4. Probabilities of events.

Parameter	Age group (years)
	18–49	50–64	65–69	70–74	75–79	80–84	85+
Probabilities related to infection and treatment
Receiving PEP*	2.10%	2.10%	1.36%	1.36%	1.36%	1.36%	1.36%
Resistance to PEP†	1.77%	1.77%	1.77%	1.77%	1.77%	1.77%	1.77%
Symptomatic influenza infection in absence of PEP, healthy,  and at-risk populations, identical for influenza A and B27	6.55%	6.55%	6.17%	6.17%	6.17%	6.17%	6.17%
Relative risk of symptomatic influenza after PEP compared  with no PEP26	22.62%	22.62%	22.47%	22.47%	22.47%	22.47%	22.47%
Seeking medical advice, healthy, and at-risk  populations10–13^,22^,29^,30	18.17%	19.22%	19.85%	19.85%	19.85%	19.85%	19.85%
Probability that medical advice is a GP visit (others are  A&E visits), healthy, and at-risk populations10^,11^,29^,30	99.73%	99.70%	99.30%	99.30%	99.30%	99.30%	99.30%
Probability of receiving neuraminidase inhibitors on seeking  medical advice, at-risk only26‡	16.00%	16.00%	11.00%	11.00%	11.00%	11.00%	11.00%
Probability of complication after GP or A&E visit without effective antiviral treatment$
Healthy10^,11^,29^,30	5.04%	5.31%	8.05%	8.05%	8.05%	8.05%	8.05%
At-risk10^,11^,29^,30	13.16%	11.88%	16.02%	16.02%	16.02%	16.02%	16.02%
Relative risk of complication after effective antiviral treatment,  at-risk only26	48.01%	47.64%	48.84%	48.84%	48.84%	48.84%	48.84%
Probability of hospitalization following any complication, %10–13^,22^,29^,30
Healthy	2.12%	4.31%	19.53%	19.53%	19.53%	19.53%	19.53%
At-risk	6.50%	16.06%	30.45%	30.45%	30.45%	30.45%	30.45%
Probability of death following hospitalization for any complication, %12^,13^,22
Healthy	1.14%	2.47%	18.16%	18.16%	18.16%	18.16%	18.16%
At-risk	2.89%	10.74%	26.77%	26.77%	26.77%	26.77%	26.77%
Probability of death following outpatient treatment for any complication, %10–13^,22^,29^,30
Healthy	0.08%	0.47%	12.85%	12.85%	12.85%	12.85%	12.85%
At-risk	0.23%	1.29%	8.46%	8.46%	8.46%	8.46%	8.46%

A&E, accident and emergency; GP, general practitioner; PEP, post-exposure prophylaxis.

*At-risk population only. Probabilities for clinical at-risk population without prior vaccination26; the same probabilities are assumed for the residential care population with or without vaccination as no distinction was made in the original source. The clinical at-risk population with previous vaccination did not receive post-exposure prophylaxis.

†Calculated as the mean of three reported values in adults28, elderly assumed to be the same as adults.

‡At-risk population only (neuraminidase treatment is not recommended for healthy individuals in the UK). Identical for GP and A&E visits. Probability of resistance to neuraminidase inhibitors 1.77% for all age groups, calculated as the mean of three reported values in adults28, elderly assumed to be the same as adults.

$Assumed that the probability is identical for patients not seeking medical advice.

Cost data

The reference year for all costs was 2012/2013. The trivalent vaccine price was updated to £6.4331, and the quadrivalent vaccine price was updated to the list price of £9.9432. It was assumed that elderly and at-risk patients would receive the vaccination during a visit to a GP for other health reasons, so no vaccine administration costs were included. The cost of neuraminidase inhibitor treatment and PEP was updated using the most recent published data31 and was £17.22 for all age groups. The cost of a visit to the GP or A&E were updated to the latest published data from the Personal Social Services Research Unit (PSSRU)33 (£37.00 and £135.00, respectively, for all age groups).

The mean costs per script for prescribed medications in each age group were obtained from a UK database study10^,11^,29^,30 and adjusted to 2013 prices using the PSSRU inflation index33. The cost of nurse visits at home was estimated from the probability of a home visit34, PSSRU unit cost data for 201333, and assuming that adult patients would receive one visit and elderly patients would receive three35. The average cost of outpatient and hospital treatment for complications in each age group were obtained from a UK database study10^,11^,29^,30 and adjusted to 2013 prices using the PSSRU inflation index33. Costs were assumed to be the same for all complications. These age-dependent cost data are shown in Table 5.

Table 5. Age-dependent cost data.

Parameter	Age group (years)
	18–49	50–64	65–69	70–74	75–79	80–84	85+
Medication cost per script10^,11^,29^,30*	£0.07	£0.14	£1.04	£1.04	£1.04	£1.04	£1.04
Reimbursed home visits by nurse33-35*†	£0.014	£0.021	£0.356	£0.356	£0.356	£0.356	£0.356
Hospitalization for complications10^,11^,29^,30	£5,017.20	£7,597.89	£10,249.62	£10,249.62	£10,249.62	£10,249.62	£10,249.62
Outpatient treatment for complications10^,11^,29^,30	£74.66	£78.06	£78.73	£78.73	£78.73	£78.73	£78.73

*Weighted average accounting for differences between healthy and at-risk individuals and the distribution of the population between healthy and at-risk.

†Cost of home visits by nurse estimated from the probability of a home visit34, unit cost data for 201333, and assuming that adult patients would receive one visit and elderly patients would receive three35.

Cost data used only in the analysis from a societal perspective

Costs for prescription fees, over-the-counter medication, private nursing care, and transportation are borne by the patient. These costs are included only in the societal perspective and are not included in the NHS perspective. Prescription fees were estimated from the proportion of people receiving any medication36, the cost of a script from Department of Health37 data, and the average number of scripts per prescription fee38. No prescription fees were included for vaccination or hospitalization. Over-the-counter medication usage was estimated from previous publications34^,38^,39, and multiplied by average cost27 adjusted to 2013 prices using the PSSRU inflation index33. It was assumed that the probability of a private nurse visit was half that of a reimbursed nursing visit, and the costs were calculated as described for reimbursed nursing visits above. These cost data are shown in Table 6.

Table 6. Age-dependent cost data used only in the analysis from a societal perspective.

Parameter	Age group (years)
	18–49	50–64	65–69	70–74	75–79	80–84	85+
Prescription fees36–38
PEP	£1.67	£1.55	£0.00	£0.00	£0.00	£0.00	£0.00
Influenza without complications	£0.94	£0.88	£0.00	£0.00	£0.00	£0.00	£0.00
Influenza with complications, outpatient treatment	£1.46	£1.36	£0.00	£0.00	£0.00	£0.00	£0.00
Over-the-counter medication costs27^,34^,38^,39
Influenza without complications and without  medical advice	£1.97	£1.97	£1.49	£1.49	£1.49	£1.49	£1.49
Influenza without complications and with medical  advice	£0.91	£0.91	£0.91	£0.91	£0.91	£0.91	£0.91
Influenza with complications treated in outpatient  setting, or after hospital discharge	£0.91	£0.91	£0.91	£0.91	£0.91	£0.91	£0.91
Private nurse visits33–35
Influenza without complications	£0.007	£0.011	£0.063	£0.063	£0.063	£0.063	£0.063
Influenza with complications treated in outpatient  setting, or after hospital discharge	£0.007	£0.011	£0.189	£0.189	£0.189	£0.189	£0.189
Indirect costs
Time away from work and reduced productivity at work10^,11^,29^,30^,42^,43
No hospital treatment	£360.49	£360.49	£0.00	£0.00	£0.00	£0.00	£0.00
Hospitalization	£664.77	£1006.65	£0.00	£0.00	£0.00	£0.00	£0.00

PEP, post-exposure prophylaxis.

Average transport costs for all influenza cases seeking medical advice were £9.21 in all age groups, derived from a published study40 and adjusted to 2013 levels using the Consumer Price Index from the ONS41.

Indirect costs include lost productivity due to time away from work (absenteeism), reduced productivity due to illness while at work (presenteeism), and lost productivity due to premature death. It was assumed that elderly individuals (aged 65 years or more) incurred no lost productivity because they would already have retired from work, and adult relatives would visit them outside working hours. A day away from work was assumed to cost 20% of average weekly earnings in 201342. The number of days off work due to influenza not requiring hospital treatment was obtained from Keech and Beardsworth43, and for influenza requiring hospitalization it was assumed to be equal to the duration of hospital stay reported in a UK database study10^,11^,29^,30. These data are shown in Table 6.

Yearly productivity loss due to premature death was derived from Menzin et al.44, calculating a weighted average between males and females using ONS data19 and adjusted to 2013 average weekly earnings42, and is shown in Table 7. Lost productivity due to premature death was not included in the total costs or in the ICER calculations.

Table 7. Age-dependent costs of yearly productivity loss due to premature death*.

Age (years)	Cost	Age (years)	Cost	Age (years)	Cost	Age (years)	Cost
18	£479,120
19	£479,120
20	£479,120
21	£479,120	41	£282,267	61	£282,267	81	£24,107
22	£479,120	42	£282,267	62	£282,267	82	£24,107
23	£479,120	43	£282,267	63	£282,267	83	£24,107
24	£479,120	44	£282,267	64	£282,267	84	£24,107
25	£479,120	45	£282,267	65	£24,107	85	£24,107
26	£479,120	46	£282,267	66	£24,107	86	£24,107
27	£479,120	47	£282,267	67	£24,107	87	£24,107
28	£479,120	48	£282,267	68	£24,107	88	£24,107
29	£479,120	49	£282,267	69	£24,107	89	£24,107
30	£479,120	50	£282,267	70	£24,107	90	£24,107
31	£479,120	51	£282,267	71	£24,107	91	£24,107
32	£479,120	52	£282,267	72	£24,107	92	£24,107
33	£479,120	53	£282,267	73	£24,107	93	£24,107
34	£479,120	54	£282,267	74	£24,107	94	£24,107
35	£282,267	55	£282,267	75	£24,107	95	£24,107
36	£282,267	56	£282,267	76	£24,107	96	£24,107
37	£282,267	57	£282,267	77	£24,107	97	£24,107
38	£282,267	58	£282,267	78	£24,107	98	£24,107
39	£282,267	59	£282,267	79	£24,107	9	£24,107
40	£282,267	60	£282,267	80	£24,107	100	£24,107

*Productivity loss due to premature death is not included in the incremental cost-effectiveness ratio.

Utility data

Baseline utility data for the healthy and at-risk population were calculated from a UK study, adjusting for the different distribution of age groups used in the model45.

Disutilities for influenza episodes not requiring hospitalization were −0.68 for adults aged 18–64 years and −0.88 for people aged ≥65 years, and disutilities for influenza requiring hospitalization were −0.80 and −0.98, respectively46^,47. The duration of an episode not requiring hospitalization was 7.5 days without effective neuraminidase treatment47, and effective neuraminidase treatment was assumed to shorten the duration by 2.5 days, with no effect on the disutility14. The duration of an episode requiring hospitalization was taken as the length of stay for influenza with any complication reported in a UK database study10^,11^,29^,30. Complicated episodes treated in an outpatient setting were assumed to have the same disutility as an uncomplicated influenza episode and the same duration as a hospitalization. These data are shown in Table 8.

Table 8. Age-dependent utility data.

Parameter	Age group (years)
	18–49	50–64	65–69	70–74	75–79	80–84	85+
Baseline utility data45
Healthy	0.96	0.93	0.93	0.92	0.90	0.88	0.82
At-risk	0.91	0.82	0.80	0.78	0.75	0.70	0.65
Influenza not requiring hospitalization
Disutility46^,47	−0.68	−0.68	−0.88	−0.88	−0.88	−0.88	−0.88
Duration without effective neuraminidase inhibitor treatment47	7.5 days	7.5 days	7.5 days	7.5 days	7.5 days	7.5 days	7.5 days
Duration with effective neuraminidase inhibitor treatment47	5 days	5 days	5 days	5 days	5 days	5 days	5 days
Influenza requiring hospitalization
Disutility46^,47	−0.80	−0.80	−0.98	−0.98	−0.98	−0.98	−0.98
Duration10^,11^,29^,30	7.0 days	10.6 days	14.3 days	14.3 days	14.3 days	14.3 days	14.3 days

Sensitivity analysis

A one-way sensitivity analysis was performed using ranges for the input parameters derived from published sources where possible. A probabilistic sensitivity analysis (PSA) was performed by recording the results of 1000 Monte Carlo simulations, each of which simultaneously sampled each of the model’s input parameters from a probability distribution. The parameters used in the sensitivity analyses are shown in Table 9. The PSA results calculated a 95% confidence interval (CI). The one-way sensitivity analysis results compared the baseline value with the change from baseline.

Table 9. Parameters used in the sensitivity analyses.

Parameter	PSA distribution	Range for one-way SA
Demographics
Percentage of population being clinically at-risk	Beta	95% CI
Percentage of population being residential	Beta	95% CI
Vaccine efficacy
TIV efficacy against influenza A in healthy and at-risk individuals	Lognormal	95% CI
QIV efficacy against influenza A in healthy and at-risk individuals		95% CI
TIV efficacy against influenza B with matching in healthy and at-risk individuals	Beta	Range
QIV efficacy against influenza B in case of mismatch in healthy and at-risk individuals	Beta	Range
Matching of B-strain in vaccine and circulation	Beta	Min–max
Circulation of B-strain	Beta	Min–max
Probabilities
Vaccination coverage rates in healthy and at-risk individuals	Beta	95% CI
Probability that ILI is influenza	Beta	95% CI
Proportion of clinically at-risk individuals without prior vaccination receiving PEP	Dependent on other parameters*	Min–max
Proportion of residential care individuals without prior vaccination receiving PEP	Dependent on other parameters*	Min–max
Proportion of residential care individuals with prior vaccination receiving PEP	Dependent on other parameters*	Min–max
Proportion of healthy and at-risk individuals resistant to PEP	Beta	95% CI
Probability of having influenza A in healthy and at-risk individuals (with no vaccination,  no PEP)	Lognormal	95% CI
Probability of having influenza B in healthy and at-risk individuals (with no vaccination,  no PEP)	Lognormal	95% CI
Relative risk of influenza following PEP in at-risk individuals	Lognormal	95% CI
Probability to seek for medical advice in healthy and at-risk individuals	Beta	95% CI
Probability to visit a GP in healthy and at-risk individuals	Beta	95% CI
Probability to receive NI after GP visit in at-risk individuals	Beta	95% CI
Probability to receive NI after A&E visit in at-risk individuals	Beta	95% CI
Probability to be resistant to NI in at-risk individuals	Beta	95% CI
Probability of complication after GP visit without effective NI treatment in healthy  and at-risk individuals	Beta	95% CI
Probability of complication after A&E visit without effective NI treatment in healthy  and at-risk individuals	Beta	95% CI
Relative risk to develop complications after NI treatment vs no NI treatment in  at-risk individuals	Lognormal	95% CI
Probability of complication without previous medical advice in healthy and  at-risk individuals	Beta	95% CI
Overall probability of a respiratory-related complication in healthy and  at-risk individuals	Beta	95% CI
Probability that a respiratory complication is pneumonia in healthy and  at-risk individuals	Beta	95% CI
Probability of cardiac complication in healthy and at-risk individuals	Beta	95% CI
Probability of renal complication in healthy and at-risk individuals	Beta	95% CI
Probability of CNS complication in healthy and at-risk individuals	Beta	95% CI
Probability of OM complication in healthy and at-risk individuals	Beta	95% CI
Probability of hospitalization having complication in healthy and at-risk individuals	Beta	95% CI
Probability of death after hospitalization for complication in healthy and  at-risk individuals	Beta	95% CI
Probability of death after outpatient treatment of complication in healthy and  at-risk individuals	Beta	95% CI
Reimbursable medical costs
Accident and Emergency visit cost	Gamma	95% CI
Cost per script	Gamma	95% CI
Cost of hospitalization with complication	Gamma	95% CI
Non-reimbursable costs (societal perspective)
Non-reimbursed OTC medication with uncomplicated influenza without medical advice  (no GP and no A&E visit)	Gamma	95% CI
Non-reimbursed OTC medication with uncomplicated influenza with medical advice  (GP and A&E visit)	Gamma	95% CI
Non-reimbursed OTC medication with outpatient treatment of complications	Gamma	95% CI
Non-reimbursed OTC medication related to a period after hospital discharge	Gamma	95% CI
Indirect costs
Absenteeism and presenteeism with uncomplicated influenza	Gamma	95% CI
Absenteeism and presenteeism because of hospitalization	Gamma	95% CI
Absenteeism and presenteeism because of treatment of an influenza complication  in an outpatient setting	Gamma	95% CI
Outcomes
Average utility of a healthy and at-risk individual	Normal	95% CI
Disutility with influenza (no NI treatment)	Gamma Lognormal	95% CI
Influenza duration (no NI treatment)	Gamma	95% CI
Disutility with influenza when NI treatment	Gamma Lognormal	Range
Influenza duration when NI treatment	Gamma	95% CI
Disutility with hospitalization of complication	Gamma Lognormal	Range
Duration of hospitalization with complication	Normal	95% CI
Disutility of outpatient treatment of complication	Gamma Lognormal	Range
Duration of outpatient treatment of complication	Normal	95% CI

*Dependent upon 8 other parameters: Probability influenza A at-risk; Probability influenza B at-risk; circulation influenza A; circulation influenza B; Probability NI treatment after GP visit; Probability of GP visit at-risk; Probability of NI treatment after A&E visit; Probability ILI is influenza.

A&E, accident and emergency; CI, confidence interval; CNS, central nervous system; GP, general practitioner; ILI, influenza-like illness; NI, neuraminidase inhibitor; OM, otitis media; OTC, over-the-counter; PEP, post-exposure prophylaxis; PSA, probabilistic sensitivity analysis; QIV, quadrivalent vaccine; SA, sensitivity analysis; TIV, trivalent vaccine.

Scenario analyses

Scenario 1 explored the effect of vaccinating only the elderly population.

Two further scenarios explored the effect of seasonal variation in the degree of matching between the trivalent vaccine and the circulating virus strains. Scenario 2 applied data from the worst matched season (2005–2006), and scenario 3 applied data from the best matched season (2002–2003). Both scenarios included the base-case vaccinated population (all people aged 65 years or more and people aged 18–64 years with clinical risk conditions) and calculated results for the first year only.

An additional scenario considered the base-case population over a 10-year time horizon.

Results

Base-case

The estimated lifetime disease burden, QALYs, costs, and ICER per QALY gained for quadrivalent vaccination compared with trivalent vaccination are shown for the base-case from the NHS perspective in Table 10. The estimated numbers of people vaccinated and receiving PEP were greater for quadrivalent vaccination than trivalent vaccination over the lifetime horizon because of the smaller number of expected deaths with quadrivalent vaccination each year. Quadrivalent vaccination would be expected to result in fewer influenza-related events and improved health outcomes (more QALYs and life-years) compared with trivalent vaccination. Quadrivalent vaccination would be expected to incur higher costs for vaccination and PEP, reflecting the larger numbers receiving these interventions and the higher cost of quadrivalent vaccination, partially offset by savings in the costs of influenza treatment and hospitalizations. The estimated ICER over the lifetime horizon was £14,645/QALY gained for quadrivalent vaccination compared with trivalent vaccination, and this could be considered cost-effective as it is below the NICE threshold range of £20,000/QALY to £30,000/QALY48.

Table 10. Base-case results from the NHS perspective in the base-case population over a lifetime.

	Trivalent vaccine	Quadrivalent vaccine	Quadrivalent compared with trivalent
Lifetime disease burden
Number vaccinated	741,827,739	741,974,804	−147,065
Number receiving PEP	3,597,241	3,597,902	−662
Number of influenza cases	120,853,828	119,440,436	1,413,392
Number seeking medical treatment for uncomplicated influenza	22,827,844	22,549,963	277,881
Number receiving antiviral treatment	590,664	568,776	21,888
Number with influenza complications	8,982,720	8,810,724	171,996
Number of hospitalizations for complications	1,245,156	1,203,375	41,780
Number of outpatient treatments for complications	7,737,564	7,607,349	130,216
Number of influenza deaths*	554,144	534,238	19,906
Health outcomes (discounted)
QALYs	956,326,785	956,378,323	51,538
Life-years	1,051,806,914	1,051,860,149	53,234
Costs (discounted)
Vaccination	£1,645,259,190	£2,543,745,127	£898,485,937
PEP	£84,612,381	£84,621,567	£9186
Treatment of uncomplicated influenza	£418,519,794	£414,666,370	−£3,853,424
Hospitalization for complications	£4,039,485,068	£3,903,387,067	−£136,098,001
Outpatient treatment of complications	£274,876,521	£271,257,173	−£3,619,347
Outpatient nurse costs	£4,675,848	£4,517,672	−£158,176
Total reimbursable medical costs	£6,467,428,803	£7,222,194,976	£754,766,174
Incremental cost-effectiveness ratio, quadrivalent vaccine vs trivalent vaccine (discounted)
£/QALY gained	–	–	£14,645
£/life-year gained	–	–	£14,178

NHS, National Health Service; PEP, post-exposure prophylaxis; QALY, quality-adjusted life-year.

*All deaths due to complications, as it was assumed that mortality from uncomplicated influenza was zero.

Table 11 shows the estimated costs and ICER from the societal perspective over a lifetime (numbers of events and health outcomes are identical for both the societal and NHS perspectives). Quadrivalent vaccination would be expected to reduce non-reimbursable, non-medical, and indirect costs, compared with trivalent vaccination. These savings would further offset the higher vaccination and PEP costs, reducing the ICER to £13,497/QALY over the lifetime horizon.

Table 11. Cost and ICER results from the societal perspective in the base-case population over a lifetime.

	Trivalent vaccine	Quadrivalent vaccine	Quadrivalent compared with trivalent
Costs (discounted)
Total reimbursable medical costs	£6,467,428,803	£7,222,194,976	£754,766,174
Non-reimbursable medical costs	£119,268,585	£118,402,781	−£865,804
Non-medical costs	£2,510,695,700	£2,509,598,167	−£1,097,533
Indirect costs, excluding costs of premature mortality	£18,764,166,876	£18,706,961,472	−£57,205,404
Total	£27,861,559,964	£28,557,157,397	£695,597,433
Premature mortality	£7,532,899,038	£7,292,117,439	−£240,781,599
Incremental cost-effectiveness ratio, quadrivalent vaccine vs trivalent vaccine (discounted)
£/QALY gained	–	–	£13,497
£/life-year gained	–	–	£13,067

ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year.

Scenario analyses

Table 12 shows the results for the scenario analyses, all conducted from the NHS perspective. A strategy of vaccinating only the elderly population (age 65 years or more) resulted in a more favourable estimated cost-effectiveness (ICER £11,998/QALY) over the lifetime horizon compared with the base-case.

Table 12. Results from scenario analyses.

	Trivalent vaccine	Quadrivalent vaccine	Quadrivalent compared with trivalent
Scenario 1. Vaccination of elderly individuals (age 65+ years) only, NHS perspective, lifetime horizon (discounted)
Costs	£6,439,273,123	£6,988,232,025	£548,958,902
QALYs	956,223,307	956,269,060	45,753
ICER (£/QALY)	–	–	£11,998
Base-case for comparison (discounted)
Costs	£6,467,428,803	£7,222,194,976	£754,766,174
QALYs	956,326,785	956,378,323	51,538
ICER (£/QALY)	–	–	£14,645
Scenarios showing seasonal variation in matching between trivalent vaccine and circulating viruses, NHS perspective, vaccination of elderly and at-risk individuals, first-year results
Scenario 2. Worst-matching season
Costs	£285,613,027	£293,446,029	£7,833,002
QALYs	45,770,225	45,772,858	2,633
Scenario 3. Best-matching season
Costs	£260,228,182	£297,516,665	£37,288,483
QALYs	45,772,483	45,772,483	0
Base-case for comparison
Costs	£270,446,038	£301,929,691	£31,483,653
QALYs	45,771,559	45,772,078	519

ICER, incremental cost-effectiveness ratio; NHS, National Health Service; QALY, quality-adjusted life-year.

Results for scenarios exploring the effect of seasonal variation in the degree of matching between the trivalent vaccine and the circulating virus strains are also shown in Table 12, for the first year only. As these are not summed over a lifetime, an ICER cannot be calculated. The base-case results are calculated using the average match over 10 seasons, and these scenarios illustrate the amount of variability that might be expected in different seasons as the degree of match varies. If matching were equivalent to the worst-matched season, the estimated incremental costs of quadrivalent vaccination over trivalent vaccination would be less than a quarter of the value in the base-case, while the estimated QALY gain would be more than 5-times higher. Conversely, with matching equivalent to the best-matched season (with 100% match between the circulating strains and the trivalent vaccine), the quadrivalent vaccine would have higher costs than the trivalent vaccine for no additional benefit.

Table 13 presents the estimated disease burden, QALYs, costs, and ICER per QALY gained for quadrivalent vaccination compared with trivalent vaccination for the base-case population from the NHS perspective over a 10-year time horizon. The estimated ICER over the 10-year horizon was £26,019/QALY gained for quadrivalent vaccination compared with trivalent vaccination.

Table 13. Results from the NHS perspective in the base-case population over a 10-year time horizon.

	Trivalent vaccine	Quadrivalent vaccine	Quadrivalent compared with trivalent
10-year disease burden
Number vaccinated	93,709, 689	93,715,972	−6283
Number receiving PEP	667,908	667,935	−27
Number of influenza cases	27,757,780	27,573,936	183,844
Number seeking medical treatment for uncomplicated influenza	5,167,678	5,131,870	35,808
Number receiving antiviral treatment	100,656	97,257	3399
Number with influenza complications	1,809,611	1,787,187	22,425
Number of hospitalizations for complications	161,228	156,357	4871
Number of outpatient treatments for complications	1,648,384	1,630,830	17,554
Number of influenza deaths*	60,115	57,973	2142
Health outcomes (discounted)
QALYs	380,399,978	380,409,294	9316
Life-years	414,236,076	414,243,774	7698
Costs (discounted)
Vaccination	£523,231,926	£808,903,912	−£285,671,986
PEP	£31,311,522	£31,312,717	−£1195
Treatment of uncomplicated influenza	£168,778,955	£167,538,856	£1,240,098
Hospitalization for complications	£1,245,541,851	£1,204,740,306	£40,801,546
Outpatient treatment of complications	£108,284,010	£107,097,774	£1,186,236
Outpatient nurse costs	£1,445,731	£1,400,194	£45,537
Total reimbursable medical costs	£2,078,593,996	£2,320,993,759	−£242,399,763
Incremental cost-effectiveness ratio, quadrivalent vaccine versus trivalent vaccine (discounted)
£/QALY gained	–	–	£26,019
£/life-year gained	–	–	£31,489

NHS, National Health Service; PEP, post-exposure prophylaxis; QALY, quality-adjusted life-year.

*All deaths due to complications, as it was assumed that mortality from uncomplicated influenza was zero.

Sensitivity analysis

Figure 1 shows the results of the one-way sensitivity analysis. Only two parameters had a substantial effect on the ICER: circulation of influenza A; and the degree of matching between the trivalent vaccine and the circulating influenza B lineages (Figure 1A). This illustrates the importance of seasonal variation as explored in the scenario analyses. This is to be expected, since the difference between the quadrivalent and trivalent vaccines is driven by the efficacy against influenza B. The effects of the 10 next most important parameters are shown on a different scale in Figure 1(B). These parameters had only modest effects on the results. The results of sensitivity analysis on disutility and vaccine price differential (varying the trivalent vaccine price while keeping the quadrivalent vaccine price fixed) are shown for the NHS perspective by age in Table 14, and the effects of different discount rates are shown for the NHS and societal perspectives in Table 15, all over the lifetime horizon. None of these parameters had a large effect on the results.

Figure 1. One-way sensitivity analysis comparing quadrivalent with trivalent vaccination. (A) Effect of the two most important parameters. (B) Effect of the next ten most important parameters. The vertical axis indicates the incremental cost-effectiveness ratio (ICER) in the base-case = £14,645/QALY. QALY, quality-adjusted life-year; TIV, trivalent vaccination; VE, vaccine efficacy.

Table 14. Effects of disutility and vaccine price in sensitivity analysis, base-case population over a lifetime.

	ICER (£/QALY gained)
	Low value	High value
Disutility for influenza with no NI treatment (low value −0.1, high value −1.6)
Age 18–49 years	£14,803	£14,400
Age 50–64 years	£14,949	£14,187
Age 65–69 years	£15,101	£14,586
Age 70–74 years	£15,074	£14,590
Age 75–79 years	£15,021	£14,596
Age 80–84 years	£14,938	£14,607
Age 85–100 years	£14,949	£14,606
Trivalent vaccine price (low value £5.22, high value £9.05)
Age 18–49 years	£15,144	£13,563
Age 50–64 years	£15,636	£12,498
Age 65–69 years	£15,745	£12,263
Age 70–74 years	£15,684	£12,395
Age 75–79 years	£15,559	£12,665
Age 80–84 years	£15,363	£13,089
Age 85–100 years	£15,389	£13,033

ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; NI, neuraminidase inhibitor.

Table 15. Effects of discount rates in sensitivity analysis, base-case population over a lifetime.

	ICER (£/QALY gained)
Discount rate	NHS perspective	Societal perspective
0%	£11.539	£10,871
1.5%	£12,812	£11,941
5%	£16,017	£14,677

ICER, incremental cost-effectiveness ratio; NHS, National Health Service; QALY, quality-adjusted life-year.

Figure 2 shows the cost-effectiveness plane from the PSA. The black square indicates the position of the deterministic value in the base-case analysis, and the diagonal line indicates the cost-effectiveness threshold of £20,000/QALY gained. Figure 3 shows the cost-acceptability curve. A total of 68% of the simulations were below a threshold of £20,000/QALY, and 87% were below a threshold of £30,000/QALY.

Figure 2. Probabilistic sensitivity analysis comparing quadrivalent with trivalent vaccination, 1000 iterations. ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; QIV, quadrivalent vaccination; TIV, trivalent vaccination.

Figure 3. Cost-effectiveness acceptability curve comparing quadrivalent with trivalent vaccination. QALY, quality-adjusted life-year; QIV, quadrivalent vaccination; TIV, trivalent vaccination.

Discussion

The present analysis has updated an earlier economic evaluation of quadrivalent influenza vaccination in the UK14. It used new information that was not available at the time of the previous analysis to replace the limited information used earlier, considered the general adult population currently recommended for influenza vaccination in the UK according to JCVI guidance (all people aged 65 years or more and people aged 18–64 years with clinical risk conditions), and excluded the pandemic year. The analysis used a lifetime, static, multi-cohort Markov model to compare a policy of vaccinating this population using quadrivalent influenza vaccine with a policy of vaccinating this population using trivalent influenza vaccine from the perspective of the NHS in the base-case. A secondary analysis was conducted from the societal perspective. The estimated ICER from the NHS perspective over a lifetime horizon was £14,645/QALY gained. From the societal perspective, the estimated ICER was £13,497/QALY over a lifetime. A strategy of vaccinating only people aged ≥65 years had an estimated ICER of £11,998/QALY over a lifetime. Sensitivity analysis indicated that the ICER was <£20,000/QALY in 68% of simulations and <£30,000/QALY in 87% of simulations. Only two parameters had a substantial effect on the ICER: matching between the trivalent vaccine and the circulating influenza B lineages; and circulation of influenza A.

The main differences from the previous economic evaluation are in the input data used in the model. There are no differences in model structure. The first difference in input data is that the present analysis included data from two large database studies in the UK that were not available at the time of the earlier analysis. The first study used the GPRD to estimate influenza-related cases and mortality attributable to the influenza A and influenza B viruses12^,13^,22. The second study used the linked GPRD and Hospital Episode Statistics databases to estimate costs associated with influenza in the UK10^,11^,29^,30. Second, the price of the quadrivalent vaccine was not available at the time of the earlier analysis and an assumption had to be used, whereas, in the present analysis, we have been able to use the current quadrivalent vaccine list price (£9.94) and a weighted average of the trivalent vaccine prices in the same reference year (£6.43). Third, the data on circulating influenza virus strains have been updated to the most recent 10 years available (from 2002–2003 to 2012–2013, excluding the pandemic year of 2009–2010). Fourth, all unit costs used in the model have been updated to a reference year of 2012–2013, and population data have been updated to 2012.

Consistent with the earlier evaluation, this updated analysis also found that quadrivalent influenza vaccination could be expected to be cost-effective, compared with trivalent influenza vaccination, in the UK. The previous analysis estimated the ICER for quadrivalent vaccination at £5299/QALY (using an assumption about the quadrivalent vaccine price) or £27,378/QALY (using the current quadrivalent vaccine list price and the 2010 trivalent vaccine price)14. The present analysis estimated the ICER at £14,645/QALY from the NHS perspective, which falls between the two previous estimates.

There are no other similar cost-effectiveness analyses comparing quadrivalent and trivalent influenza vaccination in the UK to our knowledge. The results of the present study are consistent with findings from economic analyses in the US49–51. These publications estimated that quadrivalent influenza vaccination would be expected to reduce influenza-related hospitalizations and deaths compared with trivalent vaccination51, would be cost-saving if priced at the same level as trivalent vaccination50, and would be cost-effective compared with trivalent vaccination at 2011 vaccine prices49.

The population included in the base-case of the present analysis covers elderly people aged 65 years or more and adults aged 18–64 years with clinical risk conditions. This reflects the population for which quadrivalent or trivalent vaccination would be appropriate, as live attenuated intra-nasal influenza vaccine is currently recommended for children aged 2 to <17 years5. A limitation of the present analysis is that pregnant women and workers in health and social care, although currently recommended to receive influenza vaccination, were excluded due to a lack of suitable data. A specific model may be better suited to evaluating the potential effects of different influenza vaccines in these particular groups.

Over a 10-year time horizon, quadrivalent vaccination would be expected to avoid 183,844 influenza cases, 4871 hospitalizations, and 2142 deaths, compared with trivalent vaccination. As would be expected, these estimates are considerably smaller than the estimated values over the lifetime horizon (1,413,392 influenza cases, 41,780 hospitalizations, and 19,906 deaths). The estimated ICER for quadrivalent vaccination compared with trivalent vaccination over a 10-year horizon was £26,019/QALY gained, higher than the estimated value for the lifetime horizon and still within the NICE threshold range of £20,000/QALY to £30,000/QALY48. A limitation noted in the earlier analysis was that it did not include indirect costs resulting from time lost from work due to influenza, which could have under-estimated the potential benefits of the quadrivalent vaccine14. The present study has extended the analysis to include the societal perspective, and, as predicted, the estimated ICER from the societal perspective was slightly more favourable than the estimated ICER from the NHS perspective (£13,497/QALY compared with £14,645/QALY). The relatively modest difference between the perspectives reflects our assumption that influenza cases in elderly people would not result in any time lost from work.

Another limitation noted in the earlier evaluation was that the model did not directly incorporate seasonal variations in influenza virus strain circulation and vaccine mismatch14. These seasonal variations are allowed for in the model by using average data over 10 seasons in the base-case, and are taken into account in the sensitivity analyses. In the present study, we have explored this further by running two scenarios to evaluate the effect of two different assumptions about vaccine mismatch: one using the worst-matched season in the 10-year series; and one using the best-matched season. These scenarios indicated that the seasonal variation in the degree of mismatch could have a substantial effect on the results. Using data from the worst-matched season, the estimated QALY gain for quadrivalent vaccination over trivalent vaccination would be higher than in the base-case and the estimated incremental cost would be lower. Using data from the best-matched season, the quadrivalent vaccine would be expected to have no additional benefits over the trivalent vaccine, but would have higher costs.

Other limitations noted in the earlier analysis14 also apply to this updated analysis. As a static model it cannot fully account for herd effects. However, herd effects on influenza transmission have been noted with vaccination of children49^,52. The population considered in this analysis does not include children, because current UK guidelines recommend live attenuated intra-nasal influenza vaccine for children aged 2 to <17 years5. Thus, the impact of excluding herd effect from the present study is likely to be small. Another limitation is that the study does not attempt to include any potential declines in health status that may occur after influenza infection, particularly in elderly individuals53^,54. Limitations also remain in the availability of detailed data to populate the model by age and risk status, as previously noted14. However, if more detailed data become available from future studies, they can be incorporated into the model to produce further updates to the present analysis, just as this analysis has updated the earlier analysis by incorporating more recent data.

Conclusions

This updated economic evaluation using new information indicates that quadrivalent inactivated influenza vaccination of the elderly and clinical at-risk adult population is estimated to be a cost-effective intervention compared with trivalent inactivated vaccination in the UK. Quadrivalent vaccination would be expected to reduce influenza cases, hospitalizations and deaths compared with trivalent vaccination, and the estimated ICER from the NHS perspective over a lifetime horizon was £14,645/QALY gained.

Transparency

Declaration of funding

This study was funded by GlaxoSmithKline Biologicals SA, which was involved in all stages of the study conduct and analysis including: design and operation of the model; data inputs, analysis and interpretation; manuscript preparation, review and decision to submit for publication. GlaxoSmithKline Biologicals SA also funded all costs associated with the development and the publishing of the present manuscript. All authors had full access to the data and agreed with the submission of the publication.

Declaration of financial/other relationships

GM and MG are employees of the GSK group of companies. GM holds shares in the GSK group of companies and held stock options in the GSK group of companies until July 2013. BPN reports personal fees from the GSK group of companies during the conduct of the study and outside the submitted work. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors would like to thank Carole Nadin (Fleetwith Ltd, on behalf of GSK Vaccines) for medical writing assistance and Carole Desiron (Business and Decision Life Sciences, on behalf of GSK Vaccines) for publication co-ordination. The authors also thank Ilse Van Vlaenderen and Laure-Anne Van Bellinghen (CHESS In Health, Belgium) for their contributions to the development of the model and the original publication.

Notes

*Fluarix Tetra is a trademark of the GSK group of companies.