Experimental estimates · v0.11

How the estimates work

Each event uses source counts and shared representative courses. Distinct case groups can have different courses, without counting the same experience twice. All numerical intensity assignments are provisional; the reasoning is visible here and in every report.

The common calculation

USU = count × average days × 24 × average intensity ÷ 47.25

Count

Use reported cases, episodes or a clearly dated population cohort. Keep the source definition and reporting window. Do not silently substitute all affected people for the population actually modeled.

Duration

Assign one average course in days to each counted case or person. If outcomes are separated, assign one course within each disjoint group and sum the group estimates. Clinical duration ranges can inform a representative choice; they do not establish a measured event-specific mean.

Intensity

Use one combined score from 0 to 10: 0 means no included negative experience and 10 an extreme, overwhelming experience. The assigned score averages across people and elapsed hours, including relief and sleep.

Consistency

The same condition template applies across events. Every dengue report uses five days at intensity 3 for other reported cases and seven days at intensity 5 for reported fatal cases; all flood reports use 30 days at intensity 2. Pain, fear and other concurrent symptoms are represented jointly, rather than added as separate scores.

What is one USU?

For event list v0.11, one USU is defined by a stipulated six-hour kidney-stone pain reference trajectory totaling 47.25 intensity-hours, equivalent to an average intensity of 7.875 over those six hours.

We use linear weighting (p = 1): within a course calculation, doubling its count, duration or intensity doubles that component. Mixed-event totals sum the components. Intensity scores 2, 3, 4, 5 and 6 imply approximately 25%, 38%, 51%, 63% and 76% of the reference’s average hourly intensity. These are model assumptions, not observed clinical ratios.

The reference makes the scale concrete. It does not establish that different people or forms of distress are empirically equivalent. Read about the USU research framework.

What does the average per person or episode mean?

First, we assign a duration and average intensity to an included experience. Multiplying its days by 24 and its average intensity, then dividing by 47.25, gives USU for that modeled course. It is a working assumption about the whole course, including relief, sleep and milder periods.

For a single-course report, every included person or episode receives that same allowance. When a report uses ordinary and fatal illness courses, we calculate the two groups separately. The average shown in the list is their combined USU divided by the combined count. It is weighted by the number of cases in each group, rather than taking a simple average of the two course values.

The denominator matters. Outbreak entries show USU per reported case, displacement entries show USU per person in the selected cohort, and the treated-malnutrition entry shows USU per admission. Case and admission counts do not necessarily identify unique individuals. These are cumulative course averages, not daily rates or population-wide averages.

Source coverage, reporting periods, case mix and omitted experiences still differ. A larger modeled average does not establish that every individual experienced greater suffering, and it does not measure the value of a life. See a worked example →

Count the experience once

USU can include suffering experienced before death. Death itself, lost life years and suffering after death are not assigned USU; mortality remains a separate measure. Bereavement is excluded from this version and remains unestimated.

For an illness already counted

Where a source identifies deaths within the modeled case population, remove those cases from the ordinary group and assign one complete fatal illness course. Add the two disjoint groups. Do not add a terminal-period allowance on top of the same illness.

All dengue reports use seven days at intensity 5 for reported fatal cases and five days at intensity 3 for other cases. The Ebola report applies 12 days at intensity 6 to 3,267 reported fatal cases and keeps 14 days at intensity 5 for 3,490 other cases. Other cases include unresolved outcomes. No future deaths are imputed.

For a separate fatal experience

An addition requires a defined experience, a source-supported eligible count and a check that the experience or person-time is not already counted. A whole-event death toll alone does not establish the fatal mechanism or eligibility.

The flood illustration explores three conscious minutes of drowning at intensity 8. The actual eligible count is unknown, so no fatal-suffering addition enters that event’s estimate.

The fatal courses are provisional assumptions informed by historical studies, not measured averages for these event populations. The Paraguay chikungunya report replaces the acute allowance for 52 CHIKV-attributable fatal cases with 28 days at intensity 5. The remaining 80,985 cases keep eight days at intensity 4. The deaths belong to a narrower probable/confirmed subset within the modeled total; no death rate is extrapolated. Other cases can include deaths assigned another cause, as well as unresolved outcomes. The same specified course type will use the same inputs wherever it is applied.

What is included before death?

Seven disease reports include a complete fatal illness course for their selected reported fatal cases. The Mauritius chikungunya report uses only the acute course: its source summary reports no associated deaths, with the regional-to-country inference explained in the report. All four flood reports leave fatal suffering unestimated. The Sudan malnutrition report has no matched death count, so it also leaves a separate fatal course unestimated. This is an inclusion decision, not a claim that there was no suffering.

For a new fatal experience, we need a defined population and reporting period, a count that supports the chosen course, and a clear decision about overlap. The course itself can remain simple: one assumed duration and one combined average intensity. We do not need individual patient histories.

Why is fatal suffering unestimated in these reports?

Nepal floods

The 1,367 deaths refer to the broader flood event, while this estimate covers 3,629 residents of 37 displacement sites. The report does not identify fatal cases within that site cohort or supply a cause-specific count for a separately defined fatal experience. Subtracting the broader death toll from site residents would mix populations. [S11]

Sudan: severe child malnutrition

No matched death count or outcome breakdown for the selected admission cohort was identified in the reviewed UNICEF report. We do not import a national war death toll or estimate deaths from a case-fatality rate borrowed from another population. [S25]

Mozambique floods

The selected UNICEF report gives an accommodation-centre population as of 29 January but no mortality total aligned to that snapshot. Its list of external news headlines is not an aligned UNICEF mortality figure. Neither an eligible fatal count nor overlap with the centre cohort is established. [S12]

Rio Grande do Sul floods

The 24 May 2024 briefing reports 163 deaths across floods and landslides, without a cause-specific count aligned to its displaced population. A later official review covers 47 deaths in 14 Vale do Taquari municipalities, reporting 17% drowning, 32% burial and 51% trauma. Those regional percentages cannot be applied to the statewide 163-death snapshot, and they do not establish overlap with its displacement allowance. [S5] [S23]

Pakistan floods

The selected UNHCR briefing reports at least 1,700 deaths for the whole flood event and about 7.9 million displaced people. It does not separate the deaths by mechanism or link them to the modeled displacement course. The mortality threshold is not an eligible drowning count. [S10] [S24]

Why these durations and intensities?

Sources support symptom descriptions and, where available, typical duration windows. The exact numerical intensity is our judgment in every template.

Dengue: other reported cases

5 days · average intensity 3 / 10

The existing five-day acute course is retained for cases not reported dead at the snapshot cutoff. WHO gives a 2-7-day symptom window. Five days is a representative choice, not a measured mean or a survivor-specific estimate. Other cases include unresolved outcomes and severe cases not reported dead; no future deaths are imputed. Later fatigue is excluded. [S2]

Fever, headache, body/joint aches and nausea support a meaningful acute symptom burden. Assigned intensity 3 is a working average across the whole day, including relief and sleep. It is lower than a severe-pain peak. WHO does not establish the value 3 or its ratio to the reference. [S2]

Excluded: Later fatigue, prolonged recovery, a separate severity adjustment for other reported cases, caregiver distress, bereavement, unreported infections, death itself and lost life years.

Flood displacement: initial distress

30 days · average intensity 2 / 10

We assign the same 30-day initial course to each person in the selected flood-displaced cohort. Reports of continuing displacement and difficult living conditions support a multiweek scenario. None measures a 30-day cohort mean. This is a consistency choice, not evidence of equal recovery time or a month actually spent in a shelter. [S6] [S10] [S11] [S12]

WHO describes anxiety, sadness, sleep problems and other distress after emergencies. Assigned intensity 2 represents a modest whole-day average across the selected cohort, including relief and people with little distress. This provisional mapping to the pain reference is especially weak; WHO supplies no numeric conversion. [S7]

Excluded: Injury and disease episodes, suffering before death, bereavement, later mental-health effects, distress outside the selected cohort and post-death loss.

Chikungunya: other reported cases

8 days · average intensity 4 / 10

CDC describes acute symptoms typically resolving within 7-10 days. We retain eight days for cases outside the selected CHIKV-attributable fatal group. This is a representative choice, not an observed outbreak mean or proof of survival; unresolved outcomes and other-cause deaths are not reclassified. Persistent joint symptoms are excluded from this ordinary course. [S4]

CDC describes potentially severe, debilitating joint symptoms; Paraguay's adult suspected-case definition included incapacitating joint pain or arthritis. These support testing a higher average than dengue. Assigned intensity 4, and the 4:3 ratio, remain model judgments rather than measured equivalence. [S3] [S4]

Excluded: Persistent or recurrent joint pain beyond the assigned course, extra severity profiles for other cases, caregiver distress, bereavement, unreported infections, death itself, lost life years and suffering after death.

Bundibugyo Ebola: other reported cases

14 days · average intensity 5 / 10

The original all-case assumption of 14 days is retained for cases not reported dead at the cutoff. Historical series reported courses around 9-10 days in Uganda and 15-18 days in Isiro. Fourteen days is a rounded choice, not a pooled mean, a survivor-specific estimate or an observed 2026 average. Unresolved cases remain in this group; no future deaths are imputed. [S14] [S15]

The studies describe headache, muscle and abdominal pain, gastrointestinal symptoms and marked weakness. We assign intensity 5 as a provisional combined whole-day average, allowing for treatment, relief and sleep. Neither study measured USU intensity. Choosing 5 rather than the dengue value 3 or chikungunya value 4 is a model judgment, not a demonstrated clinical ratio. [S14] [S15]

Excluded: Incubation without symptoms, persistent post-Ebola symptoms, caregiver distress, bereavement, unconfirmed or unreported illness, and post-death loss. Individual severity and time to death or recovery are not reconstructed.

Bundibugyo Ebola: fatal illness

12 days · average intensity 6 / 10

We select 12 days between the historical median onset-to-death durations of 9 days in Uganda and 15 days in Isiro. It represents the complete symptomatic course ending at death, not an extra terminal period. It is not a pooled mean or an observed 2026 average; the studies were small and treatment settings differed. [S14] [S15]

We assign intensity 6 as a combined time-average of pain and other acute distress over the whole fatal course, including relief, sleep and intervals without experience. This is a provisional judgment above the existing value 5, not a measured USU mapping. Concurrent symptoms are represented together and are not added again. No intensity is assigned after death or to known absence of experience. [S14] [S15]

Excluded: Incubation without symptoms, caregiver distress, bereavement, death itself, lost life years and suffering after death. Individual trajectories and conscious time are not measured.

Dengue: fatal illness

7 days · average intensity 5 / 10

Seven days is a provisional representative course from symptom onset to death. A Puerto Rico series of 58 fatal laboratory-positive cases reported a median of 7.1 days; in a Malaysian review of 322 deaths, 66.5% occurred on illness days 4-7. A Singapore series of 28 adults reported a longer median of 12 days. These differing populations support testing alternatives. Seven days is our choice, not a pooled mean or an observed average for the listed outbreaks. [S18] [S19] [S20]

We assign intensity 5 as one combined average for pain and other experienced distress throughout the fatal illness. Clinical descriptions of abdominal pain, vomiting and breathing difficulty motivate testing a higher average than the ordinary value 3. The exact value 5 and its ratio to 3 are judgments, not measured USU conversions. This elapsed-time average allows for earlier milder illness, treatment, relief, sleep and periods without experience; it is not a constant severe-pain score. [S17]

Excluded: Incubation without symptoms, death itself, lost life years, caregiver distress, bereavement and suffering after death. Individual symptom and consciousness trajectories are not measured; no extra terminal period is added.

Chikungunya: attributable fatal illness

28 days · average intensity 5 / 10

A Ceará, Brazil series reported a mean onset-to-death interval of 28 days and a median of 15 days across 245 laboratory-confirmed deaths. We borrow the mean for this simple aggregate model. It includes later deaths and is not a measured Paraguay average. Differences in ages, care and death attribution limit transferability. The course represents the complete symptomatic illness ending at death, not an extra terminal period. [S21]

We assign intensity 5 as a combined elapsed-time average, above the ordinary value 4. A Paraguay adult ICU study describes fever, gastrointestinal symptoms and neurological involvement, but does not measure USU intensity or a complete onset-to-death course. The exact value and 5:4 ratio are judgments, allowing for relief, sleep and intervals without experience. This is not a constant conscious pain score. No suffering is assigned after death. [S22]

Excluded: Death itself, lost life years, caregiver distress, bereavement and suffering after death. No separate joint-pain or terminal-period allowance is added. Age-specific courses, symptom-free intervals and consciousness histories are not reconstructed.

Treated severe acute malnutrition: early illness and recovery

42 days · average intensity 2 / 10

We assign a six-week course after admission. Historical outpatient programmes in South Sudan, a different country, reported 2021 mean stays of 42.3, 43.2 and 46.3 days under three adapted protocols. These provide context for a multiweek window. Forty-two days is our rounded choice, not measured suffering duration in Sudan in 2026. Nutritional treatment can continue after clinical improvement; pre-admission suffering is excluded. [S26] [S27]

Malnutrition can involve physical discomfort, digestive problems and irritability, while children eligible for outpatient care may have good appetite and no acute problems requiring hospital care. We assign 2 as a combined average across people and elapsed time, including improvement, relief, sleep and intervals without experience. The value is below the ordinary dengue value 3 because this course averages over a longer recovery window. Both the value and this comparison are judgments, not validated ratios to kidney-stone pain. Concurrent symptoms are bundled once. [S27] [S28]

Excluded: Suffering before treatment admission, untreated children, moderate malnutrition, later or recurrent episodes outside the selected courses, long-term developmental harm, separate infection or injury burdens, broader conflict and displacement distress, caregiver distress, bereavement, death itself, lost life years and suffering after death.

What does a treatment-admission estimate cover?

The Sudan report models an early treatment and recovery course after admission. Six weeks provides a simple window, informed by historical outpatient stays in South Sudan. It is not a measured duration of suffering; the assigned intensity allows for improvement, relief and sleep.

Pre-admission suffering and untreated children are excluded. Full modeled courses can extend beyond the reporting cutoff. More reported admissions can reflect better treatment access or reporting, so the number alone is not a trend in need or a measure of suffering prevented by treatment.

UNICEF's narrative and results table give conflicting counts. The report keeps one explicit working count and shows the alternative with the same course. The figures are not added or averaged. The source does not establish unique-child deduplication. Read the count decision.

How to read and compare entries

Compare the included experience

Every headline is cumulative USU for its stated cases or cohort under the common formula. Counts, reporting periods and omitted consequences differ. The table orders model outputs; it is not a comprehensive ranking of the world’s events.

A reporting period is not an assumed course

A case can be reported within a window while its modeled course extends beyond it. The flood entries assign a 30-day course to their respective reported populations; none represents 30 days of observed follow-up or a calendar-year total.

Keep deaths and missing values visible

Deaths remain separate from USU, with their source population and date. A broader event death count may accompany a narrower modeled cohort. Missing data are shown as unavailable, not zero.

Do not sum the rows

Populations and time can overlap, including Brazil’s dengue outbreak and the Rio Grande do Sul floods. No overlap adjustment is applied. Excluded suffering is unestimated; the totals are not guaranteed lower bounds.

Uncertainty and alternative assumptions

Source counts can miss or misclassify cases, and reporting practices differ across settings. Assumed duration and intensity can be substantially wrong. Multiplying averages simplifies individual variation and associations between duration and severity.

Each report includes simple alternative duration/intensity scenarios. These are not confidence intervals or a complete uncertainty analysis. Some event orderings reverse under these alternatives.

No causal baseline is fitted. Individual symptom trajectories are not reconstructed. Dengue, Ebola and chikungunya use separate representative fatal courses for their source-defined eligible deaths. Cases outside the selected fatal group are not necessarily known survivors. The figures describe assigned event-related experiences, not causally identified excess burden.

Show the inputs that matter

Every crisis report tests duration and intensity for every included course. Source counts, mortality eligibility and the reference unit stay fixed. The central scenarios reproduce the existing headlines.

The combined lower and higher scenarios change both duration and intensity for all included courses. The separate input table changes one assumption at a time, making it easier to see what drives the total. These choices are illustrative and have no assigned probabilities. They are not confidence intervals, minimum or maximum burden, or a complete uncertainty analysis.

The ordinary dengue alternatives use 2 to 7 days and intensity 2 to 4, with the central course still 5 days at intensity 3. Two days replaces the previous three-day shorter scenario and follows the short end of the symptom window cited in S2. Clinical duration evidence does not establish a numerical intensity. Fatal-course combinations use the duration and intensity alternatives already documented for each course; other profiles retain their previous endpoints.

Case underreporting, source disagreements, omitted experiences and the validity of the common scale are separate uncertainties. Where a source supplies a conflicting count, the report retains its own count comparison. A small change in the fatal component cannot establish that the whole estimate is precise.

See the Brazil dengue example · All combined scenarios (CSV) · Reproduce the scenarios (Python)

A versioned starting point

Data release 0.11 adds consistent sensitivity analysis across all thirteen reports. It tests every included course and adds downloadable scenario calculations. Source counts, central course assumptions and central USU totals are unchanged. The ordinary dengue shorter scenario now uses two days, and combined fatal scenarios vary duration and intensity together. Version 0.10 inputs remain available.

Data release 0.10 added Mauritius chikungunya in the 2026 ECDC snapshot, using the existing eight-day, intensity-4 acute course. There are thirteen reports and eight course templates. All twelve earlier entries and all course assumptions are unchanged. The new report distinguishes no reported deaths from missing mortality data and excludes persistent joint pain.

Version 0.9 added Bangladesh dengue through 5 October 2026. Version 0.9 inputs remain available.

Version 0.8 added Sudan’s early treatment and recovery estimate, with its source-count discrepancy. Version 0.8 inputs remain available. Older reports retain their original release downloads and reporting dates.

A condition-template revision applies to every linked report. Material changes receive a new method or data version and a visible note. Source counts and original definitions stay available with each release.

Sources for the event list

Estimating suffering prevented or relieved

Opportunity analyses add a causal question: how does the experience change with an intervention? We estimate avoided or reduced courses and divide by a matched cost denominator. An event’s total USU is not automatically preventable.

The Chad malaria pilot uses the same linear USU reference, with its own versioned malaria assumptions. For treatment, only the change between the treated and untreated courses can be credited. Deaths prevented remain separate.

Estimating suffering relieved

For relief, use the difference between the assumed untreated and treated intensity: people × hours × average intensity reduction ÷ 47.25. Use one combined experience so overlapping symptoms are not counted twice.

The fistula-repair opportunity demonstrates this approach with a first-year window. Its intensity reduction is a provisional judgment, not a clinical conversion. Costs, additional delivery and outcomes enter before the USU calculation. Gross relief and any unestimated treatment harms must be labeled separately.

In the Friendship Bench example, a clinical screening result provides only a proxy for improved outcomes. The intensity change and relief duration are separate WSI assumptions. A screening threshold, disability weight or wellbeing score must not be treated as a validated USU conversion.

Search WSI