The subtraction is easy; the counterfactual is harder
If 1,200 deaths are observed and 1,000 expected, the estimated excess is 200. But the expected 1,000 is not directly observed in an alternative world. It must be constructed from data and assumptions.
A simple historical average and a model accounting for population aging may produce different expectations.
Keep the outcome and attribution separate
All-cause excess mortality can capture several direct and indirect influences. It does not, by itself, assign every excess death to one cause or policy.
Cause-specific records and other evidence may help investigate explanations, but the aggregate difference is not a complete causal analysis.
Compare like with like
| Detail | Why it matters |
|---|---|
| Registration or occurrence date | Changes the time series |
| Reference years | Defines the baseline experience |
| Population adjustment | Accounts for size or age changes |
| Recent reporting completeness | Affects observed counts |
Include uncertainty and revisions
Different credible methods may disagree because they ask slightly different counterfactual questions. Explain the method rather than presenting one estimate as an exact count known without error.
Excess mortality is useful for detecting and describing unusual mortality burden. Its strength depends on making the baseline and attribution limits as visible as the headline total.
