Modelling the impact of Omicron and emerging variants on SARS-CoV-2 transmission and public health burden.

Background: SARS-CoV-2 variants of concern, such as Omicron (B.1.1.529), continue to emerge. Assessing the impact of their potential viral properties on the probability of future transmission dominance and public health burden is fundamental in guiding ongoing COVID-19 control strategies.
Methods: With an individual-based transmission model, OpenCOVID, we simulated three viral properties; infectivity, severity, and immune-evading ability, all relative to the Delta variant, to identify thresholds for Omicron's or any emerging VOC's potential future dominance, impact on public health, and risk to health systems. We further identify for which combinations of viral properties current interventions would be sufficient to control transmission.
Results: We show that, with first-generation SARS-CoV-2 vaccines and limited physical distancing in place, a VOC's potential future dominance is primarily driven by its infectivity, which does not always lead to an increased public health burden. However, we also show that highly immune-evading variants that become dominant, even in the case of reduced variant severity, would likely require alternative measures to avoid strain on health systems, such as strengthened physical distancing measures, novel treatments, and second-generation vaccines. Expanded vaccination, that includes a booster dose for adults and child vaccination strategies, is projected to have the biggest public health benefit for a highly infective, highly severe VOC with low immune-evading capacity. Conclusions: These findings provide quantitative guidance to decision-makers at a critical time while Omicron's properties are being assessed and preparedness for emerging VOCs is eminent. We emphasise the importance of both genomic and population epidemiological surveillance.

Introduction

SARS-CoV-2 has been mutating continuously since its emergence in December 2019, leading to viral variants with varying infectivity, severity, and immune-evading properties. On 26 November 2021, the World Health Organization identified Omicron (B.1.1.529) as a new variant of concern (VOC) after approximately seven months of Delta variant (B.1.617.2) dominating global transmission[1]. Within one week, 24 countries had reported cases of the Omicron variant[2], with infections also occurring in previously infected and double-vaccinated individuals[3]. By mid-January 2022, >50% of all sequenced cases in Europe were caused by Omicron[4], in which we include BA.2 as it is a sublineage of B.1.1.529. Whilst scientific research assessing the infectivity, severity, and immune-evading properties of Omicron has been ongoing, understanding the potential scope of the associated public health burden of Omicron and emerging VOCs is of top priority[5].

In many European settings, high vaccination rates—particularly among those most at risk of severe disease—have led to a reduction in disease burden. However, protection of the most vulnerable and maintaining low levels of SARS-CoV-2 transmission in the northern hemisphere has been under threat at the time of Omicron’s emergence for multiple reasons. First, indoor contacts have been increasing during the cooler season. Second, immunity of the most vulnerable population (people 65+ and those with comorbidities) had started to wane since they were primarily vaccinated in the first quarter of 2021[6]. Last, relaxation of most non-pharmaceutical interventions (NPIs) since the summer of 2021 and fatigue of COVID-19 restrictions had led to increased contacts[7]. The emergence of Omicron in December 2021 combined with these threats called into question whether the vaccination strategies identified at the time would be sufficient.

Mathematical models have been used to represent transmission dynamics of SARS-CoV-2 and have supported decision-makers throughout the pandemic on the implementation and relaxation of control strategies[8-12]. Here we further developed and applied OpenCOVID, an individual-based model of SARS-CoV-2 transmission and COVID-19 disease, which includes seasonality patterns, waning immunity profiles, vaccination and NPI strategies, and properties for multiple variants[11]. We applied the OpenCOVID model to represent a general European setting and simulated the emergence of a novel variant. We analysed disease dynamics for a wide range of infectivity, severity, and immune evasion properties—relative to the Delta variant—representing a large range of potential properties for Omicron or for any other emerging VOCs. This allowed us to determine the potential for Omicron to become the new dominant variant, and to estimate its potential future public health burden. We further identified combinations of properties for which first-generation vaccines, in the absence of strong NPIs, would be sufficient to contain transmission and public health burden (i.e., new SARS-CoV-2 infections, hospital occupancy, and COVID-19-related deaths) and identified those combinations for which additional measures would be required.

We show that, with first-generation SARS-CoV-2 vaccines and limited physical distancing in place, a VOC’s potential future dominance is primarily driven by its infectivity. Future dominance does, however, not always lead to an increased public health burden. We also show that a highly immune-evading variant that becomes dominant, even in the case of reduced variant severity, would likely require alternative measures to first-generation vaccines to avoid strain on health systems, such as strengthened physical distancing measures, face mask usage, novel treatments, and second-generation vaccines. Expanded vaccination, which includes a booster dose for adults and child vaccination strategies, is projected to have the biggest public health benefit for a highly infective, highly severe VOCs with low immune-evading capacity.

Methods

OpenCOVID individual-based model

OpenCOVID is a stochastic, discrete-time, individual-based transmission model of SARS-CoV-2 infection and COVID-19 disease[11]. The model simulates viral transmission between infectious and susceptible individuals that come in contact through an age-structured, small-world network. The probability of transmission in each exposure is influenced by the infectiousness of the infected individual, the immunity of the susceptible individual (acquired through previous infection and/or vaccination), and a background seasonality pattern (reflecting a larger proportion of contacts being in closer contact indoors with cooler temperatures). Infectiousness is a function of viral variant infectivity and time since infection (which follows a gamma distribution peaking approximately at the time of symptom onset). Once infected, a latency period is followed by a pre-symptomatic stage, after which the individual can experience asymptomatic, mild, or severe infection. Severe cases can lead to hospitalisation, ICU admission, and ultimately death. Recovery after infection leads to the development of immunity. This immunity is assumed to wane over time and can be further reduced if exposed to a novel variant with immune-evading properties. The model has the capacity to represent a number of containment measures, including non-pharmaceutical interventions such as physical distancing and facemask usage, testing strategies such as test/diagnose isolate, mass-testing, and contact tracing, and also pharmaceutical interventions such as vaccination and treatment.

Reproducibility

Detailed model descriptions and model equations are described in Shattock et al. (2022)[11]. Open access source-codes for the OpenCOVID model of all analyses presented in this study are publicly available at https://github.com/SwissTPH/OpenCOVID/tree/manuscript_december_2021/src, all figure code is also available at https://zenodo.org/record/6532404[13].

Vaccination

In this analysis, we simulate the impact of mRNA vaccines Pfizer/BioNTech and Moderna, which together make up 78% of the total number of doses secured in Europe at the time of writing[14]. Fully susceptible, partially susceptible, and infected individuals not in hospital are considered eligible to receive a vaccine. Vaccines have a two-fold effect; first, they provide protection against new infections through the development of immunity (90% reduced susceptibility). Second, once infected, vaccines reduce the probability of developing severe symptoms, leading to a 95% reduction in severe disease, impacting hospitalisations, ICU admissions, and deaths. We assume vaccination does not impact the probability of onward transmission once the individual is infected. Details regarding targeted vaccination groups and assumed durations between doses and vaccine efficacies are described in the Supplementary Methods sections 2.2 ‘Vaccine rollout’ and  2.3 ‘Vaccine-induced immunity profile’. Associated waning immunity profiles after infection and vaccination are described in Supplementary Methods sections 2.3 ‘Vaccine-induced immunity profile’ and 2.4 ‘Infection-induced immunity profile’, and Supplementary Fig. 1.

Variant properties

Delta (B.1.617.2) is assumed to be the dominant transmission variant when Omicron (B.1.1.529) emerged. A full factorial range of Omicron properties was considered: 1) infectivity (transmission multiplication factor per exposure, relative to Delta), ranging from 0 to 2, 2) disease severity (multiplication factor per infection, relative to Delta ranging from 0 to 2), and 3) immune evasion capacity from 0 to 100%. In the model, once infected with the new variant, severity influences the chance to manifest severe symptoms, rather than experiencing mild or no symptoms. One hundred per cent immune evasion means fully evading any previously naturally or vaccination-acquired immunity, making an individual fully susceptible to infection with a new variant. The immune evasion property is assumed to only relate to the individual’s previously acquired immunity (their susceptibility), thus only impacting the rate of new infections and not influencing severity once infected. Full model details on the variant properties are described in Supplementary Information section 2.5 ‘Effect of variant properties and vaccination on prognosis’, alongside probabilities of immunologically naïve infected individuals developing the severe disease in Supplementary Table 1.

Model initialisation

All model simulations were designed to be pseudo-representative of a European setting with simulations starting at the beginning of December 2021. We assume 30% of the population have been previously infected with SARS-CoV-2 over a 630-day period (representing an epidemic outbreak in Europe in March 2020). We assume the effective reproduction number () on 1 December 2021, is equal to 1.2. This represents an average scenario of increasing case numbers across Europe at the start of the winter period when the Omicron variant first emerged. This level of is lower than levels in some European countries with strongly increasing cases as of early December, however it is higher than for those that implemented strong NPIs before December 1st. The average number of daily contacts required to achieve an of 1.2 inherently considers any non-pharmaceutical interventions in place at the beginning of the winter period in Europe prior to the emergence of Omicron. Seasonality is assumed to follow a cosine function, with a peak in seasonal infectivity occurring 6 weeks from model initialisation (representing mid-winter, see Supplementary Fig. 2).

Analyses

For the full range of variant properties specified, we simulated two vaccination scenarios from the introduction of the new variant to six months into the future. For the first scenario, no future vaccinations are implemented. The second scenario is identical up to 1 December 2021, but simulates expanded vaccination with first-generation vaccines administered as third-doses in adults (six months after the second-doses) and scale-up of first- and second-doses in 5-17-year-olds. Supplementary Table 2 provides the details of the two vaccination scenarios by risk group. Each simulation provided the relative prevalence of Omicron over the next 6 months compared with Delta, as well as daily and cumulative numbers of new SARS-CoV-2 infections, conservative maximum hospital occupancy, and COVID-19-related deaths. To reflect the element of chance that naturally occurs in transmission dynamics, 10 random stochastic simulations were performed per scenario for which we present the mean. In this analysis, we do not explore the effect of varying NPI intensity over time. As such, our analysis reflects predicted disease dynamics and public health burden in the absence of strong NPIs, such as lockdowns.