Vaccine effectiveness of live attenuated and trivalent inactivated influenza vaccination in 2010/11 to 2015/16: the SIVE II record linkage study

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/34/14. The contractual start date was in October 2014. The draft report began editorial review in May 2018 and was accepted for publication in November 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Simpson et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2020 Queen’s Printer and Controller of HMSO

Description of the health problem

Globally, it is estimated that seasonal influenza is responsible for 5 million cases of severe illness and 500,000 deaths per year, with, for example, an estimated cost to the USA of US$87B per annum. 1–3 There are also 90 million new cases of influenza and 1 million cases of influenza-associated severe acute lower respiratory infection among children. 4 National influenza vaccination programmes, delivered by primary care in the community, are important to reduce influenza-related illness, and hence the considerable investment in this approach. Previously, these programmes targeted older people (i.e. those aged ≥ 65 years) and people with chronic disease (e.g. asthma) who are susceptible to serious illness from influenza. Children are also thought to be important in the transmission of influenza to the populations at risk of serious complications from influenza, and diminished circulation of the virus has been predicted to improve herd immunity. 5 Using evidence generated from epidemiological modelling,6 and following advice from the Joint Committee on Vaccination and Immunisation, from September 2013 the seasonal influenza vaccination programme has been extended. 7 In addition to the seasonal trivalent influenza vaccine (TIV), the live attenuated influenza vaccine (LAIV) is offered to all children aged 2–11 years (except children clinically severely immunocompromised owing to conditions or immunosupressive therapy or oral steroids, and children with severe asthma) by primary care clinicians in general practice (GP) and in schools in Scotland.

Existing evidence

The evidence from clinical trials of the benefits of LAIV are largely confined to healthy children aged < 7 years (mostly for children aged < 3 years). 8,9 Efforts to estimate seasonal TIV effectiveness have been largely confined to younger healthier adults (e.g. with no randomised controlled trials showing efficacy of TIV in adults aged ≥ 65 years). 9,10 Recent observational studies have attempted to estimate the vaccine effectiveness in preventing influenza-related illness in GP patients. 3,11 Further studies have examined vaccine effectiveness with hospitalisation or death; however, these studies have suffered from bias when using non-specific outcomes,8 or have been underpowered when using more specific end points (e.g. laboratory-confirmed influenza), in particular for subgroups being targeted for vaccination (e.g. older people aged ≥ 65 years, people with at-risk disease such as asthma and pregnant women). 12 Cohort studies (with nested case–control studies) or data linkage-derived estimates of vaccine effectiveness have been undertaken, with measures taken to overcome many of the confounding issues that otherwise have limited estimations of effectiveness. 13–15 There is also a need to add to the growing body of evidence with regard to the safety of these vaccines. 16 Given the ongoing controversy regarding vaccine effectiveness and, in particular, in relation to at-risk groups (e.g. those with asthma),7 there is further need for information to help evaluate new seasonal vaccine strategies.

Study design and population

Vaccine uptake is reported from a cross-sectional survey of all six influenza seasons. The test-negative design (TND) was used to measure vaccine effectiveness for the RT-PCR outcomes and a cohort study design was used for non-specific clinical outcomes (e.g. hospitalisation or death from influenza or pneumonia). 17

All practices in Scotland (n = 998) were invited to participate and 230 (self-selected) practices were recruited These represented 29 (65.9%) of 44 of the Community Health Partnership (CHP) areas in Scotland. Data were extracted on 1.25 million patients in Scotland into our study. Each patient contributed person-time to each influenza season while alive and fully registered with a participating GP (i.e. a person was included in the study if they were on a participating practice’s list of patients, including those who may have died or deregistered during the study period).

Three basic data sets for analysis were created:

1. all patients with a RT-PCR test

2. patients by age group (e.g. 2–4, 5–11, 12–17, 18–64 and ≥ 65 years)

3. patients at risk of serious influenza-like illnesses (ILIs) (e.g. asthma).

Databases

Data fields extracted from the following databases (Figure 1) were linked deterministically using the Community Health Index number; a unique identifier used by the NHS for the Scottish population. 3 The database linkage and analysis was carried out within the NHS National Services Scotland (NSS) TRE by the electronic Data Research and Innovation Service (eDRIS).

FIGURE 1.

Flow diagram for the Seasonal Influenza Vaccination Effectiveness II project. ECOSS, Electronic Communication of Surveillance in Scotland; PIPER, Pandemic Influenza Primary Care Reporting; SMR01, Scottish Morbidity Record 01.

Study period

Data from 1 September 2010 to 31 August 2016 were used. These allowed an analysis of six influenza seasons (from 2010/11 to 2015/16). Each patient contributed person-time to each influenza season while alive and registered with a participating GP. For the non-pandemic seasons, each year (i.e. 1 September to 31 August) was divided in to four periods (Figure 2). The influenza season was defined for each year using national influenza surveillance data. 26 The other periods include a pre-influenza season (starting on 1 September), a post-influenza period (which ends on 31 May each year) and a ‘non-influenza’ period (from 1 June to 31 August) (see Figure 2). Because there was a phased roll-out arrangement for influenza vaccination among children, children aged 2–3 years and primary school-aged children (for which pilots were taking place) in season 2013/14, and all preschool children aged 2–4 years and all primary school-aged children (i.e. all children aged 2–11 years) in seasons 2014/15 and 2015/16 were analysed.

FIGURE 2.

Relationship of the first influenza season (2010–11) to pre-, post- and non-influenza season periods. Reproduced from Simpson et al. 3 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

Reproduced from Simpson et al. 3 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text

Baseline characteristics for each patient was determined on 1 September each year. The earliest date of influenza vaccination varied for each influenza season, but always took place after 1 September.

Exposure definition

For people in at-risk groups, influenza vaccinations (TIV and LAIV for preschool children aged ≥ 2 years) are free and administered by general practitioners (Table 1). 28 Data on influenza vaccination carried out in GP (including Community Health Index number and date of administration) are recorded to enable reimbursement. Information on individuals receiving LAIV in schools is collated in the Child Health Services Programme/Scottish Immunisation & Recall Service database and was extracted for this analysis. 21 Vaccination was used to define exposure status when it was given at a time point between 1 September and the end of the influenza season (see Figure 2). An individual was defined as vaccinated 14 days after the seasonal influenza vaccine had been administered. 29 The time period from the first day of the influenza season to day 14 post vaccination was defined as ‘unexposed’ and the period from day 14 post vaccination until the end of the influenza season was defined as ‘exposed’. Therefore, those people vaccinated between the start of the pre-influenza period up until 14 days before the influenza season were defined as ‘exposed’ for the duration of the influenza season. 30

Study outcomes

Non-specific clinical outcomes

To determine the effect of vaccination status on influenza-related primary care consultations, hospital admissions and deaths, secondary analyses were undertaken using non-specific clinical outcomes derived from primary and secondary care. Data on ILI consultation were derived from the GP database. Data on hospitalisation and cause of death from influenza or pneumonia were derived from SMR01.

Confounding factors

Key characteristics of each identified patient characteristics present in each season of the cohort were included as confounders in the analyses. These were defined in each year on the first day of the pre-influenza season (i.e. on 1 September).

Sensitivity analyses

Simonsen et al. ’s40 framework was used to consider the role of confounding.

Unmeasured confounding

The robustness of the results were assessed by modelling the effect of an unmeasured confounder, such as frailty, on the vaccine effectiveness estimates in a sensitivity analysis; an approach adopted to help explain the role of unknown confounding in observational analyses. 40 Three factors were varied: (1) the prevalence of the confounder in the vaccinated population, (2) its prevalence in the unvaccinated population and (3) the increased risk of the outcome attributable to the confounder. 41

Instrumental variable analyses

The use of influenza vaccine coverage by geographical area has been found to be a strong and valid instrumental variable, which can be used to account for confounding. 3,42 Rather than comparing patients with respect to whether or not they received influenza vaccination, this instrumental variable behaves like natural randomisation of patients to regional vaccination groups that differ in their likelihood of receiving influenza vaccination. The NSS TRE is an important development in this respect, and permissions were received to extract granular postcode/geocoding data required to test the validity of this instrumental variable analysis in the Seasonal Influenza Vaccination Effectiveness II (SIVE II) database. Therefore, the use of vaccination uptake in geographically distinct CHP areas or other suitable health board areas as a suitable instrumental variable was explored. Because there are only 14 health boards and > 30 CHPs, the latter was used as the geographical area. To be valid, this instrumental variable needed to be related to exposure status (i.e. vaccination status) and not have an independent effect on outcome other than by ways mediated through the exposure. 43 Furthermore, the instrumental variable should not be related to any variables that confound the relationship between exposure and outcome. If an association with confounders is demonstrated, it is assumed that the instrumental variable is associated with unmeasured confounders and is therefore not valid. If the instrumental variable fulfils these criteria, it can be used in analyses to produce unbiased estimates of vaccine effectiveness by accounting for unmeasured confounding.

Adverse events associated with vaccination

A self-controlled study design was used to estimate the risk of adverse events associated with influenza vaccination. 44 The assumption underlying this design is that in the situation in which the adverse event is related to vaccination, the occurrence of an adverse event in the period after vaccination is greater than periods in the same patient that are temporally unrelated to vaccination. 45 This method has the advantage of controlling for all fixed individual-level confounders as comparisons are within the same individual, rather than between vaccinated and unvaccinated populations. The time period at risk for an adverse event (risk interval) and time period not at risk (control interval) were determined separately for each outcome. 46 For virtually all adverse events, the at-risk period was 14 days following receipt of the vaccination and the pre-risk period was the 90-day period prior to the 14 days before the vaccine (i.e. days 104 to 15 before receipt of the vaccine). The post-risk period was also 90 days and began on day 15 following vaccination. The main comparisons are with the rate of adverse events in the risk period compared with (1) the pre-risk period and (2) the post-risk period.

The self-controlled case series design uses data from only those with the adverse event and who are vaccinated. For some of the adverse events, there is the possibility of a temporal change in the risk over the ≥ 200 days of observation periods. To take this into account, data were also included from unvaccinated individuals who experienced the adverse event. These individuals were assigned a pseudo date of vaccination based on the median date of vaccination for the age and season. Interaction tests were then used to compare the rates of adverse events in (1) the risk period compared with the pre-risk period, and (2) the risk period compared with the post-risk period, among vaccinated and unvaccinated individuals. If there is evidence of a significant interaction with a higher risk ratio among vaccinated individuals, then this suggests that there is a potential adverse event associated with vaccination. Because there were a large number of adverse events being tested, the Benjamini–Hochberg false discovery rate was used to adjust for multiple testing.

The analysis of the self-controlled case series was undertaken using a stratified analysis in which the comparisons of the different risk periods were made within individuals. This was achieved by using matched logistic regression, with an offset for the length of the risk period. To avoid biases, the risk periods were not censored at death or when an individual left a practice.

Statistical analysis

A 5% significance level was used for hypothesis tests for the primary outcome. All p-values were two sided. All analyses were undertaken in R, version 3.2.3 (The R Foundation for Statistical Computing, Vienna, Austria). Ninety-five per cent confidence intervals (CIs) were also calculated. Logistic regression was used to investigate vaccine uptake. A generalised additive model was used to estimate the vaccine effect within the TND. Splines were used to model the effect of age and time within the season. A time-dependent Cox model was used to estimate the effect of vaccination on the consultation, hospitalisation and mortality end points. The receipt of vaccination was a time-dependent covariate. Summary statistics for this analysis are based on the person-time at risk.

Sample size

A final total sample size of up to 1.25 million people, from 230 practices, was expected. Using data from the Pandemic Influenza Primary Care Reporting (PIPER) 2014/15 study cohort, which had 263,000 individuals (of whom 16% were aged 2–17 years and 18% were aged ≥ 65 years),49 vaccine uptake among children aged 3–12 years was 60% and vaccine uptake among people aged ≥ 65 years was 70%. Linked to this PIPER cohort from all virology tests in Scotland were, overall, 1745 RT-PCR tests, comprising 331 RT-PCR tests among 2- to 17-year-olds and 366 RT-PCR tests among people aged ≥ 65 years. This gave a multiplier ratio of around 5 : 1 from the PIPER cohort to the SIVE II cohort, and this was used to estimate the number of RT-PCR tests expected each year. This study expected 1800 RT-PCR tests per year among people aged ≥ 65 years and 1650 laboratory tests per year among children aged 2–17 years. The study expected 630 (i.e. 1745/12 × 5) asthma patients swabbed per year, because approximately 12% of the population was treated for asthma.

Using data generated from the Seasonal Influenza Vaccine Effectiveness in the community (SIVE) project, the study estimated a vaccination rate of 60% among children targeted for receipt of LAIV and a swab positivity rate of 20% among unvaccinated children. 13 This gave 90% power to detect a vaccine effectiveness of 31% based on 1650 swabs in one season. Pooling data over two seasons gave an estimated 3300 swabs in children eligible for vaccination and a 90% power to detect a vaccine effectiveness of 22%.

For those people aged ≥ 65 years targeted for receipt of TIV, for whom there was a vaccination rate of 70% and a swab positivity rate of 10%, among the unvaccinated individuals the study anticipated an 80% power to detect a vaccine effectiveness of 39%. It was estimated that there would be a need for 1800 swabs each year in the later years. During the peak influenza activity, when swab positivity might have increased to 20%, there is a 90% power to detect a vaccine effectiveness of 31%. Approximately 1 in 12 of the population is treated for asthma and it is anticipated that 1260 swabs are needed among patients with asthma in the final two seasons. 50 Assuming that 40% are vaccinated and that the swab positivity is around 15% gives 80% power for a vaccine effectiveness of 35%. These powers do not take into account design effects for the clustering of patients within GPs. Analyses of the historic PIPER cohorts has revealed a design effect of < 7% and this serves to increase the detectable vaccine effectiveness by about 2 percentage points.

Vaccine uptake

The uptake of LAIV among the children registered with the 230 practices taking part in this project increased over the study period (Table 2), with nearly two-thirds of primary school-aged children vaccinated by 2015/16. TIV uptake among at-risk patients aged 12–65 years was highest in those aged 55–64 years, with nearly two-thirds being vaccinated. Among those aged ≥ 65 years, vaccine uptake was highest in the 75–84 years age group.

In all but two age groups (i.e. 2–4 and ≥ 85 years), more females than males received the influenza vaccine. High levels of vaccine uptake were found among the least deprived children, people living in remote and rural areas of Scotland, people with chronic diseases, people with an emergency hospital admission in the past year (specifically for preschool-aged children and 11- to 17-year-olds), and people with a prior ARI GP consultation in the past year (Table 3). The least deprived at-risk 18- to 55-year-olds were less likely to receive the vaccine. For all those people aged ≥ 65 years and eligible to receive TIV, levels of uptake were similar to levels among children, with higher levels of uptake among the least deprived groups, those people living in rural and remote areas, those people with comorbidities and a prior ARI GP consultation. Those people with an emergency hospitalisation in the last year were less likely to receive the vaccine.

Laboratory-confirmed influenza

Clinical outcomes

After adjustment for confounding, including age in the model, significant negative vaccine effectiveness was found for GP outcomes, ARI and ILI for people aged ≥ 65 years (Table 15). Significant positive vaccine effectiveness was found for emergency hospitalisation or death from influenza and pneumonia (see Table 15). Rates and relative risks can be found in Table 16.

In the sensitivity analysis, and according to Simonsen et al. ’s framework,40 vaccine effectiveness during the pre- and post-influenza season was similar to that during peak influenza season (Table 17). Little variation was found between the seasons with different poorly or well-matched vaccines, severity, or by specificity of outcome (all-cause mortality had higher vaccine effectiveness than hospitalisation for influenza and pneumonia). 37

Modelling an unmeasured confounder

Estimates from published data were used to model the effect of inadequately measured confounders on vaccine effectiveness in the cohorts for emergency hospitalisations due to pneumonia or influenza. The baseline vaccine effectiveness is that derived from pooling the estimates for the six seasons in Table 12. The pooled estimate, using a random-effects meta-analysis, is 22.9% (95% CI 19.8% to 25.8%).

It was found that a scenario of 15% prevalence for unaccounted frail individuals in the unvaccinated population, with frailty causing a fourfold increase in the risk of emergency hospitalisation and with a 5% presence of frailty among the vaccinated population, resulted in the vaccine no longer being effective (vaccine effectiveness 2.75, 95% CI –1.11 to 6.49). 9 A twofold increase in risk of hospitalisation would have relatively little impact, although it does reduce the magnitude of the vaccine effect estimate. By varying the prevalence of frailty and its risk on outcome, the change in vaccine effectiveness estimates for a number of scenarios (Table 19) were modelled. 27 The study also found that this model can be displayed effectively in graphical form (Figures 3 and 4). In Figures 4 and 5, the vaccine effectiveness is estimated given levels of the unmeasured confounder in the vaccinated population (x-axis) and unvaccinated population (each of the lines representing a different prevalence of the unmeasured confounder in the vaccinated population). At points where the prevalence of the unmeasured confounder is the same in the unvaccinated individuals and vaccinated individuals, then the baseline result is achieved, as there is no difference in the level of the unmeasured variable in the two groups.

TABLE 19 - Sensitivity analysis to quantify the effects of a hypothetical unmeasured confounder on the cohort analysis results

Increase in the risk of outcome on account of the confounder	Prevalence of confounder (%)		Emergency hospitalisation for pneumonia or influenza, adjusted vaccine effectiveness (95% CI)
	In unvaccinated population	In vaccinated population
–	0	0	22.88 (19.81 to 25.84)
Doubled	5	0	19.03 (15.80 to 22.13)
Doubled	10	5	19.21 (15.99 to 22.31)
Doubled	215	5	15.54 (12.17 to 18.77)
Quadrupled	5	0	11.31 (7.78 to 14.71)
Quadrupled	10	5	12.82 (–9.35 to 16.16)
Quadrupled	15	5	2.75 (–1.11 to 6.49)

FIGURE 3.

Unmeasured residual confounding and vaccine effectiveness (doubling of risk in emergency hospitalisations for influenza and pneumonia).

FIGURE 4.

Unmeasured residual confounding and vaccine effectiveness (quadrupling of risk in emergency hospitalisations for influenza and pneumonia).

FIGURE 5.

General practice consultations: adverse events in children with LAIV interactions. Note that dots to the left of the vertical dashed line at 1 correspond to adverse events where the RR comparing the risk period with the pre- or post-risk period is lower among those children vaccinated than in those children unvaccinated. Dots to the right of the vertical dashed line at 1, and where all of the horizontal line corresponding to the 95% CI is above 1, indicate adverse events where there is a potential association with receipt of the vaccine. Some of the estimates are based on very small numbers and the right-hand side of the plot is truncated at 10.

Vaccine safety

The LAIV was found to be safe in children, with no adverse reactions found (Figures 5 and 6). The number of adverse events recorded for children was very small and the CIs are wide. In most cases, fewer adverse events were reported among children who were vaccinated than in those who were unvaccinated, suggesting that relatively healthy children are the ones who received the LAIV.

FIGURE 6.

Hospitalisation: adverse events in children with LAIV interactions.

The TIV was found to be safe, with no association with adverse events (Figures 7 and 8). The exception to this finding was a statistically significant increase in allergic reaction (non-anaphylaxis) coded in the GP database; here, the TIV incidence RR (without the interaction term) was 1.33 (96% CI 1.13 to 1.56; p < 0.001). The Benjamini–Hochberg-adjusted p-value for this interaction test was 0.009 for GP consultations. This is the only adverse event for which there was a suggestion that the TIV was associated with a temporary increase in consultations following vaccination. In a majority of potential adverse events, those people who were vaccinated were at lower risk of GP consultation post vaccination than those people who were unvaccinated.

FIGURE 7.

General practice consultations: adverse events in adults with TIV interactions.

FIGURE 8.

Hospitalisation: adverse events in adults with TIV interactions.

Study limitations

The estimates of vaccine uptake (particularly for those people aged ≥ 65 years) are lower relative to those reported by HPS, which are based on vaccine claim counts by all GPs in Scotland. SIVE II study practices may have been located in areas with a low level of vaccine uptake (e.g. more deprived areas). However, in this study it was not possible to check this as the study did not have practice identifiers. In similar, previous studies with smaller numbers of practices,30,47,65,66 but using similar data sets, vaccine uptake among those people aged ≥ 65 years was closer to national estimates of 75%. If SIVE II practices have actual lower levels of vaccine uptake, then this study’s estimates of vaccine effectiveness will be unbiased. If this study has missing data on vaccination, then the effect of the misclassification of some vaccinated individuals as unvaccinated will have moved the estimates of vaccine effectiveness towards zero, and so the vaccine effectiveness reported in this study may be too low. However, the estimates of vaccine effectiveness are unlikely to be substantially lower, as the pooled estimates for the TND for any influenza among those people aged ≥ 65 years over the six seasons is 17.9% (95% CI 5.2% to 28.7%), whereas from the cohort for emergency hospitalisation for pneumonia or influenza the vaccine effectiveness is 22.9% (95% CI 19.8% to 25.8%). Therefore, there was little scope for big increases in the estimated vaccine effectiveness, assuming that all of the misclassified unvaccinated individuals did not experience an event. The study was unable to calculate whether or not there was any indirect effect of the LAIV. The indirect impact of LAIV, defined as reduction in cumulative disease incidence over the same period between pilot and non-pilot areas in non-target age groups (< 4 years of age and > 11 years of age), was measured in England; however, no statistically significant indirect protection to other age groups was found. 67 The work of other groups exploring this issue will continue to be monitored. Owing to the observational nature of this study, which uses routinely collected data, residual confounding may be present or unaccounted for.

A no-cost extension to this project was granted: this was due to the complexity of extracting and linking data from a number of disparate data sets (challenging in terms of putting the necessary governance permissions in place and technical processes required for data extraction and linkage). Future projects requiring data linkage should build adequate time to address governance and technical challenges.

The study recruited 230 practices, which was fewer than the 500 practices (of a total of 998 Scottish GPs) originally stated in the grant application. Substantial resource was committed to the process of recruitment (e.g. presentation to GP national users groups, GP newsletter advertisements, e-mail and telephone recruitment, and several reminders). Despite the lower than expected final number of GPs participating in this project, to date this is one of the largest individual patient-level primary care recruitment and data extractions to take place in Scotland (with one-fifth of practices participating). Only a small number of practices responded to turn down the offer to participate (almost all with no explanation). With the introduction of national strategies to use primary care data for research purposes [e.g. the Scottish Primary Care Information Resource; URL: www.spire.scot (accessed 1 October 2018)], it is hoped that ambitious recruitment targets, such as in this study, will be seen as more feasible.

Recommendations for further research

The monitoring of the LAIV programme with enhanced Sentinel swabbing of preschool- and primary school-aged children should continue. Future studies should be planned to further monitor the seasonal influenza vaccine in at-risk patients and such studies will continue to require expertise in data linkage and advanced analyses of these complex data sets. Replication of vaccine effectiveness and safety in LAIV and TIV in other countries (that have these influenza vaccine programmes) is also needed to confirm this study’s results. A future study will be required to monitor the safety and effectiveness of the adjuvanted TIV available to older people for the 2018/19 season. 61 Any future evaluation work should contain a health economics analysis to help understand at what level of vaccine effectiveness it becomes cost-effective to recommend vaccination or continue to vaccinate.

Conclusions

Using these data, it was found that LAIV was safe and effective in decreasing RT-PCR-confirmed influenza in children. TIV was safe and significantly effective (in most seasons) for those people aged 18–64 years with positive vaccine effectiveness and in most seasons for those people aged ≥ 65 years (although this was statistically significant in only one season). Higher vaccine effectiveness was found among younger adults with asthma. Few countries’ health systems allow for the integrated and accessible data recording that made this study possible and which made it feasible to centrally collate almost all hospitalisations and deaths attributed to influenza, allowing for completeness of reporting. LAIV was found to be safe, with vaccine effectiveness comparable to that found for TIV. TIV immunisation for at-risk adults in primary health-care settings is effective.

Contributions of authors

Colin R Simpson (Professor of Population Health) was the principal investigator and led the writing of this report.

Nazir I Lone (Senior Clinical Fellow, Intensive Care) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Kim Kavanagh (Senior Lecturer) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Tanya Englishby (Research Fellow) carried out the analyses and helped to design the study and write the report.

Chris Robertson (Professor of Statistics) carried out the analyses and helped to design the study and write the report.

Jim McMenamin (Consultant Epidemiologist) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Beatrix von Wissman (Specialist Trainee in Public Health) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Eleftheria Vasileiou (PhD student) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Christopher C Butler (Professor of Primary Care) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Lewis D Ritchie (Professor of Primary Care) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Rory Gunson (Consultant Clinical Scientist) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Jürgen Schwarze (Edward Clark Chair of Child Life and Health) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Aziz Sheikh (Professor of Primary Care Research and Development) helped design the study, draft and write the project protocol and revise the report for important intellectual content.

Publications

Simpson CR, Lone NI, Kavanagh K, Robertson C, McMenamin J, von Wissman B, et al. Evaluating the effectiveness, impact and safety of live attenuated and seasonal inactivated influenza vaccination: protocol for the Seasonal Influenza Vaccination Effectiveness II (SIVE II) study. BMJ Open 2017;7:e014200.

Vasileiou E, Sheikh A, Butler CC, Robertson C, Kavanagh K, Englishby T, et al. Seasonal influenza vaccine effectiveness in people with asthma: a national test-negative design case-control study. Clin Infect Dis 2020;71:e94–e104.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.