{
  "id": "case-fatality-ratio",
  "title": "Case-Fatality Ratio",
  "lang": "en",
  "source_lang": "en",
  "summary": "The case-fatality ratio is the proportion of people with a defined disease who die from it within a specified observation period.",
  "category": "health",
  "subcategory": "public-health",
  "type": "concept",
  "aliases": [
    "Case-fatality proportion",
    "Case-fatality risk",
    "Case-fatality rate",
    "CFR"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/case-fatality-ratio/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/case-fatality-ratio/",
      "json": "https://aiwiki.page/zh/case-fatality-ratio.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The case-fatality ratio (CFR) is the proportion of people with a defined disease who die from that disease within a specified observation period. It is used in epidemiology to describe disease severity among identified cases. Although commonly called the case-fatality rate, it is a proportion rather than a rate measured per unit of time. Its interpretation depends on how cases and deaths are defined, which patients are included, and whether their outcomes have been observed long enough."
    },
    {
      "heading": "Definition and calculation",
      "text": "For a defined group of cases, the case-fatality ratio is calculated as:\n\n\\[\\mathrm{CFR} = \\frac{\\text{Deaths attributable to the disease among those cases}} {\\text{Total number of cases in the group}} \\times 100\\%.\\]\n\nFor example, if 500 people meet a specified case definition and 20 of them die from the disease during adequate follow-up, the observed CFR is \\(20/500 = 4\\%\\). This is a hypothetical calculation, not an estimate for a particular disease.\n\nThe numerator must refer to deaths among the people represented in the denominator. Dividing deaths recorded during one calendar period by cases diagnosed during that same period does not necessarily meet this condition: some deaths may concern cases diagnosed earlier, while some newly diagnosed cases may die later.\n\nA case definition may include laboratory-confirmed cases, clinically diagnosed cases, or probable cases meeting specified criteria. Consequently, “CFR among confirmed cases” and “CFR among all clinically recognized cases” need not estimate the same quantity.\n\nThe observation period also matters. An in-hospital CFR, a 28-day CFR, and a CFR measured through completion of an acute illness can differ even when they concern the same disease. Some studies report all-cause deaths among cases over a specified interval rather than deaths attributed specifically to the disease; that endpoint must be distinguished explicitly."
    },
    {
      "heading": "Distinction from related measures",
      "text": "Mortality rate\n\nA mortality rate measures deaths in a population over time, generally using the population or accumulated person-time as its denominator. CFR instead uses people who have the disease.\n\nA disease can therefore have a high CFR but cause relatively few population deaths if it is rare. Conversely, a common disease with a lower CFR can cause a substantial number of deaths. Population mortality depends on disease occurrence as well as fatality among cases.\n\nInfection-fatality ratio\n\nThe infection-fatality ratio (IFR) uses all infected people as its denominator, including people whose infections are asymptomatic or never diagnosed. CFR generally uses cases identified through a clinical or surveillance definition.\n\nFor an infectious disease, an observed CFR can exceed the IFR when many mild or asymptomatic infections are absent from the case count. The relationship is not fixed, however: incomplete death ascertainment, different populations, and mismatched observation periods can also affect the estimates. Estimating IFR requires information about infections beyond routinely diagnosed cases.\n\nIncidence and prevalence\n\nIncidence describes the occurrence of new cases, while prevalence describes the proportion of a population living with a condition at a specified time or over a period. Neither directly measures the probability of death among cases.\n\nCFR complements these measures but cannot replace them. A CFR alone does not indicate how widely a disease is spreading or how much illness it causes in a population."
    },
    {
      "heading": "Estimation during an outbreak",
      "text": "During an ongoing outbreak, reported deaths and reported cases are continually changing. A simple cumulative calculation,\n\n\\[\\frac{\\text{Deaths reported to date}} {\\text{Cases reported to date}},\\]\n\nis often described as a crude or reported CFR. It may not estimate the eventual fatality proportion reliably.\n\nDelay between case identification and death\n\nDeath, when it occurs, usually follows disease onset or diagnosis after a delay. Recently reported cases therefore have unresolved outcomes. During rapid epidemic growth, including many such cases in the denominator can make the crude CFR lower than the eventual fatality proportion for those cases.\n\nA related calculation uses only reported resolved cases:\n\n\\[\\frac{\\text{Deaths}} {\\text{Deaths}+\\text{Recoveries}}.\\]\n\nThis is not automatically unbiased. Recovery and death may occur after different delays, recovery may lack a consistent definition, and recoveries may be recorded less completely than deaths.\n\nMore informative estimates can follow a defined group of cases until outcomes are known or use statistical methods that account for outcome delays. Survival analysis can incorporate incomplete follow-up, although its validity depends on assumptions about censoring, outcome ascertainment, and competing events.\n\nIncomplete case and death detection\n\nPublic health surveillance does not necessarily detect all cases. If severe cases are more likely to be tested or reported, the observed denominator disproportionately represents severe disease, potentially increasing the estimated CFR relative to that among all cases meeting a broader definition.\n\nDeaths can also be missed, reported late, or classified inconsistently. Changes in testing access, reporting practices, and death certification can alter the observed CFR without a corresponding change in the disease’s biological severity."
    },
    {
      "heading": "Factors affecting comparisons",
      "text": "CFR is not an unchanging property of a pathogen or disease. It reflects both the illness and the population and circumstances in which outcomes are measured.\n\nImportant influences include:\n\n- Patient characteristics: age, underlying illnesses, immune status, and prior immunity.\n- Disease characteristics: clinical form, stage at diagnosis, and, where relevant, pathogen variant.\n- Care setting: community cases, hospitalized patients, and intensive-care patients represent different levels of illness severity.\n- Access to care: treatment availability, timeliness of care, and health-system capacity.\n- Measurement practices: case definitions, testing strategies, follow-up duration, and rules for attributing deaths.\n\nFor example, a hospital-based CFR generally describes a selected patient population rather than all people with the disease. Comparisons between places or periods can also be distorted by differences in age distribution. Stratification and age standardization can improve comparability, but they do not remove every source of bias."
    },
    {
      "heading": "Uses and limitations",
      "text": "CFR supports public health assessment by describing fatal outcomes among recognized cases. It can help characterize outbreaks, identify groups with different outcomes, and monitor changes in disease severity or clinical management.\n\nHowever, an observed change in CFR does not by itself establish why outcomes changed. A decline could reflect more effective treatment, broader detection of mild cases, a younger case population, or several factors together. Comparisons intended to evaluate treatment effects require study designs that address differences between treated and untreated groups; a simple comparison of CFRs does not establish causation.\n\nCFR estimates also carry statistical uncertainty, especially when case counts or death counts are small. Zero observed deaths in a small group does not establish that the underlying fatality risk is zero. Confidence intervals can express sampling uncertainty, but they do not by themselves capture systematic errors from missed cases, incomplete follow-up, or misclassified deaths.\n\nA clearly reported CFR identifies the case definition, population, setting, observation period, death endpoint, counts of cases and deaths, and estimation method. It describes an average outcome in that defined group, not a precise prognosis for an individual patient."
    }
  ],
  "links": [
    "disease",
    "epidemiology",
    "case-definition",
    "mortality-rate",
    "person-time",
    "infectious-disease",
    "incidence",
    "prevalence",
    "diagnosis",
    "public-health-surveillance",
    "health-system",
    "age-standardization",
    "public-health",
    "causation"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}