ABSTRACT
Objectives
Literature in adults have demonstrated no differences in mortality based on the timing of renal replacement therapy (RRT) initiation. Pediatric studies have suggested that a longer duration between intensive care unit (ICU) admission and continuous RRT (CRRT) initiation is associated with worse outcomes. We aimed to investigate how the time between acute kidney injury (AKI) and CRRT initiation impacts outcomes in children, as this relationship remains uncertain.
Methods
Retrospective cohort study of children (<18 years) admitted to the ICU who received CRRT between 2/2014 and 2/2020. Patients with chronic kidney disease, non-AKI, or fluid overload (FO) CRRT indications were excluded. The Pediatric Organ Dysfunction Information Update Mandate criteria were used to diagnose AKI. Exposures of interest were AKI onset (time between AKI and CRRT initiation) and AKI exposure (total number of days with each AKI criteria before CRRT). Our primary outcome was RRT-free days, censored at 30 days.
Results
One hundred ninty-seven patients with a median age of 53.8 months (11.2–147.1) months received CRRT for a median of 13 (5–27) days. The median onset of AKI occurred 5 (2–7) days prior to CRRT start. For each additional day of AKI exposure before CRRT initiation there was a 0.42 decrease in RRT-free days. For every additional day between FO >15% and CRRT initiation the likelihood of dialysis liberation decreased [HR 0.82 (CI: 0.72–0.93)].
Conclusions
Earlier CRRT initiation after AKI onset was associated with more RRT-free days, while AKI, defined by FO >15%, was associated with dialysis dependence. Our findings suggest that delaying CRRT in critically ill pediatric patients may lead to worse outcomes.
Keywords: acute kidney injury, continuous renal replacement therapy, major adverse kidney events, mortality, renal recovery, timing
Graphical Abstract
Graphical Abstract.
KEY LEARNING POINTS.
What was known:
In critically ill adults, earlier dialysis initiation after severe acute kidney injury (AKI) is associated with adverse events without improving outcomes. In children, prior studies have used time interval between ICU admission and dialysis start to assess how the timing of dialysis initiation impacts outcomes.
This study adds:
Utilizing the expanded Pediatric Organ Dysfunction Information Update Mandate criteria to define AKI, we found that a shorter interval between AKI onset and CRRT initiation were associated with more RRT-free days. Fluid overload (FO) >15% was associated with a decreased likelihood of dialysis liberation.
Potential impact:
Outcomes in children may improve when CRRT is initiated soon after the development of AKI and/or FO >15%. These findings suggest that the timing of CRRT is a modifiable factor, but larger multicenter studies are needed to validate these findings.
INTRODUCTION
Severe acute kidney injury (AKI) is associated with 1.77-fold higher mortality in critically ill pediatric patients, while AKI requiring dialysis (AKI-D) is associated with 3.38-fold increased odds of mortality [1–4]. Continuous renal replacement therapy (CRRT) is the preferred dialysis modality in critically ill children as it minimizes fluctuations in osmolality and intravascular volume [5]. Renal replacement therapy (RRT) is often initiated quickly in the setting of life-threatening complications of AKI (e.g. hyperkalemia or refractory acidosis) [6]. Even without emergent indications, AKI exposes patients to homeostatic derangements and fluid overload (FO), which can impede recovery [7, 8]. Prompt initiation of CRRT after diagnosing AKI could mitigate these factors and potentially improve patient outcomes.
The timing of dialysis initiation in critically ill adults has been tested through multiple randomized trials. Two of the most notable studies are the Artificial Kidney Initiation in Kidney Injury (AKIKI) and STARRT-AKI trials which, in combination showed that an accelerated initiation of RRT is not associated with improved survival, decreased odds of major adverse kidney event (MAKE 90), or shorter length of stay compared to a standard strategy. Importantly, 40%–50% of the patients within the delayed group did not require RRT. While in the accelerated RRT initiation groups there was a higher rate of catheter-related bloodstream infections, hypotension, and hypophosphatemia [9–11]. Additionally, STARRT-AKI showed that in adults with chronic kidney disease (CKD) earlier RRT initiation was associated with higher RRT dependence at 90 days [10]. Consistent with these findings, a recent meta-analysis of 11 randomized control trials focusing on the timing of RRT initiation after AKI in critically ill adults, noted no difference in 28-day, 90-day, intensive care unit (ICU), or hospital mortality based on early versus late RRT initiation [12].
Despite adult literature demonstrating no difference in mortality between early and late RRT initiation, several observational studies in critically ill children suggest that early CRRT initiation may be beneficial. However, these pediatric studies used the time from ICU admission to CRRT initiation rather than from AKI diagnosis to CRRT initiation. Their findings indicate that later CRRT initiation is independently associated with mortality [13–15]. Complicating this relationship is the fact that FO often accompanies AKI and is itself an indication for CRRT [16]. However, the time between AKI diagnosis and/or development of FO and initiation of CRRT has not been adequately investigated in children.
Our aim was to investigate the effect of the time interval between initial clinically apparent renal insult, defined using an extended criterion incorporating serum creatinine, urine output (UOP), and/or FO, and CRRT initiation on patient outcomes [17]. Our approach differs from prior pediatric studies by anchoring time zero to earliest detectable renal insult. We hypothesized that in critically ill children without CKD, a longer interval between AKI diagnosis (using this expanded criteria) and CRRT initiation will be associated with fewer RRT-free days.
MATERIALS AND METHODS
We conducted a single-center retrospective observational study of pediatric patients (<18 years old) receiving CRRT in the cardiac and/or pediatric ICUs at Texas Children’s Hospital between February 2014 and February 2020. Patients with CKD, end-stage renal disease, or intoxication as the sole indication for CRRT were excluded. The study was approved by the Institutional Review Board of Baylor College of Medicine (protocol H-32217) and performed in accordance with the Helsinki Declaration.
Variables collected included CRRT blood flow rate, initial modality and prescribed clearance, type of filter, anticoagulation, and priming solution. Demographic and clinical data included age at hospital and ICU admission, sex, weight, height, body surface area, primary diagnostic category, primary comorbid condition, history of transplant, vasoactive support, and vasoactive inotropic score (VIS) [18]. Pediatric Logistic Organ Dysfunction-2 (PELOD-2) scores were calculated at ICU admission and CRRT start [19]. Lastly, data on concurrent extracorporeal membrane oxygenation (ECMO) were collected, and patients who received only slow continuous ultrafiltration without use of a CRRT platform were excluded.
Organ dysfunction was defined using the individual organ system scores from the PELOD-2 criteria. A score of ≥1 within each organ system was classified as organ dysfunction. MODS was defined as ≥2 organ system dysfunctions at ICU admission and/or CRRT initiation [19, 20]. Progressive MODS was defined as an increase in the number of organ dysfunctions from ICU admission to CRRT initiation [21].
AKI was defined according to the Pediatric Organ Dysfunction Information Update Mandate (PODIUM) consensus guidelines, which incorporate serum creatinine, UOP, and FO criteria [17]. For serum creatinine (AKI-Cr) and UOP (AKI-UOP) we used the Kidney Disease Improving Global Outcomes (KDIGO) guideline thresholds of serum creatinine >1.5x baseline and UOP <0.5 ml/kg/h for >6 h respectively [22, 23]. The lowest serum creatinine 90 days prior to ICU Admission was used as the patient’s baseline serum creatinine value. If no serum creatinine was available, the patient was assigned an estimated glomerular filtration rate (eGFR) of 100 ml/min/1.73 m2, and baseline creatinine was calculated using their height at admission and the modified Schwartz equation [21, 24]. Percent FO was calculated as [Fluid in (liters)—Fluid out (liters)]/ICU Admission weight (kg)] × 100 [2]. If patients were admitted to the ICU >7 days before CRRT initiation, their %FO included fluid that accumulated only during their ICU stay. FO >15% was used as the threshold to define renal dysfunction/AKI in our study population, which we based on published literature from ours and other centers [25–27]. Daily serum creatinine levels, UOP, and %FO were collected from ICU admission until CRRT initiation and censored seven days prior to CRRT initiation. AKI onset was defined as the time any AKI-Cr, AKI-UOP, or FO criteria was first met [17]. AKI exposure was defined as the total number of days with AKI-Cr, AKI-UOP, and FO prior to CRRT. For example, if a patient met the AKI-Cr criterion two days before CRRT initiation and FO criterion five days prior, AKI onset would be equal to five days with an AKI exposure of seven days. The maximum duration of AKI onset was seven days, while the maximum AKI exposure was 21 days (Supplemental Fig. S1).
Outcome
Our primary outcome was RRT-free days, defined as the absence of all dialytic therapies censored at 30 days [28, 29]. For our competing risks analysis, our cohort was divided in three subgroups. Liberation (defined as children who remained free from all dialytic modalities for > 72 h), dialysis dependence (any modality), and death all censored at 30 days post-CRRT initiation [28].
Statistical analysis
Categorical variables are presented as counts (percentages) and continuous variables as medians (interquartile ranges). Group differences were assessed using the chi-square test for categorical variables and the Kruskal–Wallis test for continuous variables. A P-value <.05 was considered statistically significant.
Multivariate linear regression was performed to analyze the association with RRT-free days. Model assumptions, including linearity, normality, and homoscedasticity, were assessed using diagnostic residual plots, which demonstrated acceptable fit (Supplemental Fig. S2). Generalized variance inflation factors (GVIF) were calculated to assess for multicollinearity, with all adjusted GVIF values below commonly accepted thresholds (Supplemental Table S1).
We fitted a multivariate competing risk Cox proportional cause-specific hazard model to analyze the association between variables and two exclusive events: liberation and mortality [30]. The proportional hazards assumption was assessed using Schoenfeld residuals, with no major violations observed (Supplemental Figs 3 and 4). Results are presented as hazard ratios (HR) with 95% confidence intervals using dialysis dependence as our reference. Cumulative incidence plots with a risk table were used to visualize renal recovery and mortality. To further explore potential nonlinear relationships, we derived partial dependence plots using a multivariate cause-specific random survival forest model with 300 trees and default parameter settings [31]. Variable importance was assessed using Shapley Explanation Plots (Supplemental Fig. S5), and partial dependence plots were generated to evaluate the nonlinear effect of AKI exposure on outcomes while adjusting for other covariates [32]. We utilized R version 4.3.1 to perform all the statistical analyses noted above.
RESULTS
One hundred and ninety-seven patients were included in this study. Ninety-three (47%) were female. The median age at ICU admission was 53.8 months (IQR 11.2–147.1), with a median PELOD-2 score of 8 (7–10). The most common indication for ICU admission was respiratory in 59 (30%) patients, while the most common primary comorbidity was hematological/oncological in 58 (29%) patients. The median VIS was 4 (0–12), and 43 (22%) required ECMO in tandem with CRRT. The median CRRT duration was 14 days (5–29) with a median % FO at CRRT initiation of 9.4% (1.2–26.2). The median RRT-free days were 16 (1–25). Thirty days after CRRT initiation, 24 (12%) patients had persistent renal dysfunction, 72 (37%) remained dialysis-dependent, 66 (34%) died, and the remaining 35 (18%) had recovery of renal function (Table 1 and Supplemental Table S2).
Table 1:
Demographics by cohort.
| Variables | Dialysis dependenta (n = 72) | Liberateda (n = 59) | Deatha (n = 66) | P value |
|---|---|---|---|---|
| Sex (Female) | 36 (50%) | 28 (47%) | 29 (44%) | .76 |
| Transplant (Yes) | 22 (31%) | 11 (19%) | 17 (26%) | .28 |
| Extracorporeal membranous oxygenation (Yes) | 14 (19%) | 13 (22%) | 16 (14%) | .81 |
| Prime (Blood) | 39 (54%) | 31 (53%) | 39 (59%) | .74 |
| Progressive multiorgan dysfunction (Yes) | 48 (67%) | 37 (63%) | 40 (61%) | .76 |
| Primary diagnosis | .05 | |||
| Respiratory | 16 (22%) | 17 (29%) | 26 (39%) | |
| Sepsis | 15 (21%) | 12 (20%) | 18 (27%) | |
| Other Shock | 14 (19%) | 10 (17%) | 9 (14%) | |
| Cardiac | 5 (7%) | 6 (10%) | 3 (5%) | |
| Trauma | 0 (0%) | 1 (2%) | 0 (0%) | |
| Pain | 9 (13%) | 0 (0%) | 2 (3%) | |
| Central nervous system | 6 (8%) | 4 (7%) | 5 (8%) | |
| Other | 7 (10%) | 9 (15%) | 3 (5%) | |
| Primary comorbidity | .08 | |||
| None | 10 (14%) | 14 (24%) | 15 (23%) | |
| Cardiac | 15 (21%) | 13 (22%) | 8 (12%) | |
| Respiratory | 1 (1%) | 1 (2%) | 2 (3%) | |
| Neurological | 1 (14) | 0 (0%) | 2 (3%) | |
| Gastrointestinal/liver | 19 (26%) | 5 (8%) | 11 (17%) | |
| Renal | 4 (5%) | 3 (5%) | 1 (2%) | |
| Hematological/oncological | 20 (28%) | 14 (24%) | 24 (36%) | |
| Metabolic | 1 (1%) | 4 (7%) | 1 (2%) | |
| Other | 1 (1%) | 5 (8%) | 2 (3%) | |
| AKI Onset Criteria | .08 | |||
| AKI | 16 (22%) | 11 (19%) | 15 (23%) | |
| UOP | 17 (24%) | 22 (37%) | 14 (21%) | |
| FO | 19 (26%) | 7 (12%) | 18 (27%) | |
| AKI-Cr & AKI-UOP | 12 (17%) | 15 (25%) | 10 (15%) | |
| AKI-Cr & FO | 0 (0%) | 2 (3%) | 1 (2%) | |
| AKI-UOP & FO | 4 (6%) | 1 (2%) | 7 (11%) | |
| All | 4 (6%) | 1 (2%) | 1 (2%) | |
| Age at ICU admission (months) | 44.8 (8.9–148.6) | 59.1 (24.6–134.5) | 41.3 (6.8–152.8) | .36 |
| Weight (kg) | 18 (9.1–41.4) | 19.5 (13–54.4) | 15.5 (7.2–49) | .13 |
| Height (cm) | 106–5 (72–143.8) | 113–5 (89.7–155) | 94.1 (65–147.2) | .07 |
| Body Surface Area (m2) | 0.73 (0.4–1.3) | 0.78 (0.6–1.5) | 0.63 (0.4–1.4) | .10 |
| Total number of organ dysfunction at CRRT start | 4 (3–5) | 3 (3–4) | 4 (3–5) | .06 |
| Vasoactive inotropic score at CRRT start | 4.25 (0–7.3) | 4.75 (0–10.0) | 5 (0–17) | .42 |
| AKI Cr Onset (days) | 2.50 (1–7) | 1 (0–3) | 1 (0–4) | .01 |
| AKI UOP Onset (days) | 3.50 (1–6.3) | 2 (0–4) | 1 (0–5.8) | .05 |
| FO onset (days) | 0.5 (0–7) | 0 (0–1) | 2 (0–8) | <.01 |
| AKI onset (days) | 7 (3–7) | 3 (2–7) | 6 (2–7.8) | <.01 |
| AKI exposure (days) | 8.5 (5–14) | 4 (2.5–8) | 8 (3–12.8) | <.01 |
Categorical variables presented as N (%) while continuous variables presented as median (Q1–Q3).
AKI, acute kidney injury; FO, fluid overload; AKI-Cr, acute kidney injury defined by serum creatinine criteria; AKI-UOP, acute kidney injury defined using urine output; UOP, urine output.
AKI onset and exposure and RRT-free days
AKI-Cr, AKI-UOP, and FO accounted for 42 (21%), 53 (27%), and 44 (22%) cases of AKI onset, respectively. Fifty-eight (30%) patients met criteria via a combination; AKI-Cr/AKI-UOP in 37 (19%), AKI-UOP/FO in 12 (6%), AKI-Cr/FO in 3 (2%), AKI-Cr/AKI-UOP/FO in 6 (3%). The median time from AKI onset and CRRT initiation was 5 days (IQR 2–7) (Table 1 and Fig. 1). The median length of AKI exposure was 7 days (IQR 3–12).
Figure 1:
Frequency of AKI onset across the observation period.
In unadjusted analysis history of organf transplant, gastrointestinal/liver comorbidities, primary diagnosis of post-operation/pain, number of organ dysfunctions at CRRT initiation, MODS at CRRT initiation, longer AKI onset, and longer AKI exposure were associated with fewer RRT-free days (Supplemental Table S3).
In our final adjusted model controlling for multiple clinical features, patients who received ECMO [β 4.82 (0.56–9.80)] and those with a higher VIS [β 0.13 (0.05–0.21)] had more RRT free days. Conversely, patients with a higher number of organ dysfunctions at CRRT initiation [β −3.55 (−6.21 to −0.90)] and a longer AKI exposure [β −0.42 (−0.70 to −0.13)] had fewer RRT-free days (Table 2).
Table 2:
Multivariable linear regression analysis, CRRT-free days.
| Variables | β −coefficient | Confidence intervals | P values |
|---|---|---|---|
| Sex (Female) | −1.19 | −4.17 to 1.80 | .43 |
| Extracorporeal membranous oxygenation (Yes) | 4.82 | 0.56 to 9.80 | .03 |
| Progressive multiorgan dysfunction (Yes) | 1.00 | −2.24 to 4.23 | .54 |
| Transplant (Yes) | −2.22 | −5.71 to 1.27 | .21 |
| Prime (normal saline) | 2.44 | −2.93 to 7.81 | .37 |
| Primary comorbid | |||
| Cardiac | −2.12 | −6.80 to 2.56 | .37 |
| Respiratory | 6.34 | −4.31 to 17.00 | .24 |
| Neurological | −6.47 | −18.45 to 5.52 | .29 |
| Gastrointestinal/liver | −1.78 | −6.97 to 3.41 | .50 |
| Renal | 1.11 | −6.84 to 9.05 | .78 |
| Hematological/oncological | −1.31 | −5.87 to 3.26 | .57 |
| Metabolic | 3.70 | −5.34 to 12.74 | .42 |
| Other | 5.76 | −2.28 to 13.80 | .16 |
| AKI exposure | −0.42 | −0.70 to −0.13 | <.01 |
| VIS | 0.13 | 0.05 to 0.21 | <.01 |
| BSA | 0.65 | −5.47 to 6.77 | .83 |
| Age at ICU admission | −0.03 | −0.07 to 0.02 | .27 |
| Total organ dysfunctions at CRRT start | −3.55 | −6.21 to −0.90 | .01 |
| PELOD-2 at CRRT start | 0.41 | −0.45 to 1.28 | .35 |
| Multiorgan dysfunction at CRRT start | −0.43 | −6.39 to 5.53 | .89 |
AKI, acute kidney injury; VIS, vasoactive inotrope score; BSA, body surface area.
AKI onset and exposure and liberation
The three cohorts; liberated, dialysis-dependent, and mortality consisted of 72 (37%), 59 (30%), and 66 (33%) patients, respectively. AKI onset was longest in the dialysis-dependent group [7 (2–7) vs. 2 (1.5–5) vs. 4 (1–7) P < .01]. Similarly, significant differences were observed in AKI exposure across the three cohorts (P ≤ .01); liberated had an AKI exposure of 4 days (IQR 2.5–8), dialysis dependent 8.5 days (IQR 5–14), and mortality 8 days (IQR 3–12.8) (Table 1). Utilizing partial effect plots, inflection points for AKI exposure ≤4 days associated with liberation and ≤1 day associated with mortality (Supplemental Figs S6 and 7).
Competing risk model
Dialysis dependence and liberation
An increase in one organ dysfunction at CRRT initiation was associated with lower hazard of liberation [HR 0.53, (CI: 0.38–0.73)]. Similarly, for each additional month increase in age, the hazard of liberation decreased [HR 0.99 (CI: 0.98–1.0)]. For everyday increase between FO onset and CRRT start [HR 0.82 (CI: 0.72–0.93)], patients were less likely to liberate (Table 3). Patients with an AKI exposure of >4 days had a significantly higher cumulative incidence of dialysis dependence (P < .01) (Fig. 2).
Table 3:
Competing risk model.
| Liberated | Dead | |||||
|---|---|---|---|---|---|---|
| Variables | HR | 95% CI | P values | HR | 95% CI | P values |
| Sex (Female) | 0.93 | 0.52–1.67 | .80 | 0.71 | 0.41–1.21 | .21 |
| Extracorporeal membranous oxygenation (Yes) | 1.63 | 0.7–3.75 | .25 | 1.90 | 0.85–4.27 | .12 |
| Progressive multiorgan dysfunction (Yes) | 1.31 | 0.71–2.44 | .39 | 0.95 | 0.53–1.69 | .85 |
| Transplant (Yes) | 0.73 | 0.34–1.56 | .41 | 0.95 | 0.49–1.85 | .89 |
| Prime (Normal Saline) | 1.41 | 0.5–3.99 | .52 | 0.94 | 0.31–2.86 | .92 |
| Primary comorbidity | ||||||
| Cardiac | 1.38 | 0.58–3.26 | .47 | 0.38 | 0.15–0.97 | .04 |
| Respiratory | 3.17 | 0.29–34.71 | .34 | 2.55 | 0.48–13.55 | .27 |
| Neurological | 0.00 | 0–Inf | 1.00 | 0.98 | 0.2–4.94 | .99 |
| Gastrointestinal/liver | 0.63 | 0.2–2.03 | .44 | 0.64 | 0.25–1.65 | .35 |
| Renal | 0.70 | 0.19–2.64 | .60 | 0.57 | 0.07–4.56 | .59 |
| Hematological/oncological | 0.97 | 0.42–2.25 | .94 | 1.02 | 0.47–2.22 | .96 |
| Metabolic | 2.76 | 0.77–9.96 | .12 | 0.53 | 0.06–4.46 | .56 |
| Other | 2.63 | 0.84–8.25 | .10 | 1.45 | 0.29–7.11 | .65 |
| AKI onset criteria | ||||||
| AKI-Cr | 0.88 | 0.77–1 | .06 | 0.91 | 0.82–1.01 | .08 |
| AKI-UOP | 0.96 | 0.84–1.09 | .49 | 0.90 | 0.81–1 | .05 |
| FO | 0.82 | 0.72–0.93 | <.01 | 1.09 | 1–1.19 | .06 |
| Vasoactive inotropic score | 1.01 | 1–1.03 | .09 | 1.01 | 1–1.02 | .01 |
| Total organ dysfunctions at CRRT start | 0.53 | 0.38–0.73 | <.01 | 1.01 | 0.75–1.35 | .96 |
| Body surface area (m2) | 1.71 | 0.58–5.04 | .33 | 0.77 | 0.23–2.55 | .67 |
| Age at ICU admission | 0.99 | 0.98–1 | .04 | 1.00 | 0.99–1.01 | .51 |
AKI, acute kidney injury; FO, fluid overload; AKI-Cr, acute kidney injury defined by serum creatinine criteria; AKI-UOP, acute kidney injury defined using urine output; UOP, urine output.
Figure 2:
Cumulative incidence plot for AKI exposure and liberation.
Dialysis dependence and death
Patients with a primary cardiac comorbidity [HR 0.38 (CI: 0.15–0.97)] and AKI-UOP [HR 0.90 (CI: 0.81–1.0)] were less likely to experience mortality. For every point increase in VIS [HR 1.01 (CI: 1.00–1.02)] patients were more likely to experience mortality (Table 3). An AKI exposure ≤1 day was no longer associated with a significant difference in the cumulative incidence of mortality in this model (P = .07) (Fig. 3).
Figure 3:
Cumulative incidence plot for AKI exposure and death.
Discussion
We utilized the expanded definition of AKI/renal dysfunction proposed by PODIUM (AKI-CR, AKI-UOP, and FO) to assess the impact of the time between AKI diagnosis and CRRT initiation on 30-day outcomes, including liberation, dialysis dependence, and mortality in critically ill children [17]. We found that a longer AKI exposure was associated with fewer RRT-free days. Interestingly, patients requiring ECMO and those with a higher VIS experienced more RRT-free days. Our competing risks model showed that patients with AKI onset due to FO were more likely to remain dialysis-dependent 30 days after CRRT initiation. Additionally, AKI exposure of ≥4 days was associated with dialysis dependence. Our findings contribute to the current literature and knowledge regarding the timing of CRRT initiation following the development of AKI in critically ill children.
Prior randomized controlled trials in critically ill adults showed no difference in mortality between early and late/standard dialysis initiation, with up to 50% of patient in the late/started initiation cohort not requiring dialysis, and more complications occurring with the early-initiation group [9, 11]. Given these findings, many clinicians advocate for delayed initiation in the absence of emergent indications for dialysis. Following these studies, AKIKI-2 showed that a further delay of RRT initiation was associated with increased 60-day mortality [33]. Currently, the data on the optimal timing of CRRT in pediatric patients remains limited.
A recent multicenter pediatric study found that each day of delay in CRRT initiation from ICU admission increased odds of MAKE 90 by 3% and mortality by 4% [13]. Another study reported a 2% increase in mortality for each day between AKI (defined by KDIGO serum creatinine and UOP criteria) and CRRT initiation [34]. Our approach differed in that we defined our exposure of interest as the time from AKI to CRRT initiation using PODIUM criteria, incorporating FO, which frequently coexists with AKI. Our findings align with previous published literature, suggesting that pediatric patients may experience worse outcomes when CRRT is delayed [13, 34]. These findings contradict previously published literature in adults, and we theorize that this could be due the immaturity of organ function across the pediatric age range leading to increased physiological vulnerability [35].
Prolonged AKI duration is associated with worse clinical outcomes. The Acute Disease Quality Initiative 16 Workgroup proposed a 48-h threshold to distinguish transient from persistent AKI [36]. Consistent with this, a secondary analysis of the AWARE cohort showed that persistent AKI was associated with ~2.6-fold higher odds of MAKE 28 [37]. These worse outcomes are theorized to result from the pathological organ cross talk that occurs in the setting of AKI [21, 38]. CRRT enables clinicians to re-establish homeostasis through electrolyte control, fluid management, and toxin removal. This re-establishment of homeostasis is theorized to improve myocardial function and gas exchange, thereby promoting organ recovery [39]. Consistent with this theory, in adjusted analyses we found that patients receiving ECMO and higher VIS had more RRT-free days, possibly reflecting improved renal perfusion and oxygen delivery during extracorporeal and vasoactive support. Adding to the existing literature, our findings suggest that in children for each day increase in AKI exposure there is a 0.42 decrease in the number of RRT-free days which is clinically meaningful when we consider how AKI exposures can compound over time.
Since the publication of the KDIGO AKI criteria, studies have shown a negative additive effect of AKI diagnosed by both UOP and serum creatinine criteria when compared to isolated UOP-AKI or elevated serum creatinine alone [40, 41]. Specifically, the AWARE investigators noted >5-fold higher mortality and >6-fold increased need for dialysis in children meeting both serum creatinine and UOP AKI criteria compared to those meeting AKI via serum creatinine or UOP criteria [42]. This is the first study to use AKI exposure to quantify cumulative AKI burden in children. Using this novel definition, we found that AKI exposure ≤4 days was associated with increased odds of liberation from dialysis. However, these findings have limited generalizability and require further validation before clinical implementation.
Development of FO during critical illness is common and occurs partly due to trans-capillary leak caused by endothelial glycocalyx dysfunction compounded by obligate fluid administration [43]. Literature has shown an association between the magnitude of FO on day 3 in ventilated children and mortality [44]. In critically ill children, FO >10% during the first 48 h of ICU admission has been linked to fewer ventilator-free days, ICU-free days, and increased mortality [45]. Our findings add to the current literature, by showing that each additional day with FO >15% before CRRT initiation increased the risk of the dialysis dependence by 18%. Although this association persisted after adjustment for illness severity (PELOD-2), FO >15% may represent a marker of disease severity rather than an independent factor. Nonetheless, we theorize this association may be due to changes to cardiac energetics, pulmonary and neurological edema, delayed wound healing, and impaired bowel function, that occur in patients with FO, and can collectively impede recovery [26, 46].
Our study has several limitations. First, it is a single-center retrospective study with a relatively small sample size and reliance on accurate UOP documentation based on diaper weights. In mixed diapers (stool and urine), UOP may be overestimated, while missed documentation could underestimate output and overestimate FO. Second, for patients without baseline creatinine, an assumed eGFR of 100 ml/min/1.73 m² was used, which may overestimate baseline kidney function in children <2 years of age, but is the recommended methodology by KDIGO guidelines [47]. Third, we only assessed AKI criteria within 7 days prior to CRRT and may have missed earlier or repeated AKI events. Fourth, we did not collect data on laboratory markers such as blood urea nitrogen, potassium, or bicarbonate that may influence dialysis initiation. Fifth, given our exclusion criteria, our findings are not generalizable across the entire heterogeneous CRRT population, specifically for patient’s receiving CRRT for non-renal indications (e.g. endo/exotoxin removal) [48]. Sixth, as an observational study looking at a salvage therapy, residual confounding and selection biases are always possible [49, 50]. Finally, we were unable to reliably determine the primary indication for CRRT due to inconsistent documentation and instead relied on study definitions; standardized criteria for RRT initiation would improve future studies.
CONCLUSION
In pediatric patients with severe AKI undergoing CRRT, a shorter interval between the onset of AKI or FO and CRRT initiation was associated with successful dialysis liberation, whereas a longer duration of FO >15% prior to CRRT initiation was associated with decreased cumulative incidence of liberation. Both AKI and FO appear to have a dose-dependent and additive effect on RRT liberation in critically ill children. Urgent studies are needed to validate whether the timing of CRRT initiation is a modifiable factor that can be targeted to improve patient outcomes.
Supplementary Material
Contributor Information
Sameer Thadani, Department of Pediatrics, Division of Critical Care Medicine, Baylor College of Medicine, Houston, TX, USA; Department of Pediatrics, Division of Nephrology, Baylor College of Medicine, Houston, TX, USA.
Harsha Jujjavarapu, Department of Pediatrics, Baylor College of Medicine, Houston TX, USA.
Xin Tan, Department of Statistics, Rice University, Houston, TX, USA.
Dana Fuhrman, Department of Pediatrics, Division of Nephrology, UPMC Children’s Hospital of Pittsburgh, Pittsburgh, PA, USA; The Center for Critical Care Nephrology, Department of Critical Care Medicine, University of Pittsburgh, Pittsburgh, PA, USA.
Michael Bell, Department of Critical Care Medicine, Children’s National Hospital, WA D.C., USA.
Poyyapakkam Srivaths, Department of Pediatrics, Division of Nephrology, Baylor College of Medicine, Houston, TX, USA.
Joseph Angelo, Department of Pediatrics, Division of Nephrology, Baylor College of Medicine, Houston, TX, USA.
Stuart Goldstein, Division of Nephrology and Hypertension, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA.
Meng Li, Department of Statistics, Rice University, Houston, TX, USA.
Ayse Akcan Arikan, Department of Pediatrics, Division of Critical Care Medicine, Baylor College of Medicine, Houston, TX, USA; Department of Pediatrics, Division of Nephrology, Baylor College of Medicine, Houston, TX, USA.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest and did not receive support from any organization for this submitted work.
FUNDING
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit centers.
DATA AVAILABILITY STATEMENT
Due to the size of the cohort, the data are not publicly available as individual participants could be easily identifiable, posing a risk to patient privacy.
REFERENCES
- 1. Chang J-W, Jeng M-J, Yang L-Y et al. The epidemiology and prognostic factors of mortality in critically ill children with acute kidney injury in Taiwan. Kidney Int. 2015;87:632–9. 10.1038/ki.2014.299 [DOI] [PubMed] [Google Scholar]
- 2. Sutherland SM, Zappitelli M, Alexander SR et al. Fluid overload and mortality in children receiving continuous renal replacement therapy: the prospective pediatric continuous renal replacement therapy registry. Am J Kidney Dis. 2010;55:316–25. 10.1053/j.ajkd.2009.10.048 [DOI] [PubMed] [Google Scholar]
- 3. Kaddourah A, Basu RK, Bagshaw SM et al. Epidemiology of acute kidney injury in critically ill children and young adults. N Engl J Med. 2017;376:11–20. 10.1056/NEJMoa1611391 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Starr MC, Gist KM, Zang H et al. Continuous kidney replacement therapy and survival in children and young adults: findings from the Multinational WE-ROCK Collaborative. Am J Kidney Dis. 2024;84:406–415.e1. S0272-6386(24)00610–3. 10.1053/j.ajkd.2023.12.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Basu RK, Wheeler DS, Goldstein S et al. Acute renal replacement therapy in pediatrics. Int J Nephrol. 2011;2011:1–8. 10.4061/2011/785392 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Vaara ST, Reinikainen M, Wald R et al. Timing of RRT based on the presence of conventional indications. Clin J Am Soc Nephrol. 2014;9:1577–85. 10.2215/CJN.12691213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Libório AB, Leite TT, Neves FMDO et al. AKI complications in critically ill patients: association with mortality rates and RRT. Clin J Am Soc Nephrol. 2015;10:21–28. 10.2215/CJN.04750514 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Rysz J, Franczyk B, Ławiński J et al. The impact of CKD on uremic toxins and gut microbiota. Toxins. 2021;13:252. 10.3390/toxins13040252 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Gaudry S, Hajage D, Schortgen F et al. Initiation strategies for renal-replacement therapy in the intensive care unit. N Engl J Med. 2016;375:122–33. 10.1056/NEJMoa1603017 [DOI] [PubMed] [Google Scholar]
- 10. Bagshaw SM, Neto AS, Smith O et al. Impact of renal-replacement therapy strategies on outcomes for patients with chronic kidney disease: a secondary analysis of the STARRT-AKI trial. Intensive Care Med. 2022;48:1736–50. 10.1007/s00134-022-06912-w [DOI] [PubMed] [Google Scholar]
- 11. Investigators S-AKI . STandard versus accelerated initiation of renal replacement therapy in acute kidney injury: study protocol for a multi-national, multi-center, randomized controlled trial. Can J Kidney Health Dis. 2019;6:2054358119852937. 10.1177/2054358119852937 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Li X, Liu C, Mao Z et al. Timing of renal replacement therapy initiation for acute kidney injury in critically ill patients: a systematic review of randomized clinical trials with meta-analysis and trial sequential analysis. Crit Care. 2021;25:15. 10.1186/s13054-020-03451-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Gist KM, Menon S, Anton-Martin P et al. Time to continuous renal replacement therapy initiation and 90-day major adverse kidney events in children and young adults. JAMA Netw Open. 2024;7:e2349871. 10.1001/jamanetworkopen.2023.49871 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Modem V, Thompson M, Gollhofer D et al. Timing of continuous renal replacement therapy and mortality in critically ill children*. Crit Care Med. 2014;42:943–53. 10.1097/CCM.0000000000000039 [DOI] [PubMed] [Google Scholar]
- 15. Cortina G, McRae R, Hoq M et al. Mortality of critically ill children requiring continuous renal replacement therapy: effect of fluid overload, underlying disease, and timing of initiation*. Pediatr Crit Care Med. 2019;20:314–22. 10.1097/PCC.0000000000001806 [DOI] [PubMed] [Google Scholar]
- 16. Ostermann M, Straaten HMO, Forni LG. Fluid overload and acute kidney injury: cause or consequence?. Crit Care. 2015;19:443. 10.1186/s13054-015-1163-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Bembea MM, Agus M, Akcan-Arikan A et al. Pediatric organ dysfunction information update mandate (podium) contemporary organ dysfunction criteria: executive summary. Pediatrics. 2022;149:S1–S12. 10.1542/peds.2021-052888B [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Gaies MG, Gurney JG, Yen AH et al. Vasoactive–inotropic score as a predictor of morbidity and mortality in infants after cardiopulmonary bypass. Pediatr Crit Care Med. 2010;11:234–8. 10.1097/PCC.0b013e3181b806fc [DOI] [PubMed] [Google Scholar]
- 19. Leteurtre S, Duhamel A, Salleron J et al. PELOD-2: an update of the PEdiatric logistic organ dysfunction score. Crit Care Med. 2013;41:1761–73. 10.1097/CCM.0b013e31828a2bbd [DOI] [PubMed] [Google Scholar]
- 20. Leteurtre S, Duhamel A, Deken V et al. Daily estimation of the severity of organ dysfunctions in critically ill children by using the PELOD-2 score. Crit Care. 2015;19:324. 10.1186/s13054-015-1054-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Thadani S, Fuhrman D, Hanson C et al. Patterns of multiple organ dysfunction and renal recovery in critically ill children and young adults receiving continuous renal replacement therapy. Crit Care Explor. 2024;6:e1084. 10.1097/CCE.0000000000001084 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. KDIGO . Section 2: AKI definition. Kidney Int Suppl. 2012;2:19–36. 10.1038/kisup.2011.32 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. KDIGO . Section 5: dialysis interventions for treatment of AKI. Kidney Int Suppl. 2012;2:89–115. 10.1038/kisup.2011.35 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Staples A, LeBlond R, Watkins S et al. Validation of the revised Schwartz estimating equation in a predominantly non-CKD population. Pediatr Nephrol. 2010;25:2321–6. 10.1007/s00467-010-1598-7 [DOI] [PubMed] [Google Scholar]
- 25. Riley AA, Watson M, Smith C et al. Pediatric continuous renal replacement therapy: have practice changes changed outcomes? A large single-center ten-year retrospective evaluation. BMC Nephrol. 2018;19:268. 10.1186/s12882-018-1068-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Thadani S, Lang A, Silos C et al. Fluid overload modifies hemodynamic impact of CRRT: evidence of a covert cardiorenal syndrome?. Shock. 2024;63:233–9. 10.1097/SHK.0000000000002483 [DOI] [PubMed] [Google Scholar]
- 27. Goldstein SL, Krallman KA, Roy J-P et al. Real-time acute kidney injury risk stratification-biomarker directed fluid management improves outcomes in critically ill children and young adults. Kidney Int Rep. 2023;8:2690–700. 10.1016/j.ekir.2023.09.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Billings IV FT, Shaw AD. Clinical trial endpoints in acute kidney injury. Nephron Clin Pract. 2014;127:89–93. 10.1159/000363725 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Stenson EK, Alhamoud I, Alobaidi R et al. Factors associated with successful liberation from continuous renal replacement therapy in children and young adults: analysis of the worldwide exploration of renal replacement outcomes collaborative in Kidney Disease Registry. Intensive Care Med. 2024;50:861–72. 10.1007/s00134-024-07336-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Ozenne B, Sørensen AL, Scheike T et al. riskRegression: predicting the risk of an event using cox regression models. R J. 2017;9:440. 10.32614/RJ-2017-062 [DOI] [Google Scholar]
- 31. Ishwaran H, Kogalur UB, Blackstone EH et al. Random survival forests. Ann Appl Stat. 2008;2:841–60. 10.1214/08-AOAS169 [DOI] [Google Scholar]
- 32. Angelini M, Blasilli G, Lenti S et al. A visual analytics conceptual framework for explorable and steerable partial dependence analysis. IEEE Trans Visual Comput Graph. 2024;30:4497–513. 10.1109/TVCG.2023.3263739 [DOI] [PubMed] [Google Scholar]
- 33. Gaudry S, Hajage D, Martin-Lefevre L et al. Comparison of two delayed strategies for renal replacement therapy initiation for severe acute kidney injury (AKIKI 2): a multicentre, open-label, randomised, controlled trial. Lancet. 2021;397:1293–300. 10.1016/S0140-6736(21)00350-0 [DOI] [PubMed] [Google Scholar]
- 34. Banigan MA, Keim G, Traynor D et al. Association of continuous kidney replacement therapy timing and mortality in critically ill children. Pediatr Nephrol. 2024;39:2217–26. . 10.1007/s00467-024-06320-w [DOI] [PubMed] [Google Scholar]
- 35. Kearns GL, Abdel-Rahman SM, Alander SW et al. Developmental pharmacology—drug disposition, action, and therapy in infants and children. N Engl J Med. 2003;349:1157–67. 10.1056/NEJMra035092 [DOI] [PubMed] [Google Scholar]
- 36. Chawla LS, Bellomo R, Bihorac A et al. Acute kidney disease and renal recovery: consensus report of the Acute Disease Quality Initiative (ADQI) 16 Workgroup. Nat Rev Nephrol. 2017;13:241–57. 10.1038/nrneph.2017.2 [DOI] [PubMed] [Google Scholar]
- 37. Ruth A, Basu RK, Gillespie S et al. Early and late acute kidney injury: temporal profile in the critically ill pediatric patient. Clin Kidney J. 2022;15:311–9. 10.1093/ckj/sfab199 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Li X, Yuan F, Zhou L. Organ crosstalk in acute kidney injury: evidence and mechanisms. J Clin Med. 2022;11:6637. 10.3390/jcm11226637 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Ronco C, Bellomo R. Acute renal failure and multiple organ dysfunction in the ICU: from renal replacement therapy (RRT) to multiple organ support therapy (MOST). Int J Artif Organs. 2002;25:733–47. 10.1177/039139880202500801 [DOI] [PubMed] [Google Scholar]
- 40. Kellum JA, Sileanu FE, Murugan R et al. Classifying AKI by urine output versus serum creatinine level. J Am Soc Nephrol. 2015;26:2231–8. 10.1681/ASN.2014070724 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Bianchi NA, Stavart LL, Altarelli M et al. Association of oliguria with acute kidney injury diagnosis, severity assessment, and mortality among patients with critical illness. JAMA Netw Open. 2021;4:e2133094. 10.1001/jamanetworkopen.2021.33094 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Kaddourah A, Basu RK, Goldstein SL et al. Oliguria and acute kidney injury in critically ill children: implications for diagnosis and outcomes*. Pediatr Crit Care Med. 2019;20:332–9. 10.1097/PCC.0000000000001866 [DOI] [PubMed] [Google Scholar]
- 43. Woodcock TE, Woodcock TM. Revised Starling equation and the glycocalyx model of transvascular fluid exchange: an improved paradigm for prescribing intravenous fluid therapy. Br J Anaesth. 2012;108:384–94. 10.1093/bja/aer515 [DOI] [PubMed] [Google Scholar]
- 44. Gomes RADS, Azevedo LF, Simões BPC et al. Fluid overload: clinical outcomes in pediatric intensive care unit. J Pediatr. 2023;99:241–6. 10.1016/j.jped.2022.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Selewski DT, Gist KM, Basu RK et al. Impact of the magnitude and timing of fluid overload on outcomes in critically ill children: a Report From the Multicenter International Assessment of Worldwide Acute Kidney Injury, Renal Angina, and Epidemiology (AWARE) Study. Crit Care Med. 2023;51:606–18. 10.1097/CCM.0000000000005791 [DOI] [PubMed] [Google Scholar]
- 46. Claure-Del Granado R, Mehta RL. Fluid overload in the ICU: evaluation and management. BMC Nephrol. 2016;17:109. 10.1186/s12882-016-0323-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. KDIGO 2026 Clinical Practice Guidelines For Acute Kidney Injury (AKI) and Acute Kidney Disease (AKD).
- 48. Subhash S, Raina R. Beyond filtration: a kidney support therapy framework for interpreting renal replacement therapy timing evidence. J Intensive Med. 2026;6:349–50. 10.1016/j.jointm.2026.01.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Landzberg E, Ogdie A, Yarnell C et al. Observational studies of early versus late salvage therapies in critical care exhibit intrinsic selection bias: two meta-analyses. Crit Care. 2025;29:411. 10.1186/s13054-025-05663-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Gist KM, Fuhrman DY, Akcan-Arikan A. Small patients, big gaps: why adult CRRT timing evidence does not translate to children. Crit Care. 2026;30:72. 10.1186/s13054-026-05893-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Due to the size of the cohort, the data are not publicly available as individual participants could be easily identifiable, posing a risk to patient privacy.