During the current pandemic, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of coronavirus disease 2019 (Covid-19), has diversified considerably. As of September 2021, the World Health Organization had defined four variants of concern (alpha [B.1.1.7], beta [B.1.351], gamma [P.1], and delta [B.1.617.2 and AY]), as well as five variants of interest (eta [B.1.525], iota [B.1.526], kappa [B.1.617.1], lambda [C.37], and mu [B.1.621]).¹

Machine learning can predict outcomes and determine variables contributing to precise prediction, and can thus classify patients with different risk factors of outcomes. This study aimed to investigate the predictive accuracy for mortality and length of stay in intensive care unit (ICU) patients using machine learning, and to identify the variables contributing to the precise prediction or classification of patients. Patients (n = 12,747) admitted to the ICU at Chiba University Hospital were randomly assigned to the training and test cohorts. After learning using the variables on admission in the training cohort, the area under the curve (AUC) was analyzed in the test cohort to evaluate the predictive accuracy of the supervised machine learning classifiers, including random forest (RF) for outcomes (primary outcome, mortality; secondary outcome, length of ICU stay). The rank of the variables that contributed to the machine learning prediction was confirmed, and cluster analysis of the patients with risk factors of mortality was performed to identify the important variables associated with patient outcomes. Machine learning using RF revealed a high predictive value for mortality, with an AUC of 0.945 (95% confidence interval [CI] 0.922–0.977). In addition, RF showed high predictive value for short and long ICU stays, with AUCs of 0.881 (95% CI 0.876–0.908) and 0.889 (95% CI 0.849–0.936), respectively. Lactate dehydrogenase (LDH) was identified as a variable contributing to the precise prediction in machine learning for both mortality and length of ICU stay. LDH was also identified as a contributing variable to classify patients into sub-populations based on different risk factors of mortality. The machine learning algorithm could predict mortality and length of stay in ICU patients with high accuracy. LDH was identified as a contributing variable in mortality and length of ICU stay prediction and could be used to classify patients based on mortality risk.

Sepsis is a leading cause of death and long-term disability in developed countries. A comprehensive report on the incidence, clinical characteristics, and evolving management of sepsis is important. Thus, this study aimed to evaluate the characteristics, management, and outcomes of patients with severe sepsis in Japan. This is a cohort study of the Focused Outcomes Research in Emergency Care in Acute Respiratory Distress Syndrome, Sepsis, and Trauma (FORECAST) study, which was a multicenter, prospective cohort study conducted at 59 intensive care units (ICUs) from January 2016 to March 2017. We included adult patients with severe sepsis based on the sepsis-2 criteria. In total, 1184 patients (median age 73 years, interquartile range (IQR) 64–81) with severe sepsis were admitted to the ICU during the study period. The most common comorbidity was diabetes mellitus (23%). Moreover, approximately 63% of patients had septic shock. The median Sepsis-related Organ Failure Assessment (SOFA) score was 9 (IQR 6–11). The most common site of infection was the lung (31%). Approximately 54% of the participants had positive blood cultures. The compliance rates for the entire 3-h bundle, measurement of central venous pressure, and assessment of central venous oxygen saturation were 64%, 26%, and 7%, respectively. A multilevel logistic regression model showed that closed ICUs and non-university hospitals were more compliant with the entire 3-h bundle. The in-hospital mortality rate of patients with severe sepsis was 23% (21–26%). Older age, multiple comorbidities, suspected site of infection, and increasing SOFA scores correlated with in-hospital mortality, based on the generalized estimating equation model. The length of hospital stay was 24 (12–46) days. Approximately 37% of the patients were discharged home after recovery. Our prospective study showed that sepsis management in Japan was characterized by a high compliance rate for the 3-h bundle and low compliance rate for central venous catheter measurements. The in-hospital mortality rate in Japan was comparable to that of other developed countries. Only one third of the patients were discharged home, considering the aging population with multiple comorbidities in the ICUs in Japan. UMIN-CTR, UMIN000019742 . Registered on 16 November 2015.

Septic shock is the most severe form of sepsis. It is widely accepted that cytokines play pivotal roles in the pathophysiology of septic shock. We reported previously that continuous hemodiafiltration (CHDF) using a polymethylmethacrylate (PMMA) membrane hemofilter removed various cytokines from blood continuously and efficiently, mainly by adsorption to membrane matrix of the hemofilter. Furthermore, in April 2000, we introduced to clinical practice a rapid assay system that determines blood levels of IL (interleukin)-6 in approximately 30 min. This enabled us to routinely measure blood IL-6 as an index of cytokine cascade activation in critically ill patients for real-time clinical monitoring of hypercytokinemia. The aim of the present cohort study was to evaluate the clinical efficacy of PMMA-CHDF in septic shock, a typical condition associated with hypercytokinemia. Forty-three patients with septic shock were assessed by monitoring of blood IL-6 level with a rapid assay system and immediate initiation of critical care including PMMA-CHDF for cytokine removal. Following initiation of PMMA-CHDF, early improvement of hemodynamics was noted, as well as an increase in urine output. PMMA-CHDF treatment improved both hypercytokinemia (assessed by measurement of blood IL-6 level) and dysoxia (assessed by measurement of blood lactate level). The present findings suggest that cytokine-oriented critical care using PMMA-CHDF might be an effective strategy for the treatment of septic shock.

There is no one-size-fits-all empiric antimicrobial therapy for sepsis because the pathogens vary according to the site of infection and have changed over time. Therefore, updating knowledge on the spectrum of pathogens is necessary for the rapid administration of appropriate antimicrobials. The aim of this study was to elucidate the current spectrum of pathogens and its variation by site of infection in sepsis. This was a prospective nationwide cohort study of consecutive adult patients with sepsis in 59 intensive care units in Japan. The spectrum of pathogens was evaluated in all patients and in subgroups by site of infection. Regression analyses were conducted to evaluate the associations between the pathogens and mortality. The study cohort comprised 1184 patients. The most common pathogen was Escherichia coli (21.5%), followed by Klebsiella pneumoniae (9.0%). However, the pattern varied widely by site of infection; for example, gram-positive bacteria were the dominant pathogen in bone/soft tissue infection (55.7%) and cardiovascular infection (52.6%), but were rarely identified in urinary tract infection (6.4%). In contrast, gram-negative bacteria were the predominant pathogens in abdominal infection (38.4%) and urinary tract infection (72.0%). The highest mortality of 47.5% was observed in patients infected with methicillin-resistant Staphylococcus aureus, which was significantly associated with an increased risk of death (odds ratio 1.88, 95% confidence interval 1.22-2.91). This study revealed the current spectrum of pathogens and its variation based on the site of infection, which is essential for empiric antimicrobial therapy against sepsis.

Accurate and early identification of infection sites might help to drive crucial decisions regarding the treatment of sepsis. We aimed to determine the clinical and etiological features of infection according to sites among patients with severe sepsis in Japan. This secondary analysis of a multicenter, prospective cohort study included 59 intensive care units (ICU) and proceeded between January 2016 and March 2017. The study cohort comprised 1184 adults (≥ 16 years) who were admitted to an ICU with severe sepsis and septic shock diagnosed according to the sepsis-2 criteria. Sites of infection diagnosed by physicians in charge at the time of arrival comprised the lung, abdomen, urinary tract, soft tissue, bloodstream, central nervous system (CNS), and undifferentiated infections. The primary outcome was in-hospital mortality. The most common sites of infection were the lungs (31.0%), followed by intra-abdominal sites (26.3%), the urinary tract (18.4%), and soft tissue (10.9%). The characteristics of the patients with severe sepsis across seven major suspected infection sites were heterogeneous. Septic shock was more frequent among patients with intra-abdominal (72.2%) and urinary tract (70.2%) infections than other sites. The in-hospital mortality rate due to severe sepsis and septic shock of a pooled sample was 23.4% (range, 11.9% [urinary tract infection] to 47.6% [CNS infection]). After adjusting for clinical background, sepsis severity, and stratification according to the presence or absence of shock, variations in hospital mortality across seven major sites of infection remained essentially unchanged from those for crude in-hospital mortality; adjusted in-hospital mortality rates ranged from 7.7% (95%CI, − 0.3 to 15.8) for urinary tract infection without shock to 58.3% (95%CI, 21.0–95.7) for CNS infection with shock in a generalized estimating equation model. Intra-abdominal and urinary tract infections were statistically associated with less in-hospital mortality than pneumonia. Infections of the CNS were statistically associated with higher in-hospital mortality rates than pneumonia in a logistic regression model, but not in the generalized estimating equation model. In-hospital mortality and clinical features of patients with severe sepsis and septic shock were heterogeneous according to sites of infection.