COVID-19

COVID-19 reshaped everyday life, put health systems under pressure and produced sharply different experiences across countries. Its impact cannot be captured by a single total. Reported deaths, excess mortality, hospital admissions, vaccination coverage and life expectancy each reveal a different part of the story. These twelve charts examine the pandemic’s early years, with comparisons anchored to clearly stated historical periods rather than a mixture of countries’ latest reports. They trace successive waves, the replacement of variants and the uneven progress of vaccination. A country reference table brings the main measures together, while the notes explain what the figures can—and cannot—tell us.

Coverage: A historical view, principally 2020–2021, with death and hospital waves through 2022 and vaccination trends through 2023. Source versions and exact dates are identified beneath each chart. These are not current pandemic totals.

1. Highest estimated excess mortality per person, 2020–2021

The largest estimated mortality burdens were not confined to the countries with the biggest populations. Once the figures are expressed per person, Bulgaria and North Macedonia head this comparison, followed by several other countries in eastern and south-eastern Europe. Bolivia and Peru also appear high in the ranking. The chart covers the combined years 2020 and 2021, so its rates should not be read as annual figures. It also excludes countries with populations below one million to keep very small populations from dominating the comparison. The uncertainty whiskers are part of the story: neighbouring positions in the ranking may not be meaningfully different where their ranges overlap. These are crude population rates, so differences in age structure remain relevant. Excess mortality looks beyond deaths officially recorded as COVID-19, but it still depends on an estimated baseline. This ranking therefore offers a broader view of the pandemic’s mortality burden rather than a definitive league table of national performance.

Sources: WHO excess-mortality estimates, May 2023; World Bank population. Top 15 among countries with a 2021 population of at least one million and matching WHO/World Bank data. Bars show WHO model means; whiskers show 95% uncertainty intervals. Two-year totals divided by 2021 population; not annual or age-standardised rates. WHO adjusts these estimates for deaths attributed to other shocks, including conflicts and natural disasters.

2. Successive death waves, 2020–2022

The pandemic did not arrive as one simultaneous global wave. The four countries shown experienced repeated surges, and their largest peaks occurred at different points in the calendar. Showing them on identical scales makes those differences visible without allowing the size of a country’s population to dominate. The United Kingdom’s early waves, Brazil’s severe 2021 period and the repeated rises in the United States and South Africa illustrate why a single end-of-period total cannot describe the whole experience. Each point represents deaths reported during a week, rather than necessarily deaths that occurred in that week. Reporting delays, corrections and backlogs can therefore make a peak sharper or move it away from the underlying outbreak. The panels are best read for broad timing and persistence. They also provide context for the vaccination and variant charts, although the overlap of events on a calendar does not, by itself, isolate the effect of any one intervention or variant.

Sources: WHO reported deaths via OWID; World Bank population. Selected countries shown on identical scales. Weekly sums of WHO reported deaths via OWID, divided by fixed 2021 World Bank populations. Reporting date rather than date of death; backlogs and revisions can create spikes. Last complete Sunday-ended week in 2022.

3. How variants replaced one another in the United States

The changing mix of variants helps explain why the pandemic’s later stages differed from its opening months. In the United States, Delta came to account for most of the estimated circulating infections before Omicron lineages rapidly replaced it. Within Omicron, successive groups then occupied larger shares of the total. The stacked areas show those substitutions particularly clearly: every period adds to approximately 100%, so an expanding band necessarily leaves less room for the others. This is a chart of composition, not the total number of infections. A variant can account for nearly every infection while overall circulation is rising or falling. The figures are CDC estimates derived from genomic surveillance, rather than an exhaustive sequencing of infections. Earlier variants are combined in the source’s other category, and separately named later lineages are grouped transparently in the notes. The chart’s strongest message is the repeated turnover in the viral population, not a ranking of variant severity.

Sources: CDC variant proportions. CDC weighted national estimates, two-week periods ending 15 May 2021–24 December 2022; release dated 20 June 2025. Aggregated published lineage categories: BA.2, BA.4 and BA.5 include names with those prefixes; separately named descendant lineages are in Other Omicron. Other/unassigned includes earlier variants. Estimated prevalence, not raw sample shares.

4. Vaccination coverage at the end of 2021

By the end of 2021, the selected countries had reached very different stages of their initial vaccination programmes. South Korea and Japan had completed initial courses for more than four-fifths of their populations in these records, while South Africa and India were at substantially lower levels. The middle segment distinguishes people who had started a course from those recorded as completing it. That distinction matters because counting doses alone can obscure how many different people have been reached. The bars use the total population as the denominator, including age groups that were not necessarily eligible at the same time in every country. They describe programme coverage at a historical checkpoint, not current immunity or protection against a later variant. Vaccine supply, eligibility, delivery capacity and uptake all influenced these patterns. The following income-group chart provides a wider context, showing how the differences evolved rather than treating one year-end snapshot as the complete vaccination story.

Sources: OWID vaccination data. 16 selected countries. Most recent report on or before 31 December 2021, no more than 30 days old; exact dates in the reference table. Initial course follows each vaccine protocol, not a uniform number of doses. Does not describe current protection or booster status.

5. COVID-19 death rates by age: United States, 2021

Age is one of the clearest dividing lines in the mortality data. In the United States during 2021, the recorded COVID-associated death rate was 0.4 per 100,000 among children aged 1–4 and 5–14, compared with 1,592.6 among people aged 85 and over. The rise through older adult groups is steep even after allowing for how many people belong to each group. A logarithmic axis makes the full range visible: equal distances represent multiplicative changes rather than equal additions. The numbers beside the dots retain the exact rates for readers who prefer direct comparisons. These are deaths relative to the population of each age group, not the probability of dying once infected. They also include COVID-19 recorded as a contributing cause, as well as the underlying cause. The chart helps explain why population age matters when comparing crude national death rates, while leaving room for the other health, exposure and social differences that shape outcomes.

Sources: CDC/NCHS mortality data. Final 2021 NCHS data as reproduced in CDC MMWR (5 May 2023). COVID-19 was an underlying or contributing cause. Age-specific rates, not infection-fatality risks. Logarithmic axis makes both low and high rates visible.

6. Reported COVID-19 deaths per capita around the world

The map shows a strong geographical contrast in the cumulative deaths reported by the end of 2021. Parts of Europe and the Americas recorded much higher totals per person than many countries in Africa and the western Pacific. That pattern is useful, but the colours need to be read alongside the excess-mortality charts. A low reported rate can reflect fewer deaths, less complete identification and registration, or a combination of the two. National definitions and reporting timetables also differ. Shading by population rate prevents the largest countries from automatically appearing most affected simply because more people live there. It does not remove differences in age structure or the completeness of surveillance. Grey areas indicate that the necessary data could not be matched, rather than that nobody died. The map is therefore an overview of the recorded geographical pattern, with the reference table supplying the exact values and the other mortality measures offering a broader perspective.

Sources: WHO reported deaths via OWID; World Bank population; Natural Earth map boundaries. Reported cumulative deaths available by 31 December 2021, divided by 2021 World Bank population. Reporting completeness and death definitions differ; low values can reflect under-reporting. Grey = no matched data, not zero. Natural Earth boundaries are illustrative.

7. Reported COVID deaths versus estimated excess mortality

The two measures of mortality tell noticeably different stories in several countries. India and South Africa have much larger modelled excess-death rates than their reported COVID death rates, while the gap is smaller in Brazil and Peru. In some countries, reported COVID deaths exceed the modelled excess total. That is not a contradiction: excess mortality measures the net difference from expected deaths across causes, while the reported COVID series counts deaths classified under a particular definition. Reductions in other causes can offset part of the increase, and the expected baseline is itself estimated. The distance between the dots should therefore not be labelled simply as missed COVID deaths. The two-year period and population denominator are aligned, but reporting lags and model uncertainty still matter. Read this chart as a comparison of two perspectives on mortality. It shows why reported totals alone are incomplete without implying that every difference has a single explanation or a precisely known cause.

Sources: WHO excess-mortality estimates, May 2023; WHO reported deaths via OWID; World Bank population. 2020–2021 WHO excess-mortality means versus cumulative deaths reported by 31 December 2021. Both divided by 2021 World Bank populations. Excess mortality includes indirect effects and changes in other causes; the difference is not a direct count of unrecorded COVID deaths. Uncertainty intervals are supplied in the data. WHO adjusts these estimates for deaths attributed to other shocks, including conflicts and natural disasters.

8. Life expectancy before and during the pandemic

The selected countries show sharply different changes in life expectancy between 2019 and 2021. Mexico and Peru experienced declines of more than four years in this World Bank series, while South Africa and India also recorded substantial falls. Australia, Japan and South Korea show small increases over the same two endpoints. These comparisons summarise the mortality conditions of each year across all ages and causes. They do not mean that every resident lost—or gained—the number of years between the two dots. Period life expectancy is a statistical measure based on that year’s mortality rates, not a forecast for a particular person. The choice of endpoints also matters: a country could experience a fall in 2020 followed by a partial recovery in 2021. This chart adds a broader demographic perspective to the death counts, while avoiding the assumption that every change in mortality during the pandemic was caused directly by the virus.

Sources: World Bank life expectancy. World Bank indicator SP.DYN.LE00.IN, both sexes, 2019 and 2021. Period life expectancy summarises mortality rates in each year; it is not a prediction of an individual lifespan. Changes reflect all causes, not COVID alone. Countries ordered by change.

9. The vaccination gap between income groups, 2021–2023

The initial vaccination rollout was highly uneven across income groups. At the end of 2021, the archived OWID series records initial-course completion at about 69% in high-income countries and 71% in upper-middle-income countries, compared with under 4% in low-income countries. The gap narrowed over the following two years as coverage increased in the lower-income groups, but it remained substantial at the end of 2023. Each line is weighted by population, so a country with many residents contributes more to its group’s result than a small country. These are not averages in which every country receives equal weight. The grouping also conceals differences between countries within the same income category. The chart describes how coverage developed; it does not isolate the relative roles of supply, delivery capacity, eligibility or willingness to receive a vaccine. Its central value is showing that a global rollout can advance rapidly in some populations while reaching others much later.

Sources: OWID vaccination data. OWID population-weighted income-group aggregates, sampled at the latest available date within each calendar month. Uses the archived dataset’s group membership and denominators; not a simple average of country percentages. Initial course only. Abrupt changes can reflect new reporting, including initial reporting from large countries.

10. When countries experienced their worst excess mortality

The heatmap places the timing of mortality shocks side by side. Each row follows one country through the same 24 months, while each column provides a cross-country view of a particular month. Stronger red cells indicate a larger estimated increase over expected deaths, making concentrated waves stand out without allowing population size to dominate. Blue cells indicate mortality below the model’s expectation. Countries with similar two-year totals can have very different patterns: one may experience a short, intense shock, while another has repeated increases spread across the period. The comparison uses a common WHO model framework, but the underlying mortality information is not equally complete everywhere. Some estimates rely more heavily on prediction than others. The colours also express percentage departures from expected deaths, not deaths per person, so this chart answers a different question from the opening ranking. Together, the two views distinguish the overall scale of mortality from when the pressure was greatest.

Sources: WHO excess-mortality estimates, May 2023. WHO May 2023 model estimates, January 2020–December 2021. Each cell = estimated excess deaths divided by expected deaths × 100. Blue means below expected; red above. Modelled values may use reported or predicted all-cause mortality inputs; input types are preserved in the data. WHO adjusts these estimates for deaths attributed to other shocks, including conflicts and natural disasters.

11. Hospital pressure through successive waves, 2020–2022

Hospital admissions reveal a dimension of the pandemic that death totals cannot fully describe. France, Spain and Denmark each experienced successive periods of increased hospital demand, with the timing and size of the rises changing across the three-year window. The panels use a shared scale and population-adjusted rates, making their patterns easier to compare. Admissions measure the flow of people entering hospital, not the number of beds occupied on a particular day. Length of stay, discharge practices and intensive-care needs also affect the pressure on services. National reporting definitions differ, including whether an admission is counted because COVID caused it or because a patient tested positive during treatment for another condition. For that reason, the strongest comparison here concerns the timing and persistence of waves, with more caution needed for exact differences in height. Missing reports remain gaps. They are not treated as evidence that admissions fell to zero or that hospital pressure disappeared.

Sources: OWID hospitalisation data. OWID hospitalisation series: rolling seven-day new hospital admissions per million divided by ten, sampled on Sundays. Missing observations are gaps, not zero. National definitions can include admissions with COVID rather than exclusively for COVID; compare timing more confidently than absolute levels.

12. Estimated global excess deaths by month, 2020–2021

WHO’s model places the largest monthly global excess-mortality estimate in May 2021, at roughly 1.74 million deaths. The cumulative estimate reaches about 14.86 million by the end of that year, with a published uncertainty interval of approximately 13.28–16.63 million. The upper panel shows how the estimated burden changed from month to month; the lower panel shows how those losses accumulated. These two perspectives are complementary. A falling monthly estimate does not reverse deaths that have already occurred, so the cumulative curve can continue rising even after the most severe period. The source supplies uncertainty bounds for the cumulative estimates, which are shown as a shaded band only in that panel. The figures cover excess mortality associated with the pandemic, including indirect effects, rather than only deaths with a confirmed COVID diagnosis. They are historical model estimates for 2020–2021 and should not be mistaken for a complete total covering every subsequent year of the pandemic.

Sources: WHO excess-mortality estimates, May 2023. WHO estimates published May 2023. The source provides uncertainty bounds for cumulative totals, not monthly totals; they appear only in the lower panel. Both panels cover all-cause excess mortality associated with the pandemic, including indirect effects. WHO adjusts these estimates for deaths attributed to other shocks, including conflicts and natural disasters.

Country reference table

All 192 countries with matching WHO excess-mortality estimates and 2021 World Bank populations. Mortality covers 2020–2021; vaccination figures use the latest available December 2021 report. A dash means unavailable. Rates are not age-standardised.

Reported totals are those available by 31 December 2021 and remain affected by reporting lags. WHO excess deaths are model means and may be negative; uncertainty bounds are included in the downloadable data. Vaccination reports more than 30 days before the cutoff are excluded. Life expectancy is World Bank period life expectancy at birth, both sexes.

Other pages

• Best Hospitals
• Current Top Scientists
• Space Exploration

Sources and image credits

Sources: WHO excess-mortality estimates, May 2023; WHO reported deaths via OWID; OWID vaccination data; CDC variant proportions; CDC/NCHS mortality data; World Bank life expectancy; World Bank population; OWID hospitalisation data; Natural Earth map boundaries. Historical coverage and source versions are specified above. Calculations by The World in Stats; assembled 10 November 2026. Body photographs are credited beside their images. Banner: SARS-CoV-2 illustration — CDC / Alissa Eckert and Dan Higgins; public domain in the United States.