Thread by @drdrtsai

1/ Our paper finally 🥴 published in @PLOSMedicine: "Social distancing to slow the US COVID-19 epidemic: Longitudinal pretest–posttest comparison group study", led by Mark Siedner (@MGHMedicine @MonganInstitute @mgh_id) & @harlingg (@UCLGlobalHealth) journals.plos.org/plosmedicine/a…

2/ If the content looks familiar, it's because we posted our @medrxivpreprint 4 months ago... medrxiv.org/content/10.110…

3/ In brief, we collected data on statewide social distancing measures that were implemented Jan 21-Mar 30 (preprint v1) when median statewide #COVID19 (reported) epidemic size was 35 cases (IQR 17-72) and avg daily case growth rate was 30% (doubling time 3.3 days)

4/ The first social distancing measures to be implemented were typically cancellations of public events and... school closures 😖😫😩😭😭😭😭😭

5/ Within one incubation period the avg daily #COVID19 case growth rate began declining by 0.9%/day such that avg daily case growth rate had declined to 12.7% by 3wk after enactment of the first social distancing measures (doubling time 7.9 days)

6/ During peer review we extended the data collection through May; +some sensitivity analyses and an event-study specification; and + #COVID19-attributed deaths as an outcome, showing that social distancing also slowed the growth in deaths with a greater lag

7/ Altogether we estimate that statewide social distancing measures reduced the total # reported #COVID19 burden by ~600,000 cases by 3wk after implementation. Earlier implementation probably would have saved even more lives.

8/ The event-study specification yielded qualitatively similar findings

9/ Reporting of race/ethnic data has been low priority in this epidemic, unfortunately. If we had had such data, probably would have seen large disparities owing to how racism serves to condition neighborhood residence, multigen households, WFH privilege, paid sick leave, etc etc

10/ We also looked at stay-at-home/shelter-in-place/"lockdown" orders, which were generally implemented a median of 11 days after the first statewide social distancing measures were already in place and the avg daily case growth rate was already declining

11/ We estimated no add'l effect of lockdowns on cases/deaths, but hard to disentangle the unique associations w/ lockdowns vs. the initial measures.
.
(I am skeptical of analyses claiming to isolate the specific effectiveness of a single type of measure, like school closures)

12/ Other biases potentially toward null: 1) some jurisdictions initiated social distancing in response to a worsening local epidemic, 2) some measures could violate SUTVA w/ intrastate spillovers, 3) testing intensified over time in some jurisdictions.

13/ Also potential confounding by political orientation of citizenry that could be assoc both w/ govt propensity (or lack thereof) to issue orders and w/ behavior that could lead to greater #COVID19 case growth, eg sciencedirect.com/science/articl…

14/ DANG IT Mark was gonna be on @DonLemonCNN_USA tonight but someone decided to pick a veep today and Mark got bumped... womp womp 🥴
.
(I'm still gonna fall in line-- but y'all best make @ewarren *something* in the cabinet...)

15/ Could be some biases _away_ from the null, namely: in some jurisdictions, social distancing behavior began to change even prior to govt orders, eg analysis from Deforche, @vercauteren_jur & @VandammeAm: medrxiv.org/content/10.110…

16/ Big scrum to publish back in March. We were the slowest 😭😭😭 (preprint Apr 8, published... Aug 11 🥉). Now people have *lots* of choices to cite! eg. Hsiang et al. preprint Mar 27, published Jun 8 in @nature... nature.com/articles/s4158…

17/ ...Flaxman & @imperialcollege COVID Response Team et al., preprint Mar 30, published Jun 8 in @nature. Honestly, by this point we were just sitting around watching everyone publish... nature.com/articles/s4158…

18/ ...@FriedsonAndrew et al, out Apr 20, published Aug 3 in @InquiryWEAI... onlinelibrary.wiley.com/doi/abs/10.111…

19/ ...@Courtemanche_CJ @aaronyelowitz et al, out Apr 29, published 🥳 May 14 🥇 in @Health_Affairs (I have to say... they've been pretty darn good at flogging reviewers to get those reviews in on time) healthaffairs.org/doi/10.1377/hl…

20/20 We have more coming out on what happens when states relaxed their social distancing measures-- stay tuned (hopefully it won't be another 4 mo)