Infections averted using our estimates for particular intervention analysis

[athowes]

• Would like to show how many infections might be averted by using our sub-national estimates rather than say national level estimates
• Alternative view would be how much higher the allocative efficiency could be [with our estimates]
• Requires choosing an intervention for illustrative purposes. Should try to choose something somewhat realistic if possible. Ask if anyone in group has expertise in a particular intervention, such that they could provide illustrative numbers
• Inspired by Allen's plot (see below). Would like to produce something similar with LHS (standard) being no district level targetting and RHS (ours) being geographic targetting

[athowes]

This is the first attempt at connecting better to policy by calculating the number of infections you can avert. This is stratifying incidence by everything (age, district, risk group) and then targetting the highest incidence first. Assume that everyone can be reached (at equal cost) and the intervention reduces incidence by 50%, then how many infections can you avert. The next versions of this plot will have facet by targetting e.g. only based on risk group, only based on district, only based on age, only based on two factors and then compare to the three factor approach. They should all be less steep than this one. Can also change the assumptions for how the intervention works. For example maybe it's not possible to reach everyone, or there are some increasing marginal costs to saturate a group.

I'd also like to think about how to present the ordering of where you should target first visually. One way would be to have a 2D (one for age, one for risk group) grid of maps. Then each district on the map (representing a district, age, risk group combination) is coloured according to it's incidence rank. I wonder if there is a way to do this that would be more concise than this.

[athowes]

This is essentially done. Improvements could be to add uncertainty (somehow) or add more complex intervention assumptions.