Thread by @douglasharris99

Is it safe to reopen schools in-person? Pleased to announce important new evid coming soon from @REACHctred. We think it’s the 1st US study to get causal effects of reopening or to look at COVID health *outcomes*. We do both w near-census of whole country. Below, I describe how…

Almost all prior work focuses on COVID positivity rates/contact tracing. Well known problems with this: rare & unsystematic COVID testing, incomplete contact tracing, & vast majority of transmission don’t result in neg health outcomes--& we care most about outcomes

We use hospitalization data from @Change_HC (covid19researchdatabase.org), which has medical claims on 170 million US residents. We focus on hospitalizations w #COVID pos test &/or COVID symptoms. The data are for every week of 2020 calendar year by county.

We combine hosp data w the vast majority of school districts in the country using data from @educationweek @BurbioCalendar @mchdata. We know start date + opening mode (in-person, remote, hybrid). Having 3 data sources is very helpful on this part.

1st part of strategy looks before-&-after schools reopen in-person. If opening more in-person affects health then should see COVID hosp rise in later weeks, relative to those opening fully remote (there should be a delay).

2nd part of strategy combines above w data on teacher bargaining power, which affects how schools reopened but shouldn’t affect hospitalizations in any other way, espec at specific date of schl reopening. (Econs call this “instrumental variables”).

Combining instrumental var w/ before-&-after is a powerful combo. Addresses the fact that school reopening is far from random (also addresses meas error in schl reopening data). So, we can get causal effects.

We have a great team @Tulane, combining health & educ economists (Engy Ziedan & me) w/ a top epidemiologist of infectious disease (Susan Hassig). This combo of expertise is important. Also, all 3 of us have also been studying #COVID in other ways before this study.

Will share results as soon as we're sure of them. We’re working as fast as we can while also going through some peer review. Possible release as early as next week or right after new year. More updates to come.