To assess how the COVID-19 pandemic altered the association between sepsis etiology and in-hospital and 90-day mortality among ICU patients with sepsis in metropolitan France from 2018 through 2022.
Approach:
Data Sources: Anonymized data were obtained from the French National Hospital Discharge Database and the outpatient healthcare consumption database, both derived from the Système National des Données de Santé (SNDS).
Study Population: The analysis included 604,517 patients aged 15 years or older admitted to an ICU with sepsis. Sepsis-related hospital stays were identified using explicit and implicit ICD-10 criteria and classified as bacterial, viral, or fungal/parasitic. For patients with multiple sepsis stays, only the last stay was included.
Statistical Analysis: Hospitalizations were divided into 3 periods: 2018-2019, 2020-2021, and 2022. Logistic regression models assessed associations between sepsis etiology and in-hospital or 90-day mortality separately by period and co-infection status, with adjustment for age, sex, comorbidities, infection sites, and septic shock.
Key Findings:
Overall, 176,688 patients (29.2%) died during hospitalization, and 205,283 (34.0%) died within 90 days of admission.
Before the pandemic, viral sepsis was associated with lower in-hospital and 90-day mortality risk than presumed bacterial sepsis in patients with and without co-infections. During the pandemic, mortality risk associated with viral sepsis increased to levels similar to or higher than those associated with bacterial sepsis.
Among co-infected patients in 2022, viral sepsis was associated with higher in-hospital mortality than bacterial sepsis (adjusted odds ratio [aOR], 1.41; 99% CI, 1.15-1.71) and higher 90-day mortality (aOR, 1.23; 99% CI, 1.01-1.49).
Fungal or parasitic sepsis generally had mortality risks similar to bacterial sepsis, except among patients without co-infection in 2020-2021, when it was associated with higher in-hospital and 90-day mortality.
Interpretation:
Before the pandemic, viral-related sepsis had an approximately 30% lower mortality risk than bacterial sepsis. During the pandemic, its in-hospital mortality risk increased to levels similar to or higher than bacterial sepsis, particularly among patients with co-infections. Findings for 90-day mortality followed a similar but attenuated pattern.
Limitations:
The study relied on medical-administrative data without clinical validation of sepsis cases, creating the potential for misclassification, particularly for implicit sepsis.
Clinical data needed to calculate Sequential Organ Failure Assessment or quick Sequential Organ Failure Assessment scores were unavailable.
The investigators could not determine which infection occurred first among co-infected patients or distinguish community-acquired from healthcare-associated infections. Data on antibiotics administered during hospitalization and antimicrobial resistance were also unavailable.
Mortality was assessed using medical-administrative data rather than death certificates.
Conclusion:
Viral-related sepsis was associated with substantial mortality, and the COVID-19 pandemic changed the relationship between sepsis etiology and mortality. Among co-infected patients, viral sepsis emerged as a high-risk condition, with increased in-hospital and 90-day mortality compared with bacterial sepsis. The findings underscore the need for heightened vigilance, more detailed coding strategies for viral sepsis, and targeted management of co-infected sepsis during future pandemics.
