We conducted a multicenter retrospective cohort study using data from the JIPAD, the largest intensive care database in Japan [15]. As of June 2023, the database included 250,672 individuals from 83 facilities. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [16].

The protocol for this research project was approved by a suitable constituted Institutional Ethics Committee (Committee of International University School of Health and Welfare Graduate School, Approval No. 23-Im-058, and the Ethics Committee of St. Luke’s International Hospital, Approval No. 23-R049), and it conforms to the provision of the Declaration of Helsinki.

This study focused on S-AKI, a subtype of AKI primarily caused by sepsis [17], including patients diagnosed with sepsis or septic shock who were considered to have developed S-AKI and required renal replacement therapy between 2015 and 2021. The exclusion criteria were as follows: age < 18 years, maintenance dialysis, cardiac arrest on admission, intensive care unit (ICU) readmissions, procedural admissions, transfer to the coronary care unit (CCU) or other ICUs/hospitals, receiving both IRRT and CRRT, unknown use of antihypertensive medications, missing height/weight data, or ICU stay < 24 h.

Patients were categorized into the IRRT or CRRT groups based on treatment during their ICU stay. The primary outcome was in-hospital mortality. Secondary outcomes were hospital and ICU length of stay.

The following data were extracted from the JIPAD: year of admission, age, sex, weight, height, ICU/hospital stay duration, chronic comorbidities (including congestive heart failure, respiratory failure, liver failure, cirrhosis, immunosuppressants use, lymphoma, acute leukemia, and metastatic cancer), ICU admission reason (planned/emergency), surgery type (planned/emergency), diagnostic codes, Acute Physiologic and Chronic Health Evaluation (APACHE) II and III scores, Simplified Acute Physiology Score II (SAPS II), and Sequential Organ Failure Assessment (SOFA) score. Data on mechanical ventilation, acute kidney injury, and the use of dopamine, noradrenaline, dobutamine, and adrenaline were also obtained. Body mass index (BMI) was calculated and categorized per the World Health Organization definitions.

Propensity score analysis was used to adjust for baseline differences between the CRRT and IRRT groups. Propensity scores were calculated using logistic regression with variables such as age, sex, BMI, comorbidities, planned surgery, surgical categories, APACHE II and III scores, SAPS II score, SOFA score, mechanical ventilation, AKI, and catecholamine levels. Propensity score analysis was used to adjust for baseline differences between the CRRT and IRRT group. The discriminatory ability of the logistic regression model to differentiate between patients receiving CRRT and IRRT was evaluated using the C-statistic. A C-statistic value between 0.6 and 0.9 was considered optimal for achieving sufficient discrimination without overfitting or insufficient separation between treatment groups [18]. Given the imbalance in group sizes, we used 1:3 nearest-neighbor matching without replacement to maximize comparability while preserving statistical power. It was performed based on estimated propensity scores, with a caliper width of 20% of the standard deviation of the logit-transformed propensity scores. Baseline characteristics balance was assessed using absolute standardized differences, with values < 10% considered balanced. Categorical variables were expressed as numbers and percentages, and continuous variables were expressed as medians with interquartile ranges (IQRs).

Risk differences and 95% confidence intervals (CIs) for in-hospital mortality were calculated after matching. The chi-square test was used for comparing groups. Median and IQR were calculated for hospital and ICU length of stay. The Wilcoxon rank-sum test was used to compare two groups. Two-sided P values < 0.05 were considered statistically significant. Regarding the sample-size calculation, based on observed mortality rates (50.0% in the CRRT group and 31.1% in the IRRT group) [19], with a significance level (α) of 0.05 and power of 80%, 209 patients in total (157 in the CRRT group and 52 in the IRRT group, at a 3:1 ratio) were required to detect a statistically significant difference in mortality between the two dialysis modalities. Analyses were conducted using SPSS software (version 29.0; SPSS Inc., Chicago, IL, USA) and R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).

We performed subgroup analyses in patients with septic shock and S-AKI to examine the effects of specific interventions and conditions.

We conducted a sensitivity analysis using inverse probability of treatment weighting (IPTW) to assess the robustness of findings for patients with S-AKI who received IRRT or CRRT, following the same methodology as the main analysis. Additionally, we estimated the average treatment effect (ATE) to further validate the robustness of our findings, using IPTW to account for potential confounding variables.