Journal of Critical Care (2012) 27, 745.e7745.

e12

Estimation of fluid status changes in critically ill patients:

Fluid balance chart or electronic bed weight?
Antoine G. Schneider MD a,b , Ian Baldwin PhD a , Elke Freitag MN a ,
Neil Glassford MBchB a , Rinaldo Bellomo MD a,b,
a
Intensive Care Unit, Department of Intensive Care, Austin Health, Heidelberg, Victoria, Australia
b
Monash University, Dept of Epidemiology and Preventive Medicine, the Alfred Center, Melbourne Victoria, Australia

Keywords:
Abstract
Fluid balance;
Purpose: Monitoring of fluid balance (FB) can be achieved by subtracting recorded fluid output from
Intensive care unit;
input or by measuring changes in body weight (BW). The latter approach is difficult in the critically ill.
Body weight;
Recently, hospital beds have become available with the ability to directly weigh patients in the intensive
Water-electrolyte balance
care unit (ICU) patients directly. We sought to compare FB estimates obtained by these 2 methods in a
cohort of critically ill patients.
Materials and Methods: Between November 2010 and May 2011, all patients admitted in our ICU for
more than 2 consecutive days and nursed on a Hill-Rom (Batesville, Ind) Total Care bed were weighed
daily at midnight hours. Fluids charting was done by electronic spreadsheet with automated 24 hours
calculation. Differences in BW and FB between 2 consecutive days were compared using correlation
and Bland-Altman analysis. Corrections for unmeasured fluids losses were performed using a
predetermined formula based on peak temperature and intubation status.
Results: We obtained complete data in 160 (31%) of 504 admissions exceeding 2 days (153 patients)
resulting in 435 data points. The change in BW over 24 hours and FB for the same period was only weakly
correlated before (r = 0.34; P b .001; Fig. 1) or after correction for insensible fluid losses (r = 0.34; P b
.001). On Bland-Altman plot, the mean bias was small (0.07 kg), but the 95% limits of agreement, very
large (5.8 and 6.0 kg). The lack of agreement increased with the magnitude of the changes.
Conclusion: Obtaining daily weights in ICU patients proved difficult. Compliance was poor. The
correlation between changes in BWs and FB was weak. Further studies are required to establish if accurate
and reproducible daily weighing of ICU patients is feasible.
2012 Elsevier Inc. All rights reserved.

1. Background and rationale

Conicts of interests: All authors stated that they had no conicts of
interest to declare.
Corresponding author. Intensive Care Unit, Austin Health, 3084
Heidelberg, Victoria, Australia.
E-mail address: antoine.schneider@austin.org.au (A.G. Schneider).
Fluid management is important in critically ill patients.
On one hand, aggressive initial uid therapy has been shown
to decrease mortality in sepsis [1]. On the other hand,
numerous studies have shown an association between
positive uid balance (FB) and increased mortality [2-7].

0883-9441/$ see front matter 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcrc.2011.12.017
745.e8
In the intensive care unit (ICU), the recording of inputs
and outputs allows the calculation of a daily FB. However,
entered data can be incomplete or inaccurate; some uids
losses such as perspiration, respiration, or diarrhea are not
taken into account or measured [8], and if the data are not
entered in a computerized system, calculation errors might
occur. These multiple error sources tend to add up, and the
cumulative calculated FB tends to be less and less accurate
over a patient ICU stay [9-11].
Because, over a short period, changes in body weight
(BW) are almost only associated with change in body uids,
measurement of BW may be a more accurate way of
estimating uid status. This approach might be suboptimal in
patients with extended ICU stay, where muscle and fat loss
as well as bone demineralization are likely to play a role in
changes in BW [12]. However, for most ICU patients and at
least probably for the rst week of ICU stay, changes in BW
are likely to be a reliable estimate of FB.
Unfortunately, weighing the critically ill patients is
technically difcult, as sedation, invasive monitoring, and
equipment make use of standard scales impossible. For
example, among 5 key studies [13-17] involving renal
replacement therapies and setting ultra ltrate rate based
on weight on admission, only 1 indicated that an actual
measurement was done, and this method was not described
[16]. Several devices have been designed and marketed,
but they are either inaccurate or very cumbersome [18].
Bioimpedance measurement and bioimpedance vector
analyses techniques are promising [19-22], but their use
in clinical practice needs to be established. Recently, beds
with built-in patient weighing scales have been made
available. According to the manufacturer [23], the
accuracy of the scale is 1% of the patient's weight, and
the repeatability is 0.3% (b75 kg) and 0.1% (N75 kg).
However, their use in the ICU has not been well described
or validated.
Accordingly, we sought (1) to evaluate the feasibility
and ease of use these beds in our ICU and (2) to test the
correlation between the FB estimated by the traditional
charting method and by evaluation of BW using the
study device.
2. Materials and methods
We designed an observational study of all consecutive
patients admitted in ICU for more than 48 hours and
occupying beds with weighing capabilities. At the time of the
study, these beds represented 14 of 20 of ICU beds. Patients
were excluded if their weight was above 181.4 kg (400 lb) or
if the bed was not tared (zeroed with linen, pillows in place)
before admission. We aimed to obtain data from a
convenience sample of more than 100 patients.
The study protocol was approved by the human research
ethics committee of our hospital, and the need for informed
consent was waived.
A.G. Schneider et al.
2.1. Study beds
We used Hill-Rom (Batesville, Ind) Avant guard 1600
and Total Care bed systems. Both these beds offer easy bed
taring and weighing capacities. Before the beginning of the
study, basic training on how to operate and tare the bed was
provided to the nursing staff. Emphasis was placed on the
importance of taring the bed before patient's admission and
to temporarily remove any extra weight from the bed.
2.2. Weighing procedure
While the ICU bay was prepared before an admission, the
bed was tared (calibrated). This tare procedure included a
standard list of beddings, pillows, and sheets. The patients
were then weighed daily at midnight to coincide with the 24-
hour FB calculation. Before pressing the weigh button, the
nurse was required to remove any extra weight not included
in the tare items list. These included urinary bags, drains, or
disinfectant bottles. Items attached to the bedside poles were
not to be touched. The head of the bed had to be tilted to
lower than 30, then the weigh button could be pressed.
Altogether, this process should take less than 5 minutes. The
results were reported in a dedicated study data collection
tool describing the procedure and its important steps, kept at
each bedside.
2.3. Fluid balance calculation
A structured query language-based electronic FB chart
(SLIC version 5.38; Incarta IT, Melbourne, Australia) was
used to calculate the FB. This software was accessible
through a monitor interface available in each ICU bay. It
requires an hourly entry for inputs (maintenance uids, uid
boluses, blood products, intravenous medications, nutri-
tion) and outputs (urinary output, drains contents,
estimation of feces volume). The FB was calculated in
real time and locked at midnight for the previous day.
2.4. Insensible fluid losses evaluation
Each study day, to allow for evaluation of insensible uid
losses (IFL), we collected intubation status and maximal
body temperature. We calculated the volume of IFL
according to a predetermined formula [8]: IFL (milliliters) =
800 + 20% 800 (maximum temperature 37). This value
was divided by 2 if the patient was intubated.
2.5. Statistical analysis
We compared changes in BW between 2 consecutive days
and the corresponding day FB. Descriptive statistics are
presented as mean and SD or median and interquartile range
(IQR) as appropriate; ordinal data are presented with number
and percentage. The FB and BW data were normally
Fluid balance chart or electronic bed weight
distributed, so Pearson correlation tests were performed for a
relationship between these 2 variables. Signicance was set
at P b .05. A Bland-Altman assessment for agreement was
used. The range of agreement was dened as mean bias 2
SD. Data were analyzed using PASW/SPSS software,
version 18 (IBM, Inc, Chicago, Ill).
3. Results
3.1. Patients' demographics and outcome
We obtained complete data in 151 patients (160 ICU
stays). As presented in Table 1, the median age of the
patients was 69 years (IQR, 23); 96 (63.6%) were male, and
85 (53.1%) were admitted under a medical unit. They stayed
in ICU for a median duration of 5.0 days (IQR, 5.9), and
their median Acute Physiology and Chronic Health
Evaluation III score was 68 (IQR, 34). Overall, 56 (35%)
of these patients were intubated at the time of study inclusion
(rst day, a weight was obtained), and 12 (7.9%) did not
survive to ICU discharge.
3.2. Compliance with the protocol
As illustrated in Fig. 1, we obtained BW measurement on
2 or more consecutive days in 160 of 504 (31.7%)
admissions exceeding 2 days (151 patients) and daily FB
data in 504 (100%) of 504. Altogether, 418 data pairs were
obtained. The compliance decreased during the study period:
the rst 3 month of the study, we obtained, respectively, 27,
33, and 27 completed patients as opposed to 16, 15, and 21 in
the next 3 months. The reasons for noncompliance with the
protocol were failure to calibrate the bed before admission or
failure to complete the procedure.
3.3. Correlation between FB and changes in BW
The correlation between daily changes in BW and FB is
presented in Fig. 2A. The correlation between these 2
parameters was weak (Pearson coefcient, 0.34; 95%
condence interval, 0.26-0.42; P b .001) and remained
unchanged after exclusion of extreme values (delta N5 L or
Table 1 Patient's characteristics
N 151 patients, 160 ICU stays
Age (median; IQR) 69.0; 23
M/F ratio (n and % males) 96/151 (63.6%)
Medical/surgical (% medical) 85/160 (53.1%)
Total ICU LOS (median; IQR) 5.0; 5.9 d
APACHE III (median; IQR) 68 34
Mechanical ventilation 56/160 (35%)
In-ICU mortality 12/151 (7.9%)
M/F indicates male/female; LOS, length of stay; APACHE, Acute
Physiology and Chronic Health Evaluation.
745.e9
1002 ICU ADMISSIONS
466
ICU LOS < 2 days
32
Not on Hill-Rom Beds*
504 Eligible admissions
344
Data incomplete
(No BW obtained)
BW and FB obtained in 160 admissions (151
patients)
Fig. 1 Flow chart. Asterisk indicates that some patients might not
have been in weighing beds and not excluded from the analysis.
LOS indicates length of stay.
5 kg per day) (r = 0.37; 95% condence interval, 0.28-
0.46). Similarly, after correction for insensible uid losses
(Fig. 2B), the correlation coefcient remained identical
(0.34; condence interval, 0.25-0.42; P b .001).
As shown by the Bland-Altman plot (Fig. 3), mean bias was
0.07 kg, but the 95% interval of agreement was wide (between
5.9 and 6.0 kg). Similarly, the bias increased with increased
change in BW. Thus, changes in FB and BW were more likely
to be in agreement for very small values (close to zero).
In addition, the measurements differed by more than 3 L
(kg) in 102 pairs (24.2%) and by more than 7 L (kg) in 20
(4.7%). Disagreement between data pairs was less than 10%
in 27 data pairs (6.9% of nonnull differences), 10% to 25% in
19 (4.8%), 25% to 50% in 36 (9.1%), 50% to 100% in 135
pairs (34.4%), and greater than 100% in 176 (44.8%).
Finally, changes in BW and FB were the same direction in
only 243 (57.4%) of 423 of observations.
4. Discussion
4.1. Summary of key findings
This study is, to the best of our knowledge, the rst
prospective study of the feasibility of weighing critically
patients using an electronic weighing bed. We found that,
even in a unit where nurses are used to taking part in clinical
trials and trained in protocol adherence and with modern
745.e10 A.G. Schneider et al.

Correlations Corrected DFB

5 5

Corrected Fluid balance

Fluid balance

-5 5 -5 5

-5 -5

Delta weight Delta weight

Fig. 2 Correlation plots. A. Delta weights and uid balance B. Delta weight and uid balance corrected for insensible uid losses
(see formula in text).

high-end beds with weighing capacities, BW measurements
were obtained in less than a third of cases. When compared
with the results of the FB, such measurement yielded only a
weak correlation and a progressively greater difference with
greater changes in BW. This correlation did not improve
after correction of the FB for insensible weight losses.
4.2. Comparison with previous studies
Several studies have explored the accuracy of the calculated
FB and compared it with a BW obtained on admission and
10.00
+1.96SD
5.00
DWDFB
discharge. Eastwood [12] assessed the accuracy of the FB by
comparing it with the changes in BW in 32 patients undergoing
cardiac surgery. The BW was obtained before the operation
and on ICU discharge. They found that FB correctly (error,
b0.25 kg) estimated changes in BW in only 9.75% of cases.
The BW changes were underestimated in 59.4% of the patients
(range, 0.26 to +6.57 kg) and overestimated in 31.2%
(range, +0.33 to +5.25 kg). In 12% of cases, there was a
difference greater than a liter. In this study, the patients were
weighed using regular scales, and patients not able to sit on
the scale were excluded. This study demonstrated the
potential limitations of cumulative FB over the entire ICU
stay (mean, 2.4 days).
More recently, Perren et al [9] performed a similar study
in a general ICU population. They weighed 147 patients
using a multicare bed on ICU admission and discharge.
Interestingly, they also obtained adequate data in only a
minority of patients (38%) admitted during the study period.
They reported a fair correlation (r2 = 0.714) between the

0.00
adjusted FB and the changes in BW but also found wide
limits of agreement. The authors also reported a high error
rate for the FB calculation (33%) with errors ranging from
3.6 to 2.0 L. Their overall conclusion was that FB is not a
-5.00
reliable tool to assess uid changes in ICU patient. In
-1.96SD agreement with this study, we also found that changes in
BW and FB had wide limits of agreement. However, in our
-10.00 report, we used a computer-based FB record and automatic
-10.00 -5.00 0.00 5.00 10.00 calculation, which although not preventing double or missed
DW+DFB/2 entries or keying errors, removes concerns about calculation
Fig. 3 Bland-Altman plot. x axis, average value for delta weight errors. Finally, our report as opposed to the 2 previously
and delta uid balance (FB). y axis, difference between delta weight cited works compared daily measurements and not cumu-
and delta FB. DW indicates delta weight; DFB, delta FB. lative FB. This difference should decrease the inuence of
Fluid balance chart or electronic bed weight
insensible uid losses that tend to accumulate during the
ICU stay.
4.3. Implications for clinicians
This study shows that even with modern technology-
based weighing beds and trained staff, obtaining reliable
weights in ICU patients is difcult. Although, as mentioned,
FB cannot be taken as a criterion standard, extreme
variations of the BW between 2 consecutive days appear
more likely to represent BW measurement errors when they
are not, at least partially correlated with a change in FB.
Obtaining reliable weight measurement, at a specic time
point, and the need to perform a bed tare before the patient is
admitted requires a long period of training and reminders
with inclusion of non-nursing staff such as care attendants
who clean and prepare the beds. This was clearly not well
achieved in our study and highlights that shift routines and
procedures take time to become part of standard care in a
given ICU. The fact that not all beds were weigh capable
may have contributed to decreased compliance by creating
uncertainty. Although further studies are required to conrm
this impression, introducing daily weight of patients is not an
easy operation and seems to require a strong educational and
behavioral change effort.
4.4. Strengths and limitations
This study is, to the best of our knowledge, the rst one to
tackle the feasibility and correlation of daily BW with FB in
ICU. All data were collected prospectively and analyzed
blindly. Our center is typically involved in multiple research
protocols, and nurses and ancillary staff are trained to new
protocols and routinely accommodate care schedule changes.
Finally, our ICU is representative of other modern and busy
ICUs in tertiary hospitals.
Fluid balance calculation, although widely used in clinical
practice and in the literature, is not a criterion standard. Data
entry is associated with potential errors. However, in our
ICU, the automated computation of FB from available data
should have eliminated calculation errors. There is no
consensus in the literature on how to calculate insensible
losses. However, to take this uncertainty into account, we
presented our results with and without correction.
Data were treated as pairs, which proportionally gives
more importance to patients with a longer stay. However,
we do not believe that would result in any bias because
our study did not focus on clinical end points but only
on day-to-day comparison and agreement between the
2 methods.
Another potential bias would be a drift in the calibration
of the bed along the duration of the ICU stay (some patients
had ICU stays N2 weeks). Maybe there is a need recalibrate
beds whenever possible (eg, during a computed tomo-
graphic scan, surgery, or when patients are mobilized
745.e11
outside the bed). However, this remains a theoretical
concern, as no evidence exists that such shifts occur. We
chose not to recalibrate beds during a patient's stay to avoid
staff confusion.
Of note, some patients not installed on weighing beds
might not have been excluded from the study. This would
falsely increase the number of missing data. However, given
the high availability of such beds in our unit and their
preferred allocation to patients with long ICU stay, it is very
unlikely that this error would change the overall conclusion
that the compliance was low.
4.5. Further studies
Studies with a modied weighing procedure with
calibration by patient service assistants may increase
compliance and appear desirable. Introduction of an
automatic screen pop up message to remind staff to perform
the BW assessment at midnight should also be studied.
Finally, the beds should probably get tested against a
criterion standard possibly in less severely ill patients in
whom use of a normal scale is feasible.
5. Conclusion
Obtaining daily BW in ICU patients is difcult. Six
months after the introduction of new beds with weighing
capacities, the correlation between changes in BWs and FB
was weak and limits of agreement wide. Further studies after
process improvement are required to establish if daily
weighing of ICU patients can be made feasible.
References
[1] Rivers E. Early goal-directed therapy in the treatment of severe sepsis
and septic shock. N Engl J Med 2001;345:1368-77.
[2] Gillespie RS, Seidel K, Symons JM. Effect of fluid overload and dose
of replacement fluid on survival in hemofiltration. Pediatr Nephrol
2004;19:1394-9.
[3] Payen D, de Pont AC, Sakr Y, et al. A positive fluid balance is
associated with a worse outcome in patients with acute renal failure.
Crit Care 2008;12:R74.
[4] Bouchard J, Mehta RL. Fluid accumulation and acute kidney injury:
consequence or cause. Curr Opin Crit Care 2009;15:509-13.
[5] Foland JA, Fortenberry JD, Warshaw BL, et al. Fluid overload before
continuous hemofiltration and survival in critically ill children: a
retrospective analysis. Crit Care Med 2004;32:1771-6.
[6] Sakr Y, Vincent JL, Reinhart K, et al. High tidal volume and positive
fluid balance are associated with worse outcome in acute lung injury.
Chest 2005;128:3098-108.
[7] Uchino S, Bellomo R, Morimatsu H, et al. Pulmonary artery catheter
versus pulse contour analysis: a prospective epidemiological study.
Crit Care 2006;10:R174.
[8] Cox P. Insensible water loss and its assessment in adult patients: a
review. Acta Anaesthesiol Scand 1987;31:771-6.
[9] Perren A, Markmann M, Merlani G, et al. Fluid balance in critically ill
patients. Should we really rely on it? Minerva Anestesiol 2011.
745.e12
[10] Grams ME, Estrella MM, Coresh J, et al. Fluid balance, diuretic use,
and mortality in acute kidney injury. Clin J Am Soc Nephrol 2011;6:
966-73.
[11] Pflaum SS. Investigation of intake-output as a means of assessing body
fluid balance. Heart Lung 1979;8:495-8.
[12] Eastwood GM. Evaluating the reliability of recorded fluid balance to
approximate body weight change in patients undergoing cardiac
surgery. Heart Lung 2006;35:27-33.
[13] Ronco C, Bellomo R, Homel P, et al. Effects of different doses
in continuous veno-venous haemofiltration on outcomes of acute
renal failure: a prospective randomised trial. Lancet 2000;356:
26-30.
[14] Saudan P, Niederberger M, De Seigneux S, et al. Adding a dialysis
dose to continuous hemofiltration increases survival in patients with
acute renal failure. Kidney Int 2006;70:1312-7.
[15] Tolwani A, Paganini E, Joannidis M, et al. Treatment of patients with
cardiac surgery associated-acute kidney injury. Int J Artif Organs
2008;31:190-6.
[16] Bellomo R, Cass A, Cole L, et al. Intensity of continuous renal-
replacement therapy in critically ill patients. N Engl J Med 2009;361:
1627-38.
A.G. Schneider et al.
[17] Palevsky PM, Zhang JH, O'Connor TZ, et al. Intensity of renal support
in critically ill patients with acute kidney injury. N Engl J Med
2008;359:7-20.
[18] Freitag E, Edgecombe G, Baldwin I, et al. Determination of body
weight and height measurement for critically ill patients admitted to the
intensive care unit: a quality improvement project. Aust Crit Care
2010;23:197-207.
[19] Sergi G, Bussolotto M, Perini P, et al. Accuracy of bioelectrical
impedance analysis in estimation of extracellular space in healthy
subjects and in fluid retention states. Ann Nutr Metab 1994;38:158-65.
[20] Kushner RF, Schoeller DA, Fjeld CR, et al. Is the impedance index
(ht2/R) significant in predicting total body water? Am J Clin Nutr
1992;56:835-9.
[21] Roos AN, Westendorp RG, Frolich M, et al. Weight changes in
critically ill patients evaluated by fluid balances and impedance
measurements. Crit Care Med 1993;21:871-7.
[22] Bracco D, Berger MM, Revelly JP, et al. Segmental bioelectrical
impedance analysis to assess perioperative fluid changes. Crit Care
Med 2000;28:2390-6.
[23] Hill-rom: Total care bed system User Manual. http://www.utmb.edu/
ces/_opmanual/totalcare.pdf.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.