COVAX Vaccine Distribution Proof of Concept

World Bank projects are increasingly using geospatial analysis to identify target areas in need of assistance. We believe the COVAX program presents a major opportunity to leverage geospatial analysis to prioritize vaccine distribution. In this note, we introduce a draft methodology to rank districts in Bangladesh according to key metrics: vulnerable population, travel time from potential distribution sites, and COVID-19 case counts. While the exact requirements for vaccine distribution are still uncertain, in this analysis we use international airports as proxies for sites with cold storage facilities.

Data Inputs

• Population and vulnerable population: Using a 1-km population grid from World Pop (2020), we apply the age incident rates from IHME to estimate vulnerable population due to age.
• Airports: International airports are sourced from the Global Infrastructure Map. Smaller airports are retrieved from OpenStreetMap.
• Global friction surface: This layer (Malaria Atlas Project, 2015) contains land-based travel time at the pixel level, enabling rapid and scalable computations of accessibility metrics.
• COVID cases: Sub-national COVID-19 case counts retrieved from MIS DGHS.

Methodology

1. Calculate the travel time between each 1-km grid cell (_origin_s) to each airport location (destinations), storing this information in an origin-destination matrix.
2. For each airport location, estimate the service area or "market shed", that is, the area of origin points for which that airport represents the closest destination.
3. For each district, summarize the number of people and vulnerable people within different travel time thresholds of airports.

Output Examples

Map of travel time to nearest airport

The following summary present two alternatives to summarize the demand for a vaccine and the difficulty of distribution. Each column represents the number of vulnerable people that can be accessed from an airport within a given hourly time threshold. The information is currently organized by airport and serviced districts (Figure 2), or district and nearest airport (Figure 3). A next step will be to integrate the estimated air travel time between international and domestic airports.

Summary Output Table

Admin 2	Closest airport (mode)	Airport type	Pop < 1hr	1-2 hr	2-3 hr	3-4 hr	4-5 hr	COVID Cases	% children who received all 8 basic vaccinations within Admin 1
Dhaka	Tejgaon Airport	domestic	248,208	8,099	32	0	0	109,183	87.4
Chittagong	Shah Amanat International Airport	international	120,056	25,337	2,062	18	0	19,708	83.3
Bogra	Bogura Airport	military	64,517	12,651	1,366	38	0	7,798	83.6
Comilla	Comilla Airport	STOLport	97,742	20,593	40	0	0	7,670	83.3
Faridpur	Barisal Airport	domestic	4,121	38,496	235	0	0	7,268	87.4
Sylhet	MAG Osmani International Airport	international	50,971	24,197	3,928	191	15	7,126	61.1

Map of Vaccination Rates