The excess.
When a catastrophe moves through a population, the honest measure is not the attributed toll — it is how many more people died than the years before said should have. This report reads that measure for the pandemic years, and it computes no baseline of its own: the excess series are the publishers' computations, cited as theirs, and the raw view shows deaths against the record's own 2015–2019 range — a range, not a model. The gap between the two counts is the finding: Egypt's excess ran 14.51 times its reported toll, Russia's 3.35 times, the United States' 1.13 — the same event, counted honestly and counted officially, on one board. And its quietest fact: Russia’s and the United States’ excess totals land 764 apart — 1,349,776 against 1,349,012. Every figure is computed from the records at render time.
published 2026-08-29 · data as published through each series' last week · edition 1 — updated only by a new dated edition · sources: OWID excess-mortality series · World Mortality Dataset · OWID reported totals
Deaths above the expected, week by week.
Seven countries' P-scores — the percentage by which each week's deaths exceeded the publisher's five-year baseline. The waves arrive at different times, at different heights, and the height of a country's wave and the size of its official count are two different facts — the next view holds them together.
Counted honestly, counted officially.
For each country: the published cumulative excess estimate beside the officially reported toll, and the ratio between them. The ratio divides two published totals — nothing more. Where it runs near one, the official count and the excess sit close together; where it runs to double digits, they part ways — and the ratio cannot say why. A registry that concealed and a registry that could not keep up with the surge look identical in this arithmetic.
One country's registry, no model at all.
The simplest version of the whole argument: Russia's monthly death registrations, 2020 through 2023, drawn over the shaded band of the highest and lowest each calendar month recorded in 2015–2019 — the record's own range. Where a pandemic-year line leaves the band, no baseline model is needed to see it.
Whose numbers, and what this page refuses.
No baseline of this page's own. An excess-death figure requires an expected value, and an expected value is a model. This page builds none: the P-scores and cumulative estimates are the publishers' computations against their stated five-year baselines, drawn as published and cited as theirs. Where this page does arithmetic, it is division of two published totals, and the exhibit says so where it happens.
The raw view is the check on everything else. Registry deaths against the record's own five-year range needs no model to read — it is the primary record with its own history as the yardstick. A reader who trusts no baseline can stand on that view alone.
What is absent, and why. No world total appears: the sources publish none for these series, and summing incomplete country coverage would manufacture one. Countries without registry data are absent, not estimated. The 1918 pandemic is not drawn: no comparable registry series meets this page's sourcing bar, and a famous number without a record is exactly what this site does not print.