90-day mortality in AKI-D and CKD5-D patients admitted to the ICU

Published Dec 16, 2025 on Dryad. https://doi.org/10.5061/dryad.sqv9s4nht

Data files

Dec 16, 2025 version files 718.51 KB

Raw_data_-_Dryad_.xls

708.61 KB
README.md

9.91 KB

Abstract

Acute kidney injury (AKI) and chronic kidney disease (CKD) are related conditions commonly encountered in intensive care. While both may require renal replacement therapy (RRT), critically ill patients with dialysis-requiring AKI have significantly higher mortality and poor clinical outcomes than those with end-stage CKD. To compare 90-day mortality between patients with KDIGO stage 3 dialysis-requiring AKI (AKI-D) and those with stage 5 dialysis-dependent CKD (CKD5-D) admitted to the ICU. Secondary objectives were to identify independent predictors of increased 90-day mortality among AKI-D and CKD5-D patients, and to identify independent predictors of RRT at 90 days among patients who experienced an episode of AKI-D. This retrospective, single-center cohort study included adult patients with AKI-D or CKD5-D admitted to the ICU of a private quaternary hospital in São Paulo, Brazil. The study was approved by the Research Ethics Committee (CEP) of the Federal University of São Paulo, with a waiver of informed consent. The primary outcome was 90-day mortality. Multivariable logistic regression was used to identify independent predictors of mortality and dialysis dependence. A total of 2,377 patients were included: 1,878 (79%) with AKI-D and 499 (21%) with CKD5-D. The overall 90-day mortality was 42.5%, significantly higher in the AKI-D (50.3% vs. 13.2%; p <0.001) compared to CKD5-D group. Independent predictors of 90-day mortality among AKI-D and CKD5-D patients included older age, the need for vasoactive drugs, mechanical ventilation and diagnosis of sepsis. Among 731 AKI-D survivors assessed at 90 days, 175 (23.9%) remained dialysis-dependent. Risk factors for persistent dialysis dependence included older age, coronary artery disease, clinical (non-surgical) admission, and higher baseline creatinine. The risk of death in critically ill patients with AKI-D was more than three times higher than in individuals with CKD5-D. Although both AKI-D and CKD5-D patients experience loss of renal function, the factors correctable by dialysis alone do not account for the mortality difference between these groups. Rather, the severity of the acute illness, reflected by the need for invasive mechanical ventilation and vasoactive drugs, emerged as key determinants of short-term outcomes in this cohort.

Dataset DOI: 10.5061/dryad.sqv9s4nht

Description of the data and file structure

90-DAY MORTALITY IN AKI-D AND CKD 5-D PATIENTS ADMITTED TO THE ICU

File structure

The de-identified data bank presented here has been used for the analyses reported in the manuscript "90-day mortality in AKI-D and CKD 5-D patients admitted to the ICU." The file contains baseline characteristics (age, gender, race), comorbidities, rrt characteristics, and ICU and hospital stay characteristics. To maintain data anonymization, time-related variables (namely, ICU length of stay and hospital length of stay) have been modified using random noise to reduce their identifiability while preserving their analytical utility. Additionally, to mask the identification of patients, their age has been categorized according to the interquartile range. Some data points may be missing due to incomplete patient records or unavailable information in the electronic health record system. Missing data have been left as empty cells.

Data description

· age (categorical): patient age at hospital admission (years) categorized according to quartiles; 1 (age < 51); 2 (51 ≤ age < 63); 3 (63 ≤ age < 76); 4 (age ≥ 76)

· gender (categorical): 1 (female); 2 (male)

· race (categorical): 1 (asian_descent); 2 (black); 3 (mixed); 4 (white);

· diabetes_mellitus (categorical): 0 (no); 1 (yes)

· systemic _arterial_hypertension (categorical): 0 (no); 1 (yes)

· hepatic_disease (categorical): 0 (no); 1 (yes)

· cardiac_insufficiency (categorical): 0 (no); 1 (yes)

· coronary_insufficiency (categorical): 0 (no); 1 (yes)

· chronic_pneumopathy (categorical): 0 (no); 1 (yes)

· neoplasia (categorical): 0 (no); 1 (yes)

· malignant_hematologic_disease (categorical): 0 (no); 1 (yes)

· bone_marrow_transplant (categorical): 0 (no); 1 (yes)

· solid_organ_transplant (categorical): 0 (no); 1 (yes)

· type_organ_transplant (categorical): 1 (kidney); 2 (liver); 3 (heart); 4 (pancreas and kidney); 5 (lung); 6 (not applicable ); 7 (liver and kidney ); 8 (multivisceral); 9 (heart and kidney); 10 (pancreas)

· apacheII: (numerical): APACHE II score at ICU admission – it was available until 2012;

· saps3 (numerical): SAPS 3 score at ICU admission – it was implemented from 2013 onward

· vasoactive_drug (categorical): 0 (no); 1(yes) – during ICU stay

· mechanical_ventilation (categorical): 0 (no); 1(yes) – during ICU stay

· cause_icu_admission (categorical): 1 (infection); 2 (cardiovascular disease); 3 (pancreatitis); 4 (bleeding); 5 (drug toxicity); 6 (urgent surgery); 7 (trauma); 8 (major burn); 9 (elective surgery ); 10 (transplant); 11 (neurologic disease/alteration); 12 (respiratory failure ); 13 (metabolic disfunction); 14 (liver failure)

· type_patient (categorical): 1 (clinical); 2 (surgical)

· first_rrt (first renal replacement therapy modality) (categorical): 1 (CVVHDF: continuous venovenous hemodiafiltration; 2 (IHD: intermittent hemodialysis)

· los_icu (numerical): ICU length of stay, in days;

· los_hosp (numerical): hospital length of stay, in days

· discharge_without_rrt (categorical): 0 (no); 1(yes)

· without_rrt_90d (categorical): 0 (no); 1(yes)

· icu_outcome (categorical): 1 (death); 2 (dialiysis dependent at icu discharge); 3 (renal recovery = dialysis independent)

· hospital_outcome (categorical): 1 (death); 2 (dialiysis dependent at hospital discharge); 3 (renal recovery = dialysis independent)

· death_90d (primary outcome) (categorical): 0 (no); 1(yes) – not alive 90 days after icu admission

· type_renal_failure (categorical): 1 (Acute Kidney Injury (AKI-D)); 3 (Chronic Kidney Disease (CKD5-D))

· baseline_creatinine (numerical): baseline creatinine in mg/dl

· creatinine_hospital (numerical): first creatinine value available at hospital, mg/dl

· rrt_90d (secondary outcome) (categorical): 0 (no); 1(yes) - dialysis dependent 90 days after icu admission

· sepsis (categorical): 0 (no); 1(yes)

· sepsis_sources (categorical): 1 (urinary tract); 2 (skin/soft tissue); 3 (abdominal); 4 (cadiovascular system); 5 (central nervous system); 6 (unknown source); 7 (respiratory); 8 (not applicable ); 9 (bloodstream infection)

· baseline_gfr_value (numerical): baseline glomerular filtration rate in ml/min;

· baseline_gfr_stage (numerical): 1 (≥ 90 mL/min); 2 (89-60 mL/min); 3a (59-45 mL/min); 3b (44-30 mL/min); 4 (29-15 mL/min); 5 (< 15 mL/min) - baseline glomerular filtration rate categorized

Code/software

R softaware (R Core Team (2024). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/) was used in all statistical analyses.

Files and variables

File: Raw_data_-Dryad.xls

Description:

Variables

·      age (categorical): patient age at hospital admission (years) categorized according to quartiles; 1 (age < 51); 2 (51 ≤ age < 63); 3 (63 ≤ age < 76); 4 (age ≥ 76)

·      gender (categorical): 1 (female); 2 (male)

·      race (categorical): 1 (asian_descent); 2 (black); 3 (mixed); 4 (white)

·      diabetes_mellitus (categorical): 0 (no); 1 (yes)

·      systemic _arterial_hypertension (categorical): 0 (no); 1 (yes)

·      hepatic_disease (categorical): 0 (no); 1 (yes)

·      cardiac_insufficiency (categorical): 0 (no); 1 (yes)

·      coronary_insufficiency (categorical): 0 (no); 1 (yes)

·      chronic_pneumopathy (categorical): 0 (no); 1 (yes)

·      neoplasia (categorical): 0 (no); 1 (yes)

·      malignant_hematologic_disease (categorical): 0 (no); 1 (yes)

·      bone_marrow_transplant (categorical): 0 (no); 1 (yes)

·      solid_organ_transplant (categorical): 0 (no); 1 (yes)

·      type_organ_transplant (categorical): 1 (kidney); 2 (liver); 3 (heart); 4 (pancreas and kidney); 5 (lung); 6 (not applicable ); 7 (liver and kidney ); 8 (multivisceral); 9 (heart and kidney); 10 (pancreas)

·      apacheII: (numerical): APACHE II score at ICU admission – it was available until 2012

·      saps3 (numerical): SAPS 3 score at ICU admission – it was implemented from 2013 onward

·      vasoactive_drug (categorical): 0 (no); 1(yes) – during ICU stay

·      mechanical_ventilation (categorical): 0 (no); 1(yes) – during ICU stay

·      cause_icu_admission (categorical): 1 (infection); 2 (cardiovascular disease); 3 (pancreatitis); 4 (bleeding); 5 (drug toxicity); 6 (urgent surgery); 7 (trauma); 8 (major burn); 9 (elective surgery ); 10 (transplant); 11 (neurologic disease/alteration); 12 (respiratory failure ); 13 (metabolic disfunction); 14 (liver failure);

·      type_patient (categorical): 1 (clinical); 2 (surgical)

·      first_rrt (first renal replacement therapy modality) (categorical): 1 (CVVHDF: continuous venovenous hemodiafiltration; 2 (IHD: intermittent hemodialysis)

·      los_icu (numerical): ICU length of stay, in days

·      los_hosp (numerical): hospital length of stay, in days

·      discharge_without_rrt (categorical): 0 (no); 1(yes)

·      without_rrt_90d (categorical): 0 (no); 1(yes)

·      icu_outcome (categorical): 1 (death); 2 (dialiysis dependent at icu discharge); 3 (renal recovery = dialysis independent)

·      hospital_outcome (categorical): 1 (death); 2 (dialiysis dependent at hospital discharge); 3 (renal recovery = dialysis independent)

·      death_90d (primary outcome) (categorical): 0 (no); 1(yes) – not alive 90 days after icu admission

·      type_renal_failure (categorical): 1 (Acute Kidney Injury (AKI-D)); 3 (Chronic Kidney Disease (CKD5-D));

·      baseline_creatinine (numerical): baseline creatinine in mg/dl

·      creatinine_hospital (numerical): first creatinine value available at hospital, mg/dl

·      rrt_90d (secondary outcome) (categorical): 0 (no); 1(yes) - dialysis dependent 90 days after icu admission

·      sepsis (categorical): 0 (no); 1(yes)

·      sepsis_sources (categorical): 1 (urinary tract); 2 (skin/soft tissue); 3 (abdominal); 4 (cadiovascular system); 5 (central nervous system); 6 (unknown source); 7 (respiratory); 8 (not applicable ); 9 (bloodstream infection0

·      baseline_gfr_value (numerical): baseline glomerular filtration rate in ml/min;

·      baseline_gfr_stage (numerical): 1 (≥ 90 mL/min); 2 (89-60 mL/min); 3a (59-45 mL/min); 3b (44-30 mL/min); 4 (29-15 mL/min); 5 (< 15 mL/min) - baseline glomerular filtration rate categorized

Code/software

Access information

Other publicly accessible locations of the data:

None

Data was derived from the following sources:

Epimed
Specific data bank

Human subjects data

The study was approved by the Research Ethics Committee (CEP) of the Federal University of São Paulo, with a waiver of informed consent. Our research was conducted on human data in compliance with the Helsinki Declaration.

Study Design

This was a retrospective, single-center cohort study approved by the Research Ethics Committee (CEP) of the Federal University of São Paulo – UNIFESP (CAAE: 27698819300005505), with a waiver of informed consent. This study is reported in accordance with the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) statement.

Setting

The study was conducted in the Adult Intensive Care Unit (ICU) of Einstein Hospital Israelita, a private quaternary care hospital located in São Paulo, Brazil. At the time of the study, the general ICU had 41 beds and provided care to both clinical and surgical patients. Through a partnership with the Brazilian Ministry of Health (Sistema Único de Saúde [SUS]), the ICU also managed patients involved in solid organ transplantation programs, including transplant candidates, those on waiting lists, and recipients of kidney, liver, lung, heart, pancreas, and multivisceral transplantation.

Study participants

Adult patients who underwent dialysis during their ICU stay between January 1, 1999, and December 31, 2015, were eligible for inclusion. Exclusion criteria included patients younger than 18 years of age, those receiving extracorporeal therapy for exogenous intoxication, patients whose initial dialysis modality was peritoneal dialysis, those transferred to other facilities, or those with incomplete data. For patients with multiple ICU admissions, only the first ICU stay was included in the analysis

Data collection and study variables

Data were obtained from the Epimed Monitor System® (Epimed Solutions, Rio de Janeiro, Brazil), an electronic structured case report form in which trained ICU case managers entered patient data prospectively and a nephrology-specific database developed by the hospital’s multidisciplinary team. All data were anonymized prior to analysis.

Collected variables included demographic characteristics, comorbidities, and history of solid organ transplantation, presence of sepsis, reason for ICU admission, baseline renal function, initial RRT modality, severity scores at ICU admission, including the Acute Physiology and Chronic Health Evaluation II (APACHE II) and the Simplified Acute Physiology Score 3 (SAPS 3). APACHE II data were available from 1999 to 2012, and SAPS 3 was adopted from 2013 onward following the implementation of the Epimed® system. Additionally, information was collected on the ICU and hospital length of stay (LOS), use of vasoactive drugs and need for mechanical ventilation during the ICU stay, and mortality (ICU, in-hospital, and 90-day post-ICU admission).

Definitions and RRT modalities

Patients were divided into two groups (AKI-D and CKD5-D) according to the type of kidney failure. Patients with pre-existing CKD5-D were defined as those with documented CKD and on regular RRT for at least three months prior to ICU admission.

Baseline renal function was defined using the most recent serum creatinine values available prior to hospitalization. When such values were unavailable, the first creatinine measurement upon admission was used. Baseline glomerular filtration rate was estimated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2021 equation. Recovery of renal function was defined as independence from RRT.

Two primary RRT modalities were used: intermittent hemodialysis (IHD) and continuous venovenous hemodiafiltration (CVVHDF). CVVHDF was directed at patients with hemodynamic instability, severe neurological injury, and/or intracranial hypertension, requiring strict metabolic and volume control. The Prisma® or Prismaflex® (Gambro Lundia AB, Lund, Sweden) systems were used, with M100 or M150 filters (AN69 dialyzer membrane) and a blood flow rate of 100 to 120 mL/min. Beginning in 1999, regional citrate anticoagulation was employed, with citrate infusion limited to 2.5 mmol/L in cases of liver dysfunction (INR ≥ 2.5). The minimum prescribed continuous dialysis dose was 25 mL/kg/h.

For IHD, the Fresenius 4008 hemodialysis machine (Fresenius Medical Care AG, Bad Homburg, Germany) was used, with a blood flow of 300–400 mL/min and dialysate flow of 500–800 mL/min. When feasible, the dialysis target was a urea reduction ratio ≥ 65 % and Kt/V ≥ 1.2 per session. Anticoagulation was not routinely used during IHD; only priming with 100 mL of 0.9 % NaCl every 15 minutes was performed.

Outcomes

Our primary outcome of interest was 90-day mortality.

Statistical Analysis

Categorical variables were expressed as absolute and relative frequencies, while continuous variables were presented as median and interquartile range (IQR). Comparisons were performed between AKI-D and CKD5-D groups. Categorical variables were compared with chi-square test or Fisher's exact test as appropriate. Continuous variables were compared using an independent t-test or Mann-Whitney U test in cases of non-normal distribution. Normality was assessed using the Shapiro-Wilk test.

To identify independent predictors of 90-day mortality, analyses were conducted in the AKI-D and CKD5-D groups separately. Variables with a p-value < 0.20 in the univariate analysis were included in the multivariate analysis, which used a multiple logistic regression model to estimate odds ratios (OR) and 95% confidence intervals (95 % CI). Multicollinearity was assessed using the variance inflation factor (VIF), with values > 5 indicating significant collinearity. The linearity assumption for continuous variables included in the model was evaluated both graphically and using the likelihood ratio test. Variable selection was performed using the stepwise method, and model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test. The analysis of predictors of dialysis dependence at 90 days was restricted to patients with AKI-D who survived, had not undergone isolated or combined kidney transplantation, and had available renal function data at 90 days post-ICU admission. This analysis followed the same methodology used for the 90-day mortality outcome.

Statistical significance was set at p < 0.05. All analyses were performed using R software, version 4.3.4 (www.r-project.org).