Sepsis is underreported in Swedish intensive care units: A retrospective observational multicentre study

Background Sepsis is a common indication for admission to the intensive care unit (ICU). Since definitions vary across studies, comparisons of prevalence and outcomes have been challenging. We aimed to compare sepsis according to ICU discharge codes with sepsis according to Sepsis-3 criteria and to investigate the epidemiology of sepsis in the ICU. We hypothesized that sepsis using discharge codes is underreported. Methods Adult ICU admissions to four ICUs in Sweden between 2015 and 2017 were screened for sepsis according to the Sepsis-3 criteria. Medical records were reviewed and data extracted from the Swedish Intensive Care Registry. Results Of 5990 adult ICU patients, 28% fulfilled the Sepsis-3 criteria on admission, but only 31% of them had sepsis as the registered main diagnosis at ICU discharge. Of the 1654 Sepsis-3 patients, 38% met the septic shock criteria. The Sepsis-3 in-hospital mortality was 26% compared to 33% in patients with septic shock. The incidence rate for ICU-treated sepsis was 81 cases per 100 000 person-years. One in four had a positive blood culture, and 44% were culture negative. Conclusion This large Swedish multicentre study showed that 28% of adult ICU patients fulfilled the Sepsis-3 criteria, but only one third of them had sepsis according to ICU discharge codes. We could confirm our hypothesis, that sepsis is severely underreported in Swedish ICUs, and we conclude that discharge codes should not be used for quality control or research purposes.