Table 4.

Summary of the studies evaluating outcomes associated with fluid overload in different pediatric populations.

Author; Year	Study design	Study sample [n; age group; inclusion criteria]	Substantive evidence (regarding association of)	Relevant findings
RENAL REPLACEMENT THERAPY
Goldstein et al. (12)	Retrospective observational Single center 1996–1998	21 children 8.8 ± 6.3 y PICU patients receiving CVVH±D	%FO at CVVH/D initiation with poor outcomes in critically ill children	- FO% was significantly lower in survivors (16.4 ± 13.8%) than nonsurvivors (34 ± 21%) after controlling for illness severity (p = 0.03)
Foland et al. (45)	Retrospective observational Single center 1997–2003	113 children 9.6 y (2.5–14.3) ICU children receiving CVVH	%FO prior to CVVH and mortality	- Median FO% was significantly lower in survivors vs. nonsurvivors [7.8 vs. 15.5%; p = 0.01] - FO% at CVVH initiation was independently associated with mortality in patients with ≥ 3 organ MODS (p = 0.01)
Gillespie et al. (44)	Retrospective observational Single center 1993–2002	77children ≤20y Patients on CVVH	%FO at the time of CVVH initiation with mortality	- High FO (≥10%) at CRRT initiation significantly increased the risk of mortality [HR 3.02, 95%CI 1.5–6.1; p = 0.002]
Goldstein et al. (43)	Prospective observational (ppCRRT) Multicenter	116 children <18 y MODS patients on CRRT	%FO prior to CRRT initiation with mortality in MODS patients	%FO at CRRT initiation was significantly lower for survivors than non-survivors even after adjusting for severity of illness Similar findings for patients receiving mechanical ventilation and vasoactive pressors, even after adjusting for illness severity (p < 0.05)
Hayes et al. (42)	Retrospective observational Single center 2000–2005	76 children 5.8 y (0–19) PICU patients with AKI needing CRRT	% FO at CRRT initiation with mortality	- Median %FO at CRRT initiation was 7.3% in survivors vs. 22.3% in nonsurvivors (p = 0.0001) - ≥20% FO at CRRT initiation was significantly associated with mortality (p = 0.006), LMV (p = 0.018), PICU stay (p = 0.0425), hospital LOS (p = 0.0123)
Sutherland et al. (41)	Prospective observational (ppCRRT) Multicenter 2001–2005	297 children <18 y ICU children on CRRT	%FO with mortality in children receiving CRRT	- FO of ≥20% was independently associated with increased mortality (aOR 8.5, 95% CI 2.8–25.7) - Presence of FO at CRRT initiation was associated with raised odds [aOR 1.03] for mortality which was significantly higher for worse degrees of FO [≥20, 10–20, <10% FO groups with 65, 43 and 29% mortality respectively]
Selewski et al. (49)	Retrospective observational Single center 2006–2010	113 children 19 m (0.2–181) PICU patients on CRRT	Different weight based FO definitions with PICU mortality	- FO% at CRRT initiation was significantly greater in non-survivors via both weight based and FB method - Univariate OR for PICU mortality was 1.056 [95%CI 1.025–1.087] by fluid balance method, 1.044 [95% CI 1.019–1.069] by the PICU admission weight-based method and 1.045 [95% CI 1.022–1.07] by hospital admission weight-based method - On multivariate analyses, all three methods significantly predicted PICU survival
de Galasso et al. (38)	Retrospective observational Single center 2000–2012	131 children 0–18 y PICU patients on CRRT	FO with mortality only in children with milder disease	- FO>10% at CRRT initiation was associated with mortality only in children with milder disease [OR 10.9,95 %CI 0.78–152.62; p = 0.07]
EXTRACORPOREAL MEMBRANE OXYGENATION
Hoover et al. (77)	Retrospective observational (Matched case control) 1992–2006 Single center	86 children 1 m−18 y Patients on ECMO vs. those on ECMO+CVVH	Improved fluid balance, caloric intake and less furosemide use with ECMO+CVVH cf ECMO alone	In ECMO survivors who received CVVH, median FB was less than that in non-CVVH survivors [25.1 vs. 40.2ml/kg/d; p = 0.028] - Use of CVVH was associated with earlier optimal caloric intake (p < 0.001) and reduced diuretic administration (p = 0.009) - Survival did not differ significantly in the two groups (p = 0.51)
Blijdorp et al. (78)	Retrospective observational (1:3 matched case- comparison) 2002–2006 Single center	61 neonates <28 d Patients on ECMO vs. those on ECMO+CVVH	Better fluid balance via HF in ECMO patients with improved outcomes	- Median time on ECMO [98 h,IQR 48–187 vs. 126 h,24–403; p = 0.02] and MV post decannulation [2.5d vs. 4.8d; p = 0.04] was significantly shorter in HF group
				- Cost per ECMO run and blood transfusion requirement was also reduced (p < 0.001) - No mortality benefit was noted in the HF group
Paden (52)	Retrospective observational 1997–2007 Single center	68 children <19 y Patients on ECMO+CRRT	Recovery of renal function and survival with HF during ECMO	- In the absence of primary renal disease, chronic renal failure did not occur following concurrent use of CRRT with ECMO - Mortality is higher in patients receiving concomitant CRRT and ECMO compared to those receiving ECMO, but is similar to patients requiring CRRT who are not on ECMO
Selewski et al. (40)	Retrospective cohort Multicenter 2007–2011	756 children <18 y Patients on ECMO	Survival associated with peak FO during ECMO and %FO at ECMO initiation	- A significantly higher hospital mortality risk was associated with each 10% rise of FO at initiation of ECMO (OR, 1.12; 95% CI, 1.04–1.19 p = 0.002). - With each 10% rise in peak FO during ECMO, there was a 9% increased odds of ECMO mortality and a 17% increased odds of hospital mortality
CARDIAC SURGERY
Grist et al. (55)	Retrospective observational Single center 2003–2009	1570 children Underwent congenital heart surgery with CPB and MUF	FO with mortality in children undergoing CPB	- 20% (314) children had a positive FB - Patients with positive FB weighed more, had higher RACHS score, longer pump time, longer LOS with an aOR for mortality of 1.73 (95% CI 1.01–2.96)[p < 0.05 for all variables]
Saini et al. (60)	Retrospective observational (1:1 matched case control) Single center 2006–2010	36 infants <1 y Post- atrioventricular septal defect repair ±PD insertion	Improved fluid balance with passive PD insertion post AVSD repair in infants	Infants with passive PD achieved negative fluid balance more rapidly (12 ± 10 vs. 27.3 ± 13 h, p < 0.0001) and to a greater extent (p = 0.002)
Hazle et al. (16)	Prospective observational Single center 2009–2010	49 infants <6 m Underwent congenital heart surgery	Postop FO with longer LOS, MV and mortality in infants undergoing cardiac surgery	- Higher mean max FO by both FB (12 ± 10 vs. 6 ± 4%, p = 0.03) and weight based (24 ± 15 vs. 14 ± 8%, p = 0.02) methods was associated with composite poor outcome - FO < 10% was associated with a good outcome by both FB and weight based methods (p = 0.02,0.01) but the association did not reach statistical significance after multivariate analysis
Sasser et al. (61)	Prospective before and after nonrandomized cohort Single center 2010–2011	52 neonates and infants Underwent congenital cardiac surgery with (25) or without (27) prophylactic PD use	Greater net negative FB with prophylactic PD placement post-CBP in infants	- Median net fluid balance was more negative in +PD at 24 and 48 h [−24 mL/kg (IQR-62,11) vs. +18 (−26, 11), p = 0.003; - 88 (−132,−54) vs.-46 (−84,–12), p = 0.004] - Duration of MV (p = 0.1), mean inotrope score (p = 0.04) and serum IL-6 and 8 levels were lower in +PD at 24h - Median time to sternal closure was lower in +PD [24 h (IQR 20–40) vs. 63 (44–72), p < 0.001]
Basu et al. (29)	Retrospective observational Single center 1997–2008	92 children 5.5 days (4–7.5) Status post arterial switch operation	Delayed AKI diagnosis with unadjusted sCr (not accounting for positive FB)	- Infants who developed AKI after surgery had higher POD1 FB [148 ± 125 vs. 115 ± 117ml/d, p = 0.016] - Correcting sCr for FO increased AKI prevalence and strengthened its association with postop morbidities
Seguin et al. (58)	Retrospective observational Single center 2005–2007	193 patients <18 y Post cardiac surgery	Early FO with LOS, MV and OI	- Early postop fluid administration was independently associated with higher D2 FO% (p = 0.0001) - D2 FO% predicted longer LOS (aHR 0.95, 95%CI 0.92–0.99, p = 0.009), longer MV (aHR 0.97, 95%CI 0.94–0.99, p = 0.03) - Higher daily FO% predicted worse daily OI in patients without cyanotic heart disease (aHR 0.16, 95%CI 0.07–0.25, p = 0.03)
Hassinger et al. (59)	Secondary analysis of prospective observational study Single center	98 children 2 wks−18 y Status post-CPB	Early postop FO with prolonged LOS, LMV, inotropic support and AKI development	- Early postop FO (≥5%) was independently associated with prolonged need for MV, LOS and inotropic support (p < 0.001) - FO was associated with post-CPB AKI; FO more often preceded than followed it but AKI was not consisitently associated with FO - Cumulative fluid administered was an excellent predictor of modified pRIFLE category [AUC = 0.96, 95%CI 0.92–1, −p = 0.002] - Patients with FO were administered higher fluid volume (p < 0.001) and had a poor urinary response to diuretics
Kwiatkowski et al. (62)	Retrospective observational (1:1 matched case control) 2007–2012	84 infants <6 m Underwent congenital heart surgery ± PDC insertion	Improved FB via elective PDC use with favorable outcomes in infants undergoing cardiac surgery	- PDC+ group had higher negative fluid balance on POD 1 and 2 (57 vs. 33%, p < 0.04; 85 vs. 61%, p < 0.01) - PDC+ group had shorter time to negative FB (16 vs. 32 h, p < 0.0001), earlier extubation (80 vs. 104h, p < 0.02), improved inotrope scores (p < 0.04), and fewer electrolyte imbalances requiring correction (p < 0.03)
Piggott et al. (57)	Retrospective observational Single center 2010–2013	95 neonates 6–29 d Underwent congenital cardiac surgery	Postop FO with prolonged MV, LOS and mortality	- AKI in 45% neonates - >15%FO was associated with prolonged LOS (p = 0.03), postop ventilator days (p < 0.001) and mortality (p < 0.001) - Certain risk factors like preop aminoglycoside use, selective cerebral perfusion, CPB time, small kidneys on US can be modified to minimize risk of AKI and perhaps FO - Prophylactic PD catheters can be placed in infants with small kidneys identified preoperatively to avert FO
Sampaio et al. (56)	Retrospective observational 2010–2013	85 children <17 y Surgery for congenital heart disease+ MV for at least 12h in PICU	FO with prolonged MV and LOS in patients post-congenital heart surgery	- Maximum CFB was associated with duration of MV (adjusted β coefficient = 0.53, CI 0.38–0.66, P < 0.001), LOS in PICU (Spearman's ρ = 0.45, P < 0.001), and presence of chest wall edema and pleural effusions on chest radiograph (p = 0.003)
Lex et al. (54)	Secondary analysis of a prospective observational study Single center 2004–2008	1,520 children <18 y Underwent open heart surgery	Early postop FO with higher mortality and morbidity	Higher FO on the day of surgery was independently associated with mortality (aOR, 1.14,95%CI 1.008–1.303; p = 0.041) and LCOS (1.21,95% CI 1.12–1.30, p = 0.001) - Higher maximum s.Cr (aOR 1.01,1.003–1.021; p = 0.009), maximum vasoactive-inotropic score (aOR 1.01,95% CI 0.005–1.029; p = 0.042) and higher blood loss on the day of the surgery (aOR1.01, 95%CI 1.004–1.025; p = 0.015) were associated with a higher risk of >5%FO
NEONATES
Ohv et al. (64)	Secondary analysis of the RCT by the Neonatal Research Network Multicenter 1999–2001	1,382 neonates ELBW newborns with birth wt between 401 and 1,000 g	Positive FB in the first 10 days of life with death/BPD	- 58% either died or developed BPD; 42% survived without BPD - Higher fluid intake (p < 0.001) or lower weight loss (p = 0.06) during first 10 d were significantly associated with death/BPD - Lower BW/GA/1.5 min Apgar scores, higher O2 requirement at 24 h of life and longer LMV were associated with death/BPD
Schmidt et al. (65)	Secondary analysis of TIPP (Randomized controlled trial of Indomethacin prophylaxis in preterms) Multicenter	999 neonates Extremely low birth weight newborns who survived to a postmenstrual age of 36 wks	- Positive FB in preemies with subsequent BPD - Uncertainty about cause-and-effect relationship between PDA and BPD	- Neonates without PDA who received prophylactic indomethacin had lower urinary output, lower weight loss (p = 0.012) and higher FiO2 requirement (p < 0.0001) by the end of first week - Incidence of BPD was similar in PDA+ neonates irrespective of indomethacin prophylaxis but was significantly higher in PDA- infants who received indomethacin (p = 0.015) - Indomethacin prophylaxis reduces the incidence of PDA but not that of BPD
Askenazi et al. (70)	Prospective observational Single center 2010–2011	58 neonates near term (≥34 wks and >2,000g) and term with Apgar score ≤7	AKI with FO and mortality in sick near term/term neonates	- AKI in 15.6% neonates - Median weight gain at D3 of life was higher in the AKI vs. non AKI cohort [8.2%, IQR 4.4–21.6% vs. −4%, IQR −6.5 to 0.0% (p < 0.001)] - Infants with AKI had lower survival rates than those without AKI [72 % vs. 100 % (p < 0.02)]
Askenazi et al. (67)	Prospective observational Single center 2012–2013	122 preterm neonates <31wks, <1,200 g	- FO with prolonged need for oxygen support, MV and mortality - AKI with BPD and mortality	- Infants with AKI (30%) had a higher max% wt change in the first 4 days of life (p = 0.05)and were at higher risk of death/MV at D28 (p < 0.03) - Although infants with FO had an increased RR to receive oxygen support/death (1.02, 95%CI 1.01–1.03; p < 0.0001) and MV/death (1.03,1.02–1.05; p < 0.0001) at D28, after adjustment this trend did not reach statistical significance (p = 0.16,0.06) - Similar finding was noted for time taken to oxygen weaning [HR 0.97 (0.9–0.99), p < 0.02; aHR 0.98 (0.9–1.01), p = 0.18]
Lee and Cho (69)	Retrospective observational Single center 200–2014	34 neonates (15 preterm, 19 term) Admitted to NICU On CRRT for ≥24 h	Higher %FO at CRRT initiation with mortality	- Neonates with ≥30% FO at the time of CRRT initiation had lower survival rates - Univariate Cox regression analysis revealed that a higher %FO at CRRT initiation and decreased urine output at the end of CRRT were associated with mortality - Multivariate Cox regression analysis showed that decreased urine output at CRRT conclusion was associated with mortality
RESPIRATORY TRACT DISEASE
Sinitsky et al. (72)	Retrospective observational Single center 2009–2013	636 children <16 y Mechanically ventilated PICU patients	Early FO with respiratory morbidity in PICU patients	- FO% had significant correlation with OI [Spearman ρ 0.318; p < 0.0001] and invasive ventilation days [ ρ 0.274; p < 0.0001] - FO% at 48 h was significant predictor of both OI (p < 0.001) and ventilation days (p = 0.002) - No association of FO% at 48 h with mortality
Flori et al. (74)	Post hoc analysis of a prospective observational study Multicenter 1996–2000	320 children <18 y Mechanically ventilated patients with ALI	FO with mortality and respiratory morbidity in children with ALI	- Positive FB (in increments of 10 mL/kg/24 h) was significantly associated with increased mortality [OR1.08, 95% CI 1.01–1.15, p = 0.02] and reduced VFDs [−0.21 (−0.39 to −0.04), p = 0.02], even after adjusting for multiple organ system failure, sepsis and the extent of oxygenation defect
Valentine et al. (73)	Retrospective observational Multicenter 2007–2010	168 children 1 m−18 y Mechanically ventilated patients with ALI	FO with fewer VFDs in children with ALI	- Higher CFB at D3 was independently associated with fewer VFDs (p = 0.02) - No association with mortality was noted (p = 0.11)
Ingelse et al. (71)	Retrospective observational Single center 2008–2014	135 children <2 y Mechanically ventilated PICU patients with viral lower respiratory tract disease	Early FO with prolonged LMV	- Mean CFB on D3 was 97.9 ± 49.2 mL/kg - Higher D3 CFB was independently associated with prolonged LMV [β = 0.166, p = 0.048] - No association found Between D3 CFB and sOSI (p = 0.7)
Willson et al. (32)	Post hoc analysis of the pediatric arm of an RCT Multicenter 2008–2010	110 children 0–18 y Mechanically ventilated children with ALI	FO with mortality, fewer VFDs and worse oxygenation	- Mean CFB in non-survivors was significantly higher than survivors [8.7 ± 9.5 vs. 1.2 ± 2.4L/m2; p < 0.001] - Higher CFB was significantly associated with fewer VFDs (p < 0.001) and higher OI [0.52 point increase in OSI for each L/m2 increase in FB; p = 0.011]
Arikan et al. (33)	Retrospective observational Single center 2004–2005	80 children 59 ± 73 months (mean ± SD) Mechanical ventilation for 24 h and presence of an indwelling arterial catheter	FO with prolonged LOS, LMV and impaired oxygenation	- Higher peak FO% predicted higher peak OI, independent of age, gender and PELOD scores (p < 0.009) - Peak FO% and severe FO% (≥15%) were both independently associated with prolonged LMV (p = 0.004 and 0.01), PICU stay (p = 0.008 and 0.01) and hospital LOS (p = 0.02 and 0.04)
SEPSIS
Abulebda et al. (37)	Retrospective observational Multicenter	317 children <10 y Septic shock patients	FO with mortality only in low risk septic patients, barring the intermediate and high risk cohort	- Increased CFB was associated with mortality in the low risk cohort (n = 204,OR 1.035, 95%CI 1.004–1.066) but not in the intermediate and high risk cohorts - Higher FB in the first 24 h was not associated with mortality
Chen et al. (75)	Retrospective observational Single center 2011–2015	202 children 1 m−18 y Admitted to PICU with severe sepsis	Early and acquired daily FO with mortality in septic children	Both early FO (aOR 1.20; 95%CI 1.08–1.33; p = 0.001; n = 202) and PICU-acquired daily FO (aOR = 5.47 per log increase; 95%CI 1.15–25.96; p = 0.032; n = 154) were independent risk factors associated with mortality even after adjusting for illness severity - Median PICU LOS increased with greater fluctuations in FO [p < 0.001] - Early FO achieved an AUC of 0.74 (95% CI 0.65–l0.82; p = < 0.001; n = 202) for predicting mortality
MISCELLANEOUS
Bhaskar et al. (76)	Retrospective observational (Matched case-control) Single center 2009–2010	114 children 0–17.4 y Admitted to PICU with shock	Early FO with mortality in shock patients	Early FO (>10% in 72h)[aOR 9.17, 95 %CI 2.22–55.57], its severity [aOR 1.11,1.05–1.19] and duration [aOR 1.61, 1.21–2.28] as independent predictors of mortality - Cases had significantly higher mortality than controls (26 vs. 6%; p = 0.003),even in the matched analysis (37 vs. 3 %; p = 0.002)
Liv et al. (34)	Prospective observational Single center 2011–2012	320 children 1 m−16 y Admitted to PICU for >24h	Early FO with AKI and mortality in critically ill children	- Early FO was independently associated with AKI (OR 1.34, p < 0.001) and mortality (OR 1.36, p < 0.001) - AUC of early FO for predicting mortality was 0.78 (p < 0.001)
Maitland et al. (79)	Open randomized controlled trial Multicenter 2009–2011	3,170 children 60 d−12 y Severe febrile illness with impaired perfusion Stratum A (3141)- saline/ albumin/no bolus Stratum B (29)– saline/albumin bolus in cases with severe hypotension	Fluid boluses with increased 48h mortality in critically ill children with impaired perfusion	- In stratum A, the 48 h mortality was 10.6,10.5,7.3% in the albumin-bolus, saline-bolus, and control group respectively - 28d mortality was 12.2, 12, and 8.7% in the three groups respectively (p = 0.004 for bolus vs. control) - In stratum B, mortality was 69% vs. 56% in the albumin vs. saline group respectively (p = 0.45)