ABSTRACT
Background: Non-pharmaceutical interventions (NPIs) and vaccines have been used by many countries to manage the dynamics of the COVID-19 pandemic. Despite numerous studies, considerable uncertainty remains about the effects of these public health interventions due to data quality issues and methodological challenges to estimating effects. However, producing accurate and precise estimates of the effects of these interventions is of utmost importance for the preparedness of any new epidemic. Method(s): We developed a population-based mechanistic compartmental model that includes the effect of NPIs on SARS-CoV-2 transmission and the effect of vaccination on the transmission and the rate of hospitalization. Our statistical approach estimated all parameters in one step, thus accurately propagating uncertainty, and representing spatial heterogeneity. We fitted the model to all available epidemiological data (hospital admissions and occupancy, cases, and deaths) from March 2020 to October 2021 in France. Hence, we estimated the time-varying transmission rate, and derived the effect of NPIs through an integrated regression model. We simulated counterfactual scenarios of the interplay of NPIs and vaccine availability and rollout with the same model. Result(s): We found that the first lockdown reduced transmission by 84% (95% CI [83-85]) and was more effective than the second and third lockdowns (reduction of 75% [72-77] and 9% [6-13], respectively). A 6pm curfew was more effective than an 8 pm curfew (transmission reduction of 69% [67-70] vs. 50% [48-53]). School closures had a smaller effect on transmission (15% [12-19]). By the end of the study period, the protection conferred by vaccines against hospitalization and against infection, considering viral variants and population vaccine coverage, ranged between 69-92% and 29-40%, respectively. In a scenario without vaccines, we predicted 209% (95% PI [34-520]) more deaths and 346% [101-798] more hospitalizations throughout the study period. Conversely, if an effective vaccine had been available after 100 days, 65% [36-80] deaths and 72% [45-84] hospitalizations could have been averted. Conclusion(s): Our results provide reliable effect and uncertainty estimates of each NPI and demonstrate that NPIs and vaccination synergistically reduced COVID-19 transmission, hospitalization, and deaths. This emphasizes the importance of stringent NPIs and a high vaccination rate to prevent further epidemic resurgences and control other emerging respiratory infectious diseases.