Methodology & reconciliation

Every figure belongs to one of two families. We never silently average across them.

1. One source per figure. Each displayed number comes from a single named publication — values are never blended.
2. Headline = international. Where a metric exists in both families, the internationally comparable estimate is shown as the headline; the Vision series is footnoted, never discarded.
3. Progress = official. Vision 2030 progress bars track the official baseline series, so the baseline, current, and target all share one definition.
4. Disclose, don’t reconcile away. When two legitimate figures exist, both are shown with their definitions; the gap is labelled methodological, not corrected.
5. Date at the metric. Each figure carries its own reference year, independent of the aggregation date (2026-08-29).

These are the discrepancies a reviewer is most likely to flag. In each case both numbers are correct on their own definition — the difference is source, year, or method.

Every underlying series is exportable as CSV or JSON so you can audit and reuse the numbers directly.

Statuses on the Race-to-2030 tracker are computed, not hand-assigned: Achieved: ≥100% of the baseline→target journey. On track: progress ≥80% of the time-expected share of the journey. At risk: below 80%. Divergences from the rule are conservative downgrades only, disclosed on the KPI. The build fails if a stored status violates this rule (scripts/validate-data.ts).

Stating the gaps is more credible than hiding them. The current ones:

• 2022 census rebasing. The pre-2022 population estimates were an overcount; the May-2022 census revised the total to 32.18M. The population series is therefore not smooth across 2021→2022 — pre-census points are flagged, and every per-capita ratio uses one canonical denominator series.
• Two budget universes. Current health expenditure (NHA, public+private) and the MOF “Health & Social Development” budget envelope (SAR 268B, which also funds non-health social programmes) are different measures. The headline uses the former; the latter is reported separately and never relabelled “total health expenditure”.
• All-sector vs MOH-sector breaks. Workforce counts are standardised to all-sector to avoid spliced series; per-1,000 ratios use the World Bank/WHO international definition, which differs from headcount÷population (both disclosed).
• Supply-chain figures are estimates. The pharmaceutical import/export and category figures are illustrative market-structure estimates flagged “estimate”, pending binding to the SFDA drug registry and GASTAT trade statistics.
• The 2030 population assumption. Projections to 2030 use a conservative ~1.4%/yr medium-variant assumption from the 2024 base, not a linear extrapolation of the 2022–24 migration surge.
• The Vision series rebase. The official life-expectancy series was rebased upward in the HSTP 2025 report (76.8 → 79.7); the tracker follows the official series and discloses the rebase so it is never read as a one-year gain.