Pretty interesting paper, although I'm not sure it's the one OP was referring to. It's a pretty short paper with a whole pile of graphs at the end, so I encourage people to read it themselves.
Quick notes from skimming:
- Treats it as an optimization problem for various levels of vaccine efficacy (10-100%), availability(10-100%), spread rate (R0 in {1.5, 2, 2.5, 3}), susceptibility to infection and symptomatic infection per age group.
- Four objective functions: symptomatic infections, deaths, non-ICU hospital usage at peak, and ICU usage at peak
- 5 age buckets: 0-19, 20-49, 50-64, 65-74, 75+
- Assumes 20% of population has immunity, and immunity lasts for a year.
- At higher efficacy and availability levels, there are some odd shifts to optimal vaccine distribution strategies. It's not strictly "oldest first" or "youngest first", there are some weird discontinuities in the middle buckets as well.
I had some questions about the wide 20-49 age bucket but it appears that it comes from a CDC planning scenario. It does look like that various curves start accelerating sharply past 50 or higher, so I guess treating the 20-49 group as one reasonably low risk group could be reasonable.
Yes, there could be many different scenarios; when vaccine supply is limited, vaccinate those at risk of death. BioNTech claims their vaccine efficacy is consistent across age groups, but we won't know for sure until large clinical trials are completed. Also, older people are less willing to be vaccinated earlier. More variables to include in the model...
- Treats it as an optimization problem for various levels of vaccine efficacy (10-100%), availability(10-100%), spread rate (R0 in {1.5, 2, 2.5, 3}), susceptibility to infection and symptomatic infection per age group.
- Four objective functions: symptomatic infections, deaths, non-ICU hospital usage at peak, and ICU usage at peak
- 5 age buckets: 0-19, 20-49, 50-64, 65-74, 75+
- Assumes 20% of population has immunity, and immunity lasts for a year.
- At higher efficacy and availability levels, there are some odd shifts to optimal vaccine distribution strategies. It's not strictly "oldest first" or "youngest first", there are some weird discontinuities in the middle buckets as well.
I had some questions about the wide 20-49 age bucket but it appears that it comes from a CDC planning scenario. It does look like that various curves start accelerating sharply past 50 or higher, so I guess treating the 20-49 group as one reasonably low risk group could be reasonable.