Infection Fatality Rate

The infection fatality rate (IFR)—also termed the infection fatality risk or infection fatality ratio—is a foundational severity parameter in infectious disease epidemiology, biostatistics, and public health modeling that quantifies the proportion of all biologically infected individuals who die from a specified pathogen over the course of their infection. Whereas the case fatality rate (CFR) restricts its denominator exclusively to clinically diagnosed or laboratory-confirmed patients (C), the IFR expands the denominator to encompass the entire clinical iceberg of infection (I_total)—summing fatal cases, hospitalized cases, mild outpatient symptomatic cases, and completely asymptomatic or subclinical infections that never seek medical care or undergo diagnostic testing:

IFR = ( D / I_total ) × 100%

Because clinical testing capacity, healthcare-seeking behavior, and syndromic case definitions vary widely across jurisdictions and epidemic phases, the CFR is heavily influenced by surveillance ascertainment bias. By anchoring severity to total biological infections, the IFR provides a surveillance-independent measure of intrinsic pathogen virulence within a given host population, serving as the critical multiplier in compartmental SIR/SEIR epidemic models to forecast hospital intensive care demand, total pandemic mortality, and the years of potential life lost averted by vaccination or non-pharmaceutical interventions.

Decomposing Total Infections Into Ascertained vs. Unascertained Strata

For most respiratory, enteric, and vector-borne viruses and bacteria (such as influenza, SARS-CoV-2, poliovirus, dengue fever, and Vibrio cholerae), clinical disease represents only the visible tip of an epidemiological iceberg. If a population of size N experiences I_total total infections during an epidemic wave, those infections partition into:

I_total = I_asymptomatic + I_{symptomatic, undiagnosed} + C_diagnosed

Let p_symp = I_symptomatic / I_total denote the symptomatic fraction (the biological probability that an infected person develops clinical symptoms), and let p_det|symp = C_diagnosed / I_symptomatic denote the case detection (ascertainment) probability among symptomatic individuals. When diagnostic testing is performed only on symptomatic patients, the overall infection-to-case ascertainment ratio is a = C_diagnosed / I_total = p_symp × p_det|symp. Consequently, the IFR, the symptomatic case fatality risk (sCFR), and the diagnosed CFR are linked by the exact algebraic identity:

IFR = p_symp × sCFR = ( C_diagnosed / I_total ) × CFR = a × CFR

For example, if an emerging respiratory virus causes clinical symptoms in 67% of infected individuals (p_symp = 0.67), and public health surveillance detects 30% of those symptomatic cases (p_det|symp = 0.30), then only a = 0.67 × 0.30 ≈ 20% (1 in 5) of total infections are captured as confirmed cases (I_total = 5 × C_diagnosed). If the observed CFR among confirmed cases is 2.5%, the true underlying IFR across all infections is 0.20 × 2.5% = 0.50%.

Because undiagnosed and asymptomatic infections are invisible to routine clinical reporting, epidemiologists must estimate the denominator I_total using three specialized study designs:

1. Population-Representative Seroprevalence Surveys and the Rogan-Gladen Estimator

The gold-standard method for quantifying cumulative population infection is a random, age- and sex-stratified seroprevalence survey measuring pathogen-specific antibodies (such as IgG or total immunoglobulin against the SARS-CoV-2 nucleocapsid or spike protein) in blood samples collected after an epidemic wave.

However, raw seropositivity proportions (p_obs) cannot be plugged directly into the IFR denominator without correcting for imperfect diagnostic assay sensitivity (Se) and specificity (Sp). Even a highly specific enzyme-linked immunosorbent assay (ELISA) or chemiluminescent immunoassay with 99.0% specificity produces a 1.0% false-positive rate; in a low-prevalence region where true cumulative infection is only 1.0%, half of all positive tests will be false positives, doubling the apparent denominator and cutting the estimated IFR in half. To remove assay error, biostatisticians apply the Rogan-Gladen estimator (1978) (or Bayesian hierarchical Stan models incorporating validation-panel uncertainty) to compute the true seroprevalence π:

π = ( p_obs + Sp − 1 ) / ( Se + Sp − 1 )

Furthermore, because serum antibody titers against mucosal respiratory viruses (espec. anti-nucleocapsid IgG) wane below the assay cutoff threshold over 4 to 9 months—a phenomenon termed seroreversion—longitudinal IFR models convolve daily incidence curves with an empirical antibody waning survival function W(τ) and account for the 10-to-14-day window between infection and seroconversion juxtaposed against the 18-to-25-day lag between infection and death:

IFR_{stratum k} = D_k(t + Δt_death) / [ N_k × π_k(t + Δt_sero) ]

2. Exhaustively PCR-Screened Closed Cohorts and Repatriation Flights

During the earliest weeks of a pandemic before serological assays are validated, researchers estimate the IFR from universally tested closed populations in which every single resident or passenger—regardless of symptoms—undergoes repeated molecular (RT-qPCR) testing:

• International Repatriation Flights and Quarantine Cohorts: In early 2020, universal PCR screening of all passengers evacuated on repatriation flights from Wuhan, China (Verity et al., 2020), alongside universal testing of the 3,711 passengers and crew aboard the Diamond Princess cruise ship (Russell et al., 2020), provided the first unbiased measurements of the asymptomatic infection fraction and age-specific IFR of ancestral SARS-CoV-2 (~0.6% to 0.7% across an all-age population pyramid).
• Comprehensive Community Census and Wastewater Integration: Combining repeated random household PCR swab surveys (such as the UK Office for National Statistics COVID-19 Infection Survey of > 150,000 biweekly swabs) with quantitative wastewater-based epidemiology provides continuous longitudinal tracking of I_total even after repeat infections and mass vaccination complicate serology.

One of the most important quantitative discoveries of modern sero-epidemiology—established across dozens of multinational seroprevalence meta-analyses by Andrew T. Levin, William P. Hanage, Nana Owusu-Boaitey, Gideon Meyerowitz-Katz, and colleagues (2020) and O'Driscoll et al. (2021)—is that for respiratory viral infections such as SARS-CoV-2, a single population-wide summary IFR is meaningless without specifying the population's age distribution.

Exponential Gompertz-Makeham Scaling With Chronological Age

Across adults aged 15 to 90+ years, the age-specific Infection Fatality Rate of ancestral SARS-CoV-2 follows a strict log-linear (exponential) relationship with age, mirroring the classical Gompertz-Makeham law of all-cause human mortality in which the probability of dying doubles approximately every 6.5 to 8.0 years of chronological age:

log₁₀( IFR(a) ) = α + β · a

where a is age in years, α ≈ −3.27, and β ≈ +0.0524 per year (corresponding to an 11.9% to 12.8% relative increase in IFR per additional year of age, or a ~3.3-fold increase per decade of life):

• Children and Adolescents (Ages 5–14 years): Lowest baseline risk, with IFR ≈ 0.001% to 0.002% (roughly 1 death per 50,000 to 100,000 infections).
• Young Adults (Ages 25–34 years): IFR ≈ 0.01% to 0.02% (1 in 5,000 to 10,000 infections).
• Middle-Aged Adults (Ages 55–64 years): IFR ≈ 0.40% to 0.75% (~1 in 130 to 250 infections).
• Older Adults (Ages 75–84 years): IFR ≈ 3.2% to 5.5% (~1 in 18 to 30 infections).
• Octogenarians and Nonagenarians (Ages 85+ years): IFR ≈ 8.5% to 15%+ (~1 in 7 to 12 infections, and up to 20%–25% among frail long-term nursing home residents).

Because age-specific IFR spans four orders of magnitude (10,000-fold) between a 10-year-old child and an 85-year-old adult, applying an age-stratified IFR vector to two countries with contrasting demographic pyramids (reflecting different stages of the demographic transition) produces vastly different population-average IFRs even when healthcare quality is held constant:

• In a youthful Stage 2/3 demographic pyramid (such as Uganda, Niger, or Kenya, where median age is 16 to 20 years and only ~2% to 3% of the population is aged ≥ 65), the population-weighted mean IFR for ancestral SARS-CoV-2 is approximately 0.15% to 0.23%.
• In an aged Stage 4/5 demographic pyramid (such as Japan, Italy, or Germany, where median age is 45 to 49 years and 23% to 29% of the population is aged ≥ 65), the exact same age-specific IFR schedule yields a population-weighted mean IFR of 1.15% to 1.70%—nearly 8-fold higher solely due to population age structure.

Biological Modifiers: Sex, Comorbidities, and Hybrid Immunity

Even within a single narrow 5-year age stratum, the IFR is further stratified by biological and immunological factors:

1. Biological Sex (Male Excess IFR): Across all adult age groups, males exhibit a 1.4- to 1.7-fold higher IFR than age-matched females during acute SARS-CoV-2 and SARS-CoV-1 infection. Mechanistic studies by Jean-Laurent Casanova and Akiko Iwasaki linked this disparity to higher rates of X-linked recessive Toll-like receptor 7 (TLR7) deficiency and neutralizing autoantibodies against Type I interferons (IFN-α2 and IFN-ω)—present in ~10% to 20% of critical COVID-19 deaths over age 70, predominantly in men—combined with stronger female innate plasmacytoid dendritic cell IFN-α responses and T-cell activation.
2. Cardiometabolic, Renal, and Immunosuppressive Comorbidities: Chronic kidney disease (stage 4–5 / dialysis), solid-organ transplantation, trisomy 21 (Down syndrome), decompensated cirrhosis, severe obesity (BMI ≥ 35 kg/m²), and poorly controlled diabetes elevate age-specific IFR by 2- to 10-fold, reflected in elevated social determinants of health disparities.
3. Vaccination and Hybrid Immunity: Primary vaccination series, booster doses, and prior infection ("hybrid immunity") elicit durable germinal-center memory B cells and cross-reactive CD4⁺/CD8⁺ T cells that reduce the population IFR by 85% to 95%+, transforming novel pandemic pathogens with high initial IFRs into endemic infections managed alongside oral antiviral essential medicines.

1. Rogan, W. J., & Gladen, B. (1978). "Estimating prevalence from the results of a screening test." American Journal of Epidemiology, 107(1), 71–76. https://doi.org/10.1093/oxfordjournals.aje.a112510
2. Verity, R., Okell, L. C., Dorigatti, I., Winskill, P., Whittaker, C., Imai, N., Cuomo-Dannenburg, G., Thompson, H., Walker, P. G. T., Fu, H., Dighe, A., Griffin, J. T., Baguelin, M., Bhatia, S., Boonyasiri, A., Cori, A., Cucunubá, Z., FitzJohn, R., Gaythorpe, K., ... Ferguson, N. M. (2020). "Estimates of the severity of coronavirus disease 2019: a model-based analysis." The Lancet Infectious Diseases, 20(6), 669–677. https://doi.org/10.1016/S1473-3099(20)30243-730243-7)
3. Levin, A. T., Hanage, W. P., Owusu-Boaitey, N., Cochran, K. B., Walsh, S. P., & Meyerowitz-Katz, G. (2020). "Assessing the age specificity of infection fatality rates for COVID-19: systematic review, meta-analysis, and public policy implications." European Journal of Epidemiology, 35(12), 1123–1138. https://doi.org/10.1007/s10654-020-00698-1
4. O'Driscoll, M., Ribeiro Dos Santos, G., Wang, L., Cummings, D. A. T., Azman, A. S., Paireau, J., Fontanet, A., Cauchemez, S., & Salje, H. (2021). "Age-specific mortality and immunity patterns of SARS-CoV-2." Nature, 590(7844), 140–145. https://doi.org/10.1038/s41586-020-2918-0
5. Brazeau, N. F., Verity, R., Jenks, S., Fu, H., Whittaker, C., Winskill, P., Dorigatti, I., Walker, P. G. T., Riley, S., Schnekenberg, R. P., Hoeltgebaum, H., Mellan, T. A., Mishra, S., Unwin, H. J. T., Watson, O. J., Cucunubá, Z. M., Baguelin, M., Whittles, L., Bhatt, S., ... Okell, L. C. (2022). "Report 34—COVID-19 infection fatality ratio: estimates from seroprevalence." Communications Medicine, 2(1), 54. https://doi.org/10.1038/s43856-022-00106-7