Back

Evaluating the impact of non-pharmaceutical interventions for SARS-CoV-2 on a global scale

Esra, R. T.; Jamesion, L.; Fox, M. P.; Letswalo, D.; Ngcobo, N.; Mngadi, S.; Estill, J. G.; Meyer-Rath, G.; Keiser, O.

2020-08-05 infectious diseases
10.1101/2020.07.30.20164939 medRxiv
Show abstract

In the absence of a viable pharmaceutical intervention for SARS-CoV-2, governments have implemented a range of non-pharmaceutical interventions (NPIs) to curb the spread of infection of the virus and the disease caused by the virus, now known as COVID-19. Given the associated social and economic costs, it is critical to enumerate the individual impacts of NPIs to aid in decision-making moving forward. We used globally reported SARS-CoV-2 cases to fit a Bayesian model framework to estimate transmission associated with NPIs in 26 countries and 34 US states. Using a mixed effects model with country level random effects, we compared the relative impact of other NPIs to national-level household confinement measures and evaluated the impact of NPIs on the global trajectory of the COVID-19 pandemic over time. We observed heterogeneous impacts of the easing of restrictions and estimated an overall reduction in infection of 23% (95% CI: 18-27%) associated with household confinement, 10% (95% CI: 1-18%) with limits on gatherings, 12% (95% CI: 5-19%) with school closures and 17% (95% CI: 6-28%) with mask policies. We estimated a 12% (95% CI: 9-15%) reduction in transmission associated with NPIs overall. The implementation of NPIs have substantially reduced acceleration of COVID-19. At this early time point, we cannot determine the impact of the easing of restrictions and there is a need for continual assessment of context specific effectiveness of NPIs as more data become available.

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.