Quality indicators (QIs) support measurement and improvement of ICU care; however, until 2017 there was no comprehensive, evidence-based QI set tailored to frontline adult ICU practice in Japan. Because evidence and practice evolve, periodic updating is required. We developed a Japanese ICU QI set in 2017–2018 and conducted a formal re-evaluation in 2025 to confirm current validity and update the core set. We used a modified RAND/UCLA appropriateness method. First, an initial list of candidate QIs was generated through a systematic literature review and major clinical guidelines. Next, a multidisciplinary panel of 14 Japanese experts (including intensivists, a nurse, a physiotherapist, and a clinical engineer) rated the appropriateness of these indicators using a nine-point Likert scale (Round 1). Following Round 1, a face-to-face consensus meeting was held to discuss, refine, add, or delete indicators based on scientific evidence, clinical importance, and feasibility, followed by Round 2 to finalize the 2018 consensus set. In 2025 (Round 3), the panel re-rated all indicators from the 2018 set using the same rating and classification framework and also rated newly proposed indicators reflecting contemporary practice. The systematic review yielded 44 initial candidate QIs. After the two-round rating process and the expert panel meeting, a 2018 consensus set of 38 QIs was established (13 structure indicators, 10 process indicators, and 15 outcome indicators). In Round 3 (2025), 37 indicators remained Appropriate, whereas one indicator was reclassified as Uncertain and was not retained in the updated core set. One newly proposed indicator was rated Appropriate and added, resulting in an updated 2025 core set of 38 indicators. All selected indicators were deemed appropriate and relevant for the Japanese ICU setting by the expert panel. Using a rigorous modified RAND/UCLA appropriateness method and a 2025 re-evaluation, we developed and updated a feasible, contextually adapted Japanese ICU QI set. The updated core set, with operational definitions and specified data sources, provides a foundation for national benchmarking and continuous quality improvement in Japanese ICUs. 330 words.

Background: Hemoglobin thresholds alone may not identify cardiovascular surgical ICU patients who derive physiologic benefit from red blood cell (RBC) transfusion. Mixed venous oxygen saturation (SvO₂) reflects the balance between oxygen delivery and consumption and may help identify patients with transfusion-responsive oxygen supply-demand mismatch. Methods: We conducted a retrospective observational study of adult cardiovascular surgical ICU patients who underwent a first 2-unit RBC transfusion episode with pre-transfusion hemoglobin (Hb) ≥ 7.5 g/dL and paired pre- and post-transfusion SvO₂ measurements. The primary outcome was SvO₂ responsiveness, defined as ΔSvO₂ ≥ 5 percentage points from pre-transfusion baseline to approximately 60 minutes after transfusion initiation. Multivariable logistic regression was used to identify predictors of response, and receiver operating characteristic analysis was used to determine the optimal pre-transfusion SvO₂ cutoff. Subgroup analyses evaluated higher Hb thresholds, and sensitivity analyses included overlap-weighted transfusion-versus-no-transfusion trajectory comparisons and multivariable linear regression using continuous ΔSvO₂. Results: Among 18,117 eligible transfusion episodes, 1,352 unique patients met the final inclusion criteria. Mean Hb increased from 9.82 ± 0.92 to 10.24 ± 0.98 g/dL, whereas mean SvO₂ changed only minimally at the cohort level (73.79 ± 9.91% to 73.86 ± 9.37%). However, lower pre-transfusion SvO₂ was associated with larger increases in SvO₂ after transfusion. In multivariable logistic regression, baseline SvO₂ was the only independent predictor of SvO₂ response (adjusted OR, 0.89 per 1% increase; 95% CI, 0.86–0.91; P < 0.001). Pre-transfusion SvO₂ predicted a response with an AUC of 0.778 (95% CI, 0.74–0.816), and the optimal cutoff was 69%. The inverse association between baseline SvO₂ and ΔSvO₂ was preserved in the Hb ≥ 9 g/dL and Hb ≥ 10 g/dL subgroups. In overlap-weighted sensitivity analyses, 0–6-hour trajectories of SvO₂, Hb, and DO₂i differed significantly between transfusion and no-transfusion groups (all group-by-time interaction P < 0.001). Conclusions: In cardiovascular surgical ICU patients with Hb ≥ 7.5 g/dL, a low pre-transfusion SvO₂ identified patients more likely to show a physiologic rise in SvO₂ after RBC transfusion. Pre-transfusion SvO₂ may complement Hb when evaluating transfusion need in this population, but prospective validation is required before physiologic SvO₂-guided transfusion can be recommended.

Red blood cell (RBC) transfusion decisions after cardiovascular surgery require integration of hemoglobin (Hb), hemodynamics, bleeding status, and oxygen supply–demand balance. Mixed venous oxygen saturation (SvO₂) reflects the global relationship between oxygen delivery and oxygen consumption, but an increase in SvO₂ does not necessarily indicate improved tissue oxygenation or clinical benefit. We evaluated acute SvO₂ changes after RBC transfusion in cardiovascular surgical ICU patients. We conducted a single-center retrospective cohort study of adult cardiovascular surgical ICU patients who received one RBC-equivalent transfusion with pre-transfusion Hb ≥ 7.5 g/dL and paired pre- and post-transfusion SvO₂ measurements. The primary outcome was an individual-level SvO₂ response, defined a priori as ΔSvO₂ ≥5

Aim:Maintaining rapid response team (RRT) response quality is difficult. A system that supports RRT assessment could potentially contribute to medical safety. Although rapid response system (RRS) triggers have been well-studied, studies on the prediction models of short-term prognosis after RRS activation are scarce. We aimed to develop a model to predict short-term outcomes after RRS activation using machine learning. Methods:This retrospective cohort study used the In-Hospital Emergency Registry in Japan, a multicentre RRS online registry. We collected data on patient demographics, treatment before RRS, RRT calls, and physiological parameters. The outcome was death within 24 h after RRS calls or unplanned transfers to an intensive care unit. To develop the eXtreme Gradient Boosted Tree Classifier (XGB) and Random Forest (RF) algorithms, a logistic regression (LR) algorithm was used. For model comparison, receiver-operating area under the curve (AUC) was evaluated and compared with those of the National Early Warning Score (NEWS) and Modified Early Warning Score (MEWS). Results:5414 cases were included in the study. The outcome occurred in 28.4% of the cases. The XGB model showed the highest AUC (0.798) compared to the RF model (0.796), LR model (0.785), NEWS (0.696), and MEWS (0.660). The most weighted feature in the XGB model was doctor activation, followed by hypotension as the activation criteria and usage of oxygen. Conclusions:We developed the first machine learning model for short-term prognosis after RRS. This model has the potential to support decision-making by RRT.

AimAlthough early detection of patients’ deterioration may improve outcomes, most of the detection criteria use on-the-spot values of vital signs. We investigated whether adding trend values over time enhanced the ability to predict adverse events among hospitalized patients.MethodsPatients who experienced adverse events, such as unexpected cardiac arrest or unplanned ICU admission were enrolled in this retrospective study. The association between the events and the combination of vital signs was evaluated at the time of the worst vital signs 0–8 hours before events (near the event) and at 24–48 hours before events (baseline). Multivariable logistic analysis was performed, and the area under the receiver operating characteristic curve (AUC) was used to assess the prediction power for adverse events among various combinations of vital sign parameters.ResultsAmong 24,509 in-patients, 54 patients experienced adverse events(cases) and 3,116 control patients eligible for data analysis were included. At the timepoint near the event, systolic blood pressure (SBP) was lower, heart rate (HR) and respiratory rate (RR) were higher in the case group, and this tendency was also observed at baseline. The AUC for event occurrence with reference to SBP, HR, and RR was lower when evaluated at baseline than at the timepoint near the event (0.85 [95%CI: 0.79–0.92] vs. 0.93 [0.88–0.97]). When the trend in RR was added to the formula constructed of baseline values of SBP, HR, and RR, the AUC increased to 0.92 [0.87–0.97].ConclusionTrends in RR may enhance the accuracy of predicting adverse events in hospitalized patients.

This study investigated the accuracy of a machine learning algorithm for predicting mortality in patients receiving rapid response system (RRS) activation. This retrospective cohort study used data from the In-Hospital Emergency Registry in Japan, which collects nationwide data on patients receiving RRS activation. The missing values in the dataset were replaced using multiple imputations (mode imputation, BayseRidge sklearn.linear model, and K-nearest neighbor model), and the enrolled patients were randomly assigned to the training and test cohorts. We established prediction models for 30-day mortality using the following four types of machine learning classifiers: Light Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting, random forest, and neural network. Fifty-two variables (patient characteristics, details of RRS activation, reasons for RRS initiation, and hospital capacity) were used to construct the prediction algorithm. The primary outcome was the accuracy of the prediction model for 30-day mortality. Overall, the data from 4,997 patients across 34 hospitals were analyzed. The machine learning algorithms using LightGBM demonstrated the highest predictive value for 30-day mortality (area under the receiver operating characteristic curve, 0.860 [95% confidence interval, 0.825–0.895]). The SHapley Additive exPlanations summary plot indicated that hospital capacity, site of incidence, code status, and abnormal vital signs within 24 h were important variables in the prediction model for 30-day mortality.

Extracorporeal cardiopulmonary resuscitation (ECPR) using veno-arterial extracorporeal membrane oxygenation (V-A ECMO) has the potential as a viable treatment for refractory out-of-hospital cardiac arrest (OHCA). While mechanical circulatory support devices, such as Impella (R) and left ventricular assist devices, are being increasingly used, initial ECPR often relies on V-A ECMO. Previous studies, including randomized controlled trials, reported the prognostic benefits of ECPR for shockable OHCA (SOHCA); however, its effectiveness for non-SOHCA (NSOHCA) remains unclear, with poorer neurological outcomes and the lower return of spontaneous circulation rates than for SOHCA being reported. The present study utilized data from the SOS-KANTO 2017 study to examine the impact of ECPR on the neurological outcomes of NSOHCA. Data from 2,502 OHCA cases were analyzed, with a focus on the relationship between ECPR and 90-day neurological outcomes. The results obtained showed significantly higher survival rates at 30 and 90 days and significantly better 90-day neurological outcomes in the ECMO attempt group than in the non-ECMO attempt group. A multivariate analysis identified ECPR as one of the significant independent predictors of favorable neurological outcomes. The prognosis of NSOHCA cases with CA was improved by ECPR using V-A ECMO, particularly in those where CPR was initiated within one minute of onset and the patient arrived at the hospital within 45 minutes. Factors associated with a favorable prognosis included a shorter time from onset to hospital arrival and the likelihood of acute coronary syndrome being the cause of CA. The present results suggest the potential of ECPR to improve the survival and the 90-day prognosis of NSOHCA, particularly when bystander CPR is initiated quickly and hospital arrival is prompt.

Abstract Aim The rapid response system (RRS) was initially aimed to improve patient outcomes. Recently, some studies have implicated that RRS might facilitate do‐not‐attempt‐resuscitation (DNAR) orders among patients, their families, and healthcare providers. This study aimed to examine the incidence and factors independently associated with DNAR orders newly implemented after RRS activation among deteriorating patients. Methods This observational study assessed patients who required RRS activation between 2012 and 2021 in Japan. We investigated patients’ characteristics and the incidence of new DNAR orders after RRS activation. Furthermore, we used multivariable hierarchical logistic regression models to explore independent predictors of new DNAR orders. Results We identified 7904 patients (median age, 72 years; 59% male) who required RRS activation at 29 facilities. Of the 7066 patients without pre‐existing DNAR orders before RRS activation, 394 (5.6%) had new DNAR orders. Multivariable hierarchical logistic regression analyses revealed that new DNAR orders were associated with age category (adjusted odds ratio [aOR], 1.56; 95% confidence interval, 1.12–2.17 [65–74 years old reference to 20–64 years old], aOR, 2.56; 1.92–3.42 [75–89 years old], and aOR, 6.58; 4.17–10.4 [90 years old]), malignancy (aOR, 1.82; 1.42–2.32), postoperative status (aOR, 0.45; 0.30–0.71), and National Early Warning Score 2 (aOR, 1.07; 1.02–1.12 [per 1 score]). Conclusion The incidence of new DNAR orders was one in 18 patients after RRS activation. The factors associated with new DNAR orders were age, malignancy, postoperative status, and National Early Warning Score 2.

Fulminant hepatic failure is a fatal complication of iron intoxication. Currently, there is no well-established treatment. A 23-year-old Japanese woman, with past medical history of iron deficiency anemia, presented with fulminant hepatic failure caused by iron intoxication. Multiple organ failure progressed despite plasma exchanges and deferoxamine therapy, and she subsequently developed acute respiratory distress syndrome. Even though Venous-Venous Extracorporeal Membrane Oxygenation was inserted, she expired. Early treatment for iron intoxication with whole bowel decontamination and deferoxamine infusion is crucial. Liver transplantation also remains a treatment option for patients with fulminant hepatic failure secondary to iron intoxication.