Access and health: getting COVID-19 vaccines to the neighborhoods that need them
When there are not enough doses to go around, who gets served first, and what does the shortage cost?
Every number here comes from a published SEAR Lab paper or from its public code, listed at the bottom with the table or sentence each one comes from. Values printed in the paper and values regenerated from the public repositories are marked differently. Where the paper has an error, the demo says so. Cite the paper, not this page.
1. Where the pandemic hit: Houston by ZIP code
The project mapped COVID-19 cases and deaths across Houston-area ZIP codes to see where the burden and vulnerable communities overlap. Switch layers, pick a region, or look up a ZIP code. Counts are totals for each ZIP code from one snapshot; they are not rates, and nothing here describes individual people.
Note on coverage. The paper's model uses 97 ZIP codes in Harris County (77002–77099, Table 2). The project's public GIS covers a wider area: 147 ZIP codes in eight regions, with 49,365 confirmed cases and 487 deaths in the snapshot. The map shows the GIS as published; the date of its snapshot is not recorded.
The paper also scores ZIP codes with a Community Health Index (health, social vulnerability and active cases). That index is built from data the Houston Health Department shared and is not in the public repository, so it is not mapped here.
2. More vaccine, or smarter distribution?
A network model shipped doses from 5 hubs through 278 providers to about 654,000 people over 60 and health care workers, at the lowest cost, with a penalty for every person left unserved. The paper ran eight scenarios: the supply available then and two, three and four times it, each shared equally or prioritized to the most vulnerable communities.
Correction. The paper says equal distribution at the supply available then served 32% of the target group. Its own penalty figures ($34M at that supply, $22M at twice it) are only consistent with about 26%: doubling supply from 32% would have cut the penalty to about $18M, not $22M. The demo shows the printed 32% with this note, and the public reconstruction, which matches the printed penalties to within 0.4%, uses 26%.
Bars are the public repository's reconstruction of the eight scenarios; circles are values the paper prints as numbers (most of its results appear only as bar charts). The reconstruction orders ZIP codes by the CDC Social Vulnerability Index as a public stand-in for the Community Health Index, and prices every unserved person at $70 (two Pfizer doses then), the top of the paper's $35–$70 range. Its "most vulnerable third of ZIP codes" is not the same group as the paper's ten prioritized communities.
What this shows
In early 2021 COVID-19 vaccines were scarce, and older adults and communities of color had been hit hardest. Getting doses to them is partly a logistics problem, especially the "last mile" from clinics and pharmacies to the people who need them.
The map shows why place matters. Cases were not spread evenly: some ZIP codes had over a thousand confirmed cases in the snapshot and others a handful. A supply chain that treats every ZIP code alike will not match where the need is.
The scenarios show the trade. Step the supply up from 1× to 4×: the shortage penalty falls to zero, while transport cost rises because more people are reached. Switch to "Both costs" to see that transport, even including the last mile, is tiny next to the cost of leaving people unserved. Then switch between equal and prioritized at 1×: total coverage barely moves (about a third, or a quarter by the corrected figure), but the paper finds that prioritizing by its health index serves the most vulnerable communities at close to 100%.
Behind this demo
Sources
- Jones, E.C., Azeem, G., Jones, E.C., Jr., Jefferson, F., Henry, M., Abolmaali, S., & Sparks, J. (2021). Understanding the last mile transportation concept impacting underserved global communities to save lives during COVID-19 pandemic. Frontiers in Future Transportation, 2, 732331. doi:10.3389/ffutr.2021.732331. Table 2 (assumptions), Table 3 (scenarios), and the Results text for the printed penalties and service levels. CC BY.
- sear-labs/covid-optsc-ffutr-2021 (public code for the paper's supply chain model, MIT licence).
github.com/sear-labs/covid-optsc-ffutr-2021,
doi:10.5281/zenodo.22309107.
results/reconstructed_scenarios.csv, and the prioritized rows regenerated with itsscripts/make_figures.pyondata/derived_paper_instance.csv. - sear-labs/houston-covid-gis (public code for the paper's Houston maps, MIT licence).
github.com/sear-labs/houston-covid-gis,
doi:10.5281/zenodo.22309109.
data/geometry/covid_regions.gpkg, per-ZIP totals only; projected and simplified offline (_build/build_zips.py).