How much suffering occurred?
Event reports estimate selected experiences within a defined population and reporting period. Shared condition templates keep the approach consistent across comparable events.
Browse crisis reports →How it works
How we estimate suffering and its relief
We combine the number of people, the intensity of their experience and its duration. Each report separates source evidence from the assumptions needed to produce a USU estimate.
Research status · 29 October 2026
The framework and the website estimates
The published v3 preprint specifies duration matching and evaluates panel summaries in simulations. It also makes linear weighting, p = 1, the primary profile specification. WSI already uses that linear calculation with the 47.25 intensity-hour reference.
WSI’s condition intensities remain explicit modeling judgments. The simulation does not calibrate those inputs or validate comparisons across people and conditions. Read what the study adds →
One unit, two questions
How much could an intervention relieve?
Opportunity reports estimate the difference between receiving an intervention and the care or experience expected without it, then relate that difference to a stated cost.
Explore ways to reduce suffering →The shared reference
What one USU means
A concrete reference makes an abstract amount of suffering easier to picture. We use a defined six-hour kidney-stone pain profile as a convention, not a measured average episode. A longer, less intense experience can accumulate substantial suffering on the same scale.
One Universal Suffering Unit is a stipulated six-hour kidney-stone pain reference totaling 47.25 intensity-hours. It defines the unit; it does not establish that different conditions or people have empirically equivalent experiences.
Event USU = people or episodes × days × 24 × average intensity ÷ 47.25
For relief, use the change in average intensity and the period over which that change is assumed to last. The current simplified WSI models use linear weighting, p = 1. Costs, intervention effects and any additionality adjustments are documented in each opportunity report.
Keep evidence and assumptions distinguishable
Published inputs
Record the source, geography, dates, population, measure and limitations. A case count, patient outcome and program budget answer different questions; none is automatically an estimate of experienced suffering.
Model choices
State average intensity, duration, coverage and other assumptions explicitly. Use simple representative courses where detailed data are unavailable. Missing information stays visible rather than becoming zero.
A symptom score or disability weight is not automatically a USU intensity. Cross-condition and interpersonal comparisons remain provisional pending empirical calibration.
What a cost-effectiveness number covers
All current on-page spending comparisons use $100. The same amount makes the arithmetic easier to follow, but does not remove differences in the cost basis, currency year, population, benefit period or harms counted.
Donation scenarios
Start with money received by an organization and explain the assumptions connecting it to additional delivery. Accepting donations does not establish a remaining funding gap or the effect of the next dollar.
Program-cost scenarios
Start with a defined delivery budget, planning-cost reference or explicitly labeled cost assumption. State its currency year, included costs, population and horizon. These examples do not establish a return on an unrestricted donation. The podoconiosis example makes its unverified cost assumption editable.
Show the review date and distinguish reported funding need, confirmed funding and unverified allocation. A historical funding gap is not a live remaining balance. Scaling an estimate in a calculator illustrates arithmetic, not unlimited delivery capacity.
Count benefits and harms transparently
Many current opportunities estimate gross benefits with adverse effects left unestimated. The trichiasis example additionally subtracts a stated treatment-pain allowance. Its result is a balance of selected pain effects, not a complete net-benefit score.
Differences in included experiences, follow-up periods, cost bases and harms can change comparisons. A larger displayed number alone does not establish a better intervention.
Deaths and fatal illness
Experienced suffering before death can be included in a fatal-illness course. A person or case receives one relevant course; fatal cases must not also be counted in a nonfatal group. Death counts and deaths prevented are reported separately. USU does not assign suffering after death or convert lost life years into suffering.
Unestimated fatal suffering is labeled, not treated as zero. The current simplified models exclude bereavement unless a future report explicitly adds and documents it.
Use scenarios to understand uncertainty
Lower, central and higher cases show how specified assumptions change the result. They are not confidence intervals or full uncertainty bounds. A central scenario is a transparent working choice, not a validated expected value.
When useful, a zero-benefit scenario shows what happens if the assumed relief fails to materialize. Any treatment burden still counted can make the balance negative.
Earlier research demonstrations
Older research pages retain their original versions and time bases, including daily or retrospective scenarios. They should be read under their own methods rather than combined with the current event list.
Follow a calculation or improve an input
From reported cases to suffering
The Brazil dengue report shows case counts, course assumptions and arithmetic.
See the event calculation →From treatment costs to relief
The eye-pain example shows later relief and a recovery deduction, with downloadable inputs and code.
See the opportunity calculation →Substantive changes receive a new model version. When a correction changes an estimate or its interpretation, we add a dated note to the affected report explaining what changed. Editorial standards · Send feedback · Research and collaboration
Overview updated 17 October 2026. Individual reports retain their own model versions and review dates.
Source data and earlier work
Technical readers can inspect the source-data series and guide to underlying measures. The research archive preserves background guides and earlier pilots with their original methods.
For the fuller USU framework and manuscript, visit sufferingunits.org.