Abstract
Non-pharmaceutical interventions (NPIs) have been implemented intermittently for more than a year in most countries to mitigate the COVID-19 epidemic. In France, as the vaccination campaign progressed in mid-2021, the government began removing major restrictions like lockdowns and curfews. However, "barrier gestures"—social distancing, mask-wearing, and hand washing—remained recommended. This study utilizes an age-structured compartmental SEIR model to estimate if these barrier gestures can be safely relaxed without causing a resurgence of severe infections. The model accounts for SARS-CoV-2 waning immunity, vaccination, and the increased transmissibility of variants of concern (Delta). Calibrated on French surveillance data, the study evaluates various relaxation scenarios from August to December 2021. Findings suggest that maintaining barrier gestures is essential to avoid exceeding healthcare capacities, and overcoming vaccine hesitancy is the critical factor for any safe return to normalcy.
Introduction
As global vaccination efforts reached significant milestones in 2021, public health officials faced the challenge of determining when and how to lift restrictions. In France, the transition from strict lockdowns to a reliance on individual "barrier gestures" raised questions about the sustainability of epidemic control. This research addresses the risk of relaxing these low-level NPIs by incorporating two critical factors often overlooked in earlier models: the waning of immunity (from both natural infection and vaccination) and the high transmissibility of the Delta variant. The goal was to quantify the impact of population compliance with these gestures on future ICU admissions.
Methods
The researchers developed an age-structured compartmental SEIVS (Susceptible-Exposed-Infectious-Vaccinated-Susceptible) model. Unlike simpler models, this approach treats immunity as a continuum, defining four distinct states of susceptibility based on the severity of previous infections and vaccination status.
A central parameter in the model is the k-value, which represents population compliance with barrier gestures. A $k$ of 1 signifies a return to pre-pandemic contact behavior, while lower values indicate higher compliance with distancing and masking. The model was calibrated using national French databases, including SI-DEP (infection surveillance), SI-VIC (hospitalization/ICU data), and VAC-SI (vaccination records). To account for biological uncertainty, the authors modeled immunity waning over both 3-year and 9-year timescales.
Key Insights and Results
The Impact of Waning Immunity and Variants
The study found that the duration of immunity is a primary driver of epidemic dynamics. If immunity wanes over 3 years instead of 9, the predicted ICU peak increases by approximately 86%. Furthermore, the Delta variant's increased transmissibility (estimated at 97% higher than the original strain) significantly raised the threshold for herd immunity, making continued NPI compliance more vital than in previous waves.
Relaxation Scenarios and ICU Capacity
The model projected the incidence of severe infections (compartment $I_4$) under various relaxation schedules. As shown in Figure 6, any significant relaxation of barrier gestures (moving toward $k=1$) between August and December 2021 was predicted to cause a resurgence that would likely overwhelm France's national ICU capacity of approximately 5,000 beds.
The Role of Vaccination and Boosters
Vaccine hesitancy emerged as the most significant bottleneck. The simulations indicated that a safe, full relaxation of measures would only have been possible if 100% of the eligible population were vaccinated. While booster campaigns for the elderly (75+) were shown to reduce ICU peaks by roughly 16%, they were insufficient on their own to allow for the complete abandonment of barrier gestures across the broader population.
Discussion and Limitations
The study concludes that maintaining barrier gestures was essential for France during the 2021 vaccination rollout to prevent healthcare system collapse. The findings emphasize that NPIs remain a necessary complement to vaccination until near-universal coverage is achieved.
Limitations of the model include the lack of explicit seasonality (which might underestimate winter peaks) and the assumption of uniform vaccine efficacy against both infection and severe disease. Future modeling should incorporate real-world data on newer variants, such as Omicron, and more granular data on booster effectiveness to further refine the relationship between behavioral compliance (the k-value) and epidemic control.
Summary updated