The leading cause of death is respiratory failure from acute respiratory distress syndrome (ARDS).[894] The overall pooled mortality rate from ARDS in COVID-19 patients is 39%; however, this varies significantly between countries (e.g., China 69%, Iran 28%, France 19%, Germany 13%).[895] Risk factors for respiratory failure include older age, male sex, cardiovascular disease, laboratory markers (such as lactate dehydrogenase, lymphocyte count, and C-reactive protein), and high viral load on admission.[896]

People <65 years of age have a very small risk of death even in pandemic epicentres, and deaths in people <65 years of age without any underlying conditions is rare.[897]

Mortality rates have decreased over time despite stable patient characteristics. In one study among patients with critical illness admitted to an intensive care unit at an academic health system in the US, the mortality rate decreased from 43.5% to 19.2% over the study period.[898] In another study in the UK, adjusted in-hospital mortality decreased from 52.2% in the first week of March 2020 to 16.8% in the last week of May 2020.[899] This may reflect the impact of changes in hospital strategy and clinical processes, and better adherence to evidence-based standard of care therapies for critical illness over time, such as high-flow nasal oxygen to avert intubation, prone positioning, and decreased use of mechanical ventilation. Further studies are needed to confirm these results and investigate causal mechanisms.

Infection fatality rate (IFR)

  • Defined as the proportion of deaths among all infected individuals including confirmed cases, undiagnosed cases (e.g., asymptomatic or mildly symptomatic cases), and unreported cases. The IFR gives a more accurate picture of the lethality of a disease compared with the case fatality rate.

  • Approximately 10% of the global population may have been infected by October 2020, with an estimated overall IFR of 0.15% to 0.2% (0.03% to 0.04% in those <70 years of age).[900]

  • The US Centers for Disease Control and Prevention’s current best estimate of the IFR, according to age (as of 10 September 2020):[136]

    • 0 to 19 years – 0.003%

    • 20 to 49 years – 0.02%

    • 50 to 69 years – 0.5%

    • ≥70 years – 5.4%.

  • Based on these figures, the overall IFR for people <70 years of age is approximately 0.18%.

  • The IFR can vary across locations. A meta-analysis reports the point estimate of the IFR to be 0.68% across populations, with high heterogeneity (as of July 2020). The rate varied across locations from 0.17% to 1.7%.[901]

  • Among people on board the Diamond Princess cruise ship, a unique situation where an accurate assessment of the IFR in a quarantined population can be made, the IFR was 0.85%. However, all deaths occurred in patients >70 years of age, and the rate in a younger, healthier population would be much lower.[902]

  • These estimates have limitations and are likely to change as more data emerge over the course of the pandemic.

Seroprevalence studies

  • Estimates of the IFR can be inferred from seroprevalence studies.

    • Worldwide seroprevalence estimates range between 0.37% and 22.1%, with a pooled estimate of 3.38% (based on data from 23 countries as of August 2020).[903]

    • UK: seroprevalence was 7.1% in the UK overall according to the first round of results of the UK Biobank COVID-19 antibody study. Previous infection was most common among people who lived in London (10.4%), and least common among those who lived in the south west of England and Scotland (4.4% in both).[904]

    • US: less than 10% of people are thought to have detectable severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antibodies based on data from July to September 2020.[905] Current seroprevalence estimates for 10 sites in the US are available. CDC: commercial laboratory seroprevalence survey data external link opens in a new window

    • China: seroprevalence was 3.2% to 3.8% in Wuhan, and decreased in other Chinese cities as the distance to the epicenter increased.[906]

  • These studies suggest that the prevalence of infections is much higher than the official case counts suggest, and therefore the virus is much less lethal than initially thought.

Case fatality rate (CFR)

  • Defined as the total number of deaths reported divided by the total number of detected cases reported. CFR is subject to selection bias as more severe/hospitalised cases are likely to be tested.

  • The World Health Organization’s current estimate of the global CFR is 2.2% (as of 21 February 2021).[907] This is much lower than the reported CFR of severe acute respiratory syndrome coronavirus (SARS), which was 10%, and Middle East respiratory syndrome (MERS), which was 37%.[46]

  • CFR varies considerably between countries.

    • In China, the overall CFR has been reported to be between 1.4% and 2.3% (0.9% in patients without comorbidities).[4][908]

  • CFR increases with age.

    • In the US, the majority of deaths were in patients aged ≥65 years. The CFR was highest among patients aged ≥85 years (10% to 27%), followed by those aged 65 to 84 years (3% to 11%), then those aged 55 to 64 years (1% to 3%), and finally those aged 20 to 54 years (<1%).[7]

    • In China, the majority of deaths were in patients aged ≥60 years.[4] The CFR was highest among patients aged ≥80 years (13.4%), followed by those aged 60 to 79 years (6.4%), and then those aged <60 years (0.32%).[908]

    • In Italy, the CFR was highest among patients aged ≥80 years (52.5%), followed by those aged 70 to 79 years (35.5%), and then those aged 60 to 69 years (8.5%).[909]

    • Deaths are rare in children.[7][21] In one study, 70% of deaths occurred in those aged 10 to 20 years, 20% in those aged 1 to 9 years, and 10% in children under 1 year of age.[910]

  • CFR increases with the presence of comorbidities.

    • In China, the majority of deaths were in patients who had pre-existing underlying health conditions (10.5% for cardiovascular disease, 7.3% for diabetes, 6.3% for chronic respiratory disease, 6% for hypertension, and 5.6% for cancer).[4]

  • CFR increases with disease severity.

    • The CFR is highest in patients with critical disease, ranging from 26% to 67% in studies.[4][911][912]

Limitations of IFR/CFR

  • Estimating the IFR and CFR in the early stages of a pandemic is subject to considerable uncertainties and estimates are likely to change as more data emerges. Rates tend to be high at the start of a pandemic and then trend downwards as more data becomes available.[913]

  • There is currently no set case definition of a confirmed case, and case definitions vary. A positive polymerase chain reaction (PCR) result is sometimes the only criterion for a case to be recognised; however, a positive PCR test does not necessarily equal a diagnosis of COVID-19, or mean that a person is infected or infectious.[914][915]

  • The number of deaths reported on a particular day may not accurately reflect the number of deaths from the previous day due to delays associated with reporting deaths. This makes it difficult to know whether deaths are falling over time in the short term.[916]

  • Patients who die 'with' COVID-19 and patients who die 'from' COVID-19 may be counted towards the death toll in some countries. For example, in Italy only 12% of death certificates reported direct causality from COVID-19, while 88% of patients who died had at least one comorbidity.[913][917]

Mortality rate by country

  • Mortality rates decreased sharply in the US over the first 6 months of the pandemic.[918]

  • The number of deaths (per 100,000 population) for different countries varies:[919]

    • South Korea – 0.7

    • Japan – 1.2

    • Australia – 3.3

    • Germany – 11.3

    • Canada – 24.6

    • France – 46.6

    • Sweden – 57.4

    • Italy – 59.1

    • US – 60.3

    • UK – 62.6

    • Spain – 65.0

    • Belgium – 86.8.

Prognostic factors

Prognostic factors that have been associated with increased risk of severe disease and mortality include:[920]

  • Increasing age

  • Male sex

  • Smoking

  • Presence of comorbidities (e.g., hypertension, diabetes, cardiovascular or cerebrovascular disease, arrhythmias, COPD, dementia, malignancy)

  • Dyspnoea

  • Tachypnoea

  • Hypoxaemia

  • Respiratory failure

  • Hypotension

  • Tachycardia

  • Lymphopenia

  • Leukocytosis

  • Neutrophilia

  • Thrombocytopenia

  • Hypoalbuminaemia

  • Liver, kidney impairment, or cardiac injury

  • Elevated inflammatory markers (C-reactive protein, procalcitonin, erythrocyte sedimentation rate)

  • Elevated lactate dehydrogenase

  • Elevated creatine kinase

  • Elevated cardiac markers

  • Elevated D-dimer

  • Elevated interleukin-6

  • Consolidative infiltrate or pleural effusion on chest imaging

  • High sequential organ failure assessment (SOFA) score.

The most common underlying diseases in deceased patients were hypertension, diabetes, and cardiovascular diseases.[921]

A ratio of arterial partial pressure of oxygen to fraction of inspired oxygen (PaO₂/FiO₂) ≤200 mmHg and respiratory failure at admission are also independently associated with an increased risk of in-hospital mortality.[922] Almost half of patients who received invasive mechanical ventilation died. The mortality rate was higher in older patients >80 years (84%) compared with younger patients ≤40 years (48%).[923]

Hospital readmission

Approximately 9% of over 106,000 patients were readmitted to the same hospital within 2 months of discharge from the initial hospitalisation. Multiple readmissions occurred in 1.6% of patients. The median time from discharge to the first readmission was 8 days. Less than 0.1% of patients died during readmission. Risk factors for readmission include:[924]

  • Age ≥65 years

  • Presence of chronic conditions (COPD, heart failure, diabetes, chronic kidney disease, obesity)

  • Hospitalisation within the 3 months preceding the first COVID-19 hospitalisation

  • Discharge to a skilled nursing facility or with home health care.


There is limited information about reinfection. Recurrent RT-PCR positivity in patients 1 to 60 days after recovery ranges between 7% to 23% in studies, with an estimated pooled rate of 12%.[925] Patients with longer initial illness and younger age were more likely to experience recurrent RT-PCR positivity, while those with severe disease, diabetes, and a low lymphocyte count were less likely.[926] It is currently unclear whether this is due to reinfection; whether it is due to factors such as the type of specimen collection and technical errors associated with swab testing, infection by mutated SARS-CoV-2, or persistent viral shedding; or whether the test result was a false-negative at the time of discharge.[927]

Studies have repeatedly reported positive RT-PCR tests for up to 90 days after initial infection; therefore, it is most likely that these cases are actually protracted initial infections. It is important to note that although persistent viral shedding has been reported for up to 90 days after the onset of infection, replication-competent virus has not been identified 10 to 20 days after the onset of symptoms (depending on disease severity).[928] A cohort study of 200 patients with past infection found that despite persistent pharyngeal RT-PCR positivity for up to 90 days after recovery, transmission to close contacts was not observed, indicating that these patients are not contagious at the post-symptomatic stage of infection.[929]

True cases of reinfection (defined as two episodes of infection at least 3 months apart by virus strains with different genomic sequences) have been reported in Hong Kong, India, Ecuador, and Belgium.[930][931][932][933] Two possible cases of reinfection have also been reported in the US; however, while different genomic variants were responsible for the two episodes in both men, the infections occurred less than 2 months apart.[934][935]

Cases of reinfection with SARS-CoV-2 variants have been reported in Brazil, the UK, and South Africa.[936][937][938][939]


The immune response, including duration of immunity, is not yet fully understood. There is evidence that suggests that infection with SARS-CoV-2 is likely to confer protective immunity against reinfection.[480][940][941][942][943] However, studies are of variable quality and comparison of findings is difficult.[944] A Public Health England study found that naturally acquired immunity, as a result of past infection, provides 83% protection against reinfection compared with people who have not had the disease previously. Protection appears to last for at least 5 months.[945] According to a large, retrospective study, people who were seropositive for SARS-CoV-2 appeared to be at lower risk for future infection, for at least several months.[946]

Emerging studies suggest that the majority of people develop a strong and broad T-cell response with both CD4+ and CD8+ T cells, and some have a memory phenotype.[947] A preprint study found that spike immunoglobulin G (IgG) was relatively stable over 6 months, spike-specific memory B cells were more abundant at 6 months than at 1 month, and CD4+ and CD8+ T cells declined with a half-life of 3 to 5 months in adults (mostly with mild disease) who recovered from COVID-19.[948] Another study in over 12,000 healthcare workers found that prior SARS-CoV-2 infection that generated antibody responses offered protection from reinfection for most people in the 6 months following infection.[949] This bodes well for potential longer-term immunity.

The immune response to SARS-CoV-2 involves both cell-mediated immunity and antibody production. Adaptive immunity to SARS-CoV-2 is thought to occur within the first 7 to 10 days of infection. A robust memory B-cell and plasmablast response is detected early in infection, with secretion of IgA and IgM antibodies by day 5 to 7, and IgG by day 7 to 10 from the onset of symptoms. IgA and IgM titres decline after approximately 28 days, and IgG titres peak at approximately 49 days. T cells are simultaneously activated in the first week of infection and SARS-CoV-2-specific memory CD4+ and CD8+ T cells peak within 2 weeks, but remain detectable for ≥100 days. Antibody and T-cell responses differ among individuals, and depend on age and disease severity. Preprint studies have found that T-cell response is likely to be present in most adults at least 6 to 8 months after primary infection.[950][951]

While there have been concerns about early declining IgG neutralising antibodies during convalescence, this is not thought to be an issue, because antibody levels always decline after the acute phase of an infection, and it is the levels of antibody titres after an infection that is important as this represents the generation of long-lived plasma cells to protect against subsequent infection.[950] Antibodies have been detected up to 8 months after infection.[952]

Analysis of a large cohort of convalescent serum donors in New York City suggests that 99.5% of patients with confirmed mild disease seroconvert 4 weeks after illness. IgG antibodies developed over a period of 7 to 50 days from symptom onset, and 5 to 49 days from symptom resolution. This suggests that people with mild disease may have the ability to develop immunity.[953] However, among patients who recovered from mild disease in China, neutralising antibody titres varied substantially.[954] There are data to suggest that asymptomatic people may have a weaker immune response to infection; however, this is yet to be confirmed.[955]

Testing of blood samples taken before the COVID-19 pandemic have shown that some people already have immune cells that recognise SARS-CoV-2. Studies have reported T-cell reactivity against SARS-CoV-2 in 20% to 50% of people with no known exposure to the virus.[956] Approximately 5% of uninfected adults and 62% of uninfected children aged 6 to 16 years had antibodies that recognise SARS-CoV-2 in one study.[957] This may be a consequence of true immune memory derived in part from previous infection with common cold coronaviruses, or from other unknown animal coronaviruses. However, further research into whether there is pre-existing immunity to SARS-CoV-2 in the human population is required.

Maternal IgG antibodies to SARS-CoV-2 have been found to transfer across the placenta after asymptomatic or symptomatic infection in pregnancy.[958]

Use of this content is subject to our disclaimer