Evaluation of influenza A/H3N2 epidemiology in England during the 2025-26 season

1 Evaluation of influenza A/H3N2 epidemiology in England during the 1 2025-26 season 2 James A Hay*,1, Punya Alahakoon†,1, Alexander Greenshields-Watson†,1, Michelle Kendall1, Mahan 3 Ghafari1,2, Chris Wymant1, Robert Hinch1, Luca Ferretti1, Jasmina Panovska-Griffiths1,3,4, Christophe 4 Fraser1 5 6 1. Pandemic Sciences Institute, Nuffield Department of Medicine, University of Oxford, Oxford, UK 7 2. Department of Biology, University of Oxford, Oxford, UK 8 3. The Queen's College, University of Oxford, Oxford, UK 9 4. UK Health Security Agency, London, UK 10 11 * Correspondence to james[email protected].uk 12 † Contributed equally 13 Abstract 14 England has experienced a high growth rate of infections caused by the influenza A/H3N2 K clade. 15 Antigenic change from the previously dominant clade, a rapid selective sweep evident in genomic 16 data, and an unusually early start to the season have raised concerns about the potential severity of 17 this year’s influenza season. We analysed publicly available surveillance data going back to the 18 2011/12 season and found that the peak growth rate of influenza infections and the peak time-varying 19 reproduction number has been largely consistent with previous severe seasons. Scenario analyses 20 using an age-stratified compartmental model compared to the previous A/H3N2 season in 2022/23 21 suggest that current trends are compatible with moderate levels of immune escape in all ages, or 22 slightly greater immune escape in children, or a 10-20% higher R0, or an earlier seed date with no 23 change in virus fitness or immune escape. Substantial immune escape appears to be unlikely given 24 current epidemiological trends. In almost all scenarios, an earlier and faster epidemic growth rate 25 leads to earlier depletion of susceptibles with a dampening effect due to the half term school holiday. 26 To support understanding and exploration of model outputs, an interactive visualisation tool was 27 developed and made available online: https://hay-idd.shinyapps.io/ModelFluUk-H3N2/. This rapid 28 analysis is intended to support situational awareness. It provides quantitative comparisons of early 29 epidemic growth rates with previous seasons and qualitative insights into plausible epidemic 30 dynamics. 31 2 Data availability: All code and data required to reproduce the analyses are available at 32 https://github.com/hay-idd/influenza_H3N2_k_clade 33 Conflict of interest: JAH, PA, AGW, LF, RH, MK, JPG, CW, MG and CF declare no competing 34 interests. 35 Funding: JAH and PA are supported by a Wellcome Trust Early Career Award (grant 36 225001/Z/22/Z). AGW is supported by a Wellcome Trust directed call (309152/Z/24/Z). MG is 37 supported by a Wellcome Trust Early Career Award (grant 309205/Z/24/Z). LF acknowledges 38 support from a PSI Career Development Fellowship, the European REA, Marie Skłodowska-Curie 39 Actions (grant agreement no. 101131463 SIMBAD), and UK Research and Innovation (UKRI) 40 under the UK government’s Horizon Europe funding guarantee (grant number EP/Y037375/1). 41 JPG's work was supported by funding from the UK Health Security Agency and the UK Department 42 of Health and Social Care (DHSC). The views expressed in this article are those of the authors and 43 not necessarily those of the UK Health Security Agency or the UK Department of Health and Social 44 Care. RH and CF were supported by research grants from CEPI. 45 Ethics statement: Ethics approval was not required for this study as all data received was 46 obtained through routine surveillance using de-identified data. 47 Use of AI: AI tools including ChatGPT and Copilot were used to assist with literature review, 48 sourcing data and developing code. 49 Acknowledgements: We thank Steven Riley, Simon Cauchemez, Ben Cowling, Oliver Eales, 50 Freya Shearer, and Juliette Paireau for helpful discussion and for providing global context for the 51 analyses. We also thank Richard Neher for helpful discussions around growth rate advantages 52 estimated from genomic data. 53 3 Introduction 54 The 2025/26 influenza season in the northern hemisphere has been dominated by an antigenically 55 drifted clade of A/H3N2 viruses. This K clade is descended from the J.2 clade on which the vaccine 56 strain selection was based and has swept to dominance from a frequency of <1% on 2025-07-02 to 57 94% in Europe as of 2025-12-13 (1). This suggests a substantial fitness advantage over other J.2 58 viruses (2). The combination of rapid growth, multiple antigenic substitutions in the haemagglutinin 59 (HA) protein, antigenic mismatch with the vaccine strain, typically higher morbidity and mortality 60 amongst the elderly during A/H3N2 seasons (3), and an unusually early start to the season have 61 raised concerns over the potential for a severe season (4–6). Furthermore, the 2025 flu season in 62 Australia was one of the worst on record (7), and Japan suffered from an early epidemic leading to 63 school closures and increased hospitalisations (8). 64 The interaction of season timing, antigenic drift, other subtype dynamics, vaccine efficacy and 65 coverage, non-HA-mediated immunity, immune waning from previous seasons, climate factors, and 66 contact pattern changes around school holidays is complex and leads to highly varied cumulative 67 and peak seasonal burden (9). The potential outlook of the 2025/26 influenza season is therefore 68 uncertain, though historical seasons with early onset have often been severe (3). 69 Real-time epidemiological assessment in England is based on weekly influenza surveillance data 70 from the Second-Generation Surveillance System (SGSS) (10), which compiles respiratory virus 71 testing data from multiple sources: the Royal College of General Practitioners Research & 72 Surveillance Centre (primary care), the Respiratory DataMart system (hospital testing) and other 73 hospital testing. The main surveillance indicator in the UK Health Security Agency (UKHSA) weekly 74 surveillance reports is the percentage of all tests in the SGSS which are positive for influenza among 75 all patients presenting with influenza-like illness (ILI). This metric increased much earlier than in other 76 recent seasons and has started declining as of the report published on 2025-12-18 using data up to 77 2025-12-14 (10). An advantage of the percentage positive metric is that it is thought to be less biased 78 by changes in overall testing volume than count data, however, it is also affected by the dynamics 79 of other ILI-causing pathogens. An ILI+ indicator, which multiplies ILI cases by the percentage of 80 tests positive for influenza, is therefore often recommended instead (11). We focus here on influenza 81 case counts and ILI+, both overall and for A/H3N2, and stratified by age for recent years, to 82 understand the transmission rate of the current clade K viruses in England based on traditional 83 measures of absolute growth rate and Rt. 84 We analysed publicly available epidemiological data from England and developed an age-stratified 85 Susceptible-Infected-Recovered model to address two questions: 86 1. Do the epidemiological data indicate a more transmissible A/H3N2 strain than previous 87 seasons? 88 4 2. Under plausible scenarios of a virus with increased transmissibility, earlier seeding, and/or 89 substantial immune evasion, what is the potential impact on cumulative and peak healthcare 90 burden for the rest of the season? 91 92 5 Methods 93 Epidemiological analyses 94 Data summary 95 We analysed weekly counts of reported influenza cases by subtype for England from the Respiratory 96 DataMart system from 2009-04-27 to 2025-12-08 (Figure S1). We also analysed weekly counts of 97 reported influenza specimens stratified by subtype for England from the WHO FluNet platform 98 between 2011-01-09 and 2025-12-14 (12). All data are aggregated by week, and we used the first 99 day of the epidemiological week as the reported date. We removed the most recent week of data to 100 avoid issues arising from reporting delays. As many influenza A cases are not subtyped, we 101 distributed unsubtyped counts into influenza A/H3N2 and A/H1N1pdm09 counts proportional to the 102 ratios of subtyped samples. Further information on all datasets and extraction methods is provided 103 in the Supplementary Material. 104 To understand age-specific dynamics, we also used data on influenza cases by age group and by 105 subtype from the Royal College of General Practitioners Research & Surveillance Centre (RCGP 106 RSC) and ILI from the RCGP RSC weekly reports (Figure S1-3). Following recommendations from 107 (11), we developed an ILI+ indicator for each age group as: 108 𝐼𝐿𝐼!=𝐼𝐿𝐼∗𝑝!"# ∗𝑁'109 Where 𝑝!"# is the proportion of all tests which are positive for influenza either overall or by subtype 110 and 𝑁 is the number of individuals in that age group. Note that imperfect case ascertainment is 111 implicit in the ILI data. 112 113 Weekly growth rate calculations 114 Weekly growth rates were calculated for each influenza season and aligned by calendar week as: 115 𝑦 =𝑙𝑜𝑔(𝑖(𝑡) 𝑖(𝑡−1))'116 Where 𝑖(𝑡) is the reported incidence over week t. We calculated weekly exponential growth rates 117 of influenza cases in England from the Respiratory DataMart using a Gaussian random walk model 118 as described by Eales et al (13). We also calculated overall weekly growth rates using data from 119 the WHO FluNet for comparison using the same method. Finally, we fitted a Generalised Additive 120 Model (GAM) with a thin-plate spline using the mgcv R package to describe weekly growth rates of 121 the age-stratified ILI+ data from the RCGP RSC data (14). 122 123 6 Time-varying reproduction number estimation 124 We estimated the time-varying reproduction number (Rt) using the EpiEstim package in R (15). Rt is 125 defined as the expected number of new infections at time t, divided by past incidence weighted by 126 relative infectiousness given by the generation interval distribution. We used the overall weekly 127 influenza incidence data for the period 2011/12 season to the 2025/26 from the UKHSA Respiratory 128 DataMart for this analysis. To approximate daily incidence, weekly counts were distributed evenly 129 across days per week and then smoothed with a 14-day rolling mean. The serial interval distribution 130 was assumed to follow a distribution with a mean of 3.6 days and a standard deviation of 1.6 days, 131 based on (16). We used a window size of 14 days and the default Gamma-distributed prior on Rt 132 with a mean of 5 and standard deviation of 5 (15). 133 Compartmental model and scenario analyses 134 Model overview 135 To explore potential epidemic scenarios distinguishing the current influenza season from the 136 previous A/H3N2 season in 2022/23, we simulated seasonal influenza transmission dynamics for 137 England using an ageand immunity-structured deterministic Susceptible-Infected-Recovered 138 model. We explored scenarios where the K clade viruses exhibit enhanced transmissibility, greater 139 immune escape, or an earlier epidemic seed time. We compared the impact of these scenarios on 140 the epidemic timing around the half-term and Christmas school holidays, the final epidemic size, the 141 final size in 65+ year olds, and peak incidence as a proxy for maximum healthcare burden. 142 Model structure 143 We divided the population into four age groups (0-4, 5-18, 19-64 and 65+ years) and two immunity 144 classes (fully susceptible and partially immune), representing eight classes in total. 145 The force of infection in group i was defined as: 146 𝜆$(𝑡)= 𝛽3 % &'( 𝐶$,&(𝑡)𝐼&(𝑡)'147 Where β is the overall transmission rate (not age-stratified), Ci,j is the contact rate between group i 148 and group j, and Ij is the number of infected individuals in group j at time t. The basic reproduction 149 number, R0, was given by the largest eigenvalue of the next-generation matrix with contact matrix 150 C, multiplied by β and the infectious period, Tg. 151 7 Transition rates between the three compartments were defined by the following set of ordinary 152 differential equations: 153 𝑑𝑆*,+ 𝑑𝑡 =−𝜂+𝑆*,+(𝑡)𝜆*,+(𝑡)'154 𝑑𝐼*,+ 𝑑𝑡 =𝜂+𝑆*,+(𝑡)𝜆*,+(𝑡)−𝐼*,+(𝑡) 𝑇,'155 𝑑𝑅*,+ 𝑑𝑡 =𝐼*,+(𝑡) 𝑇,'156 Where 𝜂$ denotes the relative susceptibility of immune class k and Tg is the infectious period. Note 157 that 𝜂$ was set to 0 for the immune population, representing all-or-nothing immunity. Immune escape 158 was modelled as a single parameter, δ, which scales the initial population immune proportion in all 159 age groups (δ=0 corresponds to complete immune escape, whereas δ=1 corresponds to no loss of 160 population immunity). We solved the model in daily timesteps using the deSolve R package (17). 161 Baseline model calibration 162 Model parameters were chosen based on standard seasonal influenza parameter values (R0 of 163 around 2, infectious period of 4-5 days, and final size of around 15% (18)). We used our intuition and 164 manual calibration to generate seasonal dynamics similar to what was seen in the 2022/23 season. 165 We note that this is a complex model with a large number of parameters, making formal model fitting 166 difficult. Some of the parameters are hard to identify and interpret, such as the overall fraction of 167 symptomatic cases reported, the symptomatic fraction by age combined with age-specific reporting 168 rates, and the level of immune escape of the seed virus. Parameter values used for the baseline 169 scenario are shown in Table S1. 170 We set the population size of the model to 60,000,000 to approximate the population size of England. 171 We distributed the population into age groups based on the age distributions in the socialmixr 172 package using the POLYMOD data (19). Each age group was stratified into the susceptible or fully 173 immune class assuming different levels of immune escape for each age group. The epidemic was 174 seeded by arbitrarily setting I19-64(0) = 1000; the seed time was varied during model calibration. 175 Contact matrices over time 176 Symmetric, age-stratified contact matrices were generated using all contact data from the 177 POLYMOD UK study using the socialmixr R package (19). We generated four contact matrices for 178 different time periods: 1) regular school term-time; 2) half-term with no school contacts and reduced 179 school contacts; 3) pre-Christmas shopping period (2022-12-01 to 2022-12-21) with an increase in 180 all non-school contacts; and 4) the Christmas school holiday period (2022-12-21 to 2023-01-05) with 181 no school contacts, a reduction in all contacts, and an increase in at-home contacts. These matrices 182 were constructed by resampling the original POLYMOD contact diary entries with replacement and 183 applying multipliers for home, work and other contact types. We smoothed the transition between 184 8 contact matrices over 3 days before and after the holiday period using a cosine function. Holiday 185 dates were based on the Oxfordshire school holiday period. For all non-school contacts, the relative 186 changes in contacts over half-term holidays and the Christmas period (shopping and holiday) were 187 extrapolated from 2021-2023 data from Kendall et al. (see Figures 5A and 7B in the paper) (20). 188 Absolute changes in overall non-school contact rates were extrapolated from the same source (see 189 Figures 1A, 7A in the paper). 190 Scenario analyses 191 Scenario analyses were chosen to illustrate potential hypotheses for the early and rapid growth of 192 A/H3N2 cases in England for the 2025/26 season. We also varied the immune escape scaling 193 parameter δ, the basic reproduction number R0, the proportion of the 0-4 and 5-18 year old 194 population initially immune, and the seed date univariably. 195 Implementation 196 All analyses were implemented and run in R version 4.2.2. The compartmental model was also 197 implemented as a Shiny app with user-friendly sliders to change key parameter values, available at: 198 https://hay-idd.shinyapps.io/ModelFluUk-H3N2/. 199 9 Results 200 The maximum growth rates of total and A/H3N2 influenza cases in the current season are 201 comparable to previous seasons 202 We propose the epidemic growth rate as an early indicator of epidemic fitness of an influenza strain, 203 as it is a compound measure of intrinsic transmissibility, R0, of immune escape, and of behavioural 204 mixing patterns. The peak growth rate occurs well before the peak of cases, at which point the growth 205 rate is zero. 206 The peak growth rate of all influenza cases combined (i.e., aggregating A/H3N2, A/H1N1pdm09, 207 unsubtyped influenza A, and influenza B) for the 2025/26 season is comparable to previous seasons 208 with a peak value so far (as of 2025-12-08) of 0.651 (posterior mean; 95% credible intervals: 0.470-209 0.838) in epidemiological week 40 (2025-09-29 to 2025-10-05) (Figure 1). The 2014/15, 2017/18, 210 2022/23 and 2023/24 seasons all had higher peak growth rates for influenza cases combined, 211 though the 2023/24 season featured a mixture of subtypes (Table 1). The 2025/26 season stands 212 out as an outlier in the timing of peak growth, which occurred in epidemiological week 40. In contrast, 213 all other seasons showed peak growth rates between week 47 and week 51. After aligning the 214 epidemic curves to the date of peak growth rate, overall trends appeared similar (Figure S4), though 215 the 2025/26 season has shown a secondary, albeit smaller rise in growth rate matching the timing 216 of other influenza seasons. 217 Restricting our analysis to only influenza A/H3N2 cases gave similar growth rate results, though the 218 2025/26 season was ranked second for peak A/H3N2 growth rate at 0.709 (0.506-0.919), surpassed 219 only by the 2014/15 season which peaked at 0.840 (0.665-1.02). Similar trends were observed using 220 the WHO FluNet data, though with a higher peak growth rate of A/H3N2 (Figure S5). 221 We observed higher peak growth rates of all influenza cases in the 0-4 and 5-18 year old age groups 222 compared to individuals aged 65+ (Figure S6). The difference in weekly growth rates between 223 children and adults was far higher than has been observed in the prior two influenza seasons, but 224 comparable to the 2022/23 season when A/H3N2 last dominated (Table 2). 225 16 287 Figure 4. Univariable sensitivity analyses of the model outputs with respect to key parameters. Each 288 row shows the impact of varying one parameter, given in the x-axis label, whilst keeping all other parameters 289 at their baseline values. The vertical dashed line shows the values used for the base scenario. Results shown 290 are for the epidemic final size (proportion of the population infected) overall and within the 65+ age group, the 291 peak epidemic growth rate after at least 1% of total infections have occurred, and the peak weekly infection 292 incidence.293 0 1000 2000 3000 4000 5000 0.0 0.5 1.0 1.5 0.0 0.1 0.2 0.3 Final size Final size 65+ Peak growth rate Peak infection incidence per 100,000 1.0 1.5 2.0 2.5 3.0 1.0 1.5 2.0 2.5 3.0 1.0 1.5 2.0 2.5 3.0 1.0 1.5 2.0 2.5 3.0 0.0 0.1 0.2 0.3 R0 Value 0 5000 10000 15000 20000 0 1 2 3 0.00 0.25 0.50 0.75 1.00 Final size Final size 65+ Peak growth rate Peak infection incidence per 100,000 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Proportion immune relative to baseline Value 0 1000 2000 3000 4000 5000 0.0 0.5 1.0 1.5 0.0 0.1 0.2 0.3 Final size Final size 65+ Peak growth rate Peak infection incidence per 100,000 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 0.0 0.1 0.2 0.3 Proportion immune relative to baseline in <18 Value 0 250 500 750 1000 0.0 0.5 1.0 1.5 0.00 0.05 0.10 0.15 0.20 Final size Final size 65+ Peak growth rate Peak infection incidence per 100,000 Aug Sep Oct Nov Aug Sep Oct Nov Aug Sep Oct Nov Aug Sep Oct Nov 0.00 0.05 0.10 0.15 0.20 Seed date (1000 infections) Value 17 Table 3. Comparison of peak disease burden and overall disease burden of model scenarios relative to base case. Peak growth rate was taken as the highest overall weekly growth 294 rate after at least 1% of all infections had occurred. All results shown, other than peak growth rate, are based on the ratio of outputs from the scenario to the base case. We do not report 295 absolute numbers due to the lack of formal model fitting (other than the epidemic final size of the base case to roughly guide interpretation of the ratios). 296 Scenario label Scenario Parameters changed from baseline Peak of weekly symptomatic cases Date of peak symptomatic cases Peak growth rate Date of peak growth rate Cumulative symptomatic cases Final size in 65+ Final size Interpretation A Base case (loosely based on the 2022/23 season) - 1.00 26/12/2022 0.83 05/12/2022 1.00 1.00 (7.57%) 1.00 (16.2%) Model parameters are roughly calibrated to match the 2022/23 influenza season, but we note that there is considerable hidden uncertainty, strong correlations between model parameters and structural identifiability issues. B Moderate immune escape Immune escape parameter set to 0.95 1.23 12/12/2022 0.70 07/11/2022 1.37 1.51 1.32 A relatively small loss of population immunity leads to a larger outbreak overall. C High immune escape Immune escape parameter set to 0.90 1.39 05/12/2022 0.82 17/10/2022 1.63 1.81 1.54 As in B, but with a much larger overall impact. D Severe immune escape Immune escape parameter set to 0.80 1.94 14/11/2022 1.30 10/10/2022 2.06 2.37 1.89 This can be ruled out, as an extremely large epidemic would have already peaked in November. E Severe reduction in immunity in children All initial immune fractions in <18 year olds reduced by 50% 0.87 17/10/2022 1.71 03/10/2022 1.13 0.77 1.26 Drastically increased transmission in children would lead to a peak before the half-term holiday and the epidemic ending by the New Year. Can be ruled out. F Minor reduction in immunity in children Initial immune fraction in 0-4 reduced to 20%; initial immune fraction in 5-18 reduced to 50% 0.88 05/12/2022 0.82 07/11/2022 1.13 1.00 1.20 Dampens the peak due to the half term circuit break. Leads to higher overall final size and lower burden in older adults. A resurgence after half term is expected. G Higher transmissibility R0 increased from 1.9 to 2.2 1.04 05/12/2022 0.79 07/11/2022 1.25 1.30 1.22 An increase in final size in all age groups and an early peak, but with a similar peak size. H Severe transmissibility R0 increased from 1.9 to 2.4 0.99 21/11/2022 1.12 10/10/2022 1.31 1.38 1.28 Similar to G but counterintuitively leads to a more drawn-out epidemic with slightly lower peak incidence. This is likely due to the timing of the half term dampener with respect to incidence, which lessens the overshoot of infections after herd immunity is reached. I Two week earlier seeding Seeding on 202508-24 rather than 2025-09-10 0.79 19/12/2022 0.62 05/12/2022 0.99 0.97 1.00 A very similar overall burden of infection and with a much lower peak burden expected in mid-December. J One month earlier seeding Seeding on 202508-01 rather than 2025-09-10 0.63 19/12/2022 0.57 03/10/2022 1.00 0.99 1.00 Similar to I, but with an overall peak growth rate seen in the first week of October rather than early December. 297 18 Discussion 298 Our analysis shows that the K clade of influenza in England has, as of 2025-12-18, an unusually 299 early peak epidemic growth rate and time-varying reproduction number but of magnitude comparable 300 to previous severe seasons over the past 15 years. The time-varying reproduction number peaked 301 at 1.49 (CI=[1.28,1.71]) on 2025-10-06 based on current data, which is slightly higher than the typical 302 reproduction number of seasonal influenza of 1.2-1.3 reported in the literature (21), but similar to 303 what we estimated using the EpiEstim Rpackage for previous seasons. There have been 304 substantially higher growth rates of A/H3N2 cases this season in children <18 years old compared 305 to adults. We also ran scenario analyses to explore possible virus changes which might be 306 compatible with the earlier start to the season. Only small reductions in population immunity due to 307 immune escape are required to drive an earlier influenza season with more cumulative infections 308 (the model parameter δ), suggesting that substantial escape from overall population immunity, not 309 just HA-mediated immunity, is incompatible with current data. Overall, our findings suggest that the 310 total infection burden in England is likely to be at the upper end of typical influenza seasons. 311 312 Antigenic escape from both the vaccine strain and the overall population immune landscape are 313 thought to be predictors of subsequent seasonal influenza burden (9). For the clade K A/H3N2 314 viruses, substantially reduced reactivity of vaccine raised ferret antisera and human sera has been 315 observed (4,22,23). Our scenario analyses suggest that the level of total immune escape in the 316 population is relatively modest, as substantially greater immune escape would generate an even 317 earlier start and markedly higher peak growth rates than those observed. 318 319 For seasonal influenza, which is likely driven by high within-age-group social mixing and greater 320 susceptibility to infection in children, changes in contact patterns during school term times and 321 holidays have a major impact on the shape of the epidemic (24). Although these changes are difficult 322 to parameterise, our base model likely represents realistic shifts in behaviour (20). Whether a new 323 strain with enhanced transmissibility or immune escape will lead to a higher winter burden partially 324 depends on the timing of peak transmission relative to school holidays – the half-term holiday can 325 act as a circuit breaker, dampening the peak and spreading cases over a longer period. 326 The analyses performed here are relatively straightforward and quick, using data which are already 327 routinely collected. Although these types of analyses were used widely during the COVID-19 328 pandemic, they have since become less common for seasonal respiratory diseases. We suggest 329 that real-time growth rate characterisation and public, interactive tools for ongoing scenario analyses 330 should be maintained regularly and updated early in the season to rule in or out different scenarios. 331 Maintaining the toolkit requires low resource but consistent input from technical specialists, and real-332 19 time access to absolute reported case numbers rather than just percentage of tests positive for 333 influenza. 334 Our analyses have several limitations. For the epidemiological analyses, we used multiple data 335 sources each with limitations in representativeness and reporting algorithms. Growth rate and Rt 336 estimates are also likely biased by changes in testing intensity and reporting rates. We note that 337 overconfident estimates from small sample sizes are a known issue with renewal equations, affecting 338 especially pre-season estimates when case numbers are very low (25). 339 The model used for scenario analyses is also caveated, as we did not perform a formal model fit due 340 to time constraints. Instead, we chose fixed parameter values based on commonly assumed 341 influenza parameters (R0, infectious period, final size), and then manually calibrated other 342 parameters to achieve a reasonable visual fit to the 2022/23 influenza incidence data. Assumptions 343 regarding age-specific immunity, immune escape, and all-or-nothing immunity in two classes rather 344 than stratified immunity all have a large impact on the projected incidence curves (26), and we 345 therefore recommend using the tool to inform a general understanding of the system rather than 346 predictions. A key omission is vaccination, which we did not include in the scenario model due to 347 challenges in parameterising age-specific vaccine efficacy against infection and disease. Our model 348 is also limited using outdated contact data and strong assumptions surrounding behaviour changes 349 in school holidays and the Christmas period. Although we are confident that these assumptions 350 capture general trends, more recent and well-calibrated parameter values would improve the 351 accuracy of the analysis. Finally, we did not consider seasonal forcing due to climate factors (27,28). 352 In summary, we have combined publicly available influenza data for England with an age-stratified 353 Susceptible-Infected-Recovered model and an interactive visualisation webtool. Our analysis shows 354 that compared to previous years, the epidemic growth rate is high but not exceptional. Modelling 355 analysis suggests that the epidemic trajectory so far is consistent with at most modest reduction in 356 population immunity compared to previous years. Increases in transmission do not always translate 357 to bigger epidemics because of the interaction between epidemic dynamics and school holidays that 358 prevent epidemic ‘overshoot’. We provide a range of scenarios and an interactive scenario-explorer 359 app, that show the impact of different assumptions on the epidemic curve for the coming months. 360 20 References 361 1. auspice [Internet]. [cited 2025 Dec 22]. Available from: https://nextstrain.org/seasonal-362 flu/h3n2/ha/2y?dmin=2024-11-29 363 2. Huddleston J, Chang J, Lee J, Steinberg P, Neher R, Bedford T. Seasonal influenza 364 circulation patterns and projections for September 2025 to September 2026 [Internet]. Zenodo; 365 2025. Available from: https://doi.org/10.5281/zenodo.17281059 366 3. Fleming DM, Elliot AJ. Lessons from 40 years’ surveillance of influenza in England and Wales. 367 Epidemiol Infect. 2008 July;136(7):866–75. 368 4. Kirsebom FC, Thompson C, Talts T, Kele B, Whitaker HJ, Andrews N, et al. Early influenza 369 virus characterisation and vaccine effectiveness in England in autumn 2025, a period 370 dominated by influenza A(H3N2) subclade K. Euro Surveill [Internet]. 2025 Nov;30(46). 371 Available from: http://dx.doi.org/10.2807/1560-7917.ES.2025.30.46.2500854 372 5. Iacobucci G. Flu in numbers: NHS faces one of worst winters ever, officials warn, amid 373 concern over mutated strain. BMJ. 2025 Nov 12;391:r2391. 374 6. Threat Assessment Brief: Assessing the risk of influenza for the EU/EEA in the context of 375 increasing circulation of A(H3N2) subclade K [Internet]. European Centre for Disease 376 Prevention and Control. 2025 [cited 2025 Dec 22]. Available from: 377 https://www.ecdc.europa.eu/en/publications-data/threat-assessment-brief-assessing-risk-378 influenza-november-2025 379 7. Dapat C, Peck H, Jelley L, Diefenbach-Elstob T, Slater T, Hussain S, et al. Extended influenza 380 seasons in Australia and New Zealand in 2025 due to the emergence of influenza A(H3N2) 381 subclade K viruses. Euro Surveill [Internet]. 2025 Dec;30(49). Available from: 382 http://dx.doi.org/10.2807/1560-7917.ES.2025.30.49.2500894 383 8. Fieldhouse R. Japan declares a flu epidemic - what this means for other nations. Nature 384 [Internet]. 2025 Oct 14 [cited 2025 Dec 22]; Available from: 385 https://www.nature.com/articles/d41586-025-03367-z 386 9. Perofsky AC, Huddleston J, Hansen C, Barnes JR, Rowe T, Xu X, et al. Antigenic drift and 387 subtype interference shape A(H3N2) epidemic dynamics in the United States [Internet]. 2024. 388 Available from: https://doi.org/10.7554/eLife.91849.1 389 10. National flu and COVID-19 surveillance report: 18 December 2025 (week 51) [Internet]. 390 Gov.uk. [cited 2025 Dec 22]. Available from: 391 https://www.gov.uk/government/statistics/national-flu-and-covid-19-surveillance-reports-2025-392 to-2026-season/national-flu-and-covid-19-surveillance-report-18-december-2025-week-51 393 11. Eales O, McCaw JM, Shearer FM. Biases in routine influenza surveillance indicators used to 394 monitor infection incidence and recommendations for improvement. Influenza Other Respi 395 Viruses. 2024 Dec;18(12):e70050. 396 12. flunetchart [Internet]. [cited 2025 Nov 7]. Available from: 397 https://worldhealthorg.shinyapps.io/flunetchart/ 398 13. Eales O, Windecker SM, McCaw JM, Shearer FM. Inferring temporal trends of multiple 399 pathogens, variants, subtypes or serotypes from routine surveillance data. Am J Epidemiol. 400 2025 June 6;(kwaf119):kwaf119. 401 14. Mgcv: Mixed GAM computation vehicle with automatic smoothness estimation [Internet]. 402 Comprehensive R Archive Network (CRAN). [cited 2024 June 24]. Available from: 403 https://cran.r-project.org/web/packages/mgcv/index.html 404 21 15. Cori A. Estimate Time Varying Reproduction Numbers from Epidemic Curves [R package 405 EpiEstim version 2.2-5] [Internet]. Comprehensive R Archive Network (CRAN). 2025 [cited 406 2025 Dec 21]. Available from: https://cran.r-project.org/web/packages/EpiEstim/index.html 407 16. Cowling BJ, Fang VJ, Riley S, Malik Peiris JS, Leung GM. Estimation of the serial interval of 408 influenza. Epidemiology. 2009 May;20(3):344–7. 409 17. Solvers for Initial Value Problems of Differential Equations (“ODE”, “DAE”, ’DDE’) [R package 410 deSolve version 1.40] [Internet]. Comprehensive R Archive Network (CRAN). 2023 [cited 2025 411 Nov 22]. Available from: https://cran.r-project.org/web/packages/deSolve/index.html 412 18. Yang W, Lipsitch M, Shaman J. Inference of seasonal and pandemic influenza transmission 413 dynamics. Proc Natl Acad Sci U S A. 2015 Mar 3;112(9):2723–8. 414 19. Funk S. Socialmixr: Social Mixing Matrices for Infectious Disease Modelling. 2018 Jan 15; 415 Available from: https://datacompass.lshtm.ac.uk/646/#.XoMX-QScPY0.mendeley 416 20. Kendall M, Ferretti L, Wymant C, Tsallis D, Petrie J, Di Francia A, et al. Drivers of epidemic 417 dynamics in real time from daily digital COVID-19 measurements. Science [Internet]. 2024 418 Aug 16 [cited 2024 Aug 21];385(6710). Available from: 419 https://pubmed.ncbi.nlm.nih.gov/38991048/ 420 21. Biggerstaff M, Cauchemez S, Reed C, Gambhir M, Finelli L. Estimates of the reproduction 421 number for seasonal, pandemic, and zoonotic influenza: a systematic review of the literature. 422 BMC Infect Dis. 2014 Sept 4;14(1):480. 423 22. Sabaiduc S, Kaweski SE, Separovic L, Gao R, Ranadheera C, Bastien N, et al. Emergence of 424 seasonal influenza A(H3N2) variants with immune escape potential warrants enhanced 425 molecular and epidemiological surveillance for the 2025–2026 season. Journal of the 426 Association of Medical Microbiology and Infectious Disease Canada [Internet]. 2025 Oct 31; 427 Available from: http://dx.doi.org/10.3138/jammi-2025-0025 428 23. Kikawa C, Huddleston J, Loes AN, Turner SA, Lee J, Barr IG, et al. Near real-time data on the 429 human neutralizing antibody landscape to influenza virus to inform vaccine-strain selection in 430 September 2025 [Internet]. bioRxivorg. 2025. Available from: 431 http://dx.doi.org/10.1101/2025.09.06.674661 432 24. Cauchemez S, Valleron A-J, Boëlle P-Y, Flahault A, Ferguson NM. Estimating the impact of 433 school closure on influenza transmission from Sentinel data. Nature. 2008 Apr 434 10;452(7188):750–4. 435 25. Němcová B, Goldstein IH, Sebastian J, Minin VM, Bracher J. Unjustified Poisson assumptions 436 lead to overconfident estimates of the effective reproductive number [Internet]. medRxiv. 437 medRxiv; 2025. Available from: http://dx.doi.org/10.1101/2025.07.31.25332479 438 26. Yuan H-Y, Baguelin M, Kwok KO, Arinaminpathy N, van Leeuwen E, Riley S. The impact of 439 stratified immunity on the transmission dynamics of influenza. Epidemics. 2017 Sept;20:84–440 93. 441 27. Shaman J, Kohn M. Absolute humidity modulates influenza survival, transmission, and 442 seasonality. Proc Natl Acad Sci U S A. 2009 Mar 3;106(9):3243–8. 443 28. van Noort SP, Águas R, Ballesteros S, Gomes MGM. The role of weather on the relation 444 between influenza and influenza-like illness. J Theor Biol. 2012 Apr 7;298:131–7. 445 446 22 Supplementary Material 447 Supplementary Methods – additional information on data sources 448 1. Second-Generation Surveillance System (SGSS) 449 Weekly influenza positivity from PCR tests by age was obtained from the UKHSA dashboard for the 450 SGSS, available at Influenza | UKHSA data dashboard. Values are shown as a percentage of people 451 with at least one positive PCR result for influenza (in a 7 day period). Further details are available at 452 Data quality report: national flu and COVID-19 surveillance report - GOV.UK (1). This dataset 453 provided us with the proportion of tests which are positive for influenza, but not the absolute number 454 of tests performed, proportion positive stratified by influenza subtype, nor stratification by age. 455 2. Royal College of General Practitioners (RCGP) Research & Surveillance Centre (RSC) 456 The RCGP RSC is a nationally representative primary care-based surveillance system. Weekly 457 communicable and respiratory diseases reports are publicly available at RSC: Public health data 458 (2). In theory, this dataset has all of the information needed to construct an ILI+ indicator, but it is 459 only publicly available in PDF reports and a live dashboard. The PDF reports did not contain the 460 granularity required for the full age-stratified ILI+ indicator, and we were unable to automate 461 digitisation of the dashboard. 462 We compiled weekly national incidence of influenza-like illness (ILI) rates (per 100,000 population) 463 for England. PDF reports were downloaded for six-month intervals from January 2021 to 464 December 2025, and Table E from each report was converted to CSV. These tables included age-465 stratified ILI rates across four age categories differing by reporting period: 466 1. 1-4 yrs, 5-14yrs, 15-64yrs and 65+yrs, and all ages from week 26, 2023 to week 44, 2025 467 2. <15 yrs, 15-64yrs, 65+yrs, and all ages for weeks 40-53, 2020, week 1, 2021 to week 5, 468 2021, and weeks 1-26, 2023. 469 This dataset provides ILI rates by age, which we convert to estimates for absolute ILI case numbers 470 by multiplying by the population size per age group. However, it does not stratify ILI into influenza 471 A/H3N2 positivity. 472 We also digitised data on the total number of samples tested and the number of samples positive for 473 influenza by age group and by subtype for the 2022/23 to 2025/26 influenza seasons. 474 23 3. Respiratory DataMart sentinel system for hospital testing. 475 The Respiratory DataMart is a sentinel laboratory surveillance system which monitors all major 476 respiratory viruses in England, compiling data from 17 contributing laboratories. Participating 477 laboratories test swabs for respiratory viruses, including influenza A viruses, using real-time 478 polymerase chain reaction (RT-PCR), though not all laboratories test for or report all viruses. The 479 absolute number of samples in this system is much lower than SGSS, but gives absolute numbers 480 of tests rather than just percentage positive. 481 We created a combined dataset spanning the period 27 April 2009 to 8th December 2025. This 482 dataset included weekly counts of laboratory-confirmed influenza cases by subtype: influenza A 483 (not subtyped), influenza A(H1N1)pdm09, influenza A(H3N2), and influenza B. Additionally, the 484 dataset provides the overall percentage of specimens testing positive for influenza. However, these 485 datasets were not stratified by age group. 486 4. World Health Organisation (WHO) FluNet 487 Weekly counts of reported influenza specimens were obtained from the WHO FluNet platform (3). 488 This is the global influenza virological surveillance system collecting weekly sentinel and non-489 sentinel (e.g., outbreak investigations, point-of-care testing) surveillance data from national 490 influenza centres. Data was extracted for the period 9th January 2011 to 14th December 2025, 491 restricted to England. The extracted dataset included the weekly number of samples classified as 492 A(H3), influenza A (not subtyped), A(H1N1pdm09) influenza B, and overall influenza cases. We 493 allocated the unsubtyped influenza A cases into A/H3N2 and A/H1N1 cases proportional to the 494 ratio of subtyped samples. 495 Aligning mismatched age groups from different datasets 496 Where age groups were misaligned, we re-distributed metrics (such as ILI or percentage positive 497 for influenza) into new age bandings by assuming that each metric is the same for all ages (in 498 years) within an age group. We then recombined datasets into new age bands by reweighting 499 these age metrics based on the age distribution of England (4). 500 24 Supplementary References 501 1. Data quality report: national flu and COVID-19 surveillance report [Internet]. Gov.uk. [cited 502 2025 Dec 21]. Available from: https://www.gov.uk/government/publications/sources-of-503 surveillance-data-for-influenza-covid-19-and-other-respiratory-viruses/data-quality-report-504 national-flu-and-covid-19-surveillance-report#about-this-report 505 2. RCGP. Public health data [Internet]. [cited 2025 Dec 21]. Available from: 506 https://www.rcgp.org.uk/representing-you/research-at-rcgp/research-surveillance-507 centre/public-health-data 508 3. flunetchart [Internet]. [cited 2025 Dec 21]. Available from: 509 https://worldhealthorg.shinyapps.io/flunetchart/ 510 4. Population estimates for England and Wales - Office for National Statistics [Internet]. Office for 511 National Statistics; 2025 [cited 2025 Dec 21]. Available from: 512 https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationesti513 mates/bulletins/populationestimatesforenglandandwales/mid2024 514 25 Table S1. Model parameters assumed for the baseline scenario. These are the default parameters 515 in the interactive web tool. 516 Parameter Assumed value R0: basic reproduction number 1.9 Tg: infectious period 4.5 days γ: immune escape multiplier 1 Seed date 19th September Seed size in age group 1 (0-4 yrs) 1000 Population size 60,000,000 Proportion of work contacts kept in school holidays 0.75 Multiplier for home contacts in school breaks 1.00 Multiplier for non-school and non-work contacts in school breaks 1.10 Multiplier for all non-school contacts in Christmas period (1-15 December) 1.25 Proportion of all contacts kept over Christmas 0.70 Multiplier for home contacts over Christmas holiday 3.00 Proportion initially immune (0-4 yrs) 0.35 Proportion initially immune (5-18 yrs) 0.60 Proportion initially immune (19-64 yrs) 0.70 Proportion initially immune (65+ yrs) 0.70 Symptomatic fraction and reporting rate (0-4 yrs) 0.05 Symptomatic fraction and reporting rate (5-18 yrs) 0.20 Symptomatic fraction and reporting rate (19-64 yrs) 0.30 Symptomatic fraction and reporting rate (65+ yrs) 0.50 517 32 551 Figure S7. Daily incidence data used for Rt estimation in Figure 2. 552 33 553 Figure S8. Comparison of symptomatic influenza incidence in 65+ from the scenario analyses shown 554 in Figure 4. 555 34 556 Figure S9. Comparison of weekly age-stratified growth rates from scenarios matching Figure 4. 557