At birth the kidney is functionally immature; function slowly improves as renal blood flow and glomerular filtration increase during the neonatal period. In pre-term newborns, nephrogenesis is also not yet complete. Due to this functional renal immaturity, neonatal fluid dynamics are different from that of older patients, and quantifying fluid balance and detecting FO can be difficult. For example, total body water (TBW), which encompasses extracellular water (ECW) and intracellular water, is high in the fetus accounting for ~95% of body weight. Throughout gestation, the proportion of body weight represented by water decreases, but even newborns at term have TBW accounting for nearly 75% of their birth weight (BW) (5). After birth, isotonic contraction of the extracellular fluid compartment occurs with loss of ECW and accompanying weight loss. This physiologic diuresis is likely mediated via atrial natriuretic peptide. Regardless of gestational age (GA), newborns lose 10–15% of their BW in the first days of life and are then expected to regain their BW over the next 2 weeks (5, 6). Excessive fluid administration can confound this process and is associated with increased incidence of bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), and patent ductus arteriosus (PDA) (7–10). In addition to measurable losses, insensible losses via the skin or respiratory tract can be considerable, especially in pre-mature neonates (6, 11). While fluid-based therapies are necessary for a variety of neonatal conditions, appropriate fluid regimens are highly debated. Fluid requirements are based on GA with differing hydration needs and nutritional goals for growth. With the expected volume contraction followed by restoration and varying insensible losses, establishing the ideal weight for use in fluid calculations can be challenging and requires ongoing careful evaluation utilizing weight changes, intake and output measurements, and serum and/or urine biochemistries.

Fluid balance management strategies often include modifying environmental factors to minimize insensible losses rather than replacing estimated fluid losses given the risk of FO if estimates are incorrect. As described, excessive fluid administration can be harmful. Fluid restriction is another strategy to prevent complications associated with FO; however, presently available studies are inconclusive. A Cochrane review evaluating fluid restriction demonstrated a decreased risk of PDA and NEC in pre-term infants, but the five randomized controlled trials (RCTs) included are outdated and likely do not reflect current practices (12). In a more recent RCT, fluid restriction reduced the duration of respiratory support for severe transient tachypnea of the newborn (13). Conversely, Nicholson et al. found no difference in outcomes with post-operative fluid restriction in a cohort of neonates after cardiac surgery, a finding that suggests fluid restriction may not be helpful in all populations (14). Moreover, fluid restriction is not without consequences, risking adverse effects including dehydration, hypotension, and decreased end-organ perfusion. Optimal fluid therapy thus should allow for adequate postnatal diuresis with adjustment for increasing post-menstrual age as the kidneys mature and the degree of insensible losses diminish.

There is no consensus on how best to define FO (4), especially in neonates. FO is typically assessed by one of two methods: (1) weight-based methods, which quantify percent change in weight from baseline, or (2) cumulative fluid balance methods, which utilize daily fluid intake and output measurements from time of intensive care unit (ICU) admission (or other start point). Selewski et al. confirmed both methods correlate well in a cohort of pediatric patients receiving KRT (15). However, both approaches have disadvantages in neonates. First, the degree of postnatal diuresis and weight loss varies based on GA and confounds weight-based methods. Second, accurate recording of fluid intake and output is challenging. Van Asperen et al. found that fluid balance charted in the medical record correlated poorly with daily weight changes, and therefore providers may be unable to rely on recorded fluid balances as a sole measure in assessing FO in neonates (16). Third, diaper outputs are often recorded as “mixed” (consisting of both urine and stool), leading to uncertainty about how much of recorded volumes represent urine.

Further complicating fluid management is the co-occurrence of AKI with FO. Altered renal function secondary to AKI hinders diuresis after significant fluid accumulation, pre-disposing to the development of FO. Just as other biomarkers have been studied to better evaluate renal function (17), FO can be a marker of AKI (18). FO can also dilute Scr, hindering the ability to detect AKI. Once the existence of both disease processes is determined, it is difficult to differentiate and isolate the effects of AKI from FO; both can have profound, independent impacts on response to fluids and kidney function (19).

When FO is detected, it is unclear at what threshold treatment should be initiated in neonates. Specific FO thresholds ranging from 10 to 20% have been identified in older pediatric populations and adults as (1) requiring interventions and (2) associated with adverse outcomes (2, 36, 37). Unfortunately, similar thresholds have not yet been identified for neonates; depending on the clinical context, differing thresholds may exist.

Treatment options for FO include diuretics, peritoneal dialysis (PD), and continuous KRT (CKRT). Diuretics are frequenty utilized in the ICU and are the mainstay of therapy for the prevention or treatment of pulmonary edema or congestive heart failure in infants with congenital heart defects (38). However, more research is needed to guide optimal diuretic dosing, timing, and type (loop vs. thiazide) in order to achieve desired outcomes without causing AKI or other organ injury. Belik et al. found diuretics improved pulmonary function in ventilated patients (39), though RCTs have failed to demonstrate improvement in outcomes in pre-term infants with respiratory distress syndrome, a precursor of BPD (40). Similarly, diuretics do not prevent the development or worsening of AKI in patients with oliguria, and the optimal use of diuretics to treat oliguria and FO in patients with or at risk for AKI is not yet clearly established (41).

Dialytic modalities including PD and CKRT are also options for fluid removal, but are not used as frequently for this indication in neonates, likely because of the lack of equipment designed specifically for small patients as well as lack of high quality data supporting optimal use. In the Assessment of Worldwide Acute Kidney Epidemiology in Neonates (AWAKEN) study, some form of dialysis was used in only 4.1% of those with AKI (42). Unlike diuretics, dialytic modalities offer the benefit of managing electrolyte imbalances while allowing for adequate nutrition provision that otherwise may be restricted in patients with oliguria and AKI. PD is often the preferred modality in neonates because of the avoidance of large fluid shifts and the need for large vascular catheters. A recent systematic review and meta-analysis by Flores et al. highlighted the challenges associated with using available data to guide clinical decision making around the use of PD (43). Their meta-analysis demonstrated an increased risk of mortality in patients who received PD post-operatively compared with those who were supported with diuretics, but a larger proportion of infants in this group came from centers that implemented PD following failed diuretic response and thus may represent a group at higher risk for poor outcomes. Outcomes by which efficacy of PD was assessed varied across studies, making cross-study comparisions difficult.

CKRT is another dialytic modality used to support neonates with FO and/or AKI. As mentioned previously, currently available machines have, until recently, been designed for adults and adapted for use in neonates. Studies assessing the outcomes of patients supported with CKRT consistently show higher mortality rates in patients <10 kg than in older and larger patients (44). The combination of technical challenges and published rates of high mortality likely contribute to hesitancy around the use of this therapy in neonates. However, newer devices with lower extracorporeal volumes are now becoming available. Menon et al. published multi-center data on the adapted use of the Aquadex FlexFlow system (CHS solutions Inc., Eden Prairie, MN) (45). In their study, FO was the indication for this therapy in 46% of their sample and FO with AKI in another 15%. Even with this smaller circuit, survival rates were lower in patients <10 kg than in the other patient groups, with 60% of patients <10 kg surviving to treatment discontinuation compared with 97–100% in older/larger patients (overall survival: 32% in patients <10 kg vs. 68–85% in older/larger patients). The Cardio Renal Pediatric Dialysis Emergency Machine (Carpediem^™) is the only machine specifically designed for neonates and recently received FDA approval for treatment of AKI and FO in patients 2.5–10 kg. Published survival outcomes for neonates with AKI and FO are higher using this machine, with 97% of the sample surviving to treatment discontinuation, and 50% overall survival of this group (46). These newer devices have significant potential to expand our therapeutic options and improve our ability to manage FO (and AKI) in neonates.

Persistence of the ductus arteriosus (i.e., PDA), the most common cardiac problem among pre-term infants, is associated with excessive fluid intake (8, 20, 21, 47). Fluid intakes on days 2 and 3 of life are independently associated with increased risk of PDA indicating that early fluid administration can be problematic even before FO develops (21). In this same study, the odds of PDA increased 22% for every 10 mL/kg of fluid received on day 3, and those who received total fluid intakes >170 mL/kg/day were 4.5 times more likely to have a PDA.

Cardiac surgery is a risk factor for FO as well as AKI. Both complications often occur with cardiopulmonary bypass (CPB) support and are associated with significant morbidity and mortality (22–25). Peri-operative fluid management is complicated by the need for blood products and diuretics, impaired renal function, and low cardiac output necessitating volume resuscitation balanced with inotropic/vasopressor support. In a retrospective cohort of neonates undergoing CPB, FO was an independent risk factor for the composite outcome of death and need for CKRT or ECLS (24). Notably, those with poor outcomes were more likely to be <3 days old at the time of the operation signifying acuity of illness but also possibly reflecting inadequate diuresis pre-operatively. The authors determined that >16% FO on post-operative day (POD) 3 held the highest predictive value for poor outcomes, suggesting POD 3 and a FO >16% could represent important therapeutic thresholds. FO can also impact recovery by prolonging the duration of mechanical ventilation, time to sternal closure, and overall length of stay in neonates after cardiac surgery (23, 24). Mah et al. have also demonstrated the independent association of FO on length of stay as well as mortality (25).

Most research evaluating the relationship between fluid balance and the lung in neonates has focused on ELBW infants and BPD-related outcomes. Multiple factors are implicated in the pathogenesis of BPD including barotrauma, oxygen toxicity, and PDA; positive fluid balance and subsequent pulmonary edema are proposed to contribute as well (20, 26). Researchers hypothesize excessive fluid administration leads to increased pulmonary blood flow (especially if PDA is present), and this excess fluid subsequently shifts from the vessels into the pulmonary interstitium. Resultant pulmonary edema negatively affects lung compliance, requiring increased respiratory support and risking potential lung injury (27, 48).

Multiple studies have found increased risk of BPD in infants who received higher total fluid intakes and less postnatal weight loss through the first 10 days of life (9, 10, 27, 28). More recent investigations explored the link between neonatal fluid balance and mechanical ventilation and found FO in the first 72 h of life was associated with higher ventilator settings and longer duration of mechanical ventilation (29). In a secondary analysis of the AWAKEN cohort (42), Selewski et al. demonstrated multiple measurements of positive fluid balance as risk factors for the need for mechanical ventilation at the end of the first week of life (30, 31); every 1% increase in peak fluid balance led to a 12–14% increased risk of requiring mechanical ventilation on postnatal day 7, suggesting even incremental fluid changes can adversely affect lung function.

Neonates requiring ECLS are the most critically ill patients in the neonatal ICU. Severe respiratory pathology is a common neonatal indication for ECLS, and FO with pulmonary edema can be detrimental to the recovery of lung function (49). FO is common in pediatric patients requiring ECLS with the majority developing >30% FO in one multi-center observational study (32). Although this study included all pediatric patients, the median patient age was 10 days. Positive fluid balance alone during ECLS was an independent risk factor for mortality, and neonates had significantly higher peak FO (35.3 vs. 26.3%; p < 0.001) than pediatric patients.

CKRT is often used in patients receiving ECLS to prevent or treat FO. In another retrospective study examining pediatric ECLS patients comprised of 62% neonates, Selewski et al. demonstrated increased risk of mortality with a higher degree of FO at CKRT initiation (33). Although CKRT may be able to improve fluid balance, the authors also showed that once FO was established, fluid removal had no impact on survival (33). This associated risk of mortality was later reiterated in an exclusively neonatal observational study using a FO threshold of 30% (34). Gorga et al. conducted a multi-center cohort study in which neonates represented 52% of the sample and also found an independent association between the degree of FO at CKRT initiation and mortality with graded increases in both ECLS and hospital mortality for every 10% increase in FO (35). However, the median change in FO from CKRT initiation to discontinuation did not seem to impact either ECLS or hospital mortality. The mortality risk with positive fluid balance on ECLS and the lack of survival benefit with fluid removal suggests earlier recognition and intervention with CKRT may be indicated prior to the development of FO. However, the threshold at which this should occur remains unknown.