Weekly Death Counts By Country Data: Dataset Structure and Field Coverage

Two named sources carry this beat, and neither sits inside an annual vital-statistics cycle.

The Human Mortality Database (HMD) publishes the Short-term Mortality Fluctuations (STMF) series alongside its annual life tables: weekly death counts running from as early as 1990 (Finland) into 2026, covering the same 41 countries and sub-populations as the main database — including split series for Germany (East/West/total), France (civilian/total), the UK (England & Wales, Scotland, Northern Ireland and a UK total) and New Zealand (Maori/non-Maori/total). Datadory scores HMD 10/10, one of only two perfect scores among the 16 primary datasets in this slice.

HMD's STMF inherits the parent database's discipline. The archive's granularity runs to single-year ages up to 109 plus an open 110+ interval on the annual side; the weekly STMF layer carries death counts by week and age band, sitting beside annual files that also break deaths out by Lexis triangle (age x birth cohort x calendar year). That structure lets a longevity model reconcile a weekly spike against the underlying age composition rather than treating it as a single national number.

The ONS weekly file answers a different question shape: registered deaths for England and Wales by week of registration, with most mortality series covering England and Wales while some extend to the whole UK, and subnational breakdowns reaching regions, local authorities and health boards. Granularity across the ONS hub spans weekly provisional deaths, monthly COVID-19 registrations and annual registered deaths, cut by single year of age or age bands, sex, cause of death and geography.

A working extraction takes minutes once you know which doors are open:

Step 6 is where most excess-mortality dashboards quietly break: two national series can disagree by a week at period boundaries without either being wrong.

Three workflows recur among insurance users of these feeds:

Excess-mortality baselining. Join STMF weekly counts to a seasonal baseline built from prior years of the same series, flag weeks above tolerance, and annotate against ONS England and Wales registrations as an independent read. Because both sources run into 2026, the current year always has a comparator.

Claims-assumption reviews. When weekly counts diverge from baseline for consecutive weeks, re-examine lapse and mortality assumptions before the next valuation cycle — using SOA MORT exports (CSV, Excel, XML or JSON after accepting the on-page disclaimer) as the statutory reference point.

Cross-country longevity screens. HMD's harmonization means a week in Finland starting 1990 and a week in Chile land in one schema; where a market falls outside the 41 countries, World Bank indicators extend coverage to 200+ economies, though only annually.