Major Adverse Kidney Events in Pediatric Continuous Kidney Replacement Therapy

This cohort study examines incidence of major adverse kidney events at 90 days and risk factors associated with them among youths treated with continuous kidney replacement therapy.


Introduction
Children, adolescents, and young adults with critical illness have a high likelihood of developing acute kidney injury (AKI) and disorders of fluid balance, including pathologic fluid overload. 1Patients with severe AKI and fluid overload are commonly treated with continuous kidney replacement therapy (CKRT) to allow for electrolyte, acid base, and fluid balance optimization.Youths treated with CKRT represent one of the most critically ill populations treated in intensive care units (ICUs).While there is significant literature highlighting the short-term morbidity and mortality, the incidence and risk factors associated with modern consensus composite outcomes that go beyond mortality have not been described in youths treated with CKRT, to our knowledge.
To standardize the reporting of cohort studies and clinical trials in critical care nephrology, the National Institute of Diabetes and Digestive Kidney Diseases workgroup on clinical trials put forth the major adverse kidney events at 90 days (MAKE-90) composite outcome.The power of the MAKE-90 outcome is that it expands outcome reporting from mortality alone to include patient-centered end points (new kidney replacement therapy and persistent kidney dysfunction). 2,3While MAKE-90 has been studied in adults treated with CKRT, it has yet to be studied in pediatric patients, to our knowledge. 4A better understanding of the incidence of MAKE-90 and patient-level risk factors associated with MAKE-90 will be a critical step to improve clinical care, develop future clinical trials, and ultimately improve outcomes.Although large studies from 2017 to 2021 [5][6][7] have aimed to investigate the optimal time to initiate CKRT, there is no consensus on when to trial CKRT liberation.
In adults with critically illness, failure to successfully liberate from CKRT is associated with an increased risk of mortality, dialysis dependence, and continued kidney dysfunction, all components of MAKE-90. 4,8,9 begin to address these knowledge gaps, we performed a planned analysis of the Worldwide Exploration of Renal Replacement Outcomes Collaborative in Kidney Disease (WE-ROCK) registry of more than 900 youths treated with CKRT.The aims of this study were to describe the incidence of MAKE-90, identify independent patient clinical characteristics, and identify independent CKRT parameters (including liberation patterns) associated with MAKE-90 in youths treated with CKRT for AKI or fluid overload.We hypothesized that the incidence of MAKE-90 would be high in this population and that there would be patient-level risk factors (ie, primary comorbidities, reason for hospital admission, and illness severity) and CKRT parameters (ie, duration, timing of initiation, and liberation pattern) associated with MAKE-90.

Methods
Each participating site in this cohort study sought institutional board approval, with a waiver of informed consent or parental permission due to the retrospective nature of the study.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline for reporting study results.

Study Design and Setting
Data for the study were collected from the WE-ROCK of pediatric and young adult patients who required CKRT.WE-ROCK was formed to create an international registry of patients receiving CKRT

Outcomes
Our primary outcome was MAKE-90.Based on previous studies, we defined MAKE-90 as a binary, composite outcome that includes any of the following: persistent kidney dysfunction, defined as a 25% or greater decline in eGFR from the reference value at 90 days; continued need for any form of KRT at 90 days; and mortality, defined as death from any cause at 90 days. 2We chose a 90-day period in accordance with the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, which define chronic kidney disease as kidney dysfunction lasting at least 90 days. 16If a patient died, the death end point was met but not persistent kidney dysfunction or dialysis dependence.If a patient met the dialysis end point, the individual was also considered to have persistent kidney dysfunction. 17Our secondary outcome was mortality.

Statistical Analysis
Continuous data are reported as medians with IQRs, while categorical data are reported as frequencies with percentages.We compared baseline characteristics between patients with and without MAKE-90 using the Wilcoxon rank-sum and χ 2 test for continuous and categorical variables, respectively.Kaplan-Meier curves were generated to depict the cumulative probability of death over 90 days for the overall cohort and for each liberation pattern group.The log-rank test was used to compare mortality differences between groups.Time zero was the day of CKRT initiation.Patients were censored if they did not experience death within 90 days of CKRT liberation.2A).There was also a significant difference in mortality comparing the patient group that successfully liberated with the group that had CKRT reinstituted (log-rank P = .006)(Figure 2B).Patients who successfully liberated had the lowest probability of MAKE-90 (33.6%; 95% CI, 25.6%-42.7%)compared with not attempted (95.6%; 95% CI, 93.3%-97.5%)and reinstituted (61.0%; 95% CI, 51.2%-69.9%)groups (eFigure 2 in Supplement 1).

Independent Factors Associated With MAKE-90
There was no association between calendar year of treatment and MAKE-90 in the primary outcome model; thus, this variable was not included in final models.The multivariable regression model to estimate odds of MAKE-90 is shown in   of comorbidities was protective against persistent kidney dysfunction or continued need for dialysis (eTable 2 in Supplement 1).

Discussion
This cohort study found that MAKE-90 outcomes were common and occurred in almost two-thirds of youths requiring CKRT.To our knowledge, this is the first study to characterize MAKE in a multicenter cohort exclusive to children, adolescents, and young adults receiving CKRT.In contrast to investigations and consensus guidelines focused on outcomes associated with ventilator liberation in patients with acute lung injury, there is a paucity of data regarding patient outcomes related to CKRT discontinuation. 18This study identified patient level-factors associated with MAKE-90, including the novel description of the association of CKRT liberation patterns with MAKE-90.We found that patients who successfully liberated from CKRT within the first 28 days after starting therapy had significantly greater odds of survival and avoidance of other adverse kidney events at 90 days.
While we confirmed that youths treated with CKRT were at high risk of mortality at 90 days (368 patients [38.0%]), the power in this study is in describing the high burden of patient-level kidney outcomes in this population.We found that 262 patients (27.0%) had persistent kidney dysfunction and 91 patients (9.4%) continued to be treated with KRT at 90 days.These data may serve to provide clinicians with information to counsel families.Furthermore, these data may be used to design clinical trials and develop standardized follow-up clinics for these patients at increased risk.
To improve outcomes in youths treated with CKRT, it is important to understand risk factors associated with MAKE-90 to identify high-risk populations and potential targets for intervention.We identified patient (eg, comorbidities) and CKRT (eg, time to initiation and liberation pattern) characteristics associated with adverse kidney outcomes 90 days after CKRT initiation.It is known that AKI is common in patients with congenital heart disease and is associated with their morbidity and mortality. 19,20In this study, we found the highest odds of MAKE-90 outcomes in patients with cardiac comorbidities.We found that patients admitted for congenital heart disease or cardiomyopathy had the highest rates of MAKE-90.Further investigation into this cohort is warranted to identify modifiable factors associated with MAKE-90.
The optimal timing of CKRT initiation remains a controversial topic in critical care nephrology across age groups.Interestingly, we found that a longer time to CKRT initiation anchored to ICU admission was independently associated with MAKE-90.There was a 7% increase in the odds of MAKE-90 for patients who started CKRT on ICU day 6 compared with those who were initiated on ICU day 1.These findings require further exploration.Might there be a clinical phenotype that should be explored to better determine which patients will benefit from earlier compared with later CKRT initiation?Results from the Standard vs Accelerated Initiation of Renal Replacement Therapy in Acute Kidney Injury (STARRT-AKI) trial 21 in adults support not starting CKRT until it is absolutely indicated.
Further exploration of this registry is necessary to speculate rationale for the association.
While significant time and effort have been spent evaluating CKRT initiation timing, 6,7,21,22 little is known about liberation patterns, practices, or the association of liberation attempts with outcomes.Notably, the worst outcomes in our study cohort occurred in patients in whom liberation was never attempted.4][25] KRT may also impair kidney recovery after AKI. 26 In our study, 33.8% of patients were successfully liberated and fulfilled the criteria of achieving more than 72 hours without KRT after therapy discontinuation.
In a single-center study examining outcomes associated with CKRT liberation in adults that similarly used a 72-hour observation period, Liu and colleagues 4 also reported a low occurrence of liberation success.It is quite possible that the high rate of liberation failure seen in both studies reflects variations in clinical practice.Currently, there are no consensus guidelines providing recommendations for clinicians deciding on the optimal timing of CKRT discontinuation.In a 2022 cross-sectional survey 27,28 of intensivists and nurses working in 20 European pediatric ICUs, investigators reported no consensus regarding decisions on how or when to liberate patients from CKRT.Our study highlights the need for a shift in the paradigm of how we study CKRT in youths, from focusing on CKRT initiation to a more wholistic approach systematically evaluating liberation.
Our study has several notable strengths.Results are representative of a contemporary, multicenter, international cohort of pediatric patients with CKRT and no prior history of dialysis dependence.To our knowledge, this is the first study in pediatrics and one of few studies overall to examine the association of liberation from CKRT with clinically meaningful outcomes.Data collection occurred over a relatively short period, decreasing the likelihood of center practice changes.

Limitations
This study has several limitations.Although we controlled for severity of illness, we recognize that it is possible that liberation pattern does not directly contribute to MAKE but instead is a consequence of disease severity.Alternatively, it is also feasible that pathophysiologic mechanisms contributing to adverse outcomes after CKRT are independent of severity of illness.This study is retrospective, and the data were self-reported by each participating center, which may have led to selection bias.
Although we included multiple markers of severity of illness in the analysis, it is possible that residual confounding remained.Given that only the first liberation attempt was documented, we may have missed patients who successfully liberated after a subsequent liberation attempt.Although smaller centers were included in the analysis, all included hospitals were tertiary or quaternary, possibly limiting generalizability to resource-limited settings.In addition, the type I error rate may have exceeded the nominal level given that no adjustments for multiple testing were performed.The definition of liberation status may have introduced periods of immortal time in which patients could not experience the outcome, resulting in some expected bias in our estimates.Additionally, only multivariable results using complete data are reported, and 34 patients with missing data were excluded from these analyses.However, the overall proportion of missing data was less than 5%.

Conclusions
In this cohort study, we found that MAKE were common 90 days after the initiation of CKRT in children, adolescents, and young adults.We found that successful liberation from therapy within 28 days was associated with lower odds of MAKE-90.
a A total of 34 patients were not included in the analysis due to missing covariate data.bThe aORs and 95% CIs were obtained by logistic regression; aORs for continuous factors are scaled to reflect the IQR OR (ie, reference = 25th percentile and contrast = 75th percentile).cThe same model was rerun with each liberation pattern individually with no change in covariate aORs.
Major Adverse Kidney Events in Pediatric Continuous Kidney Replacement Therapy JAMA Network Open.2024;7(2):e240243.doi:10.1001/jamanetworkopen.2024.0243(Reprinted) February 23, 2024 2/12 Downloaded from jamanetwork.comby guest on 03/02/2024 from 2015 to 2021, with a minority of patients included from 2015 to 2017 (67 patients).The registry includes patients from 32 centers and 7 nations. 11,12 Patients were categorized into 1 of 3 liberation categories based on the first liberation attempt during the first 28 days of CKRT.Only the first attempt at liberation was considered: (1) liberated: patients had no receipt of CKRT or other dialysis modality for 72 or more hours after discontinuing CKRT.(2)Reinstituted: patients resumed CKRT or another dialysis modality within 72 hours of a liberation attempt.(3) Not attempted: patients had no attempt at CKRT liberation within the first 28 days after CKRT initiation.A period of 72 hours has been used in numerous other investigations assessing kidney recovery as an adequate period to capture successful liberation and return of intrinsic kidney function.
White method were used to correct for the clustering of patients within hospitals.For continuous covariates, linear associations with the outcome were assumed and aORs are presented as interquartile odds ratios comparing a reference (25th percentile) with a contrast (75th percentile) value.The model performance was measured by optimism-corrected C index and Brier score.A calibration curve of the observed vs estimated probability was used to internally validate the accuracy of estimations using 1000 bootstrap resamplings.Sensitivity analysis was performed using similar methods to examine risk factors associated with persistent kidney dysfunction or dialysis dependence at 90 days.
In all analyses, a 2-sided P value < .05 was considered statistically significant.All statistical analyses were performed using R statistical software version 4.3.1 (R Project for Statistical Computing).The rms package version 6.7.1 was used to perform logistic regression, model validation, and calibration.Kaplan-Meier curves were generated using survival version 3.5.5 and survminer version 0.4.9 packages, and tables were generated using the gtsummary package version 1.7.2.The analysis was conducted from May 2 to December 14, 2023.

Table 2 .
After 34 patients with missing covariate data(3.5%)wereexcluded, the model was conducted among 935 patients.It had good discrimination and accuracy in estimating, with an optimism-corrected C index of 0.84 and a Brier score of 0.15.The calibration curve confirmed that the model's estimated probabilities for MAKE-90 were in line with

Table 1 .
Patient Characteristics and CKRT Parameters (continued) c Respondents could choose more than 1 race.