Introduction

In recent years, the deleterious impact of fluid overload in critically ill patients across the age spectrum has become clear [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]. Evidence first described in children has led the field in highlighting the adverse outcomes associated with excessive fluid accumulation in sick children. The association between greater positive fluid balance at the time of initiation of continuous kidney replacement therapy (CKRT) and increased mortality was first described in 2001 [8]. This association has been demonstrated repeatedly across a spectrum of neonatal and pediatric populations, including those who are critically ill, receiving mechanical ventilation, on extracorporeal life support, and undergoing congenital heart surgery. Thus, the importance of documenting and monitoring fluid balance and for complications attributed to fluid accumulation is vital to avoiding adverse events and improving outcomes in sick children.

Careful examination of the concept of “fluid overload” in children reveals several gaps in our understanding of the impact of fluid accumulation. In early studies, the term fluid overload was used to simply describe positive fluid accumulation in children at the time of CKRT initiation [7,8,9, 12, 17, 20, 21]. In these studies, fluid accumulation likely represented a pathologic state was equated with fluid overload, and children by extension were perceived to benefit from intervention with CKRT. Since those initial studies, the terminology in the literature describing fluid accumulation has largely used the term fluid overload to equate with a state of positive fluid balance. However, use of the term fluid overload in this way introduces bias by the inherent assumption that all fluid accumulation is detrimental, and by extension fluid removal or “negative fluid overload” is good. Furthermore, the methodology utilized to calculate fluid balance in these studies varied [8, 12,13,14, 17, 20]. To advance our understanding and generate new knowledge, a standardization of terminology and methods for reporting fluid balance is necessary.

Fluid accumulation represents an attractive target for intervention in sick children. Several strategies to prevent or mitigate a state of fluid overload have been proposed [22]. Prior to widespread implementation of such strategies, however, it is vital to appraise the source and strength of existing evidence to support the association between positive fluid balance and outcomes in sick children. The current literature has significant limitations, including being small, single center, and observational (often retrospective) in design, along with a lack of standardized reporting on measures of fluid balance (definition, timing, epidemiology) across studies [23,24,25]. Rigorous observational studies and randomized trials evaluating prospective interventions and strategies to manage fluid balance and accumulation in sick children are needed.

Acknowledging the importance of fluid accumulation and disorders of fluid balance, the meeting chairs of the first ADQI dedicated to neonates and pediatric patients (the 26th ADQI) convened a diverse expert multidisciplinary working group dedicated to fluid balance, fluid accumulation, and fluid overload. We herein summarize the consensus to describe in detail: definitions and epidemiology of disorders of fluid balance, targets for intervention, and endpoints for clinical trials.

Methods

The methodology utilized for ADQI meetings has been developed iteratively over the last two decades [26]. The aim of ADQI meetings is to provide expert-based statements, supported by evidence where applicable, and interpretation of current knowledge for use in clinical care and research endeavors. In addition, ADQI aims to identify evidence and knowledge-to-care gaps to establish future research priorities. The 26th ADQI consensus meeting included physicians and scientists (adult and pediatric nephrology and critical care, pediatric cardiology), nurse practitioners, nurse educators, clinical pharmacists, critical care dieticians, health services researchers, and patient and family advocates for a 3-day meeting held in Napa Valley, CA on November 11–14, 2021.

The preparation began over a year prior to the in-person meeting with a detailed literature review (2001–2021), topic elicitation, question development, and proposed statements. The workgroup identified several themes around “fluid balance and fluid overload” to generate concrete questions for the in-person meeting which were iteratively refined though consensus during sequential plenary sessions.

Question 1: What defines fluid balance in sick children?

Statement: Fluid balance is the difference between total input and output that can be expressed as “daily” and “cumulative” over a defined duration of time (Table 1)

Table 1 Definitions of fluid balance [27]
Full size table

Many terms have been used interchangeably to describe the impact of fluid on outcomes in sick children, including but not limited to “fluid balance,” “fluid accumulation,” and “fluid overload.” Identifying consensus terminology is critical to advance our understanding of the impact of fluid management on outcomes across populations. Such definitions will guide the development of clinical practice guidelines, epidemiological surveillance, clinical trials, and ultimately improved care for sick children and neonates.

The concept of fluid balance is at the heart of describing fluid accumulation in any population. Fluid balance is a measure of the intake and output for a given duration. This is classically expressed over a period of time as daily fluid balance (24-h) or cumulative fluid balance (since admission, previous number of days, etc.). The cumulative fluid balance is often expressed as a percent corrected for a pre-specified anchor weight. We propose utilizing the terms daily fluid balance, cumulative fluid balance, and percent cumulative fluid balance to describe the fluid status objectively in sick children for the purpose of clinical care and research (Table 1).

The two methodologies commonly utilized in the literature to describe fluid balance include cumulative fluid input and output and weight-based methodology (Table 1). The “cumulative fluid input and output methodology” has been adopted since the aforementioned 2001 study that first described the impact of “fluid overload” on outcomes in critically ill children initiating CKRT [8, 11, 15, 20, 27, 30]. This is the most frequently used methodology of quantifying fluid accumulation in the literature.

With this methodology, practitioners must be able to accurately measure all inputs (intravenous fluids, blood products, enteral feeds, intravenous flushes, etc.) and all outputs (urine, drains, dressings, stool, etc.) over a given time period. In the absence of a Foley catheter, urine output is often calculated utilizing change in diaper weight or measured voids which introduces potential limitations. Inaccuracies or missing input/output measurements are carried forward in all subsequent fluid balance calculations, propagating any potential inaccuracy. Enteral and intravenous fluid (IVF) contributions to input are generally treated as equivalent. Furthermore, this methodology does not account for insensible losses or potential insensible gains (humidified ventilatory circuits).

The second methodology to describe fluid balance utilizes a weight-based approach [4, 14, 16,17,18] as a change in patient weight from an anchor weight. The ability to reliably measure weight in a consistent and safe matter is crucial for application of this approach, which frequently requires set protocols to account for potential barriers such as mechanical ventilation, extracorporeal membrane oxygenation (ECMO), and use of bed weights. In this context, a reliable weight does not necessarily reflect the patient’s “true weight,” but instead a weight that is measured using a consistent method from day to day. This approach removes the inherent inaccuracies of accounting for daily input and output and should theoretically capture insensible and other losses. This would also enable compensation for missed or inaccurate daily measures of fluid input and output in the calculation of cumulative fluid balance.

For either method, a weight is needed in the denominator to understand the percent change in fluid balance by serving as a baseline for the equation or the anchor weight. Anchor weights used to describe percent change in fluid balance in the literature include intensive care unit (ICU) admission weight, hospital admission weight, estimated dry weight, pre-operative weight, and birthweight in neonates [4, 7,8,9, 12, 14, 17, 28, 31]. The ICU admission weight is the weight most commonly utilized as the anchor weight for fluid balance calculations in the literature. This practice likely reflects the fact that the ICU admission weight often is the first recorded weight available to clinicians. Less commonly reported in the literature is the use of hospital admission weight, which may reflect the weight in the emergency department, ICU admission weight, or ward admission weight. This may risk introducing bias due to the lack of standardization of weighing practices across different areas. The admission weight represents patients in a variety of fluid balance states ranging from hypo- to hypervolemic. Finally, some reports have used the physician estimate of “dry weight” as the baseline for weight correction. However, clinicians do not reliably predict dry weight, recognize fluid overload clinically, and the dry weight implies a healthy state with muscle mass that may significantly change over the course of a critical illness [31]. Therefore, no clear gold standard exists against which to systematically compare different approaches, highlighting a knowledge gap requiring targeted study.

A vital point for defining fluid balance is the concept of euvolemia, commonly referred to clinically as an estimated healthy dry weight. In considering fluid balance, we must acknowledge some of the inherent assumptions commonly utilized in the literature and their weaknesses. In particular, most fluid balance calculations assume even fluid balance at ICU admission with the assumption that the anchor weight represents a euvolemic state. This assumption is not supported by evidence and is subject to many potential inaccuracies (over/inadequate fluid resuscitation prior to ICU admission, fluid management on cardiopulmonary bypass, intravenous fluids administered on the general wards, etc.).

In describing the terminology and methodology, the practicality and impact of these definitions in low- and middle-income countries (LMICs) necessitate a separate discussion. In evaluating the potential resources necessary to accurately monitor fluid balance, the “cumulative fluid input and output methodology” may be more resource intensive as it requires a detailed 24-h monitoring. The weight-based methodology has the advantage of being less resource intensive. In LMIC settings, it remains important to use a standardized method to define anchor weight. Weighing neonates and small children is a well-established method in neonatal and pediatric intensive care units (PICU) in LMICs. However, weighing older children and adolescents receiving mechanical ventilation is a challenge as ICU beds capable of weighing patients are less commonly available.

Critically ill neonates warrant separate discussion, as measuring fluid balance and fluid accumulation can be particularly challenging in this population. Special considerations for neonates include increased insensible losses, insensible losses that change with gestational age, and the expected physiologic diuresis over the first 1–2 postnatal weeks. The ascertainment of fluid balance in neonates is further complicated by the fact that sick neonates rarely have Foley catheters in place. Weight-based methods have been clearly shown to be a superior measure of fluid balance in neonates [14, 18, 32,33,34]. For the neonatal population, the most common anchor weight is the birthweight in the first two postnatal weeks [32]. After this period, there has been little research describing the accurate calculation of an anchor weight.

As the importance of fluid balance has become increasingly clear, an important step is to incorporate fluid balance as a “vital sign” in sick children. As with any other vital sign, fluid balance can be incorporated into clinical rounds and monitoring on a daily basis in sick children. Furthermore, there are high-risk populations where fluid balance may need to be evaluated on a more frequent basis (e.g., CKRT, ECMO). The availability of such data serves as a quality indicator, and quality improvement methodology can be utilized to optimize this [35, 36].

Areas for research

  • Identify the optimal anchor weight needed to calculate fluid balance measurements to be used through a patient’s clinical course (e.g., premorbid weight, dry weight, admission weight, PICU admission weight).

  • Systematically study, develop, implement, and evaluate protocols for daily weights in sick children and in different resourced settings (e.g., LMICs).

  • Evaluate the optimal method and process to adjust anchor weight for patients hospitalized for extended periods of time.

  • Evaluate the optimal method and process to adjust anchor weight for neonates outside of the first 2 postnatal weeks.

Question 2: What defines fluid overload in sick children?

Statement: Fluid overload denotes a pathologic state of positive fluid balance associated with a clinically observable event(s), which may vary by age, case-mix, acuity, and phase of illness. No specific threshold of positive fluid balance alone can define fluid overload across all sick children

In order to advance the field, the term percent cumulative fluid balance should be utilized to describe the cumulative positive fluid balance over a given time period. The term fluid overload should be defined as the degree of positive fluid balance that is associated with adverse patient-centered events such as increased length of mechanical ventilation, length of stay, and/or increased mortality. The threshold and timing of fluid balance that define the pathologic state of fluid overload differ across populations, and disease processes are variable (Table 2 and Supplemental Table S1) [2, 3, 8, 10, 11, 14,15,16, 18, 20, 23, 27,28,29, 37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89]. Observational studies have identified important characteristics that impact the thresholds associated with fluid overload, including age, underlying disease process, illness acuity, temporal profile in fluid balance, and available resources. In considering the utilization of thresholds to report fluid balance relative to reporting fluid balance as a continuous variable, one must understand the potential pitfalls to this practice. The utilization of thresholds is subject to potential bias (e.g., influence of extreme values), and this must be appreciated.

Table 2 Selected studies evaluating the impact of fluid balance on outcomes
Full size table

With the paradigm shift on how to ideally define fluid balance, there must be an associated shift in how we describe its epidemiology. In defining fluid overload as positive fluid balance associated with adverse clinical events, it is implicit that not all states of positive fluid balance are in fact deleterious. Furthermore, the full spectrum of fluid balance should be considered, including negative fluid balance. It is reasonable to think that negative fluid balance is not always beneficial for the patient, and that in certain disease states, may become detrimental to the patient. This paradigm and its associations with various disease states and clinical factors are illustrated in Fig. 1.

Fig. 1
figure 1

The Spectrum of Fluid Balance [90]. There is a spectrum of fluid balance for the sick child that is a U-shaped curve. (A) This spectrum can swing between positive fluid balance (+FB) and negative fluid balance (-FB) with varying levels of clinical impact depending on the severity of the abnormal fluid balance. (B) Depending on the case-mix and/or resources available, the U-shaped curve may have more or less tolerance for a positive or negative fluid balance. These depict examples of different scenarios where a patient population may have greater or lesser tolerance of a +FB or -FB scenario. (C) For any given sick child, there are several factors (i.e., host factors, interventions, adverse outcomes) that may push them towards a greater +FB or -FB or pull them back towards a state of neutral fluid balance at the center. These may vary over the course of the hospitalization and can be constantly changing

Full size image

A major gap and limitation in the epidemiology of fluid balance is understanding the full continuum of fluid balance, particularly iatrogenic and clinically important negative fluid balance or clinical dehydration. The incidence of positive fluid balance, including temporal profile and magnitude, varies by center, definition, reported thresholds, and predetermined thresholds set as “clinically meaningful fluid overload” (Supplemental Table S1). The standardization of definitions and reporting will improve the understanding of the full spectrum of fluid balance and allow us to better define and describe fluid overload. The literature to date shows wide ranges in the incidence of fluid overload among critically ill children depending on the threshold used: 17–70% for > 5%, 4.5–56% for > 10%, 9–33% for > 20% (Supplemental Table S1).

In recent years, pediatric fluid research shifted from identifying absolute thresholds of fluid overload to understanding the impact of timing and trajectory of fluid accumulation on outcomes. In a single center, retrospective analysis of post-cardiac surgery patients, the median peak positive fluid balance was 4.5% and most commonly occurred on postoperative day 2 [91]. Among a cohort of mechanically ventilated children across five PICUs, the mean daily cumulative positive fluid balance was positive 20–35 ml/kg on hospital days 1–3 [92]. An evaluation of over 1000 critically ill children in Canada found that the positive fluid balance increased with each subsequent ICU day from 1.6% on hospital day 1 to 16.4% on hospital day 10 [2]. This data taken together emphasizes the importance of considering timing, magnitude, and trajectory in evaluating the impact of fluid balance.

A significant amount of work has gone into understanding the case mix and factors associated with derangements in fluid balance and the development of fluid overload (Supplemental Table S1). In post-cardiac surgery populations, young age, severity of AKI, longer cardiopulmonary bypass time, and surgical complexity were associated with derangements in fluid balance [11, 13, 37, 93]. In general, PICU populations, younger age, AKI, inotrope use, ventilatory support, shock diagnosis, and admission to a medical/surgical unit vs. a cardiac unit were associated with higher fluid balance categories [2, 3, 27, 94, 95]. Given the associations between positive fluid balance and adverse outcomes, including mortality, further investigation for modifiable factors as targets to enrich clinical trials and be actionable for intervention are needed.

Another important subset of patients with little data is in LMICs, like South Africa, that has reported lower proportions of critically ill children with fluid overload (3% with > 10%). The same cohort also had similar median peak positive fluid balance of 3.5% for their study period as compared to high-income settings [96]. Further work is needed to better understand whether different thresholds are associated with adverse events attributed to a positive fluid balance in often less resources settings encountered in LMICs.

Triggers and interventions

Two types of interventions can be implemented to modify fluid accumulation: limiting fluid administration and facilitating active fluid removal through pharmacologic (i.e., diuretics) and/or mechanical (i.e., kidney replacement therapy (KRT)) approaches. In sick children, administered fluids have varying degrees of necessity and include those that are life sustaining (i.e., resuscitation, blood products, medication, nutrition) and those that may be restricted (i.e., intravenous fluids, carriers for medications). A large proportion of fluid administered to critically ill children comes in forms that may be restricted, particularly maintenance fluids [97, 98]. We suggest clinicians and researchers study fluid balance in an overlapping but sequential manner: limitation, targeted diuretic therapy, and then consideration for the use of mechanical fluid removal (i.e., KRT). Figure 2 demonstrates a theoretical framework of the potential overlapping interventions based on the degree of fluid accumulation and its clinical impact.

Fig. 2
figure 2

Overlapping Interventions to Manage Fluid Balance. Interventions to manage fluid balance are overlapping based on the degree of fluid accumulation and clinical impact of the fluid. These interventions include limitation of continuous fluids, targeted diuretic therapy, and then consideration for the use of kidney support therapy. Green triangle represents limitations to continuous fluids and optimizing the osmotic gradient. Red triangle represents the need for renal replacement therapy. Purple oval outlines the role of diuretics while patients during both limitation of fluids and need for renal replacement therapy

Full size image

Another important clinical consideration is to begin to consider preventative measures or fluid stewardship programs aimed at mitigating fluid accumulation in high risk populations prior to the development of overt fluid overload [3, 97, 99]. This could be done by identifying patients at high risk for developing fluid overload and/or kidney dysfunction (i.e., predictive enrichment and clinical risk prediction) that would further magnify the accumulation of IVF and by taking a targeted approach to more aggressively limit a positive fluid balance. This may include the deployment of several tools (e.g., renal angina index, novel biomarkers, furosemide stress test, fluid overload, and kidney injury score (FOKIS)) in high-risk populations [99,100,101,102,103]. These tools can be utilized to identify high-risk patients who would most benefit from a precision approach to fluid management with the targeted use of diuretics and kidney support therapy. A clinical trial currently underway that is utilizing this approach is the “Trial in AKI using NGAL and Fluid Overload to optimize CRRT Use” (TAKING FOCUS 2). This trial is evaluating the impact of sequentially mobilizing urinary biomarkers and the furosemide stress test to identify patients at risk for the development of fluid overload and severe AKI [104, 105].

Areas for future research

  • Further understanding of the incidence of both positive and negative fluid balance in all sick children (acute, critically ill, neonates, post-cardiac surgery, oncologic patients, etc.).

  • Further delineate the thresholds of positive fluid balance whereby the development of fluid overload is most likely to occur across all sick children.

  • Understand the impact of timing and trajectory of positive fluid balance on outcomes.

  • Identify the risk factors and potential modifiable factors for positive fluid balance and fluid overload in all sick children.

  • Further study and understanding of the interplay and impact of intravascular volume status and total body volume. Further work is needed to define and standardize the terminology including but not limited to hypervolemia, euvolemia, and hypovolemia relative to total body volume and intravascular volume.

  • Develop objective and reproducible measures of edema.

  • Identify the potential associations of negative fluid balance and clinical dehydration on clinical outcomes (such as venous thromboses).

  • Identify the incidence, risk factors, and associated outcomes of negative and positive fluid balance in children in LMICs. This is urgently needed to better understand and put into clinical context the findings of the FEAST trial.

  • Identify thresholds for intervention in various case mixes of sick children.

  • Evaluate the effectiveness of different interventions on the degree of positive fluid balance and the impact on fluid overload and associated outcomes.

  • Evaluate the effectiveness of a preventative strategy or fluid stewardship interventions. (i.e., reduce fluid administration, improve cumulative fluid balance, improve physiological outcomes, and improve patient reported outcomes).

Question 3: What are the challenges to translating observational data to clinical management of fluid balance?

Statement: The literature describing the association of fluid balance on outcomes is based on observational studies. Evidence from observational data alone cannot establish a causal relationship between fluid balance and outcomes

In order to move the field forward and improve outcomes related to fluid balance, a better understanding and establishment of the causal link between fluid balance and outcomes are needed. The lack of randomized trials in hospitalized children and complexity of evaluating fluid balance in critically ill patients make it challenging to causally link fluid balance to outcomes. In order to have the highest likelihood of success, these trials would benefit from utilizing many of the previously described tools to identify the highest risk patient populations that would potentially benefit. We provide further suggestions on optimizing study design and outcomes to inform the development of trials focused on the impact of fluid balance on outcomes.

Study designs

Table 3 outlines some of the ongoing and seminal studies that have been done to date in adults and children evaluating the impact of fluid balance on clinical outcomes. To date, most data surrounding the impact of fluid balance on outcomes are derived from observational studies. Observational data are important tools that allow us to study new concepts and exposures that are difficult to randomize. The confounders that have an association with both the exposure (i.e., fluid balance) and outcomes must be realized and controlled for. A useful tool for researchers is a directed acyclic graph (DAG). A DAG is a diagrammatic tool to allow researchers to visualize confounders, mediators, and intermediaries on pathways, as well as other relevant exposures. A DAG developed from observational studies describing the interactions of fluid overload and morbidity and mortality is presented in Fig. 3. A detailed understanding of the complex relationship between patient characteristics, fluid balance, and outcomes from observational studies will serve to inform clinical trial design.

Table 3 Planned, ongoing, or recent randomized clinical trials in adult/pediatric populations focused on fluid accumulation in critical illness (with or without AKI)
Full size table
Fig. 3
figure 3

Directed Acyclic Graph of Fluid Overload and Mortality [90]. The directed acyclic graph (DAG) above is an example of the utilization of this tool to describe the complex relationship between fluid overload and mortality in sick children. Different DAGs would be necessary to evaluate positive or negative fluid balance or other fluid states and different outcomes. The variables depicted in this figure are likely not exhaustive and additional nuanced variables could be considered (e.g., genetics). The figure is meant to be as comprehensive as possible with current knowledge-to-date to demonstrate the complexity of the relationship and the importance that all factors are considered to minimize bias when trying to use observational studies to evaluate the fluid overload-mortality relationship and its potential for a causal pathway. Arrows depict a known associated risk relationship (whether that is a positive or negative relationship) E=the primary exposure of interest (fluid overload in this example) O=the primary outcome of interest (mortality in this example) White circles = represent variables that are confounders in the primary exposure-outcome relationship and would be the required minimal adjustment variables to account for in order to get a fully adjusted analysis of the primary exposure-outcome relationship (i.e., the minimal adjustment set) Pink circles = represent additional variables that are also confounders but not necessary for a minimal adjustment set evaluating the primary exposure-outcome relationship Orange circle = ancestor to primary exposure of interest

Full size image

In designing future observational studies, we suggest researchers to incorporate novel approaches to analysis to strengthen these studies. The first is to utilize appropriate sample sizes and to switch from logistic regression modelling (i.e., odds ratios) to linear or log-linear modelling (i.e., risk differences and risk ratios) when feasible. This analytic strategy will improve our understanding to look at risk ratios, risk differences, or attributable risks to determine the degree that a particular risk may be contributing to outcomes of interest.

Traditional randomized controlled trials, or explanatory trials, have been considered the gold standard to delineate causality in between an intervention and outcome. Randomized controlled trials can be challenging to design and implement, particularly for relationships as complex as outcomes associated with fluid balance. In recent years, pragmatic trials have been increasingly utilized to study approaches to management fluid, with fluid balance representing a key process and outcome measure. Pragmatic trials are characterized by simple eligibility, broad inclusion, simple data collection and use of core outcome sets, easily applied intervention, and group randomization (e.g., ward level, unit level, hospital level) [113]. Each of these characteristics aims to optimize the ease of the trial implementation, to evaluate interventions in “real world” scenarios, and to enable ease of translation of findings into practice. Several pragmatic and novel clinical trial design approaches have recently been applied (i.e., stepped-wedge implementation; cluster cross-over, quasi-experimental) in a range of clinical contexts that may be well-suited to evaluation of complex interventions for fluid management in sick children. Adaptive trial designs represent another important tool that may be utilized to study fluid balance in sick children. Simply put, such studies allow for modifications of study design that may be implemented at predefined interim analysis [114].

Given the complexity of fluid balance in sick children and our limited understanding of whether or not a direct causal pathway exists with adverse outcomes (i.e., mortality), we suggest innovative trial design be incorporated into future studies. A good example is the FLUID trial which proposes a hospital-wide (adults and children) cluster-randomized crossover trial to evaluate outcomes between 0.9% NaCl and Ringer’s lactate fluids (https://clinicaltrials.gov/ct2/show/NCT02721485, NCT02721485) [115].

Outcomes

The association of fluid overload and short-term outcomes during the initial hospitalization has been the most frequently reported study design, with mortality being the most common outcome. While mortality remains important, it is relatively uncommon, necessitating the need for alternative outcomes (Table 4). These outcomes include circumstances where fluid accumulation has led to an escalation in care due to the development of deteriorating clinical status or worsening organ function, including increased receipt of respiratory support, increased oxygenation index, increased ionotropic support, worsened cardiac dysfunction, subsequent or worsening AKI, receipt of KRT, or receipt of extracorporeal life support (ECLS). By extension, there are studies that report on outcomes related to the duration of such interventions (e.g., mechanical ventilation, ECLS, KRT) or hospitalization (ICU or total hospital length of stay) based on degrees of positive fluid balance. Additionally, ICU mortality and in-hospital mortality are commonly reported as outcomes of interest in observational studies. To date, there are few data describing the cost attributable to abnormalities in fluid balance in children.

Table 4 Summary of the spectrum of endpoints in trials focused on fluid accumulation/overload
Full size table

Recent work suggests that AKI and disorders of fluid accumulation may impact the long-term outcomes associated with other organs and multisystem disorders. Recent work by the Life After Pediatric Sepsis Evaluation (LAPSE) investigators has evaluated the association of severe AKI with the development of new comorbidities in children following critical care admission for sepsis. In this multicenter study, children with severe AKI had an increased odds of death or new substantive functional morbidity (adjusted odds ratio, 2.78; 95% CI, 1.63–4.81; p < 0.001) [122]. More recently, the functional outcomes of children treated with CKRT were assessed by the Functional Status Scale (FSS) in a single-center retrospective study of 45 children. In this study, 31 (69%) had worse FSS score at PICU discharge, and 51% of the children had new morbidity (defined by the FSS score). Furthermore, on adjusted analysis, the degree of fluid overload at CKRT initiation predicted a worse FSS score in this population [118]. In critically ill adults, the degree of positive fluid balance at ICU discharge predicted impaired mobility and discharge to another healthcare facility [123]. To date, there has not been a systematic evaluation of the impact of fluid balance on long-term outcomes in sick children. Future studies should include a multidisciplinary evaluation of outcomes including respiratory, cardiovascular, neurodevelopmental, health-related quality of life, and functional outcomes.

In recent years, the Major Adverse Kidney Events (MAKE) have been proposed to be included as a composite outcome to be utilized and reported in all effectiveness clinical trials in AKI. This composite outcome includes death, new kidney support therapy, or persistent serum creatinine greater than two times baseline. This outcome can be assessed at 30, 60, and 90 days. An important area for future research is to understand the applicability of MAKE outcomes to children and to revise it as necessary to fit pediatric and neonatal outcomes.

Areas for future research

  • Pragmatic randomized trials such as crossovers, cluster RCTs, stepped-wedge, or other quasi-experimental designs are likely needed to further understand the complex relationship of fluid balance and mortality, and whether or not a causal link exists.

  • Capitalize on existing networks such as NINJA, collaborative pediatric critical care research network to conduct multicenter randomized trials to answer the missing epidemiological questions surrounding fluid balance and mortality and other clinical outcomes.

  • Effects on outcomes of “isolated positive fluid balance” in patients with “normal” kidney function.

  • Prospective association of fluid overload and organ function (causally or temporally associated).

  • Verification if fluid overload is “continuously” associated with outcomes (i.e., FO 10% leads to an ICU stay of 7 days; 15%, 8 days; 20%, 10 days) or if there is a “dichotomic” threshold (i.e., FO above/below the clinically relevant threshold has significant outcomes implications).

  • Improved understanding of the etiology of fluid balance and the development of the state of fluid overload. This includes contributions from severity of illness, disease pathophysiology, timing, impaired fluid clearance (e.g., AKI), inadequate fluid stewardship, endothelial dysfunction, inflammation, and capillary leak.

Conclusion

The understanding of how fluid balance impacts the clinical care and outcomes of sick children is evolving. This includes a recognition that observational data alone does not determine causality. Fluid balance is an objective measure that ideally will be considered daily in patient care. Fluid overload is a pathologic state associated with positive fluid balance and clinically observable events. Armed with these important distinctions, further research will be able to better discern the relationship between fluid balance and fluid overload as well as help evaluate future clinical practices.