Hindawi Publishing Corporation
Critical Care Research and Practice
Volume 2011, Article ID 854142, 5 pages
Positive FluidBalanceIs Associated withHigherMortality
andProlonged MechanicalVentilationinPediatricPatients with
1Division of Pediatric Critical Care, Children’s Hospital and Research Center Oakland, 747 52nd Street, Oakland, CA 94609, USA
2Division of Pediatric Pulmonology, UCSF Children’s Hospital, San Francisco, CA 94143, USA
3Division of Nephrology, UCSF Medical Center, San Francisco, CA 94143, USA
4Department of Epidemiology and Biostatistics, CTSI, Children’s Hospital and Research Center Oakland, 747 52nd Street, Oakland,
CA 94609, USA
5Department of Anesthesia and Critical Care and CVRI, UCSF Medical Center, San Francisco, CA 94143, USA
Correspondence should be addressed to Heidi R. Flori, email@example.com
Received 25 November 2010; Accepted 1 April 2011
Academic Editor: Ira Cheifetz
Copyright © 2011 Heidi R. Flori et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction. We analyzed a database of 320 pediatric patients with acute lung injury (ALI), to test the hypothesis that positive
fluid balance is associated with worse clinical outcomes in children with ALI. Methods. This is a post-hoc analysis of previously
collected data. Cumulative fluid balance was analyzed in ml per kilogram per day for the first 72hours after ALI while in the PICU.
The primary outcome was mortality; the secondary outcome was ventilator-free days. Results. Positive fluid balance (in increments
of 10mL/kg/24h) was associated with a significant increase in both mortality and prolonged duration of mechanical ventilation,
independent of the presence of multiple organ system failure and the extent of oxygenation defect. These relationships remained
unchangedwhenthesubgroupofpatientswithsepticshock(n = 39)wereexcluded.Conclusions.Persistentlypositivefluidbalance
may be deleterious to pediatric patients with ALI. A confirmatory, prospective randomized controlled trial of fluid management
in pediatric patients with ALI is warranted.
large observational studies and post hoc analyses of random-
ized trials have found that positive fluid balance during the
first days after development of ALI/ARDS is independently
associated with a higher risk of death [1–4]. Smaller trials
have indicated that a reduction in extravascular lung water
correlates with improved clinical outcomes [5–7]. In 2006,
the ARDS Network published results of their Fluid and
Catheter Treatment Trial . This trial demonstrated that a
conservative fluid management strategy, aimed at achieving
normal intravascular filling pressures after resolution of
shock and resulting in relative even fluid balance over the
first 7 days of the study, resulted in a shorter duration of
mechanical ventilation and ICU stay.
Similar studies have not been conducted in children
with ALI. The Pediatric Acute Lung Injury and Sepsis
Investigators (PALISIs) network published a post hoc analysis
of their multicenter weaning modes of ventilation study .
This analysis indicated that positive fluid balance was not
associated with prolonged need for mechanical ventilation
in children intubated for acute respiratory failure of all
etiologies, but not specifically children with ALI, who were
a subset of this study population.
We conducted a prospective, observational study of
320 children admitted to two large pediatric intensive care
units (PICUs) that met the American European Consensus
Conference (AECC) definition of ALI . We hypothesized
that positive fluid balance would be associated with worse
clinical outcomes in children with ALI. We conducted the
2Critical Care Research and Practice
following post hoc analysis on data from our original cohort
to test this hypothesis.
For the parent investigation all pediatric patients admitted
to the PICU at Children’s Hospital and Research Center
Oakland (CHRCO) between July 1996 and May 2000 and
the University of California Medical Center, San Francisco
Children’s Hospital (UCSF) between July 1996 and July
1998 were prospectively evaluated. All patients who met
the AECC definition of ALI  were included in the
analyses. All patients had at least one arterial blood gas
supporting the PaO2/FiO2 <300 requirement obtained as
part of routine clinical care. Patients were excluded if they
were <36 weeks corrected gestational age or >18 years of age,
had evidence of left atrial hypertension either clinically or
by echocardiogram, or had any echocardiographic evidence
of intracardiac shunt. The study was approved by the
Institutional Review Board at CHRCO and the Committee
on Human Research at UCSF.
For the current investigation, all patients who received
exchange transfusions and those patients requiring extra-
corporeal membrane oxygenation, continuous venovenous
hemofiltration, or exchange transfusions were excluded as
the total fluid balance for these patients would skew the
results. All clinical data including fluid administration and
losses were included in the database a priori.
The primary outcome was PICU mortality. The sec-
ondary outcome was ventilator-free days, defined, as in
previous studies [8, 10], as the number of days the patient
was alive and not mechanically ventilated in the 28 days
after the onset of ALI. All patients who died while still
mechanically ventilated were assigned a value of zero; all
patients not requiring mechanical ventilation were assigned
a value of 28.
Data were first entered into a Microsoft Access relational
database for cleaning and preparation for analysis. All clin-
ically and statistically significant covariates from the parent
gation. Potential covariates were age, gender, ethnicity, diag-
nosis associated with ALI, past medical history, individual
organ system failures, blood gas variables, presence of neu-
tropenia, presence of airleak, and respiratory indices .
Cumulative fluid balance data were measured as a
continuous variable over the first 3 days after onset of ALI.
Daily fluid balance can be confounded by those patients that
do not spend 24 hours per day in the PICU, particularly on
the day of admission or in the event that the patient dies or
is discharged before 72 hours. To account for this potential
discrepancy, fluid balance data were recorded as mL per
per hour variable was adjusted to 10mL/kg/day increments
to facilitate clinical interpretation of the results.
Initial bivariate analyses included chi-square tests for
categorical data and t-tests for continuous variables. A
multivariate logistic regression model was created using a
method of forward selection to examine the associations
with mortality. In addition, a linear regression model was
Table 1: Patient clinical characteristics and outcomes, (N = 313).
Age, median (25–75% interquartile
PaO2/FiO2 at ALI onset#
Diagnoses associated with ALI, % (n)
Nonpulmonary organ system failures
at ALI onset
PRISM III at ALI onset#
Exhaled TV (cc/kg IBW)#
Ventilator free days
Mortality, % (n)
#Mean ± standard deviation.
∗Patients with any intracardiac shunts and/or left sided heart failure
3.4 yrs (1d–18 yrs)
161 ± 74
1.4 ± 1.5
10.3 ± 8.7
10.1 ± 4.3cc/kg
14.8 ± 10.6d
developed to determine the associations with number of
ventilator-free days. The coefficient of determination R2
was computed for each model with the Hosmer-Lemeshow
goodness-of-fit test computed for the logistic model. A
significance level of 0.05 was used for all statistical tests.
Analyses were completed using Stata6 statistical software
(StataCorp, College Station, TX, USA).
3.Results and Discussion
Table 1 describes the baseline characteristics of the subgroup
of 313 patients included in this analysis. As expected,
these characteristics were essentially the same as the cohort
examined in the parent study.
Bivariate logistic regression analysis demonstrated that
increasing fluid balance, in 10mL/kg/day increments, was
associated with increasing mortality (OR 1.12, 95% CI 1.06,
1.20, P < .001, Figure 1: Chi-squared statistic, P < .01).
Bivariate linear regression analysis also identified that an
increase in fluid balance, again in 10mL/kg/day increments,
was associated with fewer ventilator-free days (Coef: −0.41,
95% CI: −0.60, −0.21, P < .01). To better represent this
bivariate relationship graphically, this outcome was cate-
gorized to reflect patients requiring prolonged mechanical
ventilation, defined as 14 or more days of mechanical
ventilation (Figure 2, Chi-squared statistic, P = .47).
In multivariate analysis, increasing fluid balance was
associated with increased mortality (OR 1.08, 95% CI 1.01–
1.15, P = .02, Hosmer-Lemeshow goodness of fit P =
.87, Table 2). This effect was independent of the severity of
oxygenation defect, as measured by PaO2/FiO2, at the onset
organ system failures. Similarly, increasing fluid balance was
Critical Care Research and Practice3
Fluid balance and mortality
Fluid balance (mL/kg/d)
Figure 1: Bar graph depicting the association between cumulative
fluid balance within the first 72 hours after ALI and all-cause
Fluid balance (mL/kg/d)
Percent with ≤14
ventilator free days (%)
Fluid balance and need for
prolonged mechanical ventilation
Figure 2: Bar graph depicting the association between cumulative
fluid balance within the first 72 hours after ALI and need for
prolonged mechanical ventilation (patients requiring mechanical
ventilation for ≥14 days).
also associated with fewer ventilator-free days independent
of nonpulmonary organ system failure (Table 2). There
was no evidence of interaction between fluid balance and
cardiovascular failure, renal failure, age, or gender.
In a sensitivity analysis, multivariate models were
repeated excluding the 39 patients with ALI as a result of
sepsis. The association between increasing fluid balance and
to the primary analyses; see Table 3.
3.1. Discussion. This study represents the first large analysis
of the possible association between cumulative fluid balance
in children with ALI and increasing mortality and duration
all patients were identified prospectively and all fluid data
were collected a priori as a part of the original study.
These data are particularly relevant since there have been
no large, multicenter trials testing liberal versus conservative
fluid strategy in the treatment of pediatric ALI to date,
(a) Multivariate results for mortality
Odds ratio (95% C.I.)
0.91 (0.82, 1.00)
1.90 (1.45, 2.49)
7.46 (3.60, 15.45)
1.08 (1.01, 1.15)
(b) Multivariate results for ventilator-free days
Coefficient (95% C.I.)
0.43 (0.14, 0.72)
−2.21 (−3.07, −1.35)
−6.33 (−9.06, −3.61)
−0.21 (−0.39, −0.04)
1PaO2/FiO2 measured in 20 point increases.
2Nonpulmonary, non-CNS organ system failure.
3Fluid Balance measured in 10mL/kg/day increments.
(a) Multivariate results for mortality, excluding patients with sepsis
Odds ratio (95% C.I.)
0.88 (0.79, 0.99)
2.28 (1.60, 3.26)
6.81 (2.94, 15.79)
1.09 (1.00, 1.18)
(b) Multivariate results for ventilator-free days, excluding patients with
Coefficient (95% C.I.)
0.41 (0.11, 0.70)
−2.92 (−3.98, −1.87)
−5.56 (−8.55, −2.57)
−0.21 (−0.42, −0.01)
1PaO2/FiO2 measured in 20 point increases.
2Nonpulmonary, non-CNS organ system failure.
3Fluid Balance measured in 10 mL/kg/day increment.
management strategy in adults has been associated with an
increased number of ventilator-free days and a trend towards
decreased mortality .
In our original study, we concluded that there was
remarkable similarity in the epidemiology and clinical risk
factors associated with the clinical outcome of adults and
children with ALI . Biological marker studies completed
in both our laboratory [12, 13] and others [14, 15] also
indicate that the pathophysiology of inflammatory injury
to the alveolar epithelium and lung endothelium damage
associated with early ALI is similar in adults and children.
Optimal fluid balance in the patient with ALI/ARDS is
a dynamic process. The early phase of ALI is characterized
by increased capillary permeability and the need to maintain
intravascular volume to preserve cardiac function, in partic-
ular in patients with underlying sepsis. Fluid resuscitation
may be needed to maintain intravascular volume; this may
4Critical Care Research and Practice
worsen underlying alveolar edema and further impair gas
exchange. The question for clinicians is at what point the
fluid resuscitation becomes excessive. The nonsurvivors in
this study received over two times the amount of fluid per
kg per day as the survivors. Were the nonsurvivors destined
from the beginning to have a worse outcome based on
their presenting illness, regardless of how much fluid they
received? However, positive fluid balance was found to be
associated with mortality independent of the severity of
illness, as measured by organ system failures. This suggests
that fluid overload itself may be a risk factor for mortality,
regardless of the initial presenting severity of illness.
With this as background, is this study sufficient to
warrant a change in practice to that which is currently
recommended in adults? We believe that the answer is not
are likely to benefit children with ALI, these hypotheses must
be formally tested through randomized clinical trials first.
What next? One possibility is to complete similar
analyses in other recently completed randomized controlled
trials. Those analyses would also be subjected to the same
limitations as all post hoc, observational analyses do, in
addition to the possibility of including fewer patients than
in this study. A second option would be to complete a
prospective, multicenter observational study of fluid man-
of mechanical ventilation practices in pediatric ALI .
However, these are all observational approaches, which are
subject to confounding.
mortality than the average PICU patient, there is ultimately
a need for rigorously performed, randomized controlled
trials of both new and more routine management practices.
Ongoing collaborations between the PALISI and NHLBI
ARDS Networks are specifically focusing on testing potential
new ALI treatment targets, in both children and adults
using similar algorithms and Bayesian statistical modeling to
minimize the number of children that need to be enrolled.
Careful consideration of study design and clinical endpoints
is paramount in order to allow for multicenter, randomized,
controlled trials to complete enrollment while equipoise and
rigor can be maintained.
The major limitation of this study is that it is a post hoc
the parent investigation ended in 2000 and may not reflect
the full impact of lung protective mechanical ventilation
algorithms. Nonetheless, the mean exhaled tidal volume in
our patients was lower than the relatively injurious limb
of the adult ARMA trial (10mL/kg in this study, 12mL/kg
in ARMA). Another limitation regards the quantification
of severity of illness. At the time our study was designed,
the PRISM III  score had not yet been validated nor
validated for the first 12 to 24 hours after PICU admission
and not for the first day of diagnosis of ALI, when our
data were collected. In our cohort, approximately 30% of
patients developed ALI after their first day of ICU admission.
Therefore, organ system failure was used to measure the
severity of illness, according to current practice at that time.
Central venous pressure data and serum albumin levels
were not recorded in the original dataset and therefore not
reflected in these analyses. Lastly, although we tested for
interactions between fluid balance and cardiovascular failure
(n = 133 patients) and/or renal failure (n = 47 patients) and
found none, it is possible that interactions do exist within
subgroups of these patients.
Based on the results of recent adult trials and our obser-
vational data, we believe that positive fluid balance in the
wake of ALI may be deleterious in children as it is in adults.
management in a large, prospective, multicenter pediatric
This work is supported by NIH RR 15543 (HF), NIH RR
01271 (GG), NIH HL 51856 (MM), NIH KL2 RR024130
 Y. Sakr, J. L. Vincent, K. Reinhart et al., “High tidal volume
and positive fluid balance are associated with worse outcome
in acute lung injury,” Chest, vol. 128, no. 5, pp. 3098–3108,
 R. S. Simmons, G. G. Berdine, J. J. Seidenfeld et al., “Fluid
balance and the adult respiratory distress syndrome,” Ameri-
can Review of Respiratory Disease, vol. 135, no. 4, pp. 924–929,
 A. L. Rosenberg, R. E. Dechert, P. K. Park, R. H. Bartlett, and
of cumulative fluid balance on outcome in acute lung injury:
a retrospective review of the ARDSnet tidal volume study
cohort,” Journal of Intensive Care Medicine, vol. 24, no. 1, pp.
tance of fluid management in acute lung injury secondary to
septic shock,” Chest, vol. 136, no. 1, pp. 102–109, 2009.
 S. G. Sakka, M. Klein, K. Reinhart, and A. Meier-Hellmann,
“Prognostic value of extravascular lung water in critically III
patients,” Chest, vol. 122, no. 6, pp. 2080–2086, 2002.
 J. P. Mitchell, D. Schuller, F. S. Calandrino, and D. P. Schuster,
“Improved outcome based on fluid management in critically
 D. Schuller, J. P. Mitchell, F. S. Calandrino, and D. P. Schuster,
“Fluid balance during pulmonary edema: is fluid gain a
marker or a cause of poor outcome?” Chest, vol. 100, no. 4,
pp. 1068–1075, 1991.
 H. P. Wiedemann, A. P. Wheeler et al., “Comparison of two
fluid-management strategies in acute lung injury,” The New
England Journal of Medicine, vol. 354, no. 24, pp. 2564–2575,
 A. G. Randolph, P. W. Forbes, R. G. Gedeit et al., “Cumulative
fluid intake minus output is not associated with ventilator
weaning duration or extubation outcomes in children,” Pedi-
atric Critical Care Medicine, vol. 6, no. 6, pp. 642–647, 2005.
Critical Care Research and Practice5 Download full-text
 H. R. Flori, D. V. Glidden, G. W. Rutherford, and M. A.
Matthay, “Pediatric acute lung injury: prospective evaluation
of risk factors associated with mortality,” American Journal of
 G. R. Bernard, A. Artigas, K. L. Brigham et al., “The
North American-European consensus conference on ARDS,”
American Journal of Respiratory and Critical Care Medicine,
vol. 149, no. 3, pp. 818–824, 1994.
 H. R. Flori, L. B. Ware, D. Glidden, and M. A. Matthay, “Early
elevation of plasma soluble intercellular adhesion molecule-1
in pediatric acute lung injury identifies patients at increased
risk of death and prolonged mechanical ventilation,” Pediatric
Critical Care Medicine, vol. 4, no. 3, pp. 315–321, 2003.
 H. R. Flori, L. B. Ware, M. Milet, and M. A. Matthay, “Early
elevation of plasma von Willebrand factor antigen in pediatric
acute lung injury is associated with an increased risk of death
and prolonged mechanical ventilation,” Pediatric Critical Care
Medicine, vol. 8, no. 2, pp. 96–101, 2007.
 A. M. LeVine, A. Lotze, S. Stanley et al., “Surfactant content
in children with inflammatory lung disease,” Critical Care
Medicine, vol. 24, no. 6, pp. 1062–1067, 1996.
 B. Jacobs and V. N. Bal, “Cytokines in children with ARDS,”
Critical Care Medicine, vol. 33, article A43, 2006.
 M. Santschi, P. Jouvet, F. Leclerc et al., “Acute lung injury in
children: therapeutic practice and feasibility of international
clinical trials,” Pediatric Critical Care Medicine, vol. 11, no. 6,
pp. 681–689, 2010.
 M. M. Pollack, K. M. Patel, and U. E. Ruttimann, “PRISM
III: an updated pediatric risk of mortality score,” Critical Care
Medicine, vol. 24, no. 5, pp. 743–752, 1996.