Carbapenems are considered as last-resort antibiotics for the treatment of infections caused by multidrug-resistant Gram-negative bacteria. With the increasing use of carbapenems in clinical practice, the emergence of carbapenem-resistant pathogens now poses a great threat to human health. Currently, antibiotic options for the treatment of carbapenem-resistant Enterobacteriaceae (CRE) are very limited, with polymyxins, tigecycline, fosfomycin, and aminoglycosides as the mainstays of therapy. The need for new and effective anti-CRE therapies is urgent. Here, we describe the current understanding of issues related to CRE and review combination therapeutic strategies for CRE infections, including high-dose tigecycline, high-dose prolonged-infusion of carbapenem, and double carbapenem therapy. We also review the newly available antibiotics which have potential in the future treatment of CRE infections: ceftazidime/avibactam, which is active against KPC and OXA-48 producers; meropenem/vaborbactam, which is active against KPC producers; plazomicin, which is a next-generation aminoglycoside with in vitro activity against CRE; and eravacycline, which is a tetracycline class antibacterial with in vitro activity against CRE. Although direct evidence for CRE treatment is still lacking and the development of resistance is a concern, these new antibiotics provide additional therapeutic options for CRE infections. Finally, we review other potential anti-CRE antibiotics in development: imipenem/relebactam and cefiderocol. Currently, high-dose and combination strategies that may include the new β-lactam/β-lactamase inhibitors should be considered in severe CRE infections to maximize treatment success. In the future, when more treatment options are available, therapy for CRE infections should be individualized and based on molecular phenotypes of resistance, susceptibility profiles, disease severity, and patient characteristics. More high-quality studies are needed to guide effective treatment for infections caused by CRE.

Background: Hyperbilirubinaemia is a common complication of sepsis. Elevated bilirubin may induce inflammation and apoptosis. It was hypothesised that increased serum bilirubin on Intensive Care Unit (ICU) admission contributes to sepsis-related acute respiratory distress syndrome (ARDS). Methods: Serum bilirubin on ICU admission was measured in 1006 patients with sepsis. Serial serum bilirubin was analysed prospectively in patients with sepsis who had ARDS for a period of 28 days. The effects of clinical factors and variants of the UGT1A1 gene on serum bilirubin levels were determined. Outcomes were ARDS risk and mortality. Results: During 60-day follow-up, 326 patients with sepsis developed ARDS, of whom 144 died from ARDS. The hyperbilirubinaemia (⩾2.0 mg/dl) rate in patients with ARDS (22.4%) was higher than in those without ARDS (14.1%, p = 0.002). For each 1.0 mg/dl increase in admission bilirubin, ARDS risk and 28- and 60-day ARDS mortalities were increased by 7% (OR = 1.07; p = 0.003), 20% (OR = 1.20; p = 0.002) and 18% (OR = 1.18; p = 0.004), respectively. Compared with subjects with bilirubin levels <2.0 mg/dl, patients with hyperbilirubinaemia had higher risks of ARDS (OR = 2.12; p = 0.0007) and 28-day (OR = 2.24; p = 0.020) and 60-day ARDS mortalities (OR = 2.09; p = 0.020). In sepsis-related ARDS, serial bilirubin levels in non-survivors were consistently higher than in survivors (p<0.0001). Clinical variables explained 29.5% of the interindividual variation in bilirubin levels, whereas genetic variants of UGT1A1 contributed 7.5%. Conclusion: In sepsis, a higher serum bilirubin level on ICU admission is associated with subsequent ARDS development and mortality.

Objectives Current mortality prediction models used in the intensive care unit (ICU) have a limited role for specific diseases such as influenza, and we aimed to establish an explainable machine learning (ML) model for predicting mortality in critically ill influenza patients using a real-world severe influenza data set. Study design A cross-sectional retrospective multicentre study in Taiwan Setting Eight medical centres in Taiwan. Participants A total of 336 patients requiring ICU-admission for virology-proven influenza at eight hospitals during an influenza epidemic between October 2015 and March 2016. Primary and secondary outcome measures We employed extreme gradient boosting (XGBoost) to establish the prediction model, compared the performance with logistic regression (LR) and random forest (RF), demonstrated the feature importance categorised by clinical domains, and used SHapley Additive exPlanations (SHAP) for visualised interpretation. Results The data set contained 76 features of the 336 patients with severe influenza. The severity was apparently high, as shown by the high Acute Physiology and Chronic Health Evaluation II score (22, 17 to 29) and pneumonia severity index score (118, 88 to 151). XGBoost model (area under the curve (AUC): 0.842; 95% CI 0.749 to 0.928) outperformed RF (AUC: 0.809; 95% CI 0.629 to 0.891) and LR (AUC: 0.701; 95% CI 0.573 to 0.825) for predicting 30-day mortality. To give clinicians an intuitive understanding of feature exploitation, we stratified features by the clinical domain. The cumulative feature importance in the fluid balance domain, ventilation domain, laboratory data domain, demographic and symptom domain, management domain and severity score domain was 0.253, 0.113, 0.177, 0.140, 0.152 and 0.165, respectively. We further used SHAP plots to illustrate associations between features and 30-day mortality in critically ill influenza patients. Conclusions We used a real-world data set and applied an ML approach, mainly XGBoost, to establish a practical and explainable mortality prediction model in critically ill influenza patients.