Journal of Critical Care 29 (2014) 37–42

Contents lists available at ScienceDirect

Journal of Critical Care journal homepage: www.jccjournal.org

Outcomes

Indications and outcomes in children receiving renal replacement therapy in pediatric intensive care☆ Erin D. Boschee, BSc a, Dominic A. Cave, MBBS, FRCPC b, Daniel Garros, MD c, Laurance Lequier, MD, FRCPCP c, Donald A. Granoski, RRT c, Gonzalo Garcia Guerra, MD c, Lindsay M. Ryerson, MD, FRCPC c,⁎ a b c

Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Alberta, Canada Department of Anesthesiology, University of Alberta, Edmonton, Alberta, Canada Department of Pediatrics, University of Alberta, Edmonton, Alberta, Canada

a r t i c l e

i n f o

Keywords: Pediatric Acute kidney injury Renal replacement therapy Peritoneal dialysis

a b s t r a c t Purpose: We aimed to describe patient characteristics, indications for renal replacement therapy (RRT), and outcomes in children requiring RRT. We hypothesized that fluid overload, not classic blood chemistry indications, would be the most frequent reason for RRT initiation. Materials and Methods: A retrospective cohort study of all patients receiving RRT at a single-center quaternary pediatric intensive care unit between January 2004 and December 2008 was conducted. Results: Ninety children received RRT. The median age was 7 months (interquartile range, 1-83). Forty-six percent of patients received peritoneal dialysis, and 54% received continuous renal replacement therapy. The median (interquartile range) PRISM-III score was 14 (8-19). Fifty-seven percent had congenital heart disease, and 32% were on extracorporeal life support. The most common clinical condition associated with acute kidney injury was hemodynamic instability (57%; 95% confidence interval [CI], 46-67), followed by multiorgan dysfunction syndrome (17%; 95% CI, 10-26). The most common indication for RRT initiation was fluid overload (77%; 95% CI, 66-86). Seventy-three percent (95% CI, 62-82) of patients survived to hospital discharge. Conclusions: Hemodynamic instability and multiorgan dysfunction syndrome are the most common clinical conditions associated with acute kidney injury in our population. In the population studied, the mortality was lower than previously reported in children and much lower than in the adult population. © 2014 Elsevier Inc. All rights reserved.

1. Introduction The role of renal replacement therapy (RRT) in critically ill pediatric patients is expanding. Classic indications for RRT in the pediatric population include metabolic/electrolyte imbalance, uremia with bleeding and/or encephalopathy, fluid overload (FO) with pulmonary edema/respiratory failure, intoxications, inborn errors of metabolism (IEM), and nutritional support [1]. Renal replacement therapy encompasses intermittent hemodialysis, peritoneal dialysis (PD), and continuous RRT (CRRT). Since the late 1990s, the latter has become the modality of choice in many pediatric intensive care units (PICUs) [1]. Continuous RRT is a mainstay of treatment in PICU for acute kidney injury (AKI) due to well-maintained hemodynamic stability with slow predictable fluid and solute removal [2]. The decision to initiate RRT is influenced by patient characteristics, physician beliefs, and institution-specific factors [3]. ☆ Financial support provided by a grant from Women and Children's Health Research Institute. ⁎ Corresponding author. WMC 3A3.19, Edmonton, AB, Canada T6G 2B7. E-mail address: [email protected] (L.M. Ryerson). 0883-9441/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcrc.2013.09.008

Despite the rapid expansion of RRT in the PICU, there are no set guidelines that describe when RRT should be initiated and in whom. The inability to generalize findings from adult RRT studies to critically ill children due to differences in age, size, disease, and comorbidities necessitates further research into the current indications and outcomes of pediatric RRT [4]. We aimed to describe the population of critically ill children requiring RRT at our institution in terms of patient characteristics, degree of AKI, indications for RRT, and outcomes. A secondary objective was to identify risk factors for mortality in those patients who needed RRT. 2. Materials and methods 2.1. Patients The Health Research Ethics Board at the University of Alberta approved this study. The requirement for individual informed patient consent was waived. Data on all patients who received RRT at Stollery Children's Hospital between January 2004 and December 2008 were reviewed. Renal replacement therapy was defined as any form of acute blood purification in the PICU including PD and CRRT.

38

E.D. Boschee et al. / Journal of Critical Care 29 (2014) 37–42

Subjects who were RRT dependent before PICU admission were excluded. Patients were identified by the prospectively collected Pediatric ICU Evaluations (PICUES) database and cross-referenced with our PICU RRT patient database. Data were retrospectively gathered from the PICUES and RRT databases as well as from individual patients' medical records. Patient characteristics included age, sex, height, weight, and PRISM-III score at the time of PICU admission. The primary diagnosis at PICU admission was classified into 7 categories: post–cardiac surgery, systemic inflammatory response syndrome/sepsis/shock, respiratory failure, cardiac arrest, primary renal disease, liver disease/transplant, and drug overdose. Any diagnosis that did not fit one of these categories was labeled “other” and was detailed on the case report form. Chronic diagnoses including cancer, history of previous transplant, congenital heart disease, diabetes, chronic renal failure (not RRT dependent), and chromosomal abnormalities were also recorded. Any other major chronic diagnoses deemed as potentially contributing to the patient's PRISM-III score at the time of PICU admission were noted in a separate category termed other. Use of mechanical ventilation, inotropes and/or vasopressors, or extracorporeal life support (ECLS) was recorded. Associations with AKI were classified as hemodynamic instability, multiple-organ dysfunction syndrome (MODS), tumor lysis syndrome (TLS), hemolytic uremic syndrome/thrombotic thrombocytopenic purpura, IEM, or other. Hemodynamic instability was defined as a systolic blood pressure less than the fifth percentile for age, despite fluid resuscitation, with concurrent use of intravenous inotropes and/ or vasopressors. Multiple-organ dysfunction syndrome was defined as the presence of at least 3 failed organ systems concomitantly during the PICU admission [2]. The degree of AKI was determined by the pediatric-modified RIFLE (pRIFLE) score, which stratifies AKI into levels as Risk, Injury, Failure, Loss or End-stage, based on criteria described by Akcan-Arikan et al [5]. This was achieved by calculation of eGFR or prior 24-hour urine output [5]. Indications for RRT were deduced from the medical records and classified into the following: FO (N 10%), metabolic acidosis (pH b 7.1), hyperkalemia (N 5.5 mmol/L), symptoms of uremia, TLS, IEM, intoxication, or other. Patients may have had more than 1 indication for RRT. The percent FO (% FO) was calculated by determining fluid input and output from the time of PICU admission until RRT initiation using the following formula by Goldstein et al [6]. % FO ¼ ð½total fluid intake−total fluid output ðLÞ=admission body 

weight in kgÞ 100 Fluid overload greater than 10% was chosen as a cutoff value based on the published literature demonstrating a significant increased mortality with FO greater than 10% [7–9]. Duration of RRT, length of time in the PICU before RRT initiation, and the need for multiple runs of RRT were recorded. In addition, RRT complications were recorded including bleeding, thrombosis, infection, central nervous system events, shock requiring fluid resuscitation, and catheter malfunction. 2.2. Renal replacement therapy PRISMA and PRISMAFLEX (Cobe-Gambro Healthcare, Lakewood, Col) hemofiltration machines were used. Blood flow rates ranged from 50 to 200 mL/min. Data on mode of CRRT initially ordered: (continuous venovenous hemodialysis, continuous venovenous hemofiltration, or continuous venovenous hemodiafiltration) was collected. Anticoagulation was either heparin for patients on ECLS or citrate for all others, unless contraindicated. Peritoneal dialysis was initiated by the placement of a Tenckhoff catheter in the peritoneal cavity and subsequent dialysis. Some neonates returned to PICU after palliative cardiac surgery with a

Tenckhoff catheter in place; these patients were not included in the analysis, unless PD cycles were started in the PICU. Small volume (10 mL/kg) and continuous cycles (30-60 minutes) were used. Dialysate fluid (1.5%-4.25% glucose) was used with 25 mmol/L sodium bicarbonate and a variable potassium prescription. 2.3. Statistics The primary outcomes were 28-day mortality and survival to hospital discharge. Secondary outcomes were PICU length of stay (LOS), hospital LOS, ventilator days, and RRT dependence at hospital discharge. We explored the data for possible risk factors to identify mortality in those patients who needed RRT. Descriptive statistics were used to summarize both continuous and categorical variables. Survivors and nonsurvivors were compared for their baseline characteristics and parameters of renal function. Continuous data are presented as means with SD or median with interquartile range (IQR), where appropriate. Nonnormally distributed data were analyzed using the Mann-Whitney U test. Categorical data are presented as counts with percentages and analyzed using the Fisher exact test or the χ 2 test. Statistical analysis was performed using Stata version 10.0 software (Stata Corporation, College Station, Tex). A P value less than .05 was considered statistically significant. 3. Results Between 2004 and 2008, 98 patients were identified as having received RRT. Seven patients were excluded because of a primary diagnosis of chronic renal failure and preexisting RRT before their PICU admission. One patient was excluded because there was no available PRISM-III score at PICU admission. This left 90 patients who received RRT over the 5-year period for the analysis. The median (IQR) age was 7 (1-83) months, and the median (IQR) weight was 6.2 (3.5-20) kg. Sixty-three percent were male. Details of patient baseline demographics and clinical characteristics are presented in Table 1. The most common primary diagnosis was post– congenital cardiac surgery. These patients are presented separately (Table 2). Two patients had a primary diagnosis of acute renal disease; these patients were not RRT dependent before their PICU admission and were included in our analysis. The median (IQR) length of time in PICU before RRT initiation in all subjects was 2 days (1-4). Most patients were intubated and mechanically ventilated (94%) and received inotropes and/or vasopressors (82%). Twenty-nine patients (32%) were on ECLS. Hemodynamic instability was the most common association with AKI in 51 patients (57%). Fifteen patients (17%) had MODS. Survivors and nonsurvivors were compared. Multiple-organ dysfunction syndrome was 3 times more common in nonsurvivors (P = .021). Nonsurvivors had a median 7 days in PICU before RRT initiation compared with 2 days in survivors (P = .089; Table 1). Serum urea and eGFR were similar in both groups at the initiation of RRT. The most common indication for initiation of RRT was FO (77%). Other classic indications included uremia (22%), hyperkalemia (26%), and metabolic acidosis (22%). Intoxication, TLS, and IEM as indications for RRT were found in only a few patients. Median % FO was clinically different but not statistically different between the survivors and the nonsurvivors (P = .27). Survivors and nonsurvivors had similar numbers of patients with pRIFLE class I or F. No patients had a pRIFLE diagnosis of loss or end-stage AKI. Postoperative cardiac patients were significantly younger (P b .0001) and smaller (P b .0001) compared with the non–postoperative group (Table 2). There was no difference in the incidence of AKI diagnosed by pRIFLE between the 2 groups. The median creatinines, both at PICU admission (P = .007) and RRT initiation (P = .006), were significantly different, but this is merely reflective of the differences in

E.D. Boschee et al. / Journal of Critical Care 29 (2014) 37–42

39

Table 1 Baseline patient characteristics overall and by survival status Variable

All patients (n = 90)

Survivors (n = 66)

Nonsurvivors (n = 24)

P

Age (mo) Sex (male), n (%) Weight (kg) PRISM-III score Days in PICU pre-RRT Patients intubated and ventilated, n (%) Patients on inotropes/pressors, n (%) Patients on ECLS, n (%) % FO from PICU admission to RRT initiationa eGFR at RRT initiation (mL/min) Creatinine at PICU admission Urea at PICU admission Creatinine at RRT initiation Urea at RRT initiation Primary diagnosis Post–cardiac surgery, n (%) Sepsis, n (%) Post–cardiac arrest, n (%) Post–liver transplant, n (%) Respiratory failure, n (%) Intoxication, n (%) Primary renal disease, n (%) Otherb, n (%) pRIFLE score at RRT initiation Risk, n (%) Injury, n (%) Failure, n (%)

7 (1-83) 57 (63.3) 6.2 (3.5-20) 14 (8-19) 2 (1-4) 85 (94.4) 74 (82.2) 29 (32.2) 13.3 (4.8-22.4) 26.3 (18.1-39.3) 71.5 (45-112) 7.2 (4.6-11.2) 117 (76-206) 14.6 (8.1-27.4)

6.5 (2-73) 41 (62.1) 5.6 (3.5-20) 14.5 (10-17) 2 (1-4) 63 (95.5) 52 (78.8) 20 (30.3) 12.1 (4.8-36) 25.9 (19.3-36.8) 77.5 (49-130) 7.3 (4.8-11.4) 119 (80-231) 14.9 (10-25)

7.5 (1-82) 16 (66.7) 9.2 (3.5-16) 13.5 (7-23) 7 (1-7) 22 (91.7) 22 (91.7) 9 (37.5) 20.1 (6.4-26.8) 29.9 (13.1-50.2) 57.5 (38-105) 6.8 (4.5-10.5) 116.5 (51-148) 14.2 (6.9-30.1)

40 (44.4) 9 (10) 6 (6.7) 5 (5.6) 5 (5.6) 4 (4.4) 2 (2.2) 19 (21.1)

33 (50) 5 (7.6) 4 (6.1) 2 (3.0) 1 (1.5) 4 (6.1) 2 (3.0) 15 (22.7)

7 (29.2) 4 (16.7) 2 (8.3) 3 (12.5) 4 (16.7) 0 0 4 (16.7)

.096 .240 .656 .116 .017 .570 1.000 .520

6 (6.7) 10 (11.1) 60 (66.7)

4 (6.1) 7 (10.6) 44 (66.7)

2 (8.3) 3 (12.5) 16 (66.7)

.656 .723 1.00

.912 .807 .989 .959 .089 .606 .219 .612 .270 .216 .124 .457 .182 .779

Data presented as median (IQR) unless otherwise noted. a n = 75. b Other includes congenital heart disease awaiting corrective surgery, status epilepticus, cancer, diabetic ketoacidosis, endocarditis, bowel obstruction, and hemophagocytic lymphohistiocytosis.

age. There was no statistical difference in the median % FO or survival between the 2 groups. Both PD and CRRT subjects were also compared (Table 3). Fortyone patients (46%) were initially prescribed PD, and 49 received CRRT (54%). Of the CRRT patients, 34 had continuous venovenous hemodiafiltration, 12 had continuous venovenous hemodialysis, and 3 received some form of CRRT that was not clearly specified. In general, subjects on PD were younger (P b .0001), were smaller (P b .0001), and were post–cardiac surgery (P b .001) for congenital heart disease. The median length of time in PICU before initiation of RRT was 2 days for both PD and CRRT groups (P = .987). PRISM-III scores at admission were similar. The % FO and survival to hospital discharge were also similar. Nine patients had multiple runs of CRRT. The median duration of RRT was 7 days for both the PD and CRRT groups.

Twenty-nine patients (32%) were on ECLS as well as CRRT (Table 4). Median PRISM score of ECLS patients was 17 (IQR, 15-23). Sixteen patients (55%) of the ECLS group were post cardiacvascular surgery. Twenty-two (76%) of the ECLS group received CRRT, whereas 7 patients (24%) were on PD. The patients on ECLS who were on PD all had PD catheters in situ before cannulation; PD was attempted in this patient group before adding a PRISMA/PRISMAFLEX machine to the ECLS circuit. The primary indication for RRT in the ECLS group was FO. Complications of RRT were infrequent. Six patients (15%) with PD developed peritonitis. The incidence of peritonitis with PD varies in the literature from 6% to 38% [10–13]. One patient on PD developed bleeding from the catheter site. Five patients on CRRT had complications; 3 had problems with the CRRT catheter, including infection, bleeding, or catheter malfunction; 2 patients had a cardiac arrest at

Table 2 Postoperative cardiac patients and non–postoperative cardiac patients Variable Age (mo) Weight (kg) PRISM-III score Patients on inotropes/pressors, n (%) pRIFLE score at RRT initiation Risk, n (%) Injury, n (%) Failure, n (%) % FO from PICU admission to RRT initiation Creatinine at PICU admission Creatinine at CRRT initiation PD, n (%) Survival to discharge, n (%)

All patients (n = 90)

Postoperative cardiac (n = 40)

Non–postoperative cardiac (n = 50)

P

59.5 (7-154) 18.2 (7-45) 13 (7-20) 37 (74.0)

b.0001 b.0001 .275 .027

7 (1.0-83.0) 6.2 (3.5-20.0) 14 (8-19) 74 (82.2)

2 (0-4.5) 3.6 (2.9-5.3) 15 (12-17.5) 37 (92.5)

6 (6.7) 10 (11.1) 60.(66.7) 13.3 (4.8-22.4)

2 (5) 3 (7.5) 28 (70) 11.7 (7.6-21.2)

4 (8) 7 (14) 32 (64) 17.9 (4.8-32.5)

.689 .502 .654 .290

71.5 (45-112) 117 (76-206) 41 (45.5) 66 (73.3)

58.5 (42-78) 93 (66-127) 31 (77.5) 33 (82.5)

96.5 (47-156) 148 (86-302) 10 (20) 33 (66)

.007 .006 b .0001 .096

Data presented as median (IQR) unless otherwise noted.

40

E.D. Boschee et al. / Journal of Critical Care 29 (2014) 37–42

Table 3 PD and CRRT patients Variable

Table 5 Clinical outcomes of 90 patients PD patients (n = 41)

Age (mo) 3 (1-6) Weight (kg) 4.1 (3.2-5.7) PRISM-III score 14 (11-17) Patients intubated and ventilated, n (%) 39 (95.1) Patients on inotropes/pressors, n (%) 36 (87.8) Patients on ECLS, n (%) 7 (17.1) Primary diagnosis Post–cardiac surgery, n (%) 31 (75.6) Sepsis, n (%) 4 (9.8) % FO from PICU admission to 18.4 (4.1-18.8) RRT initiation Days in PICU pre-RRT 2 (1-4) Duration of RRT (d) 7 (4-11) Survival to hospital discharge, n (%) 33 (80.5)

CRRT patients (n = 49)

P

Outcomes

65 (3-165) 19.4 (5.5-45) 15 (7-23) 46 (93.9) 38 (77.6) 22 (44.9)

b.0001 b.0001 .792 .585 .161 .006

PICU LOS (d) Ventilator (d) Hospital LOS (d) 28-d mortality, n (%) Survival to hospital discharge, n (%) RRT dependent at PICU discharge, n (%) RRT dependent at hospital discharge, n (%)

9 (18.4) 5 (10.2) 20.1 (5.4-32.5)

b.001 1.000 .155

2 (1-5) 7 (3-18) 33 (67.3)

.987 .864 .321

Data presented as median (IQR) unless otherwise noted.

some point during their PICU admission after CRRT initiation. The etiology of the cardiac arrests was unclear, and although they may be unrelated to CRRT, they have been included for completeness. Overall, 28-day mortality was 10% and survival to hospital discharge was 73% (Table 5). The median (IQR) PICU LOS was 17 (10-36) days, and the median (IQR) hospital stay was 49.5 (29-87) days. Three patients (4.5%) were RRT dependent upon hospital discharge.

4. Discussion This retrospective cohort study evaluated a 5-year institutional experience in the provision of RRT to pediatric patients. There is a paucity of data on pediatric AKI and the benefits and risks of RRT. In critically ill adults with AKI, up to 70% require RRT [14]; this percentage is unclear among critically ill pediatric populations. Epidemiologic trends show a shift in the primary etiology of pediatric AKI from primary renal disease and hemolytic uremic syndrome to secondary renal injury due to renal ischemia, nephrotoxic medications, and sepsis [15]. Pediatric patients with AKI secondary to sepsis and complex multisystem failures have higher mortality than do patients with primary AKI [16]. Fluid overload in the setting of AKI and RRT is associated with mortality and is a target for intervention [2,5– 9,17,18]. Furthermore, mortality in pediatric patients with AKI necessitating CRRT (vs other RRT therapies) is greater than 40% [2,4,15], which is similar to that seen in adult studies [19]. Use of pRIFLE criteria may allow improved and more consistent recognition of pediatric AKI, although serum creatinine and urine output are late markers of renal injury and say little about current renal function [5]. There is no biomarker of renal injury that is currently in widespread

Table 4 Patient characteristics of ECLS and non-ECLS patients Variable Age (mo) Weight (kg) PRISM-III score CRRT, n (%) PD, n (%) Primary diagnosis Post–cardiac surgery, n (%) Sepsis, n (%) % FO from PICU admission to RRT initiation Survival to hospital discharge, n (%)

ECLS patients (n = 29)

Non-ECLS (n = 61)

P

3 (1-25) 4.3 (3.3-11.5) 17 (15-23) 22 (75.9) 7 (24.1)

9 (2-88) 7.5 (4-20.5) 13 (8-16) 27 (44.3) 34 (55.7)

.068 .082 .001 .006 .006

16 (55.2) 1 (3.5) 19.7 (8.6-22.3)

24 (39.3) 8 (13.1) 12.5 (4.1-24.1)

.006 .262 .375

20 (69.0)

Data presented as median (IQR) unless otherwise note.

46 (75.4)

.612

17 (10-36) 15 (29-87) 50 (9-29) 9 (10) 66 (73.3) 8 (10) 3 (4.5)

Data presented as median (IQR) unless otherwise noted.

use to immediately detect AKI, nor is there a “golden hour” for AKI; we argue that waiting for significant changes in serum creatinine may be too late. In our study, survival to hospital discharge was 73% for all patients including 81% survival for PD patients and 67% survival for CRRT patients (P = .321). This result signifies an improvement in survival compared with recently published pediatric studies [2,4], although our patient population includes PD as a form of RRT. A recent publication in near-term/term neonates, with unknown therapy for RRT, demonstrates a 78% survival rate [20]. This improved survival rate in neonates, compared with pediatric populations, is in accordance with our higher survival rate because almost half of our population were young infants who received PD. It is not clear in the article by Askenazi et al [20] whether their improved survival is due to a different cause of AKI, a different natural history in this age group, or a different mode of RRT because this is unknown. Our patients likely represent a very different population from other published pediatric studies, but we believe that they are representative of the case mix in any large PICU. Our patients were considerably younger and smaller compared with other recently published pediatric studies, which is a surprising finding because typically outcomes are worse for children younger than 1 year and weighing less than 10 kg who require CRRT [4]. We hypothesized that a possible reason for our improved survival was the large percentage of patients who were post–cardiac surgery for congenital heart disease; however, there was no difference in survival between postoperative cardiac patients and non–postoperative cardiac patients. The postoperative cardiac group was younger, smaller, and more likely to receive PD. These patients likely had single-system disease without associated comorbidities, but their PRISM-III scores were similar. In younger, smaller children, initiation of PD is usually preferred because large bore CRRT catheters may be difficult to place. There are a relatively increased number of solute transporters in the peritoneal membrane in neonates and infants. This combined with their relative efficiency in the setting of critical illness makes PD highly effective in this population. Sasser et al [21] describe the use of prophylactic PD in high-risk neonates and infants after CPB. They found that the use of prophylactic PD was associated with decreased incidence of severe FO as well as lower serum concentrations of inflammatory cytokines in the early postoperative period [21]. Sasser's failure to demonstrate improved survival with prophylactic PD vs peritoneal drainage despite a more negative fluid balance is in agreement with our findings that there was no difference in FO between survivors and nonsurvivors. This suggests that the mechanism for improved survival may lie elsewhere. The most significant associations with AKI requiring RRT in our PICU population were hemodynamic instability and MODS. The high incidence of hemodynamic instability as a cause of AKI is again likely linked to the significant number of cardiac surgical patients because Stollery Children's Hospital is the primary pediatric cardiac center in Western Canada. Multiple-organ dysfunction syndrome was 3 times more common in nonsurvivors than in survivors. Only 17% of our total study population had MODS. This has likely also contributed to our improved mortality because many of the

E.D. Boschee et al. / Journal of Critical Care 29 (2014) 37–42

studies in patients on CRRT had much higher numbers of patients with sepsis and MODS [2,4,7]. Another possible reason for improved survival in our patient population may have been early initiation of RRT. The median time from PICU admission until RRT initiation was 2 days, which is similar to other recent studies [2,4]. A notable difference between our survivors and nonsurvivors was the time to initiation of RRT (median 2 days vs 7 days), suggesting that the earlier initiation of RRT may improve outcomes. There are retrospective observational data in adults that support early initiation of RRT [22–24]. This has not been prospectively demonstrated in the pediatric population, although volume overload is associated with worse outcomes [25,26] so early initiation of RRT to prevent volume overload should lead to improved outcomes. Although our survival is higher than previously reported, we believe that our patients were equally unwell at the time of PICU admission. Our median PRISM-III score was 14, which is comparable with other studies [2]. Our survival was not biased by large numbers of patients with intoxication or TLS who typically have favorable outcomes. The purpose of the PRISM-III score is to predict mortality based on certain patient characteristics at the time of PICU admission. It is possible that the PRISM-III score does not accurately account for AKI, but it remains the most widely used scoring system for predicting mortality in the PICU. Akcan-Arikan et al [5] report that AKI was an independent predictor of mortality independent of the PRISM-II score. The most common indication for RRT initiation in our population was to manage FO. We defined FO as greater than 10% based on the published literature demonstrating significantly increased mortality with an FO greater than 10% [7–9]. A precise definition of clinically significant FO has not been identified. The % FO may simply represent a marker of hemodynamic instability and the subsequent need for fluid resuscitation. It has been demonstrated that % FO is associated with mortality in the PICU [2,5–9,17,18,27,28]. However, no study to date has demonstrated causality. Our study is one of the first to demonstrate no difference in FO between survivors and nonsurvivors; this may suggest a different pathophysiology in our patient population. We believe that FO is associated with mortality in the pediatric population and argue that early initiation of RRT to prevent FO is associated with improved survival. Selewski et al [28] demonstrated in a population of patients on CRRT and ECMO that prevention of significant FO is more effective at improving outcomes than attempting fluid removal once FO is established. We did not collect data on fluid removal in our patients. Our median FO was clinically different between survivors and nonsurvivors, although not statistically different because of the wide range. The specific % FO at which to intervene and the benefits of intervention remain unknown. The present study has its limitations. It was a retrospective database review with the associated shortcomings inherent of databases. Although we present 90 patients who received RRT from a single institution over a 5-year period, this is a relatively small number for a retrospective review, and the division of patients between PD and CRRT may make it difficult to draw conclusions. We also included a large number of patients who were post–cardiac surgery for congenital heart disease, although we believe that many large PICUs in North America would have a similar case mix. However, whether our results apply to other centers is unknown. Another limitation is the lack of clear records stating the actual reasons that persuaded the attending physician to start RRT in those patients who did not have classic indications.

5. Conclusions We present a diverse population of pediatric patients requiring RRT, both PD and CRRT, which we think is reflective of the case mix at large combined cardiac and medical-surgical PICUs. Hemodynamic instability and MODS are the most significant associations with AKI in

41

our PICU population. The most common indication for RRT was management of FO. The paucity of conventional indications of acid base, electrolyte disturbance, or severe uremia in the presence of significant renal dysfunction as assessed by pRIFLE criteria suggests that renal impairment is being recognized early and managed aggressively. In the population studied, the mortality was lower than previously reported in children receiving CRRT and much lower than that in the adult population. Acknowledgments The authors would like to acknowledge the Women and Children's Health Research Institute in Edmonton, Alberta, Canada, for providing financial support for this research project. References [1] Walters S, Porter C, Brophy PD. Dialysis and pediatric acute kidney injury: choice of renal support modality. Pediatr Nephrol 2009;24:37–48. [2] Hayes LW, Oster RA, Tofil NM, et al. Outcomes of critically ill children requiring continuous renal replacement therapy. J Crit Care 2009;24:394–400. [3] Gibney N, Hoste E, Burdmann EA, et al. Timing of initiation and discontinuation of renal replacement therapy in AKI: unanswered key questions. Clin J Am Soc Nephrol 2008;3:876–80. [4] Symons JM, Chua AN, Somers MJG, et al. Demographic characteristics of pediatric continuous renal replacement therapy: a report of the prospective pediatric continuous renal replacement therapy registry. Clin J Am Soc Nephrol 2007;2: 732–8. [5] Akcan-Arikan A, Zappitelli M, Loftis LL, et al. Modified RIFLE criteria in critically ill children with acute kidney injury. Kidney Int 2007;71:1028–35. [6] Goldstein SL, Currier H, Graf C, et al. Outcome in children receiving continuous venovenous hemofiltration. Pediatrics 2001;107:1309–12. [7] 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. [8] Foland JA, Fortenberry JD, Warshaw BL, et al. Fluid overload before continuous hemofiltration and survival in critically ill children: a retrospective analysis. Crit Care Med 2004;32:1771–6. [9] Gillespie RS, Seidel K, Symons JM. Effect of fluid overload and dose of replacement fluid on survival in hemofiltration. Pediatr Nephrol 2004;19:1394–9. [10] Werner HA, Wensley DF, Lirenman DS, et al. Peritoneal dialysis in children after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1997;113:64–6811. [11] Chan KL, Ip P, Chiu CS, Cheung YF. Peritoneal dialysis after surgery for congenital heart disease in infants and young children. Ann Thorac Surg 2003;76:1443–9. [12] Pedersen KR, Hjortdal VE, Christensen S, et al. Clinical outcome in children with acute renal failure treated with peritoneal dialysis after surgery for congenital heart disease. Kidney Int 2008;73:S81–S861. [13] Fleming F, Bohn D, Edwards H, et al. Renal replacement therapy after repair of congenital heart disease in children. J Thorac Cardiovasc Surg 1995;109: 322–31. [14] Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005;294:813–8. [15] Hui-Stickle S, Brewer ED, Goldstein SL. Pediatric ARF epidemiology at a tertiary care center from 1999 to 2001. Am J Kidney Dis 2005;45:96–101. [16] Bunchman TE, McBryde KD, Mottes TE, et al. Pediatric acute renal failure: outcome by modality and disease. Pediatr Nephrol 2001;16:1067–71. [17] Goldstein SL, Somers MJ, Baum MA, et al. Pediatric patients with multi-organ dysfunction syndrome receiving continuous renal replacement therapy. Kidney Int 2005;67:653–8. [18] Selewski DT, Cornell TT, Lombel RM, et al. Weight-based determination of fluid overload status and mortality in pediatric intensive care unit patients requiring continuous renal replacement therapy. Int Care Med 2011;37:1166–73. [19] Bagshaw SM, Wald R, Barton J, et al. Clinical factors associated with initiation of renal replacement therapy in critically ill patients with acute kidney injury—a prospective multicenter observational study. J Crit Care 2012;27:268–75. [20] Askenazi DJ, Koralkar R, Hundley HE, et al. Fluid overload and mortality are associated with acute kidney injury in sick near-term/term neonates. Pediatr Nephrol 2013;28:661–6. [21] Sasser WC, Dabal RJ, Askenazi DJ, et al. Prophylactic peritoneal dialysis following cardiopulmonary bypass is children is associated with decreased inflammation and improved clinical outcomes. Congenit Heart Dis 2013 In press. [22] Bent P, Tan HK, Bellomo R, et al. Early and intensive continuous hemofiltration for severe renal failure after cardiac surgery. Ann Thorac Surg 2001;71:832–7. [23] Gettings LG, Reynolds HN, Scalea T. Outcome in post-traumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Int Care Med 1999;25:805–13. [24] Seabra VF, Balk EM, Liangos O, et al. Timing of renal replacement therapy initiation in acute renal failure: a meta-analysis. Am J Kidney Dis 2008;52:272–84.

42

E.D. Boschee et al. / Journal of Critical Care 29 (2014) 37–42

[25] Bellomo R, Cass A, Cole L, et al. An observational study fluid balance and patient outcomes in the randomized evaluation of normal vs. augmented level of replacement therapy trial. Crit Care Med 2012;40:1753–60. [26] Arikan AA, Zappitelli M, Goldstein SL, et al. Fluid overload is associated with impaired oxygenation and morbidity in critically ill children. Pediatr Crit Care Med 2012;13:253–8.

[27] Hazle MA, Gajarski RJ, Yu S, et al. Fluid overload in infants following congenital heart surgery. Pediatr Crit Care Med 2013;14:44–9. [28] Selewski DT, Cornell TT, Blatt NB, et al. Fluid overload and fluid removal in pediatric patients on extracorporeal membrane oxygenation requiring continuous renal replacement therapy. Crit Care Med 2012;40:2694–9.