Introduction

Acute kidney injury (AKI) occurs commonly in critically ill children in a variety of settings with an increasing incidence and evolving epidemiology [1–5]. Several studies have shown that AKI is associated with adverse outcomes, including increased mortality, length of mechanical ventilation, and length of stay (LOS) in hospital in a variety of pediatric populations [1, 4–9]. Despite a number of advancements in the recognition and diagnosis of AKI there currently are no proven therapeutic interventions to prevent the development of or treat established AKI. Until preventive or interventional treatments are developed and validated, pediatric AKI management will be focused on modifiable factors, such as preventing additional kidney injury, optimizing nutrition, and minimizing the sequelae of electrolyte abnormalities and fluid overload (FO).

Pediatric nephrologists and intensivists have been at the forefront of recognizing the impact FO has on outcomes in critically ill children with and without AKI. FO commonly occurs in critically ill children and has been recognized as an independent predictor of adverse outcomes. The genesis of FO in critically ill children is multifactorial in nature, with contributions from overly aggressive fluid administration, systemic inflammation and capillary leak, disordered fluid homeostasis, and AKI. FO represents a modifiable target for intervention as a number of treatment strategies exist, including fluid management, medications, and renal replacement therapy (RRT). In this review we comprehensively describe the definition of FO, the impact of FO on kidney function, the impact of FO on the diagnosis of AKI, the association of FO with outcomes, the targeted therapy of FO, and the impact of the timing of RRT.

Definition

In order to monitor fluid balance and accurately identify FO, it is critical to accurately define FO. The two broad methodologies that are commonly utilized to define FO include cumulative intensive care unit (ICU) fluid balance and weight-based methods.

The cumulative fluid balance method represents the most commonly utilized method in the literature and relies on recording fluid intake and output from ICU admission as follows [10]:

$$ \%\ \mathrm{F}\mathrm{O} = \frac{\mathrm{Sum}\ \mathrm{of}\ \mathrm{Daily}\ \left(\mathrm{Fluid}\ \mathrm{in}\ \left(\mathrm{liters}\right)\hbox{-} \mathrm{Fluid}\ \mathrm{out}\ \left(\mathrm{liters}\right)\right)}{\mathrm{ICU}\ \mathrm{Admission}\ \mathrm{Patient}\ \mathrm{Weight}\ \left(\mathrm{kilograms}\right)}*100 $$

This method of calculating fluid balance was utilized initially in the seminal article by Goldstein et al. [10] and then in subsequent articles describing the association between FO and adverse outcomes at continuous RRT (CRRT) initiation in children [11, 12]. Several studies have subsequently utilized this definition to evaluate the association of FO with adverse outcomes in a variety of cardiac and general pediatric critical care populations not requiring CRRT [2, 13–15]. In an attempt to account for fluid administration prior to ICU admission, a hybrid of this definition has been utilized to include input and output up to 7 days prior the initiation of CRRT [16]. Similarly, there have been evaluations of outcomes utilizing a variety of baseline weights instead of the ICU admission weight, which have shown variation in the FO calculation, while continuing to demonstrate an association between FO and adverse outcomes [17, 18].

The weight-based method to describe FO calculates the degree of FO based on a change in weight from a baseline weight (most commonly weight at ICU admission):

$$ \%\ \mathrm{F}\mathrm{O} = \frac{\mathrm{Daily}\ \mathrm{Weight}\hbox{--} \mathrm{I}\mathrm{C}\mathrm{U}\ \mathrm{Admission}\ \mathrm{Weight}}{\mathrm{ICU}\ \mathrm{Admission}\ \mathrm{Weight}}*100 $$

Weight-based determinations of FO have consistently shown an association between the degree of FO and adverse outcomes [3, 19–22], and this methodology may be particularly useful in the neonatal population, where insensible fluid losses may have more impact on fluid balance [23]. It is imperative to utilize standardized daily weight measurement protocols to achieve precise and reproducible daily weights [24].

Impact of fluid overload on kidney function

The adequate provision of fluids is a key component to the management of children with AKI [25, 26]. One proposed three-phase model of fluid management of AKI in the critically ill includes fluid resuscitation/repletion, fluid balance maintenance, and fluid removal/recovery [26]. Early in the course of critical illness it is important to replace volume to ensure adequate end-organ perfusion, including that of the kidneys. Once a patient has been adequately resuscitated, fluid management strategies to minimize further unnecessary FO can directly impact renal function by preventing worsening edema and abdominal hypertension.

While a significant amount of focus has been placed on the impact of FO on the lungs, FO can directly impact kidney function, adversely impacting encapsulated organs, such as the kidneys, when increased venous pressure and interstitial edema develops [25, 27]. This may be particularly true in individuals with sepsis as recent adult studies have begun to show an association with central venous pressure and the development of AKI [28, 29]. Furthermore, FO and excessive fluid resuscitation is a risk factor for the development of abdominal hypertension and abdominal compartment syndrome, which may further impair renal perfusion [30, 31]. In children, abdominal hypertension is defined by a sustained or repeated pathologic increase in intra-abdominal pressure of ≥10 mm Hg [31]. At its extreme, intra-abdominal hypertension can manifest as abdominal compartment syndrome, which is defined in the clinical practice guidelines from the World Society of the Abdominal Compartment Syndrome as “Intra-abdominal pressure of greater than 10 mmHg associated with new or worsening organ dysfunction that can be attributed to elevated intra-abdominal pressure” [31]. In abdominal compartment syndrome increased abdominal pressure leads to increased renal venous pressure, ultimately impacting renal perfusion and resulting in kidney injury.

Impact of fluid overload on the diagnosis of AKI

The development of FO may complicate the diagnosis of AKI with the utilization of definitions that rely on change in serum creatinine. Serum creatinine is known to distribute into both the intracellular and extracellular fluid compartments and as a result may be subject to inaccuracies driven by fluid status [32]. This has lead to concern that the fluid status of a patient may lead to delays in diagnosis, misclassification, or missed diagnosis of patients with AKI, and there have been calls to correct for fluid balance when evaluating a patient for AKI [33].

An analysis of the implications of fluid balance on the timely diagnosis of AKI was performed as a secondary analysis of the Program to Improve Care in Acute Renal Disease (PICARD) study. In this analysis of 253 adults, FO was shown to contribute to a delay in and misclassification of AKI [33]. Liu et al. confirmed these findings in a re-analysis of the Fluid and Catheter Treatment Trial (FACTT) showing that correction for the degree of FO uncovered missed diagnoses of AKI and led to reclassification of others [34]. Furthermore, this study showed that patients recognized as having AKI only after fluid correction had significantly increased mortality compared to those without any evidence of AKI (31 vs. 12 %, p <0.001) [34]. Basu et al. performed the first study evaluating this concept in children in an evaluation of 92 patients undergoing arterial switch operations for transposition of the great arteries [35]. In this study correction of serum creatinine for the degree of fluid accumulation led to an increased recognition of AKI and resulted in a strengthened association between AKI and outcomes (length of ventilation, length of stay). These authors utilized the following equation to correct serum creatinine for fluid accumulation:

$$ \mathrm{Corrected}\ \mathrm{Creatinine}=\mathrm{Serum}\ \mathrm{Creatinine}\left[1+\left(\mathrm{N}\mathrm{e}\mathrm{t}\ \mathrm{Fluid}\ \mathrm{balance}/\mathrm{Total}\ \mathrm{body}\ \mathrm{water}\right)\right] $$
$$ \mathrm{Where}\ \mathrm{Total}\ \mathrm{Body}\ \mathrm{Water} = 0.6\ *\ \mathrm{Weight}\ \left(\mathrm{kg}\right) $$

These studies require validation in broader pediatric populations, but the findings suggest that the development of FO may confound the diagnosis of AKI based on changes in serum creatinine. An important caveat to the study of Basu et al. [35] is that the utilization of 0.6 as the constant to estimate total body water may not be accurate in all patients and may be impacted by a number of variables, including age, sex, and disease state (nephrotic syndrome, cirrhosis, heart failure).

The results of these studies suggest that the development of FO may precede the development of AKI as diagnosed by serum creatinine-based criteria. Hassinger et al. recently evaluated the timing and impact of FO and AKI in an evaluation of 98 children following cardiopulmonary bypass [13]. In this study early postoperative FO occurred commonly (31 % with fluid balance >5 % on post-operative day 1) and was associated with adverse outcomes, including increased LOS in hospital and longer time on mechanical ventilation. An interesting finding was that the development of FO predated the development of AKI and that while the development of FO predicted those patients who would develop AKI, the converse was not true [13]. This result is consistent with those from adult studies which have shown that the development of FO or a positive fluid balance predates or is associated with the subsequent development of AKI [36, 37]. Special consideration should be paid to post-surgical patients, as fluid administration and composition need to be carefully considered since transient syndrome of inappropriate antidiuretic hormone (SIADH) is common in this group and can lead to fluid overload without AKI [38, 39]. Thus, FO may be an early marker of kidney dysfunction itself and may serve as a biomarker. This concept has recently been codified as one of the components of the Renal Angina Index, which has been previously described [40].

Epidemiology and outcomes of fluid overload

A number of studies have evaluated the impact of FO in critical care patient populations, including neonates, post-cardiac surgery patients, general pediatric ICU patients, and those treated with extracorporeal membrane oxygenation (ECMO) or CRRT. In all settings, FO is common and associated with poor outcomes (Table 1). Here we review studies examining the epidemiology and impact of FO on outcomes in some exemplary patient populations.

Table 1 Studies on the impact of fluid overload in critical care patient populations, including study design, patient populations, and outcomes
Full size table

Continuous renal replacement therapy

Fluid overload currently represents the most common indication for the initiation of CRRT in critically ill children [49]. Pediatric clinician investigators have been at the forefront of identifying the impact of FO on mortality in critically care populations requiring CRRT [10–12, 16, 17, 20, 21, 41, 50]. In 1994, Lane et al. identified a FO of >10 % at initiation of hemodialysis as a risk factor for mortality in a population of bone marrow transplant patients [19]. In 2001, Goldstein et al. first described the association of FO at CRRT initiation with mortality in a cohort of 21 children [10]. Subsequently, the Prospective Pediatric Continuous Renal Replacement Therapy (ppCRRT) registry was formed from 13 centers in the USA to further evaluate practices and factors associated with outcomes in children requiring CRRT. Sutherland et al. [11] and the ppCRRT showed that children who had >20 % FO at the time of CRRT initiation had increased mortality [odds ratio (OR) 8.5]. These findings have since been confirmed in multiple population sub-group analysis of the ppCRRT data and several single-center studies in a variety of populations [12, 16, 17, 20, 21, 41]. The association of FO at the initiation of RRT with adverse outcomes has subsequently been shown in a number of adult studies [36, 51, 52]. These studies consistently show that clinicians should consider initiating CRRT at a FO of >20 %, while a FO 10–20 % requires further evaluation.

Neonates

The impact of fluid balance on the development of chronic lung disease in premature neonates has been an active area of research over the past 20 years. Several studies demonstrate an association between fluid balance in the first days to weeks of life in this population and the development of bronchopulmonary dysplasia [42, 53]. In early studies the rates of AKI were not reported. In a more recent publication, however, Askenazi et al. reported on the relationship between AKI, fluid balance, and outcomes in a prospective cohort study of 122 premature neonates with a birthweight of ≤ 1200 g [43]. Neonates with AKI had a significantly higher risk of death or oxygen requirement at 28 days of life when controlling for confounding variables including measures of fluid balance. This study also demonstrated a consistent finding of an association of fluid balance and adverse outcomes [43]. In an earlier study of 58 sick near-term and term neonates (birthweight >2000 g.), Askenazi et al. examined the independent contribution of FO to outcomes [3]. AKI occurred in 15.6 % of the patients, and those with AKI had a higher median FO at day of life 3 than those without AKI (+8.2 vs. −4 %, p < 0.001) and lower survival (72 vs. 100 %, p < 0.02).

Cardiac surgery

Acute kidney injury occurs in 10–40 % of children following cardiac surgery and is commonly associated with adverse outcomes [5, 22, 54–56]. FO is an active area of interest in this population as there are a number of potential therapeutic interventions, including diuretics, intraoperative ultrafiltration, and early RRT. Hazle et al. reported in a cohort of infants following cardiac surgery that those with poor outcomes (RRT, prolonged LOS/length of ventilation) had significantly higher FO (24 ± 15 vs. 14 ± 8 %, p = 0.02) [22]. Seguin et al. reported a comprehensive evaluation of the epidemiology and timing of FO in a large cohort of 193 children following cardiac surgery [14]. FO peaked on day 2 at 7.4 ± 11.2 % and independently predicted LOS in hospital and length of mechanical ventilation [14]. Hassinger et al. reported similar findings in a cohort of 98 children following cardiac surgery [13]. In this study, early post-operative FO (FO > 5 %) was associated with longer LOS in the hospital (3.5 days), prolonged mechanical ventilation, and the development of AKI. In recognition of the impact of FO on outcomes there have been several studies evaluating the impact of early RRT on fluid balance in this patient population. Two studies have recently shown a potential role for peritoneal dialysis catheter placement in high-risk patients following cardiopulmonary bypass. In each of these studies intraoperative peritoneal dialysis catheter placement was associated with improved fluid balance in the postoperative period [57, 58]. A small study evaluating prophylactic peritoneal dialysis catheter did not confirm these findings in children following the Norwood procedure [59].

General critical care

In recent years there have been an increasing number of studies evaluating the impact FO and fluid administration in general pediatric critical care patient populations. The Fluid Expansion as Supportive Therapy (FEAST) Trial Group study of 3141 children with sepsis in Africa evaluated fluid resuscitation strategies, randomizing patients to receive one of three fluid resuscitation strategies (5 % albumin, normal saline, or no bolus) [60]. The authors concluded that fluid boluses significantly increased 48-h mortality in critically ill children with impaired perfusion. While a significant amount of debate has followed this publication, including the role of anemia and high percentage of patients with malaria, the trial has sparked significant discussion about current fluid resuscitation standards [61, 62]. Recently, Bhaskar et al. evaluated the role of early fluid accumulation in a matched case–control study of 114 children admitted with shock (cases were children with ≥ 10 % FO during the first 72 h of admission) [44]. In this study, both early FO [adjusted OR 9.17, 95 % confidence interval (CI) 2.22–55.57] and the severity of FO (adjusted OR 1.11, 95 % CI 1.05–1.19) independently predicted mortality. FO has also been shown to contribute to increased respiratory morbidity. In two evaluations of large cohorts of ventilated children with respiratory failure, FO has been shown to be associated with oxygenation index [2, 15]. These findings are consistent with a report from the Pediatric Acute Lung Injury and Sepsis Investigator’s (PALISI) Network which demonstrated that increasing fluid balance at day 3 of ICU admission was independently associated with fewer ventilator-free days in a population of 168 children with acute lung injury [63]. An interesting finding was that children enrolled in this study had comparable fluid balance on each day of admission to the liberal arm of the adult FACTT trial [63]. This study confirms previous reports of a positive fluid balance being associated with increased mortality and prolonged mechanical ventilation [64].

ECMO

Children with ECMO support represent a unique patient population that is particularly prone to the development of FO and AKI based on the severity of their illness and the inflammatory response that accompanies exposure to the extracorporeal circuit [65, 66]. Epidemiologic studies of several pediatric ECMO populations have demonstrated that the incidence of AKI can be as high as 72 % and that AKI is independently associated with adverse outcomes [8, 9, 67, 68]. In 2000, Swaniker et al. reported FO as a predictor of outcomes in an evaluation of 128 children on ECMO for respiratory failure [68].

The role of FO in children on ECMO and its unique contributions in this patient population warrants further discussion. Early studies investigating the development of FO show an elevation in the total body water in patients on ECMO and that an improved fluid balance is associated with improved lung function and time to ECMO weaning [47, 69]. These studies mirror recent adult data which demonstrate that a positive fluid balance at day 3 of ECMO independently predicted mortality in a cohort of 115 patients [70]. Taken together, these studies suggest that FO may be a target for therapies to improve outcomes in children on ECMO. Furthermore, neonates and children who receive concomitant CRRT while on ECMO support have been shown to have improved fluid management and fluid balance [46, 71]. There have been several recent studies evaluating the impact of CRRT in children on ECMO [21, 71, 72], but these have yielded conflicting results regarding the impact of CRRT on outcomes in varied patient populations. Selewski et al. recently evaluated the association of FO with outcomes at CRRT initiation in 53 children on ECMO treated with CRRT. The median FO at CRRT initiation was significantly lower in survivors than in to non-survivors (24.5 vs. 38 %, p = 0.006) [21]. Further study on the role of CRRT in fluid management in ECMO patients is needed before firm recommendations are made.

Fluid overload as a target for intervention

A brief description of treatment strategies for FO and AKI are described in Table 2. Fluid management remains a mainstay in the treatment of critically ill patients, with appropriate fluid resuscitation representing a key component to every algorithm. The concept of “early goal directed therapy” has garnered attention across medicine. This followed the publication by Rivers et al. which showed improved outcomes utilizing goal-directed therapy over the initial 6 h following presentation in a randomized controlled trial (RCT) in adults with sepsis [81]. While these data have not been replicated in other trials [82, 83], the concept of early reversal of shock with the thoughtful administration of fluids warrants further discussion. In pediatric patients, early fluid resuscitation remains the cornerstone of therapy in critically ill patients, and adequate fluid resuscitation has been associated with improved outcomes [84]. Ackan Arikan et al. recently evaluated the implementation of a resuscitation bundle for septic shock in the emergency department and found an improvement in the rates of AKI and outcomes relative to historical controls [75]. In this study, children treated following the institution of the resuscitation bundle received more fluid earlier in emergency room course relative to historical controls. In contrast, recent studies have begun to demonstrate a clear association with fluid accumulation/FO after ICU admission and adverse outcomes in children with respiratory failure [2, 63, 64]. In an evaluation of 168 children with respiratory failure, Valentine et al. showed that fluid accumulation during day 1–7 of ICU admission was greater in these children than in those in the conservative arm of the previously mentioned FACTT trial and similar to that in those in the liberal fluid management arm [63]. Taken together these data suggest an important role for early fluid administration in shock and the importance of careful monitoring of fluid status following ICU admission.

Table 2 Treatment of acute kidney injury and fluid overload
Full size table

The have been increasing efforts to develop effective management strategies for patients once FO has developed. While diuretics remain a mainstay of the management of FO in children with AKI, it is worth noting that diuretics have never been shown to prevent AKI or improve outcomes [26, 85]. The current Kidney Disease Improving Global Outcomes (KDIGO) Clinical Practice Guideline for Acute Kidney Injury makes the following statement regarding the use of diuretics, “We suggest not using diuretics to treat AKI, except in the management of volume overload [86].” In practice, a trial of a bolus dose of furosemide early in the course of AKI may be administered in an attempt to convert oliguric AKI. The response to an initial diuretic challenge should be used to guide further intervention, and prolonged administration of ineffective diuretic therapy should be avoided. The concept of the “furosemide stress test” embodies this idea. Chawla et al. have put this forth as a functional bedside test to help predict the likelihood of progression to severe AKI and consequent need for RRT [87]. In an evaluation of adults, individuals who were more likely to progress to AKI stage 3 had significantly less urine output 2 h following furosemide therapy. This concept warrants further study in children, but functionally can be interpreted to mean that children who respond poorly to diuretics should be watched more closely for the development of FO and need for more aggressive intervention.

Recently there have been medications that have shown promise in the treatment of AKI and FO in children. Two such medications include theophylline/aminophylline and fenoldopam. Theophylline and aminophylline act by inhibiting adenosine-induced vasoconstriction. In a RCT and subsequent follow-up study, theophylline was shown to prevent the development of AKI in asphyxiated newborns [73, 74]. The KDIGO guideline makes the following statement about the use of theophylline in neonates with perinatal asphyxia, “We suggest that a single dose of theophylline may be given in neonates with severe perinatal asphyxia, who are at high risk of AKI” [86]. In a small study of 31 children with heart disease who received aminophylline after pediatric nephrology consultation, aminophylline was associated with improved renal excretory function and urine output [77]. In a recent RCT of children following cardiopulmonary bypass, prophylactic aminophylline administration for 72 h post-operatively was not shown to prevent AKI [78]. Fenoldopam is a highly selective short-acting dopamine type 1 receptor agonist that causes dilatation of the renal vasculature with increased blood flow to the renal cortex and medulla [88]. Fenoldopam has demonstrated some modest benefit in a small, single-center study of infants undergoing cardiopulmonary bypass [79]. Further trials incorporating novel biomarkers are needed to evaluate the optimal timing and patient populations for these and other therapeutic interventions.

The timing of the initiation of renal replacement therapy

As described above and in Table 1 the pediatric literature has clearly demonstrated an association between the degree of FO at CRRT initiation and outcomes, including increased mortality. The current pediatric literature supports a FO level of 10–20 % as a threshold to consider intervention. While these data suggest that earlier initiation of CRRT would improve outcomes, including mortality, there is no definitive trial to date that demonstrates this. There has been one adult RCT (n = 106) that has compared the early initiation (within 12 h of oliguria or creatinine clearance of <20 ml/min) of RRT to the late (“classic” indications) of RRT; this study did not demonstrate an impact on mortality or long-term outcomes, but it was underpowered [89]. This topic remains a priority for research across nephrology [86]. The adult literature has begun to investigate the impact of the degree of FO and the timing of RRT on renal recovery and long-term dialysis dependence. In a study of 170 adults, Heung et al. showed that the degree of FO at RRT initiation predicted renal recovery and dialysis independence at 1 year [51]. Similarly, a secondary analysis of the “Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney)” study (n = 1238) evaluating early (<2 days), delayed (2-5 days), or late (>5 days) RRT demonstrated that late initiation of RRT was associated with increased mortality, increased LOS in hospital, increased length of RRT, and increased dialysis dependence [90]. Taken together, these data suggest that the earlier initiation of RRT, prior to the development of significant FO, may improve long-term outcomes. The role and timing of interventions in the prevention/treatment of FO remains a research priority at this time.

Conclusion

In summary, FO has a direct impact on the outcomes of children with AKI. FO represents a key vital sign that should be monitored daily in all critically ill patients, via careful attention to daily weights or cumulative input and output. Furthermore, the impact of FO diluting serum creatinine may delay or even mask the diagnosis of AKI and should be accounted for when evaluating renal function. FO has been shown to be independently associated with adverse outcomes in children with and without AKI in a number of critically ill pediatric populations, including CRRT, neonatal, cardiac surgery, general critical care, and ECMO patient populations. Taken together, the literature suggests that 10–20 % FO represents a critical threshold at which outcomes are adversely impacted in a number of populations and should prompt discussions about fluid management, diuretic management, and initiation of RRT in patients unable to maintain even fluid balance with medical management.