Determinants of RSV epidemiology following suppression through pandemic contact restrictions

Abstract

Introduction: COVID-19 related non-pharmaceutical interventions (NPIs) led to a suppression of RSV circulation in winter 2020/21 in the UK and an off-season resurgence in Summer 2021. We explore how the parameters of RSV epidemiology shape the size and dynamics of post-suppression resurgence and what we can learn about them from the resurgence patterns observed so far.

Methods: We developed an age-structured dynamic transmission model of RSV and sampled the parameters governing RSV seasonality, infection susceptibility and post-infection immunity, retaining simulations fitting the UK's pre-pandemic epidemiology by a set of global criteria consistent with likelihood calculations. From Spring 2020 to Summer 2021 we assumed a reduced contact frequency, returning to pre-pandemic levels from Spring 2021. We simulated transmission forwards until 2023 and evaluated the impact of the sampled parameters on the projected trajectories of RSV hospitalisations and compared these to the observed resurgence.

Results: Simulations replicated an out-of-season resurgence of RSV in 2021. If unmitigated, paediatric RSV hospitalisation incidence in the 2021/22 season was projected to increase by 30-60% compared to pre-pandemic levels. The increase was larger if infection risk was primarily determined by immunity acquired from previous exposure rather than age-dependent factors, exceeding 90 % and 130 % in 1-2 and 2-5 year old children, respectively. Analysing the simulations replicating the observed early outbreak in 2021 in addition to pre-pandemic RSV data, we found they were characterised by weaker seasonal forcing, stronger age-dependence of infection susceptibility and higher baseline transmissibility.

Conclusion: COVID-19 mitigation measures in the UK stopped RSV circulation in the 2020/21 season and generated immunity debt leading to an early off-season RSV epidemic in 2021. A stronger dependence of infection susceptibility on immunity from previous exposure increases the size of the resurgent season. The early onset of the RSV resurgence in 2021, its marginally increased size relative to previous seasons and its decline by January 2022 suggest a stronger dependence of infection susceptibility on age-related factors, as well as a weaker effect of seasonality and a higher baseline transmissibility. The pattern of resurgence has been complicated by contact levels still not back to pre-pandemic levels. Further fitting of RSV resurgence in multiple countries incorporating data on contact patterns will be needed to further narrow down these parameters and to better predict the pathogen's future trajectory, planning for a potential expansion of new immunisation products against RSV in the coming years.

Keywords: Disease Dynamics; Epidemiology; Non-pharmaceutical interventions; RSV; Transmission modelling.

Fig. 1

A. Age-structured SIRS model of RSV transmission with reinfections, immunity waning, births with maternal immunity and deaths. B. Simulated hospitalisations for children under 5 years before the COVID-19 pandemic. Blue lines are accepted simulations (see Methods) with a negative log-likelihood lower than 1500 (corresponding to the best 15 % of fits, red dashed lines showing their median); grey lines are those discarded. Black dots show hospitalisation rates for England from SARI Watch (National flu and COVID-19 surveillance reports: 2021 to 2022 season, 2022). C. Likelihoods of accepted and rejected parameterisations. The x-axis shows the number of age groups the simulations correctly predict the attack rate. Dots highlighted in red were rejected because of a biennial seasonal pattern and those marked with ‘x’ because of less than 85% of cases within the RSV season (week 40–13).

Fig. 2

RSV resurgence as a function of epidemiological parameters. (A) Partial rank correlation coefficients (PRCC) between the sampled parameters and the proportionate change post-NPI in cumulative and peak hospitalisations. Asterisks show the correlation has a p-value above 0.05. (B) Level of cumulative hospitalisations as a function of susceptibility determined by previous exposure (blue) or age (red), captured by the ratio ${\mathrm{\kappa }}_{\mathrm{age}}$/${\mathrm{\kappa }}_{\exp }$. Statistics calculated on the relative changes from pre-pandemic years to 2021 values (epi-years from week 23 to week 22).

Fig. 3

Dynamics of post-NPI weekly hospitalitalisation incidence as a function of infection susceptibility determined by age or previous exposure. Incidence was normalised to pre-pandemic peak incidence and time to the timing of the pre-pandemic peak. Colours indicate whether RSV immunity to infection is predominantly a function of previous exposure (blue) or age (red). The lines show median values by binned values of the ratio of the two parameters (${\mathrm{\kappa }}_{\mathrm{age}}$/${\mathrm{\kappa }}_{\exp }$).

Fig. 4

Simulations of RSV resurgence from June 2021, assuming gradual recovery of contact levels from March 2021, showing the 10% of simulations with the lowest error with respect to SARI-Watch hospitalisation rates. Inset: Cumulative density functions of parameters for all simulations accepted for matching pre-NPI RSV epidemiology (grey) versus the subset of accepted simulations that also replicated the early resurgence (blue). Dashed lines show median values for the two distributions; p-values are from Kolgomorov-Smirnov tests.