The impact of spatial connectivity on NPIs effectiveness

Abstract

Background. France implemented a combination of non-pharmaceutical interventions (NPIs) to manage the COVID-19 pandemic between September 2020 and June 2021. These included a lockdown in the fall 2020 - the second since the start of the pandemic - to counteract the second wave, followed by a long period of nighttime curfew, and by a third lockdown in the spring 2021 against the Alpha wave. Interventions have so far been evaluated in isolation, neglecting the spatial connectivity between regions through mobility that may impact NPI effectiveness. Methods. Focusing on September 2020 - June 2021, we developed a regionally-based epidemic metapopulation model informed by observed mobility fluxes from daily mobile phone data and fitted the model to regional hospital admissions. The model integrated data on vaccination and variants spread. Scenarios were designed to assess the impact of the Alpha variant, characterized by increased transmissibility and risk of hospitalization, of the vaccination campaign and alternative policy decisions. Results. The spatial model better captured the heterogeneity observed in the regional dynamics, compared to models neglecting inter-regional mobility. The third lockdown was similarly effective to the second lockdown after discounting for immunity, Alpha, and seasonality (51% vs 52% median regional reduction in the reproductive number R0, respectively). The 6pm nighttime curfew with bars and restaurants closed, implemented in January 2021, substantially reduced COVID-19 transmission. It initially led to 49% median regional reduction of R0, decreasing to 43% reduction by March 2021. In absence of vaccination, implemented interventions would have been insufficient against the Alpha wave. Counterfactual scenarios proposing a sequence of lockdowns in a stop-and-go fashion would have reduced hospitalizations and restriction days for low enough thresholds triggering and lifting restrictions. Conclusions. Spatial connectivity induced by mobility impacted the effectiveness of interventions especially in regions with higher mobility rates. Early evening curfew with gastronomy sector closed allowed authorities to delay the third wave. Stop-and-go lockdowns could have substantially lowered both healthcare and societal burdens if implemented early enough, compared to the observed application of lockdown-curfew-lockdown, but likely at the expense of several labor sectors. These findings contribute to characterize the effectiveness of implemented strategies and improve pandemic preparedness. Keywords. COVID-19, NPIs, modeling, curfew, restrictions.

Planning and adjusting the COVID-19 booster vaccination campaign to reduce disease burden

“…These include spatial differences in variants' penetration 48,49 and the application of social distancing restrictions in our scenario analyses. We note however that non-pharmaceutical interventions were always applied nationwide in France, with limited exceptions 50 .…”

Pandemic preparedness: On the efficacy of non‐pharmaceutical interventions in COVID‐19 and about approaches to predict future pandemic viruses

“…A later curfew starting at 8 pm was less effective, but both types of curfew lost efficacy with time of application, suggesting pandemic control fatigue in the population. Their calculation showed that stop‐and‐go lockdowns (intercut with periods of no restrictions) could have achieved a 40% reduction of hospitalizations for a similar number of effective days under restrictions (Sabbatini et al., 2024). The lower efficacy of curfew (5% case reduction) compared to lockdown (90% case reduction) was also observed in Luxembourg (Thompson & Wattam, 2021).…”