Define the measure
Record the source, period, geography, population, unit, and important limitations.
Methodology
How WSI handles data and experimental estimates.
The source result comes first. Any WSI calculation is labeled and documented separately.
1. Start with the source
Keep missing values visible
Zero, missing, suppressed, censored, and no coverage are not treated as the same thing.
Show the source result
Counts, rates, prevalence, or severity groups appear before any WSI model.
2. Label what WSI calculates
Transparent arithmetic, editorial comparison, and experimental USU estimates are kept distinct.
Every derived value identifies its formula, denominator, time basis, and source inputs. A model range is not called a confidence interval unless it is one.
3. USU Research
Cumulative episode model
Burden is accumulated across reported episodes over a defined period.
Daily-equivalent model
A snapshot population is modeled for one day, producing a rate rather than a multi-day total.
Retrospective annualized scenario
A representative-year burden is modeled using explicit days-at-severity assumptions.
Deaths remain separate
The current pilot standard does not convert death into USU.
Versions and corrections
Substantive changes create a new version. Material errors are recorded publicly rather than silently replaced.