Hyperfiltration in Critically Ill Older Adults: Incidence, Risk Factors, Time Course, and Predictive Performance of...

This study aimed to evaluate the incidence, associated risk factors, time course, and predictive performance of kidney function estimation equations for hyperfiltration in critically ill older adults.

This retrospective observational study evaluated 325 patients (median age, 76 years) admitted to the intensive care unit (ICU) of a tertiary-care university hospital, for a total of 2934 patient-days. The hyperfiltration threshold was defined using measured creatinine clearance (Clcr) > -0.883 × age + 167.398. Independent factors associated with hyperfiltration were identified via a multivariable logistic regression model, and time to onset and duration were evaluated using Kaplan-Meier curves. Additionally, the predictive performance of the measured Clcr was assessed using the Cockcroft-Gault equation. Bias was analyzed using the Bland-Altman analysis, and accuracy was evaluated using the percentage within 30% of the measured Clcr (P30).

Hyperfiltration occurred in 56% of the patients. Risk factors included male sex, high body mass index, trauma-related admission, vasopressor use, and low serum creatinine levels. The median time from ICU admission to hyperfiltration onset was 5 days, and the median duration of hyperfiltration episodes was 8 days. The Cockcroft-Gault equation substantially underestimated measured Clcr in patients with hyperfiltration. Even among those whose measured Clcr did not meet the conventional augmented renal clearance (ARC) threshold (> 130 mL/min), the equation exhibited a clinically significant negative bias (-35 mL/min) with limited accuracy (P30: 39%). This underestimation was even more pronounced in patients meeting the ARC criteria (bias: -62 mL/min; P30: 27%).

Hyperfiltration is highly prevalent in critically ill older adults. The Cockcroft-Gault equation substantially underestimated actual kidney function, even without meeting the ARC criteria. Continuous monitoring of the measured Clcr is recommended to avoid inappropriate medication dose reductions due to underestimation.