Sepsis was involved in 31.5% of all deaths worldwide in 2021

An analysis published in The Lancet Global Health on 21 October estimates 166 million sepsis cases (95% uncertainty interval 135–201 million) and 21.4 million (20.3–22.5 million) all-cause sepsis-related deaths worldwide in 2021, representing 31.5 percent of all global deaths [s1].

Nearly a third of everyone who died that year died with sepsis involved. That is the finding, and the reason it needs careful handling is in the phrase "sepsis-related."

What sepsis-related means

Sepsis is life-threatening organ dysfunction caused by a dysregulated host response to infection [s1]. It is a syndrome, not a disease with its own aetiology, and it typically appears on a death certificate alongside an underlying cause — the pneumonia, the stroke, the cancer.

That is why sepsis has historically been hard to count. The new analysis captures both explicit and implicit sepsis cases, estimating age-, location- and sex-specific fractions of sepsis-related deaths across 195 underlying causes of death and 22 infectious syndromes from 1990 to 2021 [s1]. The underlying data represent 149 million deaths and 4,290 location-years, drawn from multiple cause-of-death records, hospital data, minimally invasive tissue sampling, and linked death certificate and hospital records [s1]. Incidence was estimated by modelling case-fatality rates from 250 million hospital admissions and 7.82 million deaths across 1,310 location-years, then applying those rates to the death estimates [s1].

So the 21.4 million figure counts deaths in which sepsis was involved, not deaths for which sepsis was the underlying cause. Both numbers are legitimate; they answer different questions.

Why it is so much larger than the last estimate

The previous GBD sepsis analysis, published in 2020, estimated 48.9 million incident cases (38.9–62.9 million) and 11.0 million sepsis-related deaths (10.1–12.0 million) in 2017, representing 19.7 percent of all global deaths [s2].

Three things separate the two figures, and they should not be conflated.

The pandemic. The new analysis finds sepsis-related deaths decreased between 1990 and 2019, then surged in 2020 and 2021 [s1]. Sepsis-related deaths from infectious underlying causes fell from 11.8 million in 1990 to 8.34 million in 2019, then rose 86.4 percent to 15.5 million in 2021 [s1].

Method. The 2020 analysis used multiple cause-of-death data from 109 million individual death records and 8.7 million hospital records [s2]; the new one uses a substantially larger and more varied evidence base and a different modelling approach [s1]. A change in method changes the number independently of any change in the world.

The comparison year. 2021 was the second pandemic year. It is the worst available year for establishing a baseline.

Read together, the honest statement is that sepsis burden rose sharply during the pandemic from a level that had been falling, and that the 2017 and 2021 estimates are not directly subtractable.

Where the burden concentrates

People aged 15 and over saw increases in both incidence, up 230 percent, and mortality, up 26.3 percent, since 1990 [s1]. Those aged 70 and over had the highest sepsis-related mortality in 2021, at 9.28 million deaths (8.74–9.86 million) [s1].

The 230 percent incidence increase in adults is striking and should be read as partly a measurement effect. Sepsis incidence is estimated by applying modelled case-fatality to mortality [s1]; better hospital survival mechanically raises estimated incidence for the same number of deaths.

The shift away from infection as the underlying cause

The finding with the clearest clinical implication is about what sepsis is complicating.

Sepsis-related deaths from non-infectious underlying causes rose from 4.69 million in 1990 to 5.81 million in 2021 [s1]. The leading non-infectious underlying causes were stroke, chronic obstructive pulmonary disease, and cirrhosis [s1].

In 2021 the two most prominent infectious syndromes complicating sepsis-related deaths from non-infectious underlying causes were lower respiratory infections inclusive of COVID-19, and bloodstream infections inclusive of HIV and malaria — the latter at 3.08 million (2.83–3.35 million). The authors describe this as a consistent pattern since 1990 [s1].

The authors' interpretation is that sepsis is increasingly a complication of non-infectious causes of death [s1]. Practically, that points at patients admitted for a stroke or a COPD exacerbation who then acquire an infection — a hospital-acquired-infection and critical-care problem rather than a community infectious disease problem.

A discrepancy worth flagging

The published abstract reports lower respiratory infections inclusive of COVID-19 at 11.33 million with an uncertainty interval of 1.20–1.47 million [s1]. The point estimate falls outside its own stated interval, which indicates a transcription error somewhere in the reported figures. Readers should treat that specific number with caution until the record is clarified; the bloodstream infection figure of 3.08 million (2.83–3.35 million) is internally consistent [s1].

What to watch

The question this analysis raises but cannot answer is whether the 2020–21 surge recedes. The 1990–2019 trend was downward [s1]. Whether 2022 onward returns to that trajectory or settles at a higher level determines whether sepsis burden is a pandemic artefact or a durable change, and that will take several more years of data to establish.

Sources

• Global, regional, and national sepsis incidence and mortality, 1990–2021: a systematic analysis — The Lancet Global Health, 21 October 2025
• Global, regional, and national sepsis incidence and mortality, 1990–2017: analysis for the Global Burden of Disease Study — The Lancet, January 2020