Urine Output Criteria for Kidney Injury May Need Recalibration, Study Suggests

Now, a team of researchers in Sweden reports that the urine output thresholds used to stage the condition may be badly out of step with the mortality risk they are supposed to signal. In an analysis of nearly 5,700 intensive care admissions at a single Swedish hospital, the investigators found that the current urine output cut-offs underestimate death risk at the early stages of kidney injury and overestimate it at the most severe stage, when compared with the mortality associated with the corresponding creatinine-based stages.

The findings, published as a Correspondence in the journal Critical Care, come at a particularly consequential moment. The Kidney Disease: Improving Global Outcomes definition, known universally as KDIGO, remains the standard for diagnosing and staging acute kidney injury, and the proposed 2026 revision of that definition leaves the urine output criterion unchanged. The authors of the new analysis argue that this is a missed opportunity, because their data suggest the urine output component of the definition does not align with the prognostic weight carried by the serum creatinine component at the same stage.

The core of the problem is that creatinine and urine output, the two pillars of the KDIGO staging system, do not perform uniformly across disease severity. Previous work has shown that in stage 1 and stage 2 acute kidney injury, the mortality risk defined by urine output does not correspond to the mortality risk defined by creatinine. In other words, a patient classified as having mild kidney injury by urine output criteria may carry a very different risk profile than a patient with the same stage assigned by creatinine, undermining the assumption that stages are equivalent regardless of which criterion is used.

The research team, led by Annelie Barrueta Tenhunen of Uppsala University Hospital together with Rasmus Mossberg, Miklos Lipcsey, Mikael Eriksson and Michael Marks-Hultström, had previously found that mortality began to rise at urine output thresholds substantially lower than the current stage 1 cut-off of 0.5 millilitres per kilogram per hour, in a statistical model that included the Simplified Acute Physiology Score 3, a general measure of illness severity. But that score itself incorporates creatinine, which the authors note likely understates the contribution of urine output as a standalone risk factor, the very way the KDIGO definition is designed to use it.

To resolve this, the team reanalysed their dataset to identify urine output thresholds that produce mortality equivalent to creatinine-derived mortality. The cohort comprised 5,698 intensive care admissions from 4,999 unique adult patients, all aged 18 or older, admitted to the general intensive care unit at Uppsala University Hospital in Sweden between October 6, 2015 and April 30, 2024. Patients with pre-existing chronic kidney disease were excluded. Each admission was treated as an independent event, so patients with multiple stays could contribute more than once. The median age was 64 years, with an interquartile range of 49 to 73, 39 percent of patients were women, and most were emergent non-surgical admissions. The median estimated mortality risk from the physiology score was 28 percent, and observed mortality was 22 percent.

The mismatch with current criteria was striking. Compared with creatinine-defined staging in a recently published multicentre cohort, mortality by urine output stage according to the current KDIGO thresholds was underestimated at stage 1 by 37 percent and at stage 2 by 48 percent, while it was overestimated at stage 3 by 34 percent. These same creatinine-derived mortality estimates served as the reference points against which the team searched for matching mortality levels defined by the lowest sustained urine output recorded during each patient’s intensive care stay.

Methodologically, the analysis was deliberately thorough in how it characterised urine production. The lowest sustained urine output was calculated across all overlapping three-, six-, twelve- and twenty-four-hour windows using a moving-average calculation. For each window duration, the minimum average was taken as the lowest sustained value. Missing urine output recordings were set to zero unless followed by a measurement, in which case that volume was averaged over the entire preceding gap. The lowest sustained urine output was then used as the predictor in logistic regression models for fourteen-day mortality, adjusted for sex and age, using each of the four window lengths. Because the relationship between urine output and mortality is not a straight line, the team modelled it with penalized thin-plate regression splines, implemented as generalized additive models with a binomial link function. Missing data were minimal, and no imputation was performed.

The results point to thresholds far below the values clinicians currently use. Age- and sex-adjusted fourteen-day mortality across minimum sustained urine output identified mortality-equivalent thresholds for each creatinine-defined KDIGO stage at the three-, six- and twelve-hour windows. At twenty-four hours, thresholds could be identified for stage 1 and stage 2, but not for stage 3. All identified thresholds fell well below the current stage 1 criterion of 0.5 millilitres per kilogram per hour. Most notably, a urine output below 0.24 millilitres per kilogram per hour sustained for only three hours was associated with mortality equivalent to creatinine-defined stage 1 acute kidney injury, suggesting that even brief periods of oliguria, reduced urine production, carry prognostic significance that the current definition ignores.

The authors frame these findings as part of a broader effort to bring coherence to the markers used in kidney injury staging. The work extends their previous study, in which they established cystatin C levels with mortality equivalent to the creatinine-defined KDIGO stages, and it adds to earlier research highlighting the limitations of the urine output criteria in the present definition. Taken together, the results allow severity to be aligned across creatinine, cystatin C and urine output, three of the markers proposed in the updated KDIGO criteria. The authors note that a corresponding analysis has not yet been performed for markers of tissue injury, which they identify as an important area for future research.

Important caveats accompany the findings. The analysis was conducted at a single tertiary-care intensive care unit, so external validation in other healthcare systems and intensive care populations is needed before the thresholds could inform practice. Adjustment was limited to age and sex, meaning residual confounding cannot be excluded, and the authors emphasise that the thresholds should be regarded as prognostic rather than causal. The study also did not account for factors that influence urine output, such as diuretic administration, fluid status and continuous renal replacement therapy. Finally, the thresholds are reported at the precision supported by the data to provide a reference for future validation, but the authors acknowledge that rounded values may be more practical for clinical guidelines. Even with these limitations, the analysis delivers a pointed message to the developers of the revised KDIGO criteria: if the urine output criterion is left unchanged, the stages it assigns may continue to mean something quite different from the stages assigned by creatinine, at every level of severity.

Subject of Research: Recalibration of urine output thresholds for staging acute kidney injury to match creatinine-derived mortality

Article Title: Recalibrating the urine output criteria for acute kidney injury: stage-specific thresholds matched to creatinine-stage-derived mortality

Article References: Tenhunen, A. B., Mossberg, R., Lipcsey, M., Eriksson, M., & Marks-Hultström, M. (2026). Recalibrating the urine output criteria for acute kidney injury: stage-specific thresholds matched to creatinine-stage-derived mortality. Critical Care, 30(1), 551. https://doi.org/10.1186/s13054-026-06436-5

Image Credits: AI Generated

DOI: 10.1186/s13054-026-06436-5

Keywords: acute kidney injury, KDIGO criteria, urine output, creatinine, intensive care, mortality, oliguria, staging, SAPS3, cystatin C, Uppsala University Hospital, prognosis