Nigeria ITN Urban Microstratification  

A ward-level workflow for prioritizing insecticide-treated net (ITN) distribution in Nigeria, combining an 80% urban population rule with PfPR (malaria parasite prevalence) ranking to identify which urban wards should receive nets and which are lower priority.

Built for NMEP (National Malaria Elimination Programme) HBHI 2026 post-hoc stratification.

Why

ITN campaigns traditionally distribute nets at LGA (district) level. In rapidly urbanizing states, this over-nets low-transmission urban cores while under-resourcing high-transmission peri-urban/rural fringes. This workflow pushes the decision down to the ward level and adds two layers of nuance that a flat urban/rural split misses:

1. Not every "urban" ward has low transmission — some still have moderate/high PfPR.
2. Even a correctly prioritized ward may be operationally hard to reach.

Workflow

Step	Script	What it does
1	scripts/01_create_urban_wards.R	Classifies each ward as Urban or Rural using the population-weighted share of urban pixels (GHS-SMOD ≥ 21) inside the ward polygon. A ward is Urban if ≥ 80% of its population sits in urban-classified pixels.
2	scripts/02_pfpr_ranking_itn_priority.R	Extracts population-weighted mean PfPR per ward (MAP/INLA raster), bands it into WHO/NMEP endemicity categories (very low / low / moderate / high), and ranks urban wards by PfPR within each state.
2b	scripts/02_pfpr_ranking_itn_priority_unweighted.R	Sensitivity variant of step 2 — same logic, but PfPR is a simple areal mean (no population weighting), and prioritization also checks the top-80%-by-rank cutoff within state.
3	scripts/03_map_itn_priority.R / 03_map_itn_priority_unweighted.R	Renders one map per state showing which urban wards are prioritized vs. deprioritized; rural wards are shown as muted background context.
4	scripts/04_operational_feasibility_accessibility.R	Takes only the wards flagged "Prioritized (urban moderate/high PfPR)" and overlays travel time to the nearest city (MAP/Weiss accessibility raster) to assign a deployment feasibility tier.
5	scripts/05_map_itn_feasibility.R	Maps deployment tier and travel time per state, plus a national overview of prioritized urban wards by tier.

Prioritization rule

Rural ward                                  -> Prioritized (rural), always
Urban ward, PfPR category moderate/high      -> Prioritized (urban moderate/high PfPR)
Urban ward, PfPR category very low/low       -> Deprioritized (urban very-low/low PfPR)

PfPR endemicity bands (population-weighted mean PfPR, ages 2-10):

Band	Range
Very low	< 1%
Low	1–10%
Moderate	10–35%
High	> 35%

Repo layout

scripts/    R scripts, run in numeric order (01 -> 05)
docs/       Workflow diagram (editable SVG) and abstract write-ups
figures/    A curated sample of output maps (priority + feasibility), by state

Inputs (not included in this repo)

The scripts expect the following under a local data/ directory (paths are set at the top of each script and will need to be updated to your environment):

• shapefiles/ — state, LGA, and ward boundaries (GRID3 / NMEP)
• population/allage_population_2025.tif — WorldPop 2025 all-age population
• urban_extent/ghs_smod.tif — GHSL Settlement Model (urban classification)
• malaria/pfpr_INLA_mean.tif — MAP/INLA modelled PfPR
• accessibility/accessibility_to_cities_2015.tif — MAP/Weiss travel time to nearest city

These rasters are large (multi-GB) and not versioned here; see the header comment in each script for the exact expected path and source.

Requirements

install.packages(c("sf", "terra", "tidyverse", "exactextractr", "ggspatial", "scales"))

Outputs

Each script writes a national .gpkg (all wards + attributes) and per-state CSVs, so results can be reviewed either in GIS or in a spreadsheet. See the header comment of each script for the exact output paths.

Status

Prepared as part of NGA ITN Urban Microstratification (HBHI 2026 post-hoc analysis); abstract submitted to ASTMH 2026 (see docs/).