Definition of Acute Kidney Injury by Urine Output in Critically Ill Patients

Acute Kidney Injury (AKI) is a common occurrence in critically ill patients in intensive care units and is strongly associated with an increased risk of complications, longer hospital stay and mortality. Research in this area was long hampered by the lack of a unified definition, but since 2012, there has been a widely accepted consensus definition for AKI: the KDIGO Clinical Practice Guideline for Acute Kidney Injury. This has facilitated comparisons of different studies and interventions.

According to this definition, AKI is based on either a) an acute elevation in blood creatinine levels (a breakdown product normally excreted in the urine) and/or b) reduced urine output. Kidney injury is further graded into three stages (1-3) based on increasing creatinine levels or decreasing urine output, where higher stages indicate worse kidney function and an increased risk of mortality.

However, the current definition of AKI also has weaknesses and limitations. Creatinine, a breakdown product primarily produced in the muscles, increases in the blood with kidney injury but reflects not the injury itself, but rather the impaired kidney function resulting from the injury. Thus, creatinine serves as a marker of kidney function rather than kidney injury, with elevated levels often appearing one or more days after the injury occurs. Additionally, creatinine levels are influenced by other factors unrelated to kidney function, such as an individual's muscle mass.

Urine output changes earlier than creatinine levels in cases of kidney injury and is easy to measure in intensive care patients, most of whom have urinary catheters. Reduced urine output is associated with an increased risk of mortality in intensive care patients, even when creatinine levels are normal. KDIGO defines reduced urine output as less than 0.5 milliliters of urine per kilogram of body weight per hour. However, this definition, which is essentially based on older studies of normal physiology, has been questioned in recent years. Reduced urine output is a normal physiological response to stress, such as surgery, making it challenging to distinguish physiological reductions in urine output from AKI. Several modern studies in surgical and intensive care contexts suggest that urine output lower than 0.5 ml/kg/h does not increase the risk of AKI or mortality. Based on these findings, lower thresholds have been proposed as a definition of AKI. More knowledge and studies that consider factors beyond urine output are needed to validate these findings. For instance, we lack knowledge about whether the relationship between urine output and patient outcomes is influenced by diuretic therapy, the amount of fluid administered, or fluid balance.

This research project aims to identify a possible threshold for urine output that reliably indicates AKI. The goal is also to examine the relationship between urine output and the severity of AKI, the need for renal replacement therapy (dialysis), long-term effects on kidney function, and mortality risk.

All adult patients treated in the intensive care unit at Uppsala University Hospital between 2016 and 2024 (up to the start of the study) will be included. Patients are identified through the Patient Administrative System for Intensive Care Units (PasIva). PasIva is an administrative support system used to report data to the Swedish Intensive Care Registry (SIR). Local data may need to be supplemented with information from SIR if necessary. The system provides data on reasons for ICU admission, severity of illness upon admission, interventions during intensive care, and 30- and 90-day mortality. Data in PasIva cannot be uploaded unless it is registered and formatted according to SIR's templates, ensuring validity.

From the local system Metavision, an electronic medical record system used in anesthesia and intensive care (Patient Data Management System, PDMS), data is retrieved on physiological variables such as urine output, fluid balance, body weight, medication, and organ-supportive treatments like renal replacement therapy (dialysis). Data is entered into Metavision hourly by ICU staff, ensuring high-resolution records. All medications administered in the relevant units are prescribed in Metavision.

From the medical record system COSMIC and databases managed by the Academic Laboratory at Uppsala University Hospital, data is collected on previous medical conditions, blood test results such as creatinine, and findings from diagnostic examinations. The Academic Laboratory is quality assured and accredited by SWEDAC according to SS-EN ISO 15189.

Patient data from the above sources will be merged to enable analysis. Data will be reviewed using standard procedures (e.g., measures of dispersion, plots) to detect anomalies indicative of errors in registration. We plan to divide the cohort into a subset for analysis and another for validation of the results. Multivariable statistical models will be applied to address the scientific questions described above.