Urine output on the day of continuous renal replacement therapy discontinuation predicts long-term outcomes in patients with severe acute kidney injury
Article information
Abstract
Background
Continuous renal replacement therapy is crucial for treating critically ill patients with acute kidney injury. This study aimed to ascertain the association between urine output at continuous renal replacement therapy discontinuation and 1-year all-cause mortality and the development of end-stage renal disease (ESRD) in patients with acute kidney injury.
Methods
This study included 851 patients with acute kidney injury who underwent continuous renal replacement therapy at three tertiary hospitals from 2012 to 2020 and categorized them into four groups based on 24-hour urine output (mL/day) on the day of continuous renal replacement therapy discontinuation: <100, 100–499.9, 500–1,499.9, and ≥1,500.
Results
During the 1 year after continuous renal replacement therapy discontinuation, 333 (39.1%) of 851 patients died, and 150 (30.1%) of 499 patients progressed to ESRD. Multivariable analysis revealed that decreased urine output was associated with a higher risk for all-cause mortality (<100 mL/day: hazard ratio [HR] 3.07, 95% confidence interval [CI] 2.22–4.24, p < 0.01; 100–499.9 mL/day: HR 2.25, 95% CI 1.58–3.19, p < 0.01; 500–1,499.9 mL/day: HR 1.67, 95% CI 1.16–2.41, p < 0.01), and for the development of ESRD (<100 mL/day: odds ratio [OR] 16.01, 95% CI 7.54–33.99, p < 0.01; 100–499.9 mL/day: OR 7.17, 95% CI 3.30–15.56, p < 0.01; 500–1,499 mL/day: OR 3.11, 95% CI 1.40–6.93, p = 0.01).
Conclusion
This study highlights the importance of monitoring urine output in predicting the long-term risks of ESRD and mortality in patients with acute kidney injury undergoing continuous renal replacement therapy.
Introduction
Continuous renal replacement therapy (CRRT) is the preferred dialysis modality over intermittent hemodialysis for the treatment of critically ill patients with acute kidney injury (AKI) in the intensive care unit (ICU). Compared with IHD, CRRT offers the advantage of maintaining hemodynamic stability, thereby minimizing ischemic insults to vital organs, and has been associated with a greater likelihood of kidney function recovery [1–3]. Despite these benefits and advances in critical care management, some patients with severe AKI requiring CRRT do not fully recover renal function after CRRT discontinuation. Many of these patients remain dependent on maintenance dialysis following hospital discharge, highlighting the serious and persistent burden in the critically ill patients with AKI [4–6].
Incomplete recovery of kidney function after AKI has recently emerged as a potential risk factor for the development of end-stage renal disease (ESRD) and overall mortality [7–9]. However, in the absence of standardized post-AKI management guidelines, even patients who experienced severe AKI requiring CRRT are frequently discharged to outpatient clinics and eventually lost to follow-up [10]. To improve long-term prognosis in this high-risk population, comprehensive post-AKI management should include periodic kidney function assessment, pharmacological and non-pharmacological interventions to prevent further deterioration, and long-term planning for maintenance dialysis or referral for kidney transplantation [11,12]. Furthermore, early identification of risk factors for long-term outcomes, particularly during hospitalization, is essential for determining which patients would benefit from close follow-up and for promoting kidney function recovery in reversible cases.
Because an adequate urine output (UO) is imperative for maintaining euvolemia and electrolyte balance [13], UO should be considered when attempting to discontinue CRRT in patients with severe AKI [14]. Notably, UO at the time of CRRT discontinuation is a clinically valuable tool for monitoring kidney excretory function in real-time and deciding whether to re-administer kidney replacement therapy. Previous studies revealed that UO at CRRT discontinuation is a significant predictor of successful CRRT discontinuation [15–17]. However, these studies were limited by short-term follow-up durations, mostly within 1 week. Therefore, we investigated the value of UO on the day of CRRT discontinuation as a prognostic factor for long-term clinical outcomes, including overall mortality and the development of ESRD, in critically ill patients with AKI.
Methods
Ethics considerations
This study was approved by the Institutional Review Board of College of Medicine, The Catholic University of Korea (No. XC20RIDI0198). The requirement for informed consent was waived owing to the retrospective design of the study. All clinical investigations were conducted in accordance with the principles of the Declaration of Helsinki.
Study population and design
This retrospective cohort initially enrolled 891 critically ill patients with AKI older than 20 years who survived to discharge after undergoing CRRT for at least 3 days in the ICU at three tertiary hospitals between July 2012 and December 2020. Patients with ESRD who were on maintenance dialysis before CRRT initiation were excluded. Four patients who underwent CRRT during the perioperative period of nephrectomy or kidney transplantation and 36 patients with missing data on survival status were excluded. Finally, we included 851 in the mortality analysis. Except for those whose data on kidney outcomes were unavailable because of death or loss to follow-up, 499 of 851 patients were enrolled to investigate the development of ESRD. We divided the patients into four groups based on 24-hour UO on the day of CRRT discontinuation: <100 mL/day (anuria), 100–499.9 mL/day (oliguria), 500–1,499.9 mL/day, and ≥1,500 mL/day. Supplementary Fig. 1 (available online) shows the study design and a flowchart of study enrollment.
CRRT discontinuation was based on the achievement of hemodynamic stability and euvolemic status, improvement in metabolic derangements such as hyperkalemia and acidosis, resolution of the underlying cause of AKI, and the degree of kidney function recovery. Patients who did not recover kidney function after CRRT discontinuation reinitiated CRRT or transited to intermittent dialysis. At the time of CRRT discontinuation, diuretics were prescribed according to the established protocol. After an intravenous dose of 20 mg of furosemide was administered, the continuous infusion rate of furosemide was adjusted to achieve the UO of 0.5 mL/kg/hr or greater. Conversion to oral furosemide and the use of other diuretics, such as thiazide and spironolactone, were determined at the discretion of the treating physician to improve the fluid status of patients.
Data collection
Demographic, clinical, and laboratory data were collected from patient medical records. When patients required reinitiation of CRRT after a prior discontinuation, data from the first CRRT discontinuation attempt were included. Demographic data included age, sex, body mass index (BMI), comorbidities, cause of AKI, type of surgery, ICU length of stay, and duration of CRRT administration. Comorbidities included diabetes mellitus (DM), hypertension, cerebrovascular disease, liver cirrhosis, and malignancy. The causes of AKI were categorized into sepsis, cardiorenal syndrome, liver failure, ischemia, and nephrotoxin. Multiple choices were allowed for the causes of AKI. Types of surgery included cardiovascular, liver transplantation, abdominal, neurological, and fracture surgeries. Laboratory and clinical data, including mean blood pressure (MBP), 24-hour UO, and therapeutic interventions, such as the use of diuretics, vasopressors, or ventilators, were collected on the day of CRRT discontinuation.
Outcomes
The study endpoints were all-cause mortality and the development of ESRD following 1 year after CRRT discontinuation. ESRD was defined as the requirement for chronic maintenance dialysis for more than 3 months. Outcomes were identified by reviewing patient electronic medical records. The development of ESRD was assessed from medical records documented in the 3 months before and after the 1-year mark.
Statistical analysis
Continuous variables are expressed as means with standard deviations for normally distributed data and medians with interquartile ranges for non-normally distributed data. Categorical variables are expressed as numbers and percentages. Comparisons of continuous variables between groups were performed using analysis of variance and the Kruskal-Wallis test, as appropriate. The chi-square test was used to compare categorical variables.
A Kaplan-Meier curve was plotted for the cumulative incidence of death, and statistical differences were analyzed using the log-rank test. A Cox proportional hazards model was used to investigate risk factors for all-cause mortality, and logistic regression analysis was conducted to elucidate risk factors of ESRD. Multivariable analysis was adjusted for variables that were significant in univariable analysis (p < 0.10) or were clinically relevant to outcomes. The variables included in the multivariable Cox regression analysis were age, sex, BMI, MBP, comorbidities, causes of AKI, surgery, therapeutic interventions, duration of CRRT administration, laboratory data, and UO. Regarding the risk of ESRD, multivariable logistic analysis was adjusted for the use of diuretics and variables included in the multivariable Cox analysis.
Using receiver operating characteristic (ROC) curves, we evaluated the predictive ability of UO for study outcomes and determined the optimal cutoff values using Youden’s index. We used restricted spline functions to explore the linear relationship between UO and study outcomes using multivariable models. Statistical significance was set at p < 0.05. The data were analyzed using IBM SPSS version 22.0 (IBM Corp.) and R Statistical software, version 4.3.1 (The R Foundation for Statistical Computing).
Results
Baseline characteristics of the participants
Baseline characteristics of a total of 851 patients are presented in Table 1. The mean age of the participants was 63.6 ± 16.0 years, and 523 (61.5%) were male. The most common comorbidity was hypertension (33.8%), followed by malignancy (29.8%) and DM (21.6%). Sepsis was the leading cause of AKI (57.5%). Of 851 participants, 265 (31.1%), 184 (21.6%), 168 (19.7%), and 234 patients (27.5%) had UO <100 mL/day, 100–499.9 mL/day, 500–1,499.9 mL/day, and ≥1,500 mL/day, respectively.
Baseline characteristics of patients
Patients with UO <1,500 mL/day were significantly older (p < 0.01) and more likely to be female (p = 0.01) than patients with UO ≥1,500 mL/day. As patients had lower UO, the prevalence of DM and hypertension as comorbidities was higher (p = 0.04 and p = 0.03, respectively), and the prevalence of cirrhosis was lower (p < 0.01). Patients with lower UO were also more likely to have sepsis (p < 0.01), less likely to have liver failure as the cause of AKI (p < 0.01), and less likely to undergo surgery (p < 0.01). As patients had lower UO, the proportion of diuretic and vasopressor use significantly decreased (p < 0.01 and p = 0.01, respectively), and the duration of CRRT administration increased (p = 0.01). Patients with UO <100 mL/day had a higher serum creatinine level than patients with UO ≥1,500 mL/day (p < 0.01). Supplementary Table 1 (available online) shows the baseline characteristics of the 499 patients included in the analysis of the development of ESRD. The baseline characteristics of the total and subgroup participants were similar, but there were no significant differences among UO groups in the prevalence of DM, cardiovascular surgery, and sepsis as the cause of AKI in the subgroup of the 499 participants.
Incidence and risk factors of 1-year mortality
Among a total of 851 participants, 333 (39.1%) died within 1 year after CRRT discontinuation. Patients who had UO ≥1,500 mL/day had longer overall survival than those who had UO of 500–1,499.9 mL/day, 100–499.9 mL/day, and <100 mL/day (p < 0.01) (Fig. 1). In the multivariable Cox regression analysis, age, male sex, malignancy, sepsis, ventilator use, and lower UO were independently associated with a higher risk of 1-year all-cause mortality, whereas MBP and serum creatinine levels were independently associated with a lower risk of 1-year all-cause mortality after adjusting for other confounding factors (Table 2). When UO ≥1,500 mL/day was used as the reference category, UO <100 mL/day (hazard ratio [HR], 3.07; 95% confidence interval [CI], 2.22–4.24; p < 0.01), 100–499.9 mL/day (HR, 2.25; 95% CI, 1.58–3.19; p < 0.01), and 500–1,499.9 mL/day (HR, 1.67; 95% CI, 1.16–2.41; p < 0.01) were independently associated with all-cause mortality.
Kaplan-Meier survival curve for all-cause mortality according to urine output.
NS, no significance.
Factors associated with 1-year all-cause mortality
Incidence and risk factors of the development of end-stage renal disease
Among the 499 patients, 150 (30.1%) progressed to ESRD 1 year after discontinuation. The incidence of ESRD significantly decreased with increasing UO (UO <100 mL/day vs. 100–499.9 mL/day vs. 500–1,499.9 mL/day vs. ≥1,500 mL/day: 56.8% vs. 41.0% vs. 26.0% vs. 6.9%, p < 0.01). In the multivariable logistic regression analysis, age, DM, hypertension, serum creatinine levels, and lower UO were associated with a significantly higher risk of ESRD. Among these variables, UO was the strongest risk factor of ESRD. UO <100 mL/day (odds ratio [OR], 16.01; 95% CI, 7.54–33.99; p < 0.01), 100–499.9 mL/day (OR, 7.17; 95% CI, 3.30–15.56; p < 0.01), and 500–1,499.9 mL/day (OR, 3.11; 95% CI, 1.40–6.93; p = 0.01) were significantly associated with the development of ESRD when compared to UO ≥1,500 mL/day (Table 3).
Factors associated with the development of end-stage renal disease at 1 year
Predictive accuracy of kidney function parameters for long-term clinical outcomes
Next, we assessed the accuracy of UO and serum creatinine levels in predicting long-term clinical outcomes. The area under the ROC curve was 0.63 (95% CI, 0.59–0.67) for UO and 0.42 (95% CI, 0.38–0.46) for serum creatinine to predict all-cause mortality (Supplementary Fig. 2A, available online). For predicting the development of ESRD, the area under the ROC curve was 0.78 (95% CI, 0.74–0.82) for UO and 0.62 (95% CI, 0.57–0.68) for serum creatinine (Supplementary Fig. 2B, available online). The optimal cutoff value of UO was 807 mL/day, which was the same for both outcomes. Spline curves revealed that decreased UO was linearly associated with mortality (p for nonlinearity = 0.28) (Fig. 2A) and ESRD risk (p for nonlinearity = 0.08) (Fig. 2B). The optimal cutoff value of serum creatinine was 0.40 mg/dL and 1.72 mg/dL for the risks of mortality and ESRD, respectively.
Association between 24-hour urine output on the day of continuous renal replacement therapy discontinuation and long-term clinical outcomes.
Restricted cubic spline curves for risks of (A) mortality and (B) end-stage renal disease. The solid blue lines represent the hazard ratios or odds ratios, and the blue areas correspond to 95% confidence intervals (CIs) in the multivariable analysis. The horizontal dotted lines indicate the hazard ratio or odds ratio of 1.0, and the references for urine output were set at the cutoff values estimated using the receiver operating characteristic curve in each group.
Subgroup analysis
We performed subgroup analyses stratified by age, sex, DM, hypertension, sepsis, surgery, and diuretic use to explore whether clinical conditions affected the association between UO and the risk of ESRD (Fig. 3). UO was inversely associated with ESRD risk in all subgroups, with statistically significant trends observed (all p for trend <0.01). In subgroups of female, DM, non-sepsis, surgery, and diuretics use, only the UO range <500 mL/day showed a significantly elevated risk of ESRD, whereas UO 500–1,499.9 mL/day was not significantly different from the reference. No significant interaction was observed between UO and the clinical variables in relation to ESRD risk (all p for interaction >0.05).
Subgroup analysis of the associations between 24-hour urine output on the day of CRRT discontinuation and the risk of end-stage renal disease.
Adjusted ORs and p-values were estimated with the use of logistic regression model. Model included age, sex, urine output, body mass index, diabetes mellitus (DM), hypertension, sepsis, cardiorenal syndrome, liver failure, surgery, mean blood pressure, use of diuretics, duration of CRRT administration, hemoglobin, albumin, serum creatinine, and arterial pH.
CI, confidence interval; CRRT, continuous renal replacement therapy; OR, odds ratio.
Discussion
The present study demonstrated that UO on the day of CRRT discontinuation independently predicted long-term clinical outcomes, including 1-year all-cause mortality and the development of ESRD, in patients with severe AKI. Notably, UO can provide risk stratification for mortality and ESRD throughout a wide range of UO. UO remained significantly associated with ESRD risk regardless of age, sex, the presence of DM, hypertension or sepsis, undergoing surgery, and diuretics use.
Risk factors for all-cause mortality and kidney outcomes after CRRT discontinuation in patients with severe AKI have been consistently investigated. Most epidemiological studies have focused on short-term mortality within 90 days and successful CRRT liberation. UO on the day of CRRT discontinuation has been suggested to be a significant predictor of short-term kidney outcomes, defined as patients being free from dialysis for at least 7 days after CRRT discontinuation [15,16,18]. However, long-term kidney outcomes have rarely been investigated. Regarding long-term mortality, a single-center study suggested that prolonged ICU stay, history of malignancy, prolonged prothrombin time, and lower kidney function at discharge were risk factors for increased 3-year mortality after CRRT discontinuation [19]. However, UO was not considered as a variable of risk factors in the previous studies. This study is a novel attempt to find the value of UO on the day of CRRT discontinuation as an early prognostic indicator of the development of ESRD, the most clinically relevant outcome of long-term kidney replacement therapy, along with all-cause mortality.
Discontinuation of CRRT is the optimal time to assess early recovery of kidney function since hemodynamic stability has been achieved, further kidney insult cannot be expected [14], and also the adverse effects of CRRT administration itself have disappeared at this time [20,21]. Previous studies have investigated the association of kidney biomarkers measured around the time of CRRT discontinuation with clinical outcomes in patients with severe AKI. Stads et al. [22] suggested that changes in serum creatinine levels over 2 days after CRRT discontinuation could predict the restart of dialysis within 3 months. Kim et al. [23] found that the serum Cystatin C level measured at CRRT discontinuation was an independent predictor of the need for dialysis within 14 days after CRRT discontinuation. However, these studies mainly focused on the need for dialysis within days to a few months after CRRT discontinuation. Furthermore, serum creatinine could be confounded by factors such as muscle mass and CRRT clearance [24–26]. Our study suggests that UO on the day of CRRT discontinuation, another kidney function parameter, can serve as a better prognostic factor for long-term outcomes than serum creatinine in patients with AKI.
Non-recovery of kidney function after AKI events is an important determinant of long-term clinical consequences, including kidney disease progression and mortality [27,28]. Increasing evidence indicates that AKI and chronic kidney disease (CKD) are closely interconnected conditions with overlapping pathophysiological mechanisms. Maladaptive repair processes following ischemic injury, such as nephron loss and progressive fibrosis, have been proposed as underlying mechanisms of AKI-to-CKD transition [29]. In our study, patients with lower UO on the day of CRRT discontinuation had a higher risk of 1-year development of ESRD and also had lower kidney function at discharge (data not shown). These findings highlight a potential temporal relationship between immediate non-recovery of kidney function after CRRT discontinuation, subsequent renal decline, and progression to CKD. To further elucidate the mechanism by which UO at CRRT discontinuation serves as a predictor of long-term clinical outcomes, future studies should investigate kidney function trajectories and assess structural changes after CRRT discontinuation.
Oliguria and anuria, which are the criteria for AKI classification and highly specific markers of severe AKI, are associated with a higher risk of acute dialysis and in-hospital mortality in patients [30–32]. Oliguria was also found to have an association with kidney function recovery in patients who underwent CRRT for severe AKI. Wu et al. [4] found that oliguria on the day of dialysis discontinuation, age over 65 years, high Sequential Organ Failure Assessment score, and longer dialysis duration were risk factors for reinitiation of dialysis within 30 days after weaning from postoperative dialysis, including CRRT. However, this study was limited by its short follow-up duration and dichotomous UO stratification by oliguria. Our study stratifies UO into multiple levels and highlights the applicability of UO stratification to predicting long-term clinical outcomes after severe AKI requiring CRRT. Specifically, the non-oliguric range of UO, 500–1,499.9 mL/day, as well as anuria and oliguria on the day of CRRT discontinuation, were significantly associated with increased risks of 1-year mortality and ESRD. Furthermore, this study is noteworthy for identifying the UO cutoff value of 807 mL/day to predict long-term outcomes after AKI events and revealing that UO could provide risk stratification throughout a wide range of UO.
This study has several limitations. First, its retrospective and observational design may introduce bias due to missing data and unmeasured confounding factors. Our analysis did not account for preexisting CKD, one of the risk factors for the progression to ESRD after AKI [33,34]. Second, we could not provide the temporal association between UO and kidney function recovery because of the absence of kidney function data during follow-up. Finally, the prescription of CRRT, such as the mode or dose, mainly determines changes in laboratory findings, but could not be included due to a lack of information. On the other hand, in patients whose data were collected after CRRT discontinuation, the time lag from CRRT discontinuation to laboratory data measurement could not assure steady state values. Despite these limitations, the strength of this study is that it demonstrated the clinical significance of UO on the day of CRRT discontinuation in predicting long-term outcomes after severe AKI in a relatively larger number of patients.
In conclusion, UO on the day of CRRT discontinuation was independently associated with 1-year all-cause mortality and the development of ESRD in critically ill patients with AKI. The present study underscores that UO on the day of CRRT discontinuation could provide risk stratification for long-term clinical outcomes in the early stages of kidney function recovery throughout a wide range of UO, helping to guide physicians in planning for long-term kidney outcomes.
Supplementary Materials
Supplementary data are available at Kidney Research and Clinical Practice online (https://doi.org/10.23876/j.krcp.25.200).
Notes
Conflicts of interest
All authors have no conflicts of interest to declare.
Funding
This work was supported by a National Research Foundation of Korea (NRF) grant, funded by the Korean government (MSIT) (No. 2021R1C1C1012208).
Data sharing statement
The data presented in this study are available from the corresponding author upon reasonable request.
Authors’ contributions
Conceptualization, Project administration, Supervision: YAH
Data curation, Visualization: SK
Formal analysis: EJK, HDK, SK
Investigation, Resources: SY, JM
Methodology: YAH, SK
Validation: ESK, BHC
Writing–original draft: SK
Writing–review & editing: ESK, BHC, YAH
All the authors read and approved the final manuscript.