Uptake and impact of vaccinating primary school children against influenza: Experiences in the fourth season of the live attenuated influenza vaccination programme, England, 2016/2017

Background In the 2016/2017 influenza season, England was in its fourth season of the roll‐out of a live‐attenuated influenza vaccine (LAIV) targeted at healthy children aged two to less than 17 years. For the first time, all healthy children aged 2 to 8 years were offered LAIV at national level in 2016/2017. Since the commencement of the programme in 2013/2014, a series of geographically discrete pilot areas have been in place where quadrivalent LAIV was also offered to all school age children. In 2016/2017, these were children aged 8 to 11 years, other than those targeted by the national programme. Methods We evaluated the overall and indirect impact of vaccinating primary school age children, on the population of England, by measuring vaccine uptake levels and comparing cumulative disease incidence through various influenza surveillance schemes, in targeted and non‐targeted age groups in pilot and non‐pilot areas in 2016/2017. Results Our findings indicate that cumulative primary care influenza‐like consultations, primary and secondary care swab positivity, influenza confirmed hospitalisations and emergency department attendances in pilot areas were overall lower than those observed in non‐pilot areas; however, significant differences were not always observed in both targeted and non‐targeted age groups. Excess mortality was higher in pilot areas compared with non‐pilot areas. Conclusions These results are similar to earlier seasons of the programme indicating the importance and continuing support of vaccinating all primary school children with LAIV to reduce influenza related illness across the population, although further work is needed to understand the differences in excess mortality.

and Immunisation (JCVI) made the recommendation to implement the programme based on a range of evidence, including an analysis of the burden of influenza by age group 3 and mathematical modelling predicting the reduction of the burden of influenza and its future benefits, alongside an economic evaluation. 4 Epidemiological and mathematical modelling studies have supported the concept that vaccinating children can play a key role in reducing the burden of influenza across all ages; however, there has only been limited evidence on the impact of such programmes at national level in other countries to date. [5][6][7][8][9][10][11] The 2016/2017 influenza season was the fourth season of introducing the LAIV vaccine in the UK. England extended the childhood vaccine programme nationally to include all healthy children aged 2 to 4 years as well as to children of school years 1-3 (aged 5 to 8 years). 12 In addition to the national programme, five geographically discrete pilot areas in England vaccinated the remaining healthy children of primary school age in school years 4 to 6 (aged 8 to 11 years) in 2016/2017. The 2016/2017 season was dominated by the circulation of influenza A(H3N2), the impact of which predominately affects older persons. There were increased care home outbreaks, hospital admissions and excess mortality in this age group. 13  An established and standardised web-based portal, Immform, was used to collate and report vaccine administration data by locally commissioned data providers, to Public Health England (PHE). 15 The end of season vaccination uptake was calculated as the number of children in the target population who received at least one dose of influenza vaccine in the period from 1 September 2016 until 31 January 2017. Healthy children and those children identified to be at-risk with no contraindication against the LAIV vaccine were offered the quadrivalent LAIV vaccine. Children identified to be at-risk in whom LAIV was contraindicated, were offered the quadrivalent inactivated vaccine instead.

| Measuring impact
The 2016/2017 LAIV impact study ran from weeks 402 016 to 142 017, by which time influenza transmission in the community had returned to well below baseline levels. 13 To denote the overall impact of the childhood LAIV programme, cumulative disease incidence in non-pilot areas was compared against the cumulative disease incidence in pilot areas for a number of virological and clinical respiratory end points in primary and secondary care surveillance schemes, during the period of the study. In addition, the impact of the LAIV programme on excess all-cause mortality was also studied.
As the dominant circulating influenza A subtype during the study period 2016/2017 was influenza A(H3N2) which is known to affect the older age groups, the previously described 17+ age group was split into two further age groups. 8 Thus, the overall impact of vaccinating primary school age children was evaluated across five age groups: 5-10 years, to measure the direct impact and <5, 11-16,   Service sentinel GP network. 16 Through this network, 162 GP practices participated in non-pilot areas and 11 practices in pilot areas.
Respiratory swabbing in primary care is undertaken through GP practices within the RCGP RSC network as well as an additional sentinel swabbing network, the Specialist Microbiology Network (SMN). 17 Through both schemes, a total of 88 GP swabbing practices participated in non-pilot areas, and four GP swabbing practices participated in pilot areas. In all areas, swabs were taken from patients presenting with ILI, regardless of the patient's vaccination status. The weekly number of laboratory confirmed influenza hospital admissions and ICU/HDU admissions, collated through the two USISS (sentinel and mandatory) schemes were used to calculate confirmed influenza hospital and ICU/HDU admission rates by the specified age groups and pilot and non-pilot areas, using estimated hospital catchment population per 100 000 population as the denominator. 19 The Emergency Department Syndromic Surveillance System (EDSSS), as previously described, 20 monitored the proportion of all weekly respiratory related emergency department (ED) attendances against all ED attendances with a diagnosis by the specified age group and by pilot (one ED) and non-pilot (25 EDs) areas.

| Secondary care
Respiratory swabbing in secondary care is monitored through the Respiratory DataMart system (RDMS), as over 90% of samples from this scheme are collated from patients in secondary care settings. 21 Overall swab positivity for all influenza reversetranscription-polymerase chain reaction (RT-PCR) respiratory swab results in a network of 14 PHE and NHS laboratories in England was compared by age group and by pilot area. Each patient's postcode of residence was used to assign patient samples to a pilot or nonpilot area.

| Excess all-cause mortality
Excess all-age all-cause mortality and all-age respiratory deaths was the primary cause of death. 23

| Statistical methods
The methods used to evaluate and measure the impact of the LAIV vaccine on the various surveillance schemes were the same as those used in previous seasons. 7,8 Surveillance schemes where the cumulative disease incidence To measure the impact of the LAIV programme, the non-pilot areas were set as reference, and odds ratios and 95% CI were calculated by age-group and surveillance scheme. For each scheme, adjusting for clustering at the reporting unit level (e.g., GP practice, hospital or laboratory), data were converted to binomial individual level and random effects logistic regression undertaken.
For excess mortality monitoring, cumulative excess mortality rates were calculated by summing the difference between observed and expected weekly deaths over the time of the study by pilot area and using the resident population respective to non-pilot or pilot areas.

| Vaccine programme impact
From weeks 402 016 to 142 017, cumulative GP ILI consultation rates and swab positivity in primary and secondary (RDMS) care were lower in pilot areas compared with rates in non-pilot areas across all age groups.
Such differences were however less marked in secondary care schemes in particular ED respiratory attendances (EDSSS) and ICU/HDU flu confirmed rates (USISS) in the older age groups  and 65+ years) (Figures 2 and 3). Examination of pre-vaccination data for those schemes for which data were available (RCGP RSC ILI and influenza confirmed hospitalisations and ICU admissions), provided a mixed pattern.
Cumulative ICU/HDU admission and hospitalisation rates for influenza were similar in primary school pilot compared to non-pilot areas for the two seasons prior to the start of the childhood vaccination programme, whereas for the three seasons since, they have been consistently lower. The differences were less marked for GP ILI consultation rates, where pre-introduction ILI rates were generally lower in pilot compared to non-pilot areas, which may reflect underlying differences between the pilot and non-pilot areas ( Figure 5).

| Adjusted impact for primary and secondary care schemes
For primary care schemes, significant reductions were observed in cumulative GP ILI consultation rates in adults in pilot areas compared to non-pilot areas when adjusting for clustering, as a result of vaccinating primary school aged children (Table 2). Non-significant reductions were also noted in children of primary school age and in under 5 year old children for GP ILI consultation rates in pilot compared to non-pilot. Non-significant reductions were also observed in cumulative swab positivity in both adults and children ( ( Table 3). Non-significant reductions were also noted in all but one (11 to 16 year olds) age groups for influenza confirmed hospitalisations (USISS sentinel) and RDMS positivity (Table 3). A significantly higher rate for influenza confirmed hospitalisations in pilot areas for 11 to 16 year olds were noted through the USISS sentinel scheme (Table 3).

| CONCLUSION
This study evaluates the uptake and impact of the childhood LAIV Our finding that excess all-cause mortality was significantly higher in pilot areas is one that has been noted in previous seasons and can be likely explained by the pre-existing higher all-cause excess seen in the pilot areas prior to the introduction of the LAIV programme, which may reflect differences in the underlying health and sociodemographic profile of the populations in pilot and non-pilot areas. 7,8 The higher excess respiratory mortality observed in the pilot areas in our study was surprising as the pattern was reversed in previous postprogramme seasons where higher excess respiratory mortality was noted in non-pilot areas. 6,8 Absence of reductions in pneumonia and influenza mortality associated with a school children's vaccination programme has also been reported elsewhere. 30 This observation may be due to the lack of study power or to the increasingly narrowing gap between the pilot and non-pilot areas described further below and warrants further investigation.
A number of strengths have been highlighted from this study.
First, this study uses data collected from a wide range of wellestablished surveillance systems which cover healthcare service utilisation across the disease spectrum of influenza. Second, the methods used in this study to assess the uptake and impact of the childhood programme have been developed over the past three seasons which has enabled us to confidently assess its findings. Third, vaccine uptake is measured at population level which allows for direct comparisons to previous published studies. [6][7][8] There are some potential limitations to this study, including potential differential reporting. New GP practices and hospitals are recruited each season in pilot and non-pilot areas which may contribute to higher differential reporting and in turn introduce less sensitive case detection amongst these new entities compared to long-standing participating practices. There was only one emergency department site in the pilot areas compared to 25 in the non-pilots areas, although this is taken into account in the model. As the national vaccination programme continues to roll out to children in other school years and the gap between pilot and non-pilot areas decreases, the ability to carry out such comparisons is diminishing, which is evident when looking at the historical comparisons.
In conclusion, this study is the fourth of its kind in a series of evaluations of the LAIV programme to find continuing positive outcomes in support of the roll out of the national childhood LAIV programme.
Other approaches to estimate the population impact of the programme may need to be considered as the narrowing differences between uptake between pilot and non-pilot areas means that the methods used in this study may not be suitable in future seasons. Lack of impact of the programme on excess all-cause mortality also warrants further investigation.