Brooke Nichols is a health economist and infectious disease mathematical modeler at Boston University’s School of Public Health. Her research focuses on HIV treatment and prevention, and more recently, on resource allocation and testing strategies for COVID-19.

Across the United States and the world, governments have utilized non-pharmaceutical interventions (NPIs) to limit the spread of SARS-CoV-2 (the virus that causes COVID-19). The basic principle behind all NPIs is to reduce the mixing of individuals, which in turn reduces how many people one infected person will go on to infect. Research has shown NPIs can be effective in slowing the spread, but as we head into the winter in the northern hemisphere and try to make the most of coming vaccines, understanding which NPIs work best and how well they work will be crucial.

A research article published this week in Science by Oxford University’s Jan M. Brauner and others looked at the relative effectiveness of eight different NPIs implemented between January and the end of May in 41 mostly European countries. It used the differences in which countries implemented which NPIs at which time and the case and death counts in each country to estimate how effective each NPI was.

As with other studies looking at the relative effectiveness of NPIs, the main difficulty is parsing out the effect of each intervention independent of one another given that there was frequently simultaneous implementation of multiple interventions. This study also didn’t assess all factors that could affect viral spread, like mask wearing. Still, the authors were able to use advanced statistical analyses to account for much of this uncertainty and estimate the effects of individual and combined NPIs.

They found that the most effective interventions in order of relative effect were: limiting all gatherings to <10 people, closing schools and universities, limiting gatherings to <100 people, and closing most non-essential businesses (although they found this last one to be only slightly more effective than more targeted closures of bars, restaurants and nightclubs). Interestingly, they found stay-at-home orders offered limited benefits when added on top of a combination of the other NPIs. The authors also noted that a key limitation of their analysis was the inability to parse out whether schools themselves were the source of ongoing transmission, or whether the closure of schools acted as a signal to society of the gravity of the situation and affected behavior of the public more generally. (To understand the combined effect of interventions, you can check out the calculator that the authors have made available here.) 

While, as the authors note, these data should be interpreted cautiously, hopefully this analysis and others like it can help guide policymakers at local, state and national levels to implement evidence-based NPIs through the tough winter months ahead.