Counts
Cases, deaths, tests, ACT courses, people protected by IRS and domestic expenditure are summed across the countries in view. The count of reporting countries is displayed alongside every aggregate.
Methodology
How every figure on this platform is produced
Each number in the dashboard is traceable to a published source observation for a single country-year. This page documents the pipeline, indicator definitions, aggregation rules, assumptions and limitations so that any analyst can reproduce — or challenge — what is displayed.
- Research Preview v0.1
- 2026-08-04
- 46
- 2000–2024
Scope and unit of analysis
The unit of analysis is the country-year. Nothing is displayed at a finer resolution than the source data supports.
46 sub-Saharan African countries, 2000–2024. Sub-Saharan Africa carries approximately 95% of the global malaria burden, so the platform focuses there rather than diluting comparison across low-transmission regions. WHO publishes at national level only; no sub-national estimate is shown, implied or interpolated.
Data pipeline
Five stages run offline to produce a versioned snapshot. Presentation-time aggregation is deliberately separated from data preparation.
- → Raw country-year observations
Acquisition
Indicator series are retrieved from the WHO Global Health Observatory OData API and the World Bank World Development Indicators API by country and year. Each request is recorded with its indicator code and retrieval date.
- → Harmonised long-format table
Harmonisation
Country identifiers are normalised to ISO-3166 alpha-3, sub-regional and income groupings are attached from World Bank classifications, and each indicator is mapped to a single internal metric key with an explicit unit.
- → Validated series + exclusion log
Validation
Series are checked for duplicate country-years, unit drift, implausible year-on-year discontinuities and percentage values outside 0–100. Failures are logged and excluded rather than corrected silently.
- → Versioned analysis snapshot
Assembly
Validated series are pivoted to fixed year vectors with explicit nulls, joined to annual population denominators and simplified national boundary geometry, and written to a single versioned snapshot bundle.
- → Dashboard, tables, CSV export
Presentation
The application performs aggregation at render time using the documented rules below. No figure is precomputed for display, so any aggregate can be reproduced from the snapshot independently.
Aggregation rules
Most comparability failures in public health dashboards occur here, not in acquisition.
Rates
Incidence per 1 000 at risk and mortality per 100 000 are never averaged across countries. They are recomputed as summed numerator divided by summed population for the countries that actually reported in that year.
Coverage percentages
ITN access, ITN use and IPTp3 are combined as population-weighted means, so a small country cannot move a regional figure as far as a large one.
Missing values
A country-year with no source observation remains null. It is excluded from both numerator and denominator — never zero-filled, carried forward or interpolated.
Indicator definitions
Fifteen indicator series, each with a single documented definition, unit and source classification.
| Indicator | Unit | Source classification | Definition |
|---|---|---|---|
| Estimated malaria cases | cases | WHO GHO (modelled) | Estimated total malaria cases, all species, national. |
| Estimated malaria deaths | deaths | WHO GHO (modelled) | Estimated malaria deaths, all ages, national. |
| Incidence per 1 000 at risk | per 1 000 | Derived | Estimated cases ÷ population at risk × 1 000. |
| Mortality per 100 000 | per 100 000 | Derived | Estimated deaths ÷ population × 100 000. |
| Confirmed cases | cases | WHO GHO (reported) | Cases confirmed by RDT or microscopy. |
| Presumed cases | cases | WHO GHO (reported) | Cases treated on clinical suspicion without confirmation. |
| Reported cases (total) | cases | WHO GHO (reported) | Confirmed plus presumed cases reported by the programme. |
| Suspects tested by RDT | tests | WHO GHO (reported) | Suspected cases tested by rapid diagnostic test. |
| Suspects tested by microscopy | tests | WHO GHO (reported) | Suspected cases tested by microscopy. |
| Cases treated with ACTs | courses | WHO GHO (reported) | Cases treated with artemisinin-based combination therapy. |
| ITN access (modelled) | % | WHO GHO (modelled) | Population with access to an insecticide-treated net. |
| ITN use, previous night | % | Household surveys via WHO | Population sleeping under an ITN the previous night. |
| People protected by IRS | people | WHO GHO (reported) | People protected by indoor residual spraying. |
| IPTp3 coverage in pregnancy | % | Household surveys via WHO | Pregnant women receiving three or more doses of IPTp. |
| Domestic malaria expenditure | US$ | WHO GHO (reported) | Domestic government malaria expenditure, current US$. |
Comparability over time
Time-series visuals are constructed so that visible change corresponds to change in the data.
Map colour breaks are computed once over all years and held fixed, rather than recomputed per frame. A country changing colour during year scrubbing therefore reflects a real movement in burden, not a rescaled legend. Where series with different units are compared on one axis — burden against financing, for example — both are indexed to a 2015 baseline of 100 and labelled as an index rather than a level.
Stated assumptions
- Population at risk is approximated by total national population; sub-national transmission strata are not published in the source series.
- Where an indicator is reported by a country in one year and not the next, the gap is treated as unobserved rather than unchanged.
- Sub-regional and income groupings follow World Bank classifications as at the snapshot date and are not back-cast to earlier definitions.
- Financing figures are nominal current US$ and are not deflated or adjusted for purchasing power.
- Reported case counts are treated as a measure of programme detection, not of transmission alone.
Known limitations
Published in full. A platform that conceals its limitations cannot be used for decisions.
- WHO burden figures are modelled estimates published with uncertainty intervals that are not rendered on this platform; single-year country movements should not be read as measured change.
- Coverage, diagnostic, treatment and financing indicators are national programme reports and generally begin in 2015, so long-run trend analysis is possible only for burden.
- A rise in reported cases can reflect improved surveillance and testing rather than rising transmission; the confirmation rate in the clinical lens should be read alongside it.
- Household-survey indicators (ITN use, IPTp3) are observed only in survey years, so apparent plateaus can be an artefact of survey timing.
- Five small island states in scope have no boundary in the simplified map geometry but are included in every table, aggregate and export.
- No sub-national, facility-level or species-disaggregated analysis is possible from the current source series.
Reproducibility
The snapshot is the unit of reproducibility, not the interface.
The platform ships a single versioned data bundle containing every country-year observation, population denominator and boundary used for display. Every table in the dashboard exports to CSV with the indicator, country, year and value as displayed, so a third party can recompute any aggregate with the rules documented above and confirm the result. Indicator codes and retrieval dates for each source series are listed on the data provenance page.
Version history
| Version | Date | Change |
|---|---|---|
| v0.1 | 2026-08-04 | Initial research preview. WHO GHO burden, intervention, diagnostic, treatment and financing series for 46 sub-Saharan African countries with World Bank population denominators; executive, programme and clinical lenses; CSV export. |
Update cadence: the surveillance layer is refreshed following each annual WHO World Malaria Report release, with interim refreshes when source indicator series are revised.
Planned methodological extensions
Documented in advance so that future outputs are assessed against a stated standard.
Environmental covariates
Rainfall, land-surface temperature and vegetation indices aligned to national boundaries and lagged against transmission, with the alignment method published before any covariate is displayed.
Risk forecasting
Short-horizon national risk models with out-of-sample validation, published error metrics, prediction intervals and per-prediction feature attribution. No forecast is shown without its validation record.
Scenario simulation
A calibrated digital twin for intervention and financing scenarios, reporting the calibration period, the parameters varied and the sensitivity of each conclusion.