Serology Adjusted IFR

Estimate IFR from assay-adjusted seroprevalence and aligned population deaths; assumes representative sampling and compatible time cohorts.

Description

Correct observed seropositivity for assay performance, estimate population infections, and calculate a matched infection-fatality percentage.

An observed positive proportion includes both true positives and false positives, while some infections may test negative. The Rogan–Gladen correction estimates the underlying prevalence from that proportion when sensitivity and specificity are supplied. The adjustment requires an assay that distinguishes the condition better than chance in this formulation.^[1]

An infection-fatality proportion divides attributable deaths by all infections in a matched population and period, including infections absent from confirmed-case counts. Using serology for that denominator additionally assumes that the assay target and sampling scheme identify the infections of interest.

Serology Adjusted IFR combines assay correction, population scaling, and a death-to-infection ratio. It reports the adjusted fraction, estimated infections, and fatality percentage, rejecting incompatible values rather than clipping them.

When to check a serology-based population estimate

Use the tool for a transparent point-estimate calculation after choosing representative serology data and deaths aligned to the infection cohort. It cannot establish the sampling or timing assumptions from the entered numbers.

Compare observed and assay-adjusted infection fractions.
Scale a valid adjusted fraction to a defined population.
Check whether supplied death counts are compatible with the resulting denominator.

How to supply serology counts and calculate IFR

Enter Seropositive count and Sample count as integers with 0 ≤ positives ≤ sample and sample greater than zero.
Enter sensitivity and specificity as fractions from 0 to 1 whose sum exceeds 1.
Enter a positive integer Population size and nonnegative integer Deaths for the matched cohort.
Run the calculation and read the adjusted fraction, estimated infections, and Infection fatality percent together.

pObserved = seropositiveCount/sampleCount; pAdjusted = (pObserved + specificity − 1)/(sensitivity + specificity − 1)

The denominator is the assay Youden index. This correction assumes the supplied assay performance applies to the sampled population.

estimatedInfections = pAdjusted × populationSize; infectionFatalityPercent = 100 × deaths/estimatedInfections

Estimated infections are not rounded to an integer.

json

{
  "input": {
    "seropositiveCount": 180,
    "sampleCount": 1000,
    "sensitivityFraction": 0.9,
    "specificityFraction": 0.9,
    "populationSize": 100000,
    "deaths": 100
  },
  "output": {
    "adjustedInfectionFraction": 0.1,
    "estimatedInfections": 10000,
    "infectionFatalityPercent": 1
  }
}

Adjusted prevalence must be greater than zero and at most one. Deaths cannot exceed population size or estimated infections. Zero deaths with a valid positive denominator gives 0%; incompatible counts or correction results produce input errors.

Limitations of serology-based IFR arithmetic

Representative sampling, matching geography and infection periods, and appropriate death attribution are assumed. Infection-to-death delay and unresolved outcomes are not modeled.
Antibody waning, vaccination-derived antibodies, repeat infections, and incomplete seroconversion can break the interpretation of seropositivity as cumulative infection.
No sampling interval, assay-parameter uncertainty, age standardization, or correction for missing deaths is supplied.
A Youden index near zero amplifies small input changes. Rejecting out-of-range adjusted prevalence does not resolve the underlying estimation problem.

Alternative or Complementary fatality and test summaries

Keep a case-fatality denominator distinct from an infection denominator. Predictive-value calculations explain the effect of assay performance at a specified prevalence, while a seropositivity-waning summary can describe one source of denominator change without correcting it automatically.

References

Estimating prevalence from the results of a screening test — American Journal of Epidemiology; W. J. Rogan and B. Gladen (1978)

Similar or alternative tools

Confirmed Case Outcome Fatality
Calculate crude and resolved-case fatality percentages from a confirmed-case cohort; these use different denominators.
Prevalence Predictive Values
Calculate test-result predictive values from assumed prevalence and assay sensitivity/specificity.