CA has plan for staged re-opening of businesses. There are several issues w/ key criteria: some disincentivize good testing behavior; others are so stringent they are unlikely to be met by most populated counties for the foreseeable future (none are now: latimes.com/california/sto…).
Criteria (p9: gov.ca.gov/wp-content/upl…):
1) no deaths in prev 14 days & <1 case/10k ppl;
2) testing capacity 1.5/1000 ppl;
3) 15 contact tracers/100k ppl &15% of homeless;
4) support for essential workers inc. PPE
1) no deaths in prev 14 days & <1 case/10k ppl;
2) testing capacity 1.5/1000 ppl;
3) 15 contact tracers/100k ppl &15% of homeless;
4) support for essential workers inc. PPE
Criteria 3&4 are useful; minor tweaks could help tailor CT capacity as needed;
#2 testing capacity is good start, but key aspect is turn-around time for results; testing capacity that returns results >72 hrs later is nearly worthless for effective contact tracing.
#2 testing capacity is good start, but key aspect is turn-around time for results; testing capacity that returns results >72 hrs later is nearly worthless for effective contact tracing.
Criteria #1 is most problematic. Deaths reflect transmission that occurred avg~27d ago (@jLewnard medrxiv.org/content/10.110…), so a death from 14d ago could reflect transmission 41d ago. That makes a very long time b/w action and consequence.
The same is true in reverse: a rapid rise in transmission might not be picked up as deaths for 3-4 wks, and even cases are lagging indicator: people seek hospital care ~13.5d after infection (@jLewnard medrxiv.org/content/10.110…)
The case count criteria is most problematic b/c it disincentivizes increased testing!!! If a county increases testing and starts detecting mild and asymptotic cases this is a good thing & should be encouraged.
When testing capacity was low, 1 case/10k people likely reflected real incidence of ~5-30/10k. But w/ increased testing underestimate could be much lower&key is to incentivize testing!
Better criteria than static incidence of cases would be decrease (i.e. R<1) in # hospitalized, or, if testing is high and w/ consistent criteria, falling # of cases. If testing includes asymptomatic people delay b/w infection & detection would be much shorter than cases/hosp.
Real-time estimation of R is now relat. easy to implement
epiforecasts.io/covid/posts/na…
w/ EpiEstim R package cran.r-project.org/web/packages/E…, academic.oup.com/aje/article/17…
Slight delays b/c of estimation methods & reporting/detection of infections but better than using deaths from up to 41d ago
epiforecasts.io/covid/posts/na…
w/ EpiEstim R package cran.r-project.org/web/packages/E…, academic.oup.com/aje/article/17…
Slight delays b/c of estimation methods & reporting/detection of infections but better than using deaths from up to 41d ago
CA re-opening criteria should be based trends in recent (not distant past) infection, and should not disincentivize testing.
Can folks working in other states/countries (or who understand basis for CA criteria) share criteria being used for relaxing social distancing? @jLewnard @taaltree @bolkerb @joshuasweitz @michaelmina_lab @mlipsitch @SRileyIDD @NAChristakis @MackayIM
@jzelner What criteria is MI using?
