Thread by @DiseaseEcology

Who should be vaccinated next?
1st batch Pfizer/BioNTech vaccine is shipping & will go to HCW + nursing homes as it should. But next tier is debated (essential workers? elderly? pre-existing morbidities?).
Model suggests elderly for decreasing deaths but more info needed
Thread

https://twitter.com/DanLarremore/status/1336469777129242624

FDA & CDC have given green light for Pfizer/BioNTech vaccine. Supplies are very limited initially & transmission is raging so it's important to choose carefully in who to vaccinate first. How can we determine what is best? Mathematical models!

A very nice paper by @bubar_kate @DanLarremore @StephenKissler @sarahcobey @yhgrad @mlipsitch tries to determine which age group should be prioritized to minimize deaths, years of life lost, and total incidence.
medrxiv.org/content/10.110…

There are a few key unknowns (most importantly, does vaccine reduce infection/infectiousness or just symptoms?

https://twitter.com/DiseaseEcology/status/1336446219283955714

) so paper has to examine multiple possibilities & see if recommendations are robust to uncertainty.

Paper compares 5 age group prioritizations (<20; 20-49; 20+; 60+, uniform) under 3 scenarios: 1) 1% of pop vaccinated per day w/ R0=1.3; 2) 1%, R0=2.6; 3) vaccination before transmission w R0=2.6). Fig shows 2 outcomes: reduction in deaths, infections w/ no age-dependent efficacy

Targeting 60+ is always best or nearly best for reducing COVID-19 deaths under these scenarios;
Reducing infections is best achieved by targeting 20-49 (possibly b/c contact & transmission is higher in 20-49 than older groups? paper doesn't say or show contact matrix).

Answer is simple then, right? Not quite. Of the 3 scenarios, scenario 1 R0=1.3 is much closer to reality & needs to be adjusted further: namely, R0/Rt in most of the world is presently at or below 1.3 (epiforecasts.io/covid/posts/gl…). Figs show a few states, countries; others same.

Why does this matter? Isn't allocation to elderly robust to variation in R0?
It's not clear. Can @bubar_kate @DanLarremore update models & show new scenarios w/ R0<1.3?

(Note, can replace some current scenarios in paper: data shows vaccine efficacy doesn't depend on age;
DOI: 10.1056/NEJMoa2034577)

Specifically, what does model suggest for R0 = 1.05-1.3 (where most of world is now?). Fig suggests that targeting young *might* be better. Also, what if vaccine supply = 70% rollout rate match projected vaccine availability?

I'm *guessing* model will still suggest targeting 60+ is best for reducing mortality. If so, plan to prioritize others essential workers before elderly will lead to many additional deaths. Obviously can consider more than deaths but additional deaths likely very large.

(Note: I'm also puzzled why NAS prioritization puts co-morbidities higher than elderly. I'm pretty sure 75 yr old has much higher (10x? 50x) chance of death than 30 yr old w/ 2 pre-existing conditions. If I'm wrong - link to paper showing opposite). Also, age much easier to know.

Also, @bubar_kate @DanLarremore what is best for reducing hospitalizations & ICU cases? I'd argue these are much more important than total infections. (Note: age specific estimates for hosp, ICU available in 10.1126/science.abc3517
doi.org/10.1016/S1473-…)

Key additional point: Model makes it clear that prioritizing those w/out previous infection can save many lives, especially in pops w/ large frac previously infected. Would need reliable rapid test to screen for robust immune response, but benefit is big.

Conclusion: Paper suggests prioritizing elderly & seronegative people best for reducing deaths.
But need more analyses to examine R0<1.3 & which strategy minimizes hospitalizations, ICU. Also, still need actual data on vaccine effects on infectiousness!(

https://twitter.com/DiseaseEcology/status/1336446219283955714

)