USU research
Research behind the estimates
Universal Suffering Units provide a proposed common reference for experienced suffering. Version 3 adds a simulation study of the duration-matching procedure.
Published 29 October 2026 · Version 3 · Not peer reviewed
Universal Suffering Units: A Proposed Framework with a Simulation Study of Duration Matching
The study asks how estimates change when people complete a duration match, fall outside the search range or give unresolved responses. It uses simulated respondents with known underlying values.
What the study found
Across 444 scenarios with 2,000 repetitions each, using only completed matches can bias the result. Retaining one-sided response bounds helps under suitable assumptions, but does not eliminate instability or bias under all response models.
What this means for WSI
Version 3 makes linear weighting the primary profile specification, which WSI already uses. It does not provide participant-derived calibration for our condition templates. The website’s existing estimates retain their stated assumptions and values.
The simulation evaluates a procedure under specified models. It does not establish interpersonal or cross-condition comparability. No participant-derived calibration has been conducted.
Cite the preprint
WSI applications
From research framework to event estimates
The event list applies one simple formula to selected dated events, with source counts, reported deaths and explicit assumptions. The four earlier pilots below retain their original units and methods.
The experimental unit
People or episodes × intensity × time
The raw burden is divided by a specified six-hour kidney-stone pain reference trajectory.
People or episodes×Modeled intensity×Duration→USU
The reference defines a reporting scale. It does not prove that pain, hunger, fear, and illness are already directly comparable.
Completed pilots
Four earlier research pilots
Dengue in the Americas
45.4 million USU18.2–95.5 million USU
Experimental cumulative USU pilot
Open pilotAcute food insecurity
109 million USU/day82.5–136 million USU/day
Experimental daily-equivalent USU pilot
Open pilotChild wasting
55.9 million USU/day globally27.9–83.8 million USU/day globally
Experimental daily-equivalent USU pilot
Open pilotFIES food insecurity
2.66 million annualized USU per 100,000Median range: 263,721–5,732,063 USU per 100,000
Experimental retrospective annualized USU scenario
Open pilotLooking for questions about the current event list? Visit the main FAQ →
Questions
USU and pilot methods
What is USU Research?
USU Research is the experimental part of WSI. It develops and tests Universal Suffering Units, mapping standards, time models, and related measurement ideas while keeping the original public data visible.
Do USU results determine Data rankings?
No. USU does not currently weight, order, or break ties in WSI data pages. The four pilots use different time bases and are not one league table.
Why publish experimental work?
The pilots show both the promise and failure modes of a common unit. Assumptions, ranges, maturity labels, and technical files make the limitations inspectable.
What is a Universal Suffering Unit?
A USU expresses modeled intensity-time burden relative to a specified six-hour kidney-stone pain trajectory. The reference makes the scale concrete. It does not by itself prove that different harms are fully comparable.
Does the kidney-stone reference prove that different harms are comparable?
No. The reference defines the reporting scale. Whether pain, hunger, fear, illness, and other experiences can be compared on that scale remains a research question.
Are the four earlier pilot headline values directly comparable?
No. Dengue reports cumulative burden across reported 2025 episodes. Acute food insecurity and child wasting report one-day equivalent burden rates. FIES reports a retrospective annualized scenario per 100,000 people. The results form a research portfolio, not a cross-domain ranking.
What does the deterministic mapping range mean?
It is the span between the approved low and high mapping scenarios. It shows how the result changes when provisional intensity or duration assumptions change. It is not a statistical confidence interval.
What do P1, P2, and D1 mean?
P1 is an early illustrative intensity mapping. P2 is informed by relevant empirical or clinical evidence. D1 means the duration values are illustrative scenarios rather than source-measured durations.
Are deaths included in USU?
No. The current pilot standard covers experienced suffering during lived time. Deaths are reported separately when the source provides a comparable mortality measure.
What would need to happen before USU could guide major decisions?
The source data, state mappings, time basis, and comparison rules would need stronger validation. Results would also need to remain stable under reasonable alternatives and show value beyond simpler indicators.
Why use a kidney-stone episode as the 1.0 USU anchor?
Renal colic is intense, time-bounded, commonly assessed on clinical pain scales, and supported by published clinical data. It is a calibration convention, not a claim that kidney-stone pain is the most important form of suffering.
Why not measure pain alone?
Serious negative experience can also involve breathlessness, nausea, fatigue, hunger, panic, fear, grief, or disruption of basic living conditions. Each non-pain state still needs a clear definition and evidence basis.
Why not rely only on cases, deaths, or DALYs?
Case counts show how many people or episodes were reported, but not how intense or prolonged the experience may have been. Death counts remain essential and are separate. DALYs and QALYs are established measures with different purposes.
Why focus on one condition or event at a time?
A focused series can keep the source, definitions, period, and model consistent. Many focused series can build a broader library without forcing them prematurely into one total score.
How are total and per-capita burden different?
Total burden reflects the aggregate modeled burden in the included population. A per-capita value reflects concentration relative to a stated denominator. The two views can produce different orderings.
How are intensity and duration estimated?
The event list uses shared representative courses, with separate fatal and other case groups where explicitly modeled. The shared method documents the inputs and prevents duplicate counting. The earlier pilots above use their original low, central and high assumptions and time models.
Where do the data come from?
WSI primarily uses official public-health, humanitarian, demographic, and research sources. Each series identifies its source, reporting period, scope, and release version.
What does it mean when a country is absent from a series?
Absence does not mean absence of suffering. A country may fall outside source coverage, use an incompatible definition, have a different reporting period, or lack enough data.
How does WSI handle uncertainty?
The current pilot standard uses approved low, central, and high mapping scenarios. The deterministic range shows model sensitivity under those assumptions; full reports also describe source limits that the range does not capture.
How is psychological suffering represented?
Psychological suffering can be modeled only when the state is clearly defined and supported by credible evidence. The FIES pilot includes food-access worry and distress but does not provide a general psychological-suffering scale.
How are overlapping harms handled?
The current pilot standard does not apply an overlap adjustment. Categories within one model must be mutually exclusive. Separate series are not added into one country total when people and time periods may overlap.
How subjective is the method?
Source counts come from the source organization; the USU mapping adds judgment about intensity and sometimes duration. WSI makes those choices visible through scenarios, maturity labels, and technical notes.
What does USU measure, and what remains separate?
USU estimates modeled experienced suffering associated with the states included in a specific series. It does not measure a person’s worth, dignity, rights, productivity, or overall value of life.
Could the framework be extended to non-human suffering?
The structure may be relevant to animal-welfare research, but the present program is human-centered. A non-human application would need species-specific evidence, expertise, and separate validation.