- Email: [email protected]
Assessment of hydration status using bioelectrical impedance vector analysis in critical patients with acute kidney injury
Accepted Manuscript Assessment of hydration status using bioelectrical impedance vector analysis in critical patients with acute kidney injury Ana Cláudia da Rosa Hise, Maria Cristina Gonzalez PII:
S0261-5614(17)30066-3
DOI:
10.1016/j.clnu.2017.02.016
Reference:
YCLNU 3064
To appear in:
Clinical Nutrition
Received Date: 20 October 2016 Revised Date:
12 February 2017
Accepted Date: 14 February 2017
Please cite this article as: Hise ACdR, Gonzalez MC, Assessment of hydration status using bioelectrical impedance vector analysis in critical patients with acute kidney injury, Clinical Nutrition (2017), doi: 10.1016/j.clnu.2017.02.016. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1 2 3
Assessment of hydration status using bioelectrical impedance vector
5
analysis in critical patients with acute kidney injury
RI PT
4
6 7
SC
8 9
Ana Cláudia da Rosa Hise1 (corresponding author – [email protected])
11
Maria Cristina Gonzalez1 ([email protected])
12
1
13
Pelotas, RS, Brazil.
14
R. Gonçalves Chaves 377, sala 411. Pelotas, RS, Brazil. CEP 960515-560
18 19
TE D
17
EP
16
Post-graduation Program in Health and Behavior – Catholic University of
AC C
15
M AN U
10
ACCEPTED MANUSCRIPT ABSTRACT
21
BACKGROUND & AIMS: The state of hyperhydration in critically ill patients
22
with acute kidney injury (AKI) is associated with increased mortality.
23
Bioelectrical impedance vector analysis (BIVA) appears to be a viable method
24
to access the fluid status of critical patients but has never been evaluated in
25
critical patients with AKI. The objective of this study is to evaluate the hydration
26
status measured using BIVA in critical patients under intensive care at the time
27
of AKI diagnosis and to correlate this measurement with mortality.
28
METHODS: We assessed the fluid status measured using BIVA in 224 critical
29
patients at the time of AKI diagnosis and correlated it with mortality. To interpret
30
the results, BIVA Software 2002 was used to plot the data from the patients
31
studied on the 95% confidence ellipses of the RXc plane for comparisons
32
between groups (non-survivors, survivors). Variables such as mechanical
33
ventilation, vasoactive drug, and sepsis, among others, were collected.
34
RESULTS: The impedance vector analysis conducted using BIVA Software
35
2002 indicated changes in the body compositions of patients according to the
36
95% confidence ellipse between the vectors R/H and Xc/H of the group of
37
survivors and the group of deceased patients. Hotelling’s test (T² = 21.2) and
38
the F test (F = 10.6) revealed significant differences (p < 0.001) between the
39
two groups. These results demonstrate that patients who died presented with a
40
greater hydration volume at the time of AKI diagnosis compared with those who
41
survived. In addition to the hydration status measured using BIVA, the following
42
were also correlated with death: diagnosis at hospitalization, APACHE II score,
43
length of hospital stay, RIFLE score, maximum organ failure, sepsis type,
44
hemoglobin, and AF.
AC C
EP
TE D
M AN U
SC
RI PT
20
ACCEPTED MANUSCRIPT CONCLUSIONS:
The
fluid
status
46
significantly demonstrated the difference in hydration between survivors and
47
non-survivors among critically ill patients with AKI.
48
Keywords: acute kidney injury; critical patient; bioelectrical impedance
49
analysis; fluid status; prognostic factor.
measured using BIVA
AC C
EP
TE D
M AN U
SC
50
assessment
RI PT
45
ACCEPTED MANUSCRIPT 51
INTRODUCTION Acute kidney injury (AKI) is related with severely ill patients in intensive
53
care units (ICUs) and has a prevalence that varies from 1.45% to 25.9%,
54
depending on the ICU and the type of patient analyzed 1. The hospital mortality
55
rate of these patients remains high despite technological advances, reaching
56
approximately 60.3% 1. The presence of AKI is considered an independent
57
prognostic factor for mortality among critical patients in the ICU 2. More recently,
58
studies have shown an association between hyperhydration status and
59
increased mortality in these patients, with fluid balance being an independent
60
predictor of mortality
61
hyperhydrated at the time of AKI diagnosis are more seriously ill and that the
62
hyperhydration status at this time is associated with low survival 4.
SC
RI PT
52
3, 4
M AN U
. Bouchard et al. found that patients who are
Different parameters with varying levels of accuracy can be used to
64
estimate the effective volume status of patients, but all have difficulty predicting
65
fluid responsiveness 5. Bioelectrical impedance vector analysis (BIVA), using
66
the resistance and reactance values obtained directly from bioelectrical
67
impedance analysis (BIA), is effective in assessing the hydration status in
68
various clinical situations
69
viable for evaluating the hydration of critically ill patients in the ICU
70
the hyperhydration status of these patients, measured using BIVA, is a predictor
71
of mortality
72
measured using BIVA of critically ill patients with AKI and the associated
73
mortality.
AC C
EP
TE D
63
6-9
. More recently, it was demonstrated that BIVA is 10
and that
11
. To date, there is no study evaluating the hydration status
ACCEPTED MANUSCRIPT 74
The objective of this study was to assess the hydration status measured
75
using BIVA in critical patients in intensive care at the time of AKI diagnosis and
76
its association with mortality.
77
METHODS A longitudinal prospective study was conducted between March and
79
September 2010 in the ICUs of the São Francisco de Paula University Hospital
80
and Santa Casa de Misericórdia de Pelotas, in the city of Pelotas, Rio Grande
81
do Sul (RS), Brazil. All the patients admitted during this period were screened
82
and all who responded to both the inclusion and exclusion criteria were included
83
in the study, resulting in a sample of 224 critically ill patients. Patients older than
84
18 years and with changes in kidney function occurring after ICU admission
85
were included. Patients with chronic AKI, dialysis-dependent chronic kidney
86
injury, AKI prior to ICU admission, kidney transplant, morbid obesity and limb
87
amputation were excluded. The study was submitted to the Ethics Committees
88
of the institutions involved, and data were collected after the signing of an
89
informed consent form by the individual responsible for the patient. The project
90
is registered in ClinicalTrials.gov as NCT02936440. All patients with normal
91
kidney function (creatinine up to 1.3 mg/dl) admitted to the ICU were monitored
92
daily using the glomerular filtration rate (GFR) calculated with the Cockcroft-
93
Gault formula
94
female), until the occurrence of a change in kidney function. In patients who had
95
no prior creatinine value and without history of renal dysfunction, the basal GFR
96
was considered to be equal to 75 ml/min/1.73 m2. The AKI diagnosis and
97
classification were made following the Risk, Injury, Failure, Loss of kidney
AC C
EP
TE D
M AN U
SC
RI PT
78
12
: 140 – age x weight (kg) / 72 x creatinine (mg/dl) x 0.85 (if
ACCEPTED MANUSCRIPT 98
function, and End-stage kidney disease (RIFLE) criteria, using the GFR on the
99
first day of kidney function change. On the first day of AKI diagnosis, the EB test was performed using a
101
Q101 tetrapolar plethysmograph (RJL Systems). The patient was fasted for at
102
least four hours and was placed in a 180° supine po sition to obtain the
103
resistance and reactance values
104
results, a graphical method also known as RXc, where the resistance and
105
reactance values normalized by height (H) (R/H and Xc/H, respectively) give
106
rise to an impedance vector and its ellipse of confidence. The hydration status
107
is inversely proportional to the size of the vector, with longer vectors indicating
108
less hydration. With the use of BIVA Software 2002, vectors were generated
109
with their respective 95% confidence ellipses in the RXc plane for comparisons
110
between groups (non-survivors X survivors) using data from the patients
111
studied. To avoid treatment bias, the doctors responsible for the fluid
112
management of the patients were unaware of the BIVA results.
13
RI PT
100
TE D
M AN U
SC
. BIVA was used for interpretation of the
Starting on the first day of AKI diagnosis, the following data were
114
collected daily: Acute Physiology and Chronic Health Evaluation II (APACHE II)
115
score, GFR (for progression time in days of AKI and kidney function recovery),
116
presence and severity of sepsis, presence of organ failure and number of
117
organs affected, use of mechanical ventilation and nephrotoxic drugs, serum
118
hemoglobin, and need for and type of dialysis. All patients were monitored daily
119
during their hospital stay, even after leaving the ICU, until hospital discharge or
120
death.
AC C
EP
113
121
The fluid balance (FB) is the total fluids administered minus the total
122
fluids eliminated over a period of time, usually 12 or 24 hours; nurses calculate
ACCEPTED MANUSCRIPT this metric in millimeters. The FB was collected daily and was classified as
124
follows: FB pre-AKI, fluid balance from ICU admission until the day of AKI
125
diagnosis; FB in AKI, fluid balance from the day of AKI diagnosis until kidney
126
function recovery or ICU discharge or death; and total FB, fluid balance from
127
ICU admission to ICU or discharge or death.
RI PT
123
The statistical analysis was conducted using Stata v. 12.0. The
129
continuous variables studied are presented as the means and standard
130
deviations, according to their normal distribution. The categorical variables are
131
presented as frequencies, and their association with death was tested using the
132
χ2 test and the linearity test for ordinal variables. For BIVA, Pearson’s linear
133
correlation coefficient (r) was first estimated between the variables R/H and
134
Xc/H of the two groups (non-survivors and survivors). The vectors generated for
135
each group were analyzed using the T2 Hotelling test and univariate analysis (F
136
test) to identify their statistical significance. The FB data were analyzed using a
137
parametric test for equality of medians. A 95% confidence interval and 5%
138
significance level were adopted for all analyses.
139
RESULTS
EP
TE D
M AN U
SC
128
The study included 224 patients, of which 88.8% were white, 57.1% were
141
male and 67.9% were elderly (≥ 60 years). The most prevalent chronic diseases
142
were systemic arterial hypertension (52.2%) and cancer (25.9%). The most
143
prevalent diagnoses during ICU stay were surgical diagnosis (39.7%), sepsis
144
(17%) and acute respiratory failure (15.6%). The death rate was 46.4%.
AC C
140
145
Table 1 shows the distribution of the clinical variables of the population
146
studied that were significantly associated with death, according to their clinical
147
progression. Surgical diagnosis during ICU stay and length of hospital stay ≥ 13
ACCEPTED MANUSCRIPT days were protective factors. Among the variables that were positively
149
associated with death, APACHE II score ≥ 15, failure classification according to
150
RIFLE in the AKI diagnosis, septic etiology of AKI, vasoactive drug use and the
151
need for dialysis increased the risk of death approximately 2-fold. The presence
152
of sepsis increased the risk 3.5-fold, with severe sepsis increasing the risk of
153
death by 51% compared with sepsis without hemodynamic compromise. The
154
need for mechanical ventilation increased the risk of death almost 4-fold. The
155
failure of three organs (in addition to the kidney) in the AKI diagnosis increased
156
the risk of death more than 5-fold, whereas the failure of three or more organs
157
during AKI increased almost 9-fold the risk of death when compared to no other
158
organ failure (except the kidneys). The other variables studied (gender, marital
159
status, age, smoking and/or drinking habit, length of ICU stay, chronic diseases,
160
use of nephrotoxic drugs, time of vasoactive drug use, mechanical ventilation,
161
dialysis and sepsis) had no associations with death.
TE D
M AN U
SC
RI PT
148
Table 2 presents the overall FB volume and the FB volume according to
163
the progression of patients. The pre-AKI FB demonstrates the accumulation of
164
liquids by patients from the time of admission to the ICU until the diagnosis of
165
AKI. This volume represents the hydration status of patients, in numerical
166
terms, at the time of the AKI diagnosis. Of the 224 patients, 140 presented AKI
167
diagnoses on the first day of ICU admission, and the pre-AKI FB was
168
disregarded, as there was no prior control of FB outside of the ICU. Assessing
169
the 84 patients who presented AKI diagnoses at least 24 hours after admission
170
to the ICU (presenting, therefore, a calculated pre-AKI FB), the non-survivors
171
group ( n = 37) had a pre-AKI FB volume that was significantly greater than that
172
of the survivors group (n = 47). Larger FB volumes (pre-AKI, during AKI, and
AC C
EP
162
ACCEPTED MANUSCRIPT 173
total) were significantly (p < 0.001) associated with the group of patients who
174
died. The impedance vector analysis performed in BIVA Software 2002
176
indicated changes in the hydration of patients according to clinical progression.
177
Table 3 shows the means and standard deviations for R/H for the non-survivors
178
and survivors groups (220.5 ± 74.6 and 246.6 ± 69.6, respectively), indicating
179
that the non-survivors group had greater hydration (reflected by significantly
180
smaller R/H values). Figure 1 presents the confidence ellipses between the R/H
181
and Xc/H vectors for the survivors group (black vector) and the non-survivors
182
group (dotted vector). The Hotelling’s and F tests showed significant differences
183
(p < 0.001) between the two groups. These results demonstrate that patients
184
who died had greater hydration volume (reflected by shorter vectors and smaller
185
R/H values) at the time of AKI diagnosis than patients who survived (time when
186
the EB was performed).
TE D
M AN U
SC
RI PT
175
Figure 2 shows the comparison of the confidence ellipses of the studied
188
patients divided into four groups: group 1 (patients diagnosed with AKI on the
189
first day of ICU admission who survived, n = 73), group 2 (patients diagnosed
190
with AKI at least 24 hours after the first day of ICU admission who survived, n =
191
47), group 3 (patients diagnosed with AKI on the first day of ICU admission who
192
died, n = 67), and group 4 (patients diagnosed with AKI at least 24 hours after
193
the first day of ICU admission who died, n = 37). Table 4 shows the mean R/H
194
values of each group and the comparisons between them.
195
DISCUSSION
AC C
EP
187
196
In critical patients, proper fluid management is essential for the treatment
197
of conditions such as shock, sepsis, heart failure and AKI. Fluid management is
ACCEPTED MANUSCRIPT 198
typically performed based on the estimated intravascular volume. Over the last
199
decades, several studies have proposed guidelines for fluid management in
200
critical patients. One of these studies demonstrated that an aggressive
201
premature resuscitation strategy including proper administration of fluids could
202
improve outcomes in patients with sepsis
203
excessive fluid accumulation may not be simply considered a consequence of
204
fluid resuscitation or severe AKI but possibly a mediator of adverse results
205
15-20
206
significant relationships with negative outcomes, including increased mortality
207
and reduced kidney function recovery rate
208
50% of hemodynamically unstable patients in the ICU respond to fluid
209
replacement, resulting in unnecessary fluid retention 22.
14
RI PT
. However, other studies found that
3, 4,
4, 21
SC
. In relation to AKI, several studies have shown that hyperhydration has
M AN U
. In addition, only approximately
In our study, we used BIVA to demonstrate a significant difference in the
211
hydration status between patients who died and who survived, where the non-
212
survivors showed greater hydration volume than the survivors. Similar findings
213
were reported by Payen et al. 3, who analyzed 1120 patients with AKI in a
214
multicenter observational cohort study of 198 ICUs. The mortality rates at 60
215
days were 36% in patients with AKI and 16% in patients without AKI (p <
216
0.001), and mean fluid balance was one of the independent risk factors for
217
mortality (0.98 ± 1.5 l/24 h versus 0.15 ± 1.06 l/24 h, p < 0.001, for non-
218
survivors and survivors, respectively).
AC C
EP
TE D
210
219
In our study, we assessed the hydration status at the time of AKI
220
diagnosis and demonstrated that greater hydration present at that time is
221
associated with lower survival. This early assessment would enable fluid
222
management control and better treatment results. Our study revealed variables
ACCEPTED MANUSCRIPT 223
correlated with death (mechanical ventilation, organ failure, septic shock,
224
vasoactive drug, among others) that are expected in patients with this level of
225
disease severity, demonstrating that our study population is in accordance with
226
what other studies have reported. Bouchard et al.
227
between hyperhydration and increased mortality in patients with AKI and also
228
found that patients with hyperhydration at the time of AKI diagnosis were more
229
critically ill than those with a lesser degree of fluid accumulation, indicated by
230
greater frequencies of respiratory failure, mechanical ventilation, and sepsis,
231
even after adjusting for all variables.
established the association
SC
RI PT
4
Although several studies point to the importance of hyperhydration in the
233
mortality of patients with AKI, we must not forget the fundamental role of the
234
kidney in the homeostasis of body fluids. Kidney dysfunction requires special
235
attention in relation to fluid replacement and hyperhydration, as fluid
236
replacement is the only effective treatment in patients at risk of developing AKI.
237
In addition, hypovolemia is the strongest causal factor of morbidity during
238
dialysis and can contribute to increase kidney damage
239
importance of fluid management, we know that the choice of method for
240
measuring fluid status is controversial. Different parameters with varying levels
241
of accuracy can be used to estimate the actual volume status of patients 24. Two
242
parameters are currently commonly used in intensive care: central venous
243
pressure (CVP) and fluid balance (FB) recorded by nurses. Regarding CVP,
244
several studies have shown its ineffectiveness in predicting the response to
245
volume replacement
246
conventional FB is not predictive of the current changes in the weights of
247
critically ill patients and that, consequently, it reflects fluctuations in total body
23
. Regarding the
AC C
EP
TE D
M AN U
232
5
. Regarding FB, Perren et al.
25
concluded that
ACCEPTED MANUSCRIPT 248
fluid inadequately. Nevertheless, FB is still the most viable method in the ICU
249
that offers a quantitative parameter of the hydration status of patients.
250
In our study, we used BIVA to measure the fluid state of patients, a
251
method that has been valid for accessing the hydration status in various clinical
252
situations
253
to assess the hydration status, also positively correlate it with mortality and
254
hospital readmission
255
body composition calculations are not possible due to changes in the hydration
256
status 33.
and recommend that BIVA be applied in cases where
As in our study, Basso et al.
11
SC
9, 32
RI PT
. Some studies, in addition to demonstrating that BIVA is able
and Chen et al.
34
M AN U
257
6, 7, 26-31
recently demonstrated
258
that hyperhydration measured by BIVA is a significant predictor of mortality in
259
critical patients.
Our choice for this mode of BIVA evaluation (i.e., the confidence ellipse)
261
is based on studies that showed not only that hyperhydration increases the
262
mortality of patients with AKI but also that hyperhydration at the time of AKI
263
diagnosis is associated with low survival
264
prospective observational study in 61 patients with serial measurements of
265
BIVA and found a good association between the hydration status classifications
266
obtained using BIVA and FB. They concluded that BIVA could serve as an
267
additional method for measuring the hydration status in critical patients,
268
providing important information for fluid management in these patients. Similar
269
to Jones et al., our study found an association between hydration status and
270
death, despite having used another methodology.
3, 4
. Jones et al.
10
conducted a
AC C
EP
TE D
260
271
BIVA by itself demonstrates the statistical significance through Hotelling’s
272
test. However, to confirm the results found using BIVA, we compared the results
ACCEPTED MANUSCRIPT 273
with the cumulative FB; the most commonly used current method for
274
quantitatively measuring the hydration of critical patients. BIVA was compatible
275
with the FB cumulative results because patients who died were more hydrated
276
than patients who survived, which is compatible with the cumulative FB. In patients with an AKI diagnosis on the first day of admission to the ICU,
278
the BIVA showed no significant difference in the hydration status of non-
279
survivors and survivors. As demonstrated by Piccoli et al.
280
BIVA is not able to detect variations in volume below 2,000 ml, suggesting that
281
the difference in balance between these two groups was less than this volume.
RI PT
277
and Jones et al.
10
,
SC
30
BIVA demonstrated a significant difference between patients who died
283
after 24 hours of admission to the ICU. This finding may suggest that
284
hyperhydration is a determining factor for death in critical patients with AKI and
285
that fluid management in the ICU is of fundamental importance, as it may
286
influence the outcomes of these patients, as also evidenced by other studies
287
4
288
in the ICU could identify greater hydration in patients as a possible risk factor for
289
death. In patients who present AKI in the first 24 hours of ICU admission, the
290
performance of BIVA at least 24 hours after the AKI diagnosis could also assist
291
in identifying greater hydration.
TE D
3,
EP
. Therefore, BIVA at the time of AKI diagnosis developed during hospitalization
AC C
292
M AN U
282
Some aspects of this study strengthen the reliability of the results: the
293
doctors who managed the patients had no access to the results of the BIVA,
294
which prevented treatment bias; both groups of patients had different clinical
295
severity characteristics, which are expected in patients with different hydration
296
statuses; and the BIVA results were compatible with the FB results.
ACCEPTED MANUSCRIPT Therefore, this study is pioneering in assessing the hydration status
298
measured using BIVA in critically ill patients with AKI. This study demonstrated
299
that BIVA could be a useful instrument in assessing the hydration of patients
300
who develop AKI in the ICU; however, further studies are needed to confirm our
301
findings.
302
Statement of autorship: ACRH and MCG participated in the conception and
303
design of the study, analysis and interpretation of the data, drafting of the article
304
and final approval of the version of the sutdy. ACRH was responsible for the
305
data collection.
306
Conflict of interest and founding sources: The authors have no conflict of
307
interests to declare. This research did not receive any specific grant from
308
funding agencies in the public, comercial, or not-for-profit sectors.
M AN U
SC
RI PT
297
AC C
EP
TE D
309
ACCEPTED MANUSCRIPT 310
References
311
1.
312
patients: a multinational, multicenter study. JAMA. 2005; 294: 813-8.
313
2.
314
CA. Acute kidney injury criteria predict outcomes of critically ill patients. Crit
315
Care Med. 2008; 36: 1397-403.
316
3.
317
positive fluid balance is associated with a worse outcome in patients with acute
318
renal failure. Crit Care. 2008; 12: R74.
319
4.
320
and recovery of kidney function in critically ill patients with acute kidney injury.
321
Kidney Int. 2009; 76: 422-7.
322
5.
323
fluid responsiveness? A systematic review of the literature and the tale of seven
324
mares. Chest. 2008; 134: 172-8.
325
6.
326
patients with different hydration status. Kidney Int. 2004; 65: 1050-63.
327
7.
328
al. Body fluid volume and nutritional status in hemodialysis: vector bioelectric
329
impedance analysis. Clin Nephrol. 2010; 73: 300-8.
330
8.
331
vectorial analysis of fluid status and prediction of contrast-induced acute kidney
332
injury. J Am Coll Cardiol. 2014; 63: 1387-94.
Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill
RI PT
Barrantes F, Tian J, Vazquez R, Amoateng-Adjepong Y and Manthous
SC
Payen D, de Pont AC, Sakr Y, Spies C, Reinhart K and Vincent JL. A
M AN U
Bouchard J, Soroko SB, Chertow GM, et al. Fluid accumulation, survival
TE D
Marik PE, Baram M and Vahid B. Does central venous pressure predict
EP
Piccoli A. Bioelectric impedance vector distribution in peritoneal dialysis
AC C
Espinosa Cuevas MA, Navarrete Rodriguez G, Villeda Martinez ME, et
Maioli M, Toso A, Leoncini M, et al. Pre-procedural bioimpedance
ACCEPTED MANUSCRIPT 333
9.
Nunez J, Mascarell B, Stubbe H, et al. Bioelectrical impedance vector
334
analysis and clinical outcomes in patients with acute heart failure. J Cardiovasc
335
Med (Hagerstown). 2014.
336
10.
337
analysis in critically ill patients: a prospective, clinician-blinded investigation. Crit
338
Care. 2015; 19: 290.
339
11.
340
care unit: bioelectrical impedance vector analysis as a tool to assess hydration
341
status and optimal fluid balance in critically ill patients. Blood Purif. 2013; 36:
342
192-9.
343
12.
344
serum creatinine. Nephron. 1976; 16: 31-41.
345
13.
346
monitoring body fluid variation by bioimpedance analysis: the RXc graph.
347
Kidney Int. 1994; 46: 534-9.
348
14.
349
treatment of severe sepsis and septic shock. N Engl J Med. 2001; 345: 1368-
350
77.
351
15.
352
fluid-management strategies in acute lung injury. N Engl J Med. 2006; 354:
353
2564-75.
354
16.
355
intravenous fluid restriction on postoperative complications: comparison of two
356
perioperative fluid regimens: a randomized assessor-blinded multicenter trial.
357
Ann Surg. 2003; 238: 641-8.
RI PT
Jones SL, Tanaka A, Eastwood GM, et al. Bioelectrical impedance vector
M AN U
SC
Basso F, Berdin G, Virzi GM, et al. Fluid management in the intensive
Cockcroft DW and Gault MH. Prediction of creatinine clearance from
TE D
Piccoli A, Rossi B, Pillon L and Bucciante G. A new method for
EP
Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the
AC C
Wiedemann HP, Wheeler AP, Bernard GR, et al. Comparison of two
Brandstrup B, Tonnesen H, Beier-Holgersen R, et al. Effects of
ACCEPTED MANUSCRIPT 358
17.
Foland JA, Fortenberry JD, Warshaw BL, et al. Fluid overload before
359
continuous hemofiltration and survival in critically ill children: a retrospective
360
analysis. Crit Care Med. 2004; 32: 1771-6.
361
18.
362
of replacement fluid on survival in hemofiltration. Pediatr Nephrol. 2004; 19:
363
1394-9.
364
19.
365
Outcome
366
Pediatrics. 2001; 107: 1309-12.
367
20.
368
organ dysfunction syndrome receiving continuous renal replacement therapy.
369
Kidney Int. 2005; 67: 653-8.
370
21.
371
Fluid overload at initiation of renal replacement therapy is associated with lack
372
of renal recovery in patients with acute kidney injury. Nephrol Dial Transplant.
373
2012; 27: 956-61.
374
22.
375
arterial waveform derived variables and fluid responsiveness in mechanically
376
ventilated patients: a systematic review of the literature. Crit Care Med. 2009;
377
37: 2642-7.
378
23.
379
achieving fluid balance in acute renal failure. Curr Opin Crit Care. 2002; 8: 535-
380
43.
381
24.
382
kidney injury: emerging concepts. Nephron Clin Pract. 2013; 123: 238-45.
RI PT
Gillespie RS, Seidel K and Symons JM. Effect of fluid overload and dose
Goldstein SL, Currier H, Graf C, Cosio CC, Brewer ED and Sachdeva R. children
receiving
continuous
venovenous
hemofiltration.
SC
in
M AN U
Goldstein SL, Somers MJ, Baum MA, et al. Pediatric patients with multi-
TE D
Heung M, Wolfgram DF, Kommareddi M, Hu Y, Song PX and Ojo AO.
AC C
EP
Marik PE, Cavallazzi R, Vasu T and Hirani A. Dynamic changes in
Mehta RL, Clark WC and Schetz M. Techniques for assessing and
Godin M, Bouchard J and Mehta RL. Fluid balance in patients with acute
ACCEPTED MANUSCRIPT 383
25.
Perren A, Markmann M, Merlani G, Marone C and Merlani P. Fluid
384
balance in critically ill patients. Should we really rely on it? Minerva Anestesiol.
385
2011; 77: 802-11.
386
26.
387
assessment. Contrib Nephrol. 2010; 164: 143-52.
388
27.
389
assessment and management in the emergency department. Contrib Nephrol.
390
2010; 164: 227-36.
391
28.
392
emergency department: a look at B-type natriuretic peptide and bioimpedance
393
vector analysis. Contrib Nephrol. 2011; 171: 187-93.
394
29.
395
biomarkers and bio-impedance in evaluating hydration status in patients with
396
acute heart failure. Clin Chem Lab Med. 2012; 50: 2093-105.
397
30.
398
of nutrition and hydration in dialysis patients with bioelectrical impedance vector
399
analysis (BIVA). Clin Nutr. 2014; 33: 673-7.
400
31.
401
and Facila L. Prognostic value of analysing the bioimpedance vector for patients
402
hospitalised for acute decompensated heart failure: A validation cohort. Rev
403
Clin Esp. 2016.
404
32.
405
galectin-3 and BIVA assessments in predicting short- and long-term events in
406
patients admitted for acute heart failure. Biomed Res Int. 2014; 2014: 983098.
A.
Bioelectric
impedance
measurement
for
fluid
status
RI PT
Piccoli
SC
Di Somma S, Gori CS, Grandi T, Risicato MG and Salvatori E. Fluid
M AN U
Tuy T and Peacock F. Noninvasive volume assessment in the
TE D
Di Somma S, Navarin S, Giordano S, et al. The emerging role of
EP
Piccoli A, Codognotto M, Piasentin P and Naso A. Combined evaluation
AC C
Trejo-Velasco B, Fabregat-Andres O, Montagud V, Morell S, Nunez J
De Berardinis B, Magrini L, Zampini G, et al. Usefulness of combining
ACCEPTED MANUSCRIPT 407
33.
Norman K, Stobaus N, Pirlich M and Bosy-Westphal A. Bioelectrical
408
phase angle and impedance vector analysis--clinical relevance and applicability
409
of impedance parameters. Clin Nutr. 2012; 31: 854-61.
410
34.
411
renal replacement therapy is associated with poorer clinical condition and
412
outcome: a prospective observational study on the combined use of
413
bioimpedance vector analysis and serum N-terminal pro-B-type natriuretic
414
peptide measurement. Crit Care. 2015; 19: 135.
AC C
EP
TE D
M AN U
SC
RI PT
Chen H, Wu B, Gong D and Liu Z. Fluid overload at start of continuous
ACCEPTED MANUSCRIPT Figure and Table legends:
416
Figure 1: Comparison of the confidence ellipses using bioelectrical vector
417
analysis (n = 224), according to the outcome (survivors and non-survivors).
418
Figure 2: Comparison of the confidence ellipses using bioelectrical vector
419
analysis (n = 224), according to the outcome (survivors and no-survivors) and
420
the moment of acute kidney injury diagnosis (1st day of intensive care unit
421
admission or after 1st day of intensive care unit admission).
AC C
EP
TE D
M AN U
SC
RI PT
415
ACCEPTED MANUSCRIPT 422
Table 1. Clinical variables distribution according to patients’ outcome. Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
RR (CI 95%)
p value
a
≤ 15
89 (71.2)
36 (28.8)
1.00 (ref)
> 15
31 (31.1)
68 (68.7)
No
89 (58.5)
63 (41.4)
Yes
31 (43.1)
41 (56.9)
RI PT
APACHE II
Total length of
< 13 days
52 (45.6)
62 (54.4)
> 13 days
68 (61.8)
42 (38.2)
57 (42.2)
78 (57.8)
TE D
M AN U
hospitalization
63 (70.8)
26 (29.2)
107 (57.5)
79 (42.5)
Surgical diagnosis at admission
Sepsis
No
EP
No
Yes
2.38 (1.76;3.24)
1.00 (ref)
1.37 (1.04;1.81)
1.00 (ref)
1.00 (ref)
1.00 (ref)
Risk
80 (67.8)
38 (32.8)
Injury
34 (38.2)
55 (60.4)
1.84 (1.35;2.51)
Failure
6 (35.3)
11 (64.7)
1.98 (1.28;3.06)
AC C
<0.001
0.51 (0.35;0.72)
25 (65.8)
RIFLE
0.02
0.70 (0.52;0.94)
13 (34.2)
Yes
0.03
SC
Direct admission in ICU
<0.001
0.01
1.55 (1.16;2.06)
1.00 (ref)
b
<0.001
Etiology
423
a
Ischemic
83 (65.3)
44 (34.6)
Others
18 (58.1)
14 (42.4)
1.22 (0.77; 1.95)
Septic
18 (28.1)
46 (71.9)
2.07 (1.56; 2.76)
χ2 test. b Linear trend for χ2 test.
1.00 (ref)
<0.001
ACCEPTED MANUSCRIPT 424
Table 1. Clinical variables distribution according to patients’ outcome (cont.). Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
≤ 3 days
91 (64.0)
51 (35.9)
> 3 days
29 (35.4)
53 (64.6)
None
78 (83.9)
15 (16.1)
<3
39 (39.4)
60 (60.6)
≥3
3 (9.4)
29 (90.6)
RR (CI 95%)
p value
1.00 (ref)
<0.001
a
AKI duration
M AN U
Maximum number of organ failure
RI PT
b
1.00 (ref)
3.76 (2.30;6.13)
5.62 (3.49;9.05)
None
70 (89.7)
8 (10.3)
<3
46(46.9)
52 (53.1)
5.17 (2.61;10.24)
≥3
4 (8.3)
44 (91.7)
8.94 (4.61;17.33)
Hemoglobin level (minimum)
86 (60.6)
TE D
≥9mg/dl <9mg/dl
Sepsis
56 (39.4)
34 (41.5)
48 (58.5)
86 (79.6)
22 (29.6)
EP
Yes
1.00 (ref)
1.00 (ref)
1.00 (ref)
82 (70.7)
Sepsis
23 (45.1)
28 (54.9)
Severe sepsis
11 (16.9)
54 (83.1)
1.51 (1.15;1.96)
No
34 (20.4)
82 (70.7)
3.47 (2.35;5.13)
No
94 (72.3)
36 (27.7)
Yes
26 (27.6)
68 (72.3)
AC C
Type of sepsis
<0.001
c
0.008
1.48 (1.13;1.95)
34 (20.4)
No
c
<0.001
SC
Number of organ failure
1.80 (1.37;2.36)
<0.001
3.47 (2.35;5.13)
1.00 (ref)
0.002
Use of vasoactive drugs
425
a
χ2 test. b at AKI diagnosis. c Linear trend for χ2 test.
1.00 (ref) 2.61 (1.93;3.54)
< 0.001
ACCEPTED MANUSCRIPT 426
Table 1. Clinical variables distribution according to patients’ outcome (cont.). Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
No
88 (81.5)
20 (18.5)
Yes
32 (27.6)
84 (72.4)
No
118 (57.2)
88 (42.7)
Yes
2 (11.1)
16 (88.9)
RR (CI 95%)
p value
1.00 (ref)
<0.001
a
Mechanic ventilation
1.00 (ref)
2.08 (1.66;2.61)
TE D
M AN U
SC
χ2 test.
EP
a
AC C
427
RI PT
Dialysis
3.91 (2.59;5.9)
< 0.001
ACCEPTED MANUSCRIPT 428
Table 2. Fluid balance volume in milliliters (median and interquartile range), before, during
429
acute kidney injury and total, according to patients’ outcome.
FB before AKId
Non-survivors
0 (0;1064.5)
0 (0; 5340)
a
Survivorsb
pc
0 (0;0)
< 0.001
4135 (550.5;10204.5) 6906.5 (3016;16725) 1557.5 (97;6408) < 0.001
Total FBf
4483 (571.5;11051.5) 7565.5 (3243;19551) 1557.5 (55;7125) < 0.001
RI PT
FB during AKIe
a
Non-survivors: n = 37, for FB before AKI, and n = 104, for FB during AKI and total FB. Survivors: n = 47, for FB before AKI, and n = 120, for FB during AKI and total FB. c p value for median comparison between non-survivor and survivor’s fluid balance (MannWhitney test). d FB before AKI: fluid balance since ICU admission until AKI diagnosis. N = 84 (140 patients had AKI in the 1st day, and for this reason, they do not have this FB). e FB during AKI: fluid balance since AKI diagnosis until renal recovery, ICU discharge or death. N = 224 f Total FB total: fluid balance since ICU admission until ICU discharge or death. N = 22
EP
TE D
M AN U
SC
b
AC C
430 431 432 433 434 435 436 437 438
All
ACCEPTED MANUSCRIPT 439
Table 3. Resistance and reactance values, adjusted for height (R/H and Xc/H,
440
respectively), according to patient´s outcome. R/H
Xc/H
(mean ± standard deviation) (mean ± standard deviation) 20.5 ± 7.5a
Non-survivors
220.5 ± 74.6
15.8 ± 7.7
TE D
M AN U
SC
p < 0.05
EP
a
RI PT
246.6 ± 69.6a
AC C
441
Survivors
ACCEPTED MANUSCRIPT 442
Table 4. P values of the comparison of resistance mean values adjusted for height
443
(R/H) of patients, grouped according to the timing of acute kidney injury (AKI)
444
diagnosis and outcome.
Group 2b (R/H = 243.5)
0.05
Group 3c (R/H = 233.8)
Group 4d (R/H = 195.8) <0.001 <0.001 0.007
Group 1: Survivors with AKI in 1st day of Intensive Care Unit (n = 73). bGroup 2: Survivors with AKI after 1st day of Intensive Care Unit (n = 47). cGroup 3: Non-survivors with AKI in 1st day of Intensive Care Unit (n = 67). dGrupo 4: Non-survivors with AKI after 1st day of Intensive Care Unit (n = 37).
EP
TE D
M AN U
SC
a
AC C
445 446 447 448
Group 3c (R/H = 233.8) 0.12
RI PT
Group 1a (R/H = 249.7)
Group 2b (R/H = 243.5) 0.72
ACCEPTED MANUSCRIPT 449
Figure 1
EP AC C
451
TE D
M AN U
SC
RI PT
450
ACCEPTED MANUSCRIPT 452
Figure 2
M AN U
SC
RI PT
453
454
AC C
EP
TE D
455
S0261-5614(17)30066-3
DOI:
10.1016/j.clnu.2017.02.016
Reference:
YCLNU 3064
To appear in:
Clinical Nutrition
Received Date: 20 October 2016 Revised Date:
12 February 2017
Accepted Date: 14 February 2017
Please cite this article as: Hise ACdR, Gonzalez MC, Assessment of hydration status using bioelectrical impedance vector analysis in critical patients with acute kidney injury, Clinical Nutrition (2017), doi: 10.1016/j.clnu.2017.02.016. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1 2 3
Assessment of hydration status using bioelectrical impedance vector
5
analysis in critical patients with acute kidney injury
RI PT
4
6 7
SC
8 9
Ana Cláudia da Rosa Hise1 (corresponding author – [email protected])
11
Maria Cristina Gonzalez1 ([email protected])
12
1
13
Pelotas, RS, Brazil.
14
R. Gonçalves Chaves 377, sala 411. Pelotas, RS, Brazil. CEP 960515-560
18 19
TE D
17
EP
16
Post-graduation Program in Health and Behavior – Catholic University of
AC C
15
M AN U
10
ACCEPTED MANUSCRIPT ABSTRACT
21
BACKGROUND & AIMS: The state of hyperhydration in critically ill patients
22
with acute kidney injury (AKI) is associated with increased mortality.
23
Bioelectrical impedance vector analysis (BIVA) appears to be a viable method
24
to access the fluid status of critical patients but has never been evaluated in
25
critical patients with AKI. The objective of this study is to evaluate the hydration
26
status measured using BIVA in critical patients under intensive care at the time
27
of AKI diagnosis and to correlate this measurement with mortality.
28
METHODS: We assessed the fluid status measured using BIVA in 224 critical
29
patients at the time of AKI diagnosis and correlated it with mortality. To interpret
30
the results, BIVA Software 2002 was used to plot the data from the patients
31
studied on the 95% confidence ellipses of the RXc plane for comparisons
32
between groups (non-survivors, survivors). Variables such as mechanical
33
ventilation, vasoactive drug, and sepsis, among others, were collected.
34
RESULTS: The impedance vector analysis conducted using BIVA Software
35
2002 indicated changes in the body compositions of patients according to the
36
95% confidence ellipse between the vectors R/H and Xc/H of the group of
37
survivors and the group of deceased patients. Hotelling’s test (T² = 21.2) and
38
the F test (F = 10.6) revealed significant differences (p < 0.001) between the
39
two groups. These results demonstrate that patients who died presented with a
40
greater hydration volume at the time of AKI diagnosis compared with those who
41
survived. In addition to the hydration status measured using BIVA, the following
42
were also correlated with death: diagnosis at hospitalization, APACHE II score,
43
length of hospital stay, RIFLE score, maximum organ failure, sepsis type,
44
hemoglobin, and AF.
AC C
EP
TE D
M AN U
SC
RI PT
20
ACCEPTED MANUSCRIPT CONCLUSIONS:
The
fluid
status
46
significantly demonstrated the difference in hydration between survivors and
47
non-survivors among critically ill patients with AKI.
48
Keywords: acute kidney injury; critical patient; bioelectrical impedance
49
analysis; fluid status; prognostic factor.
measured using BIVA
AC C
EP
TE D
M AN U
SC
50
assessment
RI PT
45
ACCEPTED MANUSCRIPT 51
INTRODUCTION Acute kidney injury (AKI) is related with severely ill patients in intensive
53
care units (ICUs) and has a prevalence that varies from 1.45% to 25.9%,
54
depending on the ICU and the type of patient analyzed 1. The hospital mortality
55
rate of these patients remains high despite technological advances, reaching
56
approximately 60.3% 1. The presence of AKI is considered an independent
57
prognostic factor for mortality among critical patients in the ICU 2. More recently,
58
studies have shown an association between hyperhydration status and
59
increased mortality in these patients, with fluid balance being an independent
60
predictor of mortality
61
hyperhydrated at the time of AKI diagnosis are more seriously ill and that the
62
hyperhydration status at this time is associated with low survival 4.
SC
RI PT
52
3, 4
M AN U
. Bouchard et al. found that patients who are
Different parameters with varying levels of accuracy can be used to
64
estimate the effective volume status of patients, but all have difficulty predicting
65
fluid responsiveness 5. Bioelectrical impedance vector analysis (BIVA), using
66
the resistance and reactance values obtained directly from bioelectrical
67
impedance analysis (BIA), is effective in assessing the hydration status in
68
various clinical situations
69
viable for evaluating the hydration of critically ill patients in the ICU
70
the hyperhydration status of these patients, measured using BIVA, is a predictor
71
of mortality
72
measured using BIVA of critically ill patients with AKI and the associated
73
mortality.
AC C
EP
TE D
63
6-9
. More recently, it was demonstrated that BIVA is 10
and that
11
. To date, there is no study evaluating the hydration status
ACCEPTED MANUSCRIPT 74
The objective of this study was to assess the hydration status measured
75
using BIVA in critical patients in intensive care at the time of AKI diagnosis and
76
its association with mortality.
77
METHODS A longitudinal prospective study was conducted between March and
79
September 2010 in the ICUs of the São Francisco de Paula University Hospital
80
and Santa Casa de Misericórdia de Pelotas, in the city of Pelotas, Rio Grande
81
do Sul (RS), Brazil. All the patients admitted during this period were screened
82
and all who responded to both the inclusion and exclusion criteria were included
83
in the study, resulting in a sample of 224 critically ill patients. Patients older than
84
18 years and with changes in kidney function occurring after ICU admission
85
were included. Patients with chronic AKI, dialysis-dependent chronic kidney
86
injury, AKI prior to ICU admission, kidney transplant, morbid obesity and limb
87
amputation were excluded. The study was submitted to the Ethics Committees
88
of the institutions involved, and data were collected after the signing of an
89
informed consent form by the individual responsible for the patient. The project
90
is registered in ClinicalTrials.gov as NCT02936440. All patients with normal
91
kidney function (creatinine up to 1.3 mg/dl) admitted to the ICU were monitored
92
daily using the glomerular filtration rate (GFR) calculated with the Cockcroft-
93
Gault formula
94
female), until the occurrence of a change in kidney function. In patients who had
95
no prior creatinine value and without history of renal dysfunction, the basal GFR
96
was considered to be equal to 75 ml/min/1.73 m2. The AKI diagnosis and
97
classification were made following the Risk, Injury, Failure, Loss of kidney
AC C
EP
TE D
M AN U
SC
RI PT
78
12
: 140 – age x weight (kg) / 72 x creatinine (mg/dl) x 0.85 (if
ACCEPTED MANUSCRIPT 98
function, and End-stage kidney disease (RIFLE) criteria, using the GFR on the
99
first day of kidney function change. On the first day of AKI diagnosis, the EB test was performed using a
101
Q101 tetrapolar plethysmograph (RJL Systems). The patient was fasted for at
102
least four hours and was placed in a 180° supine po sition to obtain the
103
resistance and reactance values
104
results, a graphical method also known as RXc, where the resistance and
105
reactance values normalized by height (H) (R/H and Xc/H, respectively) give
106
rise to an impedance vector and its ellipse of confidence. The hydration status
107
is inversely proportional to the size of the vector, with longer vectors indicating
108
less hydration. With the use of BIVA Software 2002, vectors were generated
109
with their respective 95% confidence ellipses in the RXc plane for comparisons
110
between groups (non-survivors X survivors) using data from the patients
111
studied. To avoid treatment bias, the doctors responsible for the fluid
112
management of the patients were unaware of the BIVA results.
13
RI PT
100
TE D
M AN U
SC
. BIVA was used for interpretation of the
Starting on the first day of AKI diagnosis, the following data were
114
collected daily: Acute Physiology and Chronic Health Evaluation II (APACHE II)
115
score, GFR (for progression time in days of AKI and kidney function recovery),
116
presence and severity of sepsis, presence of organ failure and number of
117
organs affected, use of mechanical ventilation and nephrotoxic drugs, serum
118
hemoglobin, and need for and type of dialysis. All patients were monitored daily
119
during their hospital stay, even after leaving the ICU, until hospital discharge or
120
death.
AC C
EP
113
121
The fluid balance (FB) is the total fluids administered minus the total
122
fluids eliminated over a period of time, usually 12 or 24 hours; nurses calculate
ACCEPTED MANUSCRIPT this metric in millimeters. The FB was collected daily and was classified as
124
follows: FB pre-AKI, fluid balance from ICU admission until the day of AKI
125
diagnosis; FB in AKI, fluid balance from the day of AKI diagnosis until kidney
126
function recovery or ICU discharge or death; and total FB, fluid balance from
127
ICU admission to ICU or discharge or death.
RI PT
123
The statistical analysis was conducted using Stata v. 12.0. The
129
continuous variables studied are presented as the means and standard
130
deviations, according to their normal distribution. The categorical variables are
131
presented as frequencies, and their association with death was tested using the
132
χ2 test and the linearity test for ordinal variables. For BIVA, Pearson’s linear
133
correlation coefficient (r) was first estimated between the variables R/H and
134
Xc/H of the two groups (non-survivors and survivors). The vectors generated for
135
each group were analyzed using the T2 Hotelling test and univariate analysis (F
136
test) to identify their statistical significance. The FB data were analyzed using a
137
parametric test for equality of medians. A 95% confidence interval and 5%
138
significance level were adopted for all analyses.
139
RESULTS
EP
TE D
M AN U
SC
128
The study included 224 patients, of which 88.8% were white, 57.1% were
141
male and 67.9% were elderly (≥ 60 years). The most prevalent chronic diseases
142
were systemic arterial hypertension (52.2%) and cancer (25.9%). The most
143
prevalent diagnoses during ICU stay were surgical diagnosis (39.7%), sepsis
144
(17%) and acute respiratory failure (15.6%). The death rate was 46.4%.
AC C
140
145
Table 1 shows the distribution of the clinical variables of the population
146
studied that were significantly associated with death, according to their clinical
147
progression. Surgical diagnosis during ICU stay and length of hospital stay ≥ 13
ACCEPTED MANUSCRIPT days were protective factors. Among the variables that were positively
149
associated with death, APACHE II score ≥ 15, failure classification according to
150
RIFLE in the AKI diagnosis, septic etiology of AKI, vasoactive drug use and the
151
need for dialysis increased the risk of death approximately 2-fold. The presence
152
of sepsis increased the risk 3.5-fold, with severe sepsis increasing the risk of
153
death by 51% compared with sepsis without hemodynamic compromise. The
154
need for mechanical ventilation increased the risk of death almost 4-fold. The
155
failure of three organs (in addition to the kidney) in the AKI diagnosis increased
156
the risk of death more than 5-fold, whereas the failure of three or more organs
157
during AKI increased almost 9-fold the risk of death when compared to no other
158
organ failure (except the kidneys). The other variables studied (gender, marital
159
status, age, smoking and/or drinking habit, length of ICU stay, chronic diseases,
160
use of nephrotoxic drugs, time of vasoactive drug use, mechanical ventilation,
161
dialysis and sepsis) had no associations with death.
TE D
M AN U
SC
RI PT
148
Table 2 presents the overall FB volume and the FB volume according to
163
the progression of patients. The pre-AKI FB demonstrates the accumulation of
164
liquids by patients from the time of admission to the ICU until the diagnosis of
165
AKI. This volume represents the hydration status of patients, in numerical
166
terms, at the time of the AKI diagnosis. Of the 224 patients, 140 presented AKI
167
diagnoses on the first day of ICU admission, and the pre-AKI FB was
168
disregarded, as there was no prior control of FB outside of the ICU. Assessing
169
the 84 patients who presented AKI diagnoses at least 24 hours after admission
170
to the ICU (presenting, therefore, a calculated pre-AKI FB), the non-survivors
171
group ( n = 37) had a pre-AKI FB volume that was significantly greater than that
172
of the survivors group (n = 47). Larger FB volumes (pre-AKI, during AKI, and
AC C
EP
162
ACCEPTED MANUSCRIPT 173
total) were significantly (p < 0.001) associated with the group of patients who
174
died. The impedance vector analysis performed in BIVA Software 2002
176
indicated changes in the hydration of patients according to clinical progression.
177
Table 3 shows the means and standard deviations for R/H for the non-survivors
178
and survivors groups (220.5 ± 74.6 and 246.6 ± 69.6, respectively), indicating
179
that the non-survivors group had greater hydration (reflected by significantly
180
smaller R/H values). Figure 1 presents the confidence ellipses between the R/H
181
and Xc/H vectors for the survivors group (black vector) and the non-survivors
182
group (dotted vector). The Hotelling’s and F tests showed significant differences
183
(p < 0.001) between the two groups. These results demonstrate that patients
184
who died had greater hydration volume (reflected by shorter vectors and smaller
185
R/H values) at the time of AKI diagnosis than patients who survived (time when
186
the EB was performed).
TE D
M AN U
SC
RI PT
175
Figure 2 shows the comparison of the confidence ellipses of the studied
188
patients divided into four groups: group 1 (patients diagnosed with AKI on the
189
first day of ICU admission who survived, n = 73), group 2 (patients diagnosed
190
with AKI at least 24 hours after the first day of ICU admission who survived, n =
191
47), group 3 (patients diagnosed with AKI on the first day of ICU admission who
192
died, n = 67), and group 4 (patients diagnosed with AKI at least 24 hours after
193
the first day of ICU admission who died, n = 37). Table 4 shows the mean R/H
194
values of each group and the comparisons between them.
195
DISCUSSION
AC C
EP
187
196
In critical patients, proper fluid management is essential for the treatment
197
of conditions such as shock, sepsis, heart failure and AKI. Fluid management is
ACCEPTED MANUSCRIPT 198
typically performed based on the estimated intravascular volume. Over the last
199
decades, several studies have proposed guidelines for fluid management in
200
critical patients. One of these studies demonstrated that an aggressive
201
premature resuscitation strategy including proper administration of fluids could
202
improve outcomes in patients with sepsis
203
excessive fluid accumulation may not be simply considered a consequence of
204
fluid resuscitation or severe AKI but possibly a mediator of adverse results
205
15-20
206
significant relationships with negative outcomes, including increased mortality
207
and reduced kidney function recovery rate
208
50% of hemodynamically unstable patients in the ICU respond to fluid
209
replacement, resulting in unnecessary fluid retention 22.
14
RI PT
. However, other studies found that
3, 4,
4, 21
SC
. In relation to AKI, several studies have shown that hyperhydration has
M AN U
. In addition, only approximately
In our study, we used BIVA to demonstrate a significant difference in the
211
hydration status between patients who died and who survived, where the non-
212
survivors showed greater hydration volume than the survivors. Similar findings
213
were reported by Payen et al. 3, who analyzed 1120 patients with AKI in a
214
multicenter observational cohort study of 198 ICUs. The mortality rates at 60
215
days were 36% in patients with AKI and 16% in patients without AKI (p <
216
0.001), and mean fluid balance was one of the independent risk factors for
217
mortality (0.98 ± 1.5 l/24 h versus 0.15 ± 1.06 l/24 h, p < 0.001, for non-
218
survivors and survivors, respectively).
AC C
EP
TE D
210
219
In our study, we assessed the hydration status at the time of AKI
220
diagnosis and demonstrated that greater hydration present at that time is
221
associated with lower survival. This early assessment would enable fluid
222
management control and better treatment results. Our study revealed variables
ACCEPTED MANUSCRIPT 223
correlated with death (mechanical ventilation, organ failure, septic shock,
224
vasoactive drug, among others) that are expected in patients with this level of
225
disease severity, demonstrating that our study population is in accordance with
226
what other studies have reported. Bouchard et al.
227
between hyperhydration and increased mortality in patients with AKI and also
228
found that patients with hyperhydration at the time of AKI diagnosis were more
229
critically ill than those with a lesser degree of fluid accumulation, indicated by
230
greater frequencies of respiratory failure, mechanical ventilation, and sepsis,
231
even after adjusting for all variables.
established the association
SC
RI PT
4
Although several studies point to the importance of hyperhydration in the
233
mortality of patients with AKI, we must not forget the fundamental role of the
234
kidney in the homeostasis of body fluids. Kidney dysfunction requires special
235
attention in relation to fluid replacement and hyperhydration, as fluid
236
replacement is the only effective treatment in patients at risk of developing AKI.
237
In addition, hypovolemia is the strongest causal factor of morbidity during
238
dialysis and can contribute to increase kidney damage
239
importance of fluid management, we know that the choice of method for
240
measuring fluid status is controversial. Different parameters with varying levels
241
of accuracy can be used to estimate the actual volume status of patients 24. Two
242
parameters are currently commonly used in intensive care: central venous
243
pressure (CVP) and fluid balance (FB) recorded by nurses. Regarding CVP,
244
several studies have shown its ineffectiveness in predicting the response to
245
volume replacement
246
conventional FB is not predictive of the current changes in the weights of
247
critically ill patients and that, consequently, it reflects fluctuations in total body
23
. Regarding the
AC C
EP
TE D
M AN U
232
5
. Regarding FB, Perren et al.
25
concluded that
ACCEPTED MANUSCRIPT 248
fluid inadequately. Nevertheless, FB is still the most viable method in the ICU
249
that offers a quantitative parameter of the hydration status of patients.
250
In our study, we used BIVA to measure the fluid state of patients, a
251
method that has been valid for accessing the hydration status in various clinical
252
situations
253
to assess the hydration status, also positively correlate it with mortality and
254
hospital readmission
255
body composition calculations are not possible due to changes in the hydration
256
status 33.
and recommend that BIVA be applied in cases where
As in our study, Basso et al.
11
SC
9, 32
RI PT
. Some studies, in addition to demonstrating that BIVA is able
and Chen et al.
34
M AN U
257
6, 7, 26-31
recently demonstrated
258
that hyperhydration measured by BIVA is a significant predictor of mortality in
259
critical patients.
Our choice for this mode of BIVA evaluation (i.e., the confidence ellipse)
261
is based on studies that showed not only that hyperhydration increases the
262
mortality of patients with AKI but also that hyperhydration at the time of AKI
263
diagnosis is associated with low survival
264
prospective observational study in 61 patients with serial measurements of
265
BIVA and found a good association between the hydration status classifications
266
obtained using BIVA and FB. They concluded that BIVA could serve as an
267
additional method for measuring the hydration status in critical patients,
268
providing important information for fluid management in these patients. Similar
269
to Jones et al., our study found an association between hydration status and
270
death, despite having used another methodology.
3, 4
. Jones et al.
10
conducted a
AC C
EP
TE D
260
271
BIVA by itself demonstrates the statistical significance through Hotelling’s
272
test. However, to confirm the results found using BIVA, we compared the results
ACCEPTED MANUSCRIPT 273
with the cumulative FB; the most commonly used current method for
274
quantitatively measuring the hydration of critical patients. BIVA was compatible
275
with the FB cumulative results because patients who died were more hydrated
276
than patients who survived, which is compatible with the cumulative FB. In patients with an AKI diagnosis on the first day of admission to the ICU,
278
the BIVA showed no significant difference in the hydration status of non-
279
survivors and survivors. As demonstrated by Piccoli et al.
280
BIVA is not able to detect variations in volume below 2,000 ml, suggesting that
281
the difference in balance between these two groups was less than this volume.
RI PT
277
and Jones et al.
10
,
SC
30
BIVA demonstrated a significant difference between patients who died
283
after 24 hours of admission to the ICU. This finding may suggest that
284
hyperhydration is a determining factor for death in critical patients with AKI and
285
that fluid management in the ICU is of fundamental importance, as it may
286
influence the outcomes of these patients, as also evidenced by other studies
287
4
288
in the ICU could identify greater hydration in patients as a possible risk factor for
289
death. In patients who present AKI in the first 24 hours of ICU admission, the
290
performance of BIVA at least 24 hours after the AKI diagnosis could also assist
291
in identifying greater hydration.
TE D
3,
EP
. Therefore, BIVA at the time of AKI diagnosis developed during hospitalization
AC C
292
M AN U
282
Some aspects of this study strengthen the reliability of the results: the
293
doctors who managed the patients had no access to the results of the BIVA,
294
which prevented treatment bias; both groups of patients had different clinical
295
severity characteristics, which are expected in patients with different hydration
296
statuses; and the BIVA results were compatible with the FB results.
ACCEPTED MANUSCRIPT Therefore, this study is pioneering in assessing the hydration status
298
measured using BIVA in critically ill patients with AKI. This study demonstrated
299
that BIVA could be a useful instrument in assessing the hydration of patients
300
who develop AKI in the ICU; however, further studies are needed to confirm our
301
findings.
302
Statement of autorship: ACRH and MCG participated in the conception and
303
design of the study, analysis and interpretation of the data, drafting of the article
304
and final approval of the version of the sutdy. ACRH was responsible for the
305
data collection.
306
Conflict of interest and founding sources: The authors have no conflict of
307
interests to declare. This research did not receive any specific grant from
308
funding agencies in the public, comercial, or not-for-profit sectors.
M AN U
SC
RI PT
297
AC C
EP
TE D
309
ACCEPTED MANUSCRIPT 310
References
311
1.
312
patients: a multinational, multicenter study. JAMA. 2005; 294: 813-8.
313
2.
314
CA. Acute kidney injury criteria predict outcomes of critically ill patients. Crit
315
Care Med. 2008; 36: 1397-403.
316
3.
317
positive fluid balance is associated with a worse outcome in patients with acute
318
renal failure. Crit Care. 2008; 12: R74.
319
4.
320
and recovery of kidney function in critically ill patients with acute kidney injury.
321
Kidney Int. 2009; 76: 422-7.
322
5.
323
fluid responsiveness? A systematic review of the literature and the tale of seven
324
mares. Chest. 2008; 134: 172-8.
325
6.
326
patients with different hydration status. Kidney Int. 2004; 65: 1050-63.
327
7.
328
al. Body fluid volume and nutritional status in hemodialysis: vector bioelectric
329
impedance analysis. Clin Nephrol. 2010; 73: 300-8.
330
8.
331
vectorial analysis of fluid status and prediction of contrast-induced acute kidney
332
injury. J Am Coll Cardiol. 2014; 63: 1387-94.
Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill
RI PT
Barrantes F, Tian J, Vazquez R, Amoateng-Adjepong Y and Manthous
SC
Payen D, de Pont AC, Sakr Y, Spies C, Reinhart K and Vincent JL. A
M AN U
Bouchard J, Soroko SB, Chertow GM, et al. Fluid accumulation, survival
TE D
Marik PE, Baram M and Vahid B. Does central venous pressure predict
EP
Piccoli A. Bioelectric impedance vector distribution in peritoneal dialysis
AC C
Espinosa Cuevas MA, Navarrete Rodriguez G, Villeda Martinez ME, et
Maioli M, Toso A, Leoncini M, et al. Pre-procedural bioimpedance
ACCEPTED MANUSCRIPT 333
9.
Nunez J, Mascarell B, Stubbe H, et al. Bioelectrical impedance vector
334
analysis and clinical outcomes in patients with acute heart failure. J Cardiovasc
335
Med (Hagerstown). 2014.
336
10.
337
analysis in critically ill patients: a prospective, clinician-blinded investigation. Crit
338
Care. 2015; 19: 290.
339
11.
340
care unit: bioelectrical impedance vector analysis as a tool to assess hydration
341
status and optimal fluid balance in critically ill patients. Blood Purif. 2013; 36:
342
192-9.
343
12.
344
serum creatinine. Nephron. 1976; 16: 31-41.
345
13.
346
monitoring body fluid variation by bioimpedance analysis: the RXc graph.
347
Kidney Int. 1994; 46: 534-9.
348
14.
349
treatment of severe sepsis and septic shock. N Engl J Med. 2001; 345: 1368-
350
77.
351
15.
352
fluid-management strategies in acute lung injury. N Engl J Med. 2006; 354:
353
2564-75.
354
16.
355
intravenous fluid restriction on postoperative complications: comparison of two
356
perioperative fluid regimens: a randomized assessor-blinded multicenter trial.
357
Ann Surg. 2003; 238: 641-8.
RI PT
Jones SL, Tanaka A, Eastwood GM, et al. Bioelectrical impedance vector
M AN U
SC
Basso F, Berdin G, Virzi GM, et al. Fluid management in the intensive
Cockcroft DW and Gault MH. Prediction of creatinine clearance from
TE D
Piccoli A, Rossi B, Pillon L and Bucciante G. A new method for
EP
Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the
AC C
Wiedemann HP, Wheeler AP, Bernard GR, et al. Comparison of two
Brandstrup B, Tonnesen H, Beier-Holgersen R, et al. Effects of
ACCEPTED MANUSCRIPT 358
17.
Foland JA, Fortenberry JD, Warshaw BL, et al. Fluid overload before
359
continuous hemofiltration and survival in critically ill children: a retrospective
360
analysis. Crit Care Med. 2004; 32: 1771-6.
361
18.
362
of replacement fluid on survival in hemofiltration. Pediatr Nephrol. 2004; 19:
363
1394-9.
364
19.
365
Outcome
366
Pediatrics. 2001; 107: 1309-12.
367
20.
368
organ dysfunction syndrome receiving continuous renal replacement therapy.
369
Kidney Int. 2005; 67: 653-8.
370
21.
371
Fluid overload at initiation of renal replacement therapy is associated with lack
372
of renal recovery in patients with acute kidney injury. Nephrol Dial Transplant.
373
2012; 27: 956-61.
374
22.
375
arterial waveform derived variables and fluid responsiveness in mechanically
376
ventilated patients: a systematic review of the literature. Crit Care Med. 2009;
377
37: 2642-7.
378
23.
379
achieving fluid balance in acute renal failure. Curr Opin Crit Care. 2002; 8: 535-
380
43.
381
24.
382
kidney injury: emerging concepts. Nephron Clin Pract. 2013; 123: 238-45.
RI PT
Gillespie RS, Seidel K and Symons JM. Effect of fluid overload and dose
Goldstein SL, Currier H, Graf C, Cosio CC, Brewer ED and Sachdeva R. children
receiving
continuous
venovenous
hemofiltration.
SC
in
M AN U
Goldstein SL, Somers MJ, Baum MA, et al. Pediatric patients with multi-
TE D
Heung M, Wolfgram DF, Kommareddi M, Hu Y, Song PX and Ojo AO.
AC C
EP
Marik PE, Cavallazzi R, Vasu T and Hirani A. Dynamic changes in
Mehta RL, Clark WC and Schetz M. Techniques for assessing and
Godin M, Bouchard J and Mehta RL. Fluid balance in patients with acute
ACCEPTED MANUSCRIPT 383
25.
Perren A, Markmann M, Merlani G, Marone C and Merlani P. Fluid
384
balance in critically ill patients. Should we really rely on it? Minerva Anestesiol.
385
2011; 77: 802-11.
386
26.
387
assessment. Contrib Nephrol. 2010; 164: 143-52.
388
27.
389
assessment and management in the emergency department. Contrib Nephrol.
390
2010; 164: 227-36.
391
28.
392
emergency department: a look at B-type natriuretic peptide and bioimpedance
393
vector analysis. Contrib Nephrol. 2011; 171: 187-93.
394
29.
395
biomarkers and bio-impedance in evaluating hydration status in patients with
396
acute heart failure. Clin Chem Lab Med. 2012; 50: 2093-105.
397
30.
398
of nutrition and hydration in dialysis patients with bioelectrical impedance vector
399
analysis (BIVA). Clin Nutr. 2014; 33: 673-7.
400
31.
401
and Facila L. Prognostic value of analysing the bioimpedance vector for patients
402
hospitalised for acute decompensated heart failure: A validation cohort. Rev
403
Clin Esp. 2016.
404
32.
405
galectin-3 and BIVA assessments in predicting short- and long-term events in
406
patients admitted for acute heart failure. Biomed Res Int. 2014; 2014: 983098.
A.
Bioelectric
impedance
measurement
for
fluid
status
RI PT
Piccoli
SC
Di Somma S, Gori CS, Grandi T, Risicato MG and Salvatori E. Fluid
M AN U
Tuy T and Peacock F. Noninvasive volume assessment in the
TE D
Di Somma S, Navarin S, Giordano S, et al. The emerging role of
EP
Piccoli A, Codognotto M, Piasentin P and Naso A. Combined evaluation
AC C
Trejo-Velasco B, Fabregat-Andres O, Montagud V, Morell S, Nunez J
De Berardinis B, Magrini L, Zampini G, et al. Usefulness of combining
ACCEPTED MANUSCRIPT 407
33.
Norman K, Stobaus N, Pirlich M and Bosy-Westphal A. Bioelectrical
408
phase angle and impedance vector analysis--clinical relevance and applicability
409
of impedance parameters. Clin Nutr. 2012; 31: 854-61.
410
34.
411
renal replacement therapy is associated with poorer clinical condition and
412
outcome: a prospective observational study on the combined use of
413
bioimpedance vector analysis and serum N-terminal pro-B-type natriuretic
414
peptide measurement. Crit Care. 2015; 19: 135.
AC C
EP
TE D
M AN U
SC
RI PT
Chen H, Wu B, Gong D and Liu Z. Fluid overload at start of continuous
ACCEPTED MANUSCRIPT Figure and Table legends:
416
Figure 1: Comparison of the confidence ellipses using bioelectrical vector
417
analysis (n = 224), according to the outcome (survivors and non-survivors).
418
Figure 2: Comparison of the confidence ellipses using bioelectrical vector
419
analysis (n = 224), according to the outcome (survivors and no-survivors) and
420
the moment of acute kidney injury diagnosis (1st day of intensive care unit
421
admission or after 1st day of intensive care unit admission).
AC C
EP
TE D
M AN U
SC
RI PT
415
ACCEPTED MANUSCRIPT 422
Table 1. Clinical variables distribution according to patients’ outcome. Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
RR (CI 95%)
p value
a
≤ 15
89 (71.2)
36 (28.8)
1.00 (ref)
> 15
31 (31.1)
68 (68.7)
No
89 (58.5)
63 (41.4)
Yes
31 (43.1)
41 (56.9)
RI PT
APACHE II
Total length of
< 13 days
52 (45.6)
62 (54.4)
> 13 days
68 (61.8)
42 (38.2)
57 (42.2)
78 (57.8)
TE D
M AN U
hospitalization
63 (70.8)
26 (29.2)
107 (57.5)
79 (42.5)
Surgical diagnosis at admission
Sepsis
No
EP
No
Yes
2.38 (1.76;3.24)
1.00 (ref)
1.37 (1.04;1.81)
1.00 (ref)
1.00 (ref)
1.00 (ref)
Risk
80 (67.8)
38 (32.8)
Injury
34 (38.2)
55 (60.4)
1.84 (1.35;2.51)
Failure
6 (35.3)
11 (64.7)
1.98 (1.28;3.06)
AC C
<0.001
0.51 (0.35;0.72)
25 (65.8)
RIFLE
0.02
0.70 (0.52;0.94)
13 (34.2)
Yes
0.03
SC
Direct admission in ICU
<0.001
0.01
1.55 (1.16;2.06)
1.00 (ref)
b
<0.001
Etiology
423
a
Ischemic
83 (65.3)
44 (34.6)
Others
18 (58.1)
14 (42.4)
1.22 (0.77; 1.95)
Septic
18 (28.1)
46 (71.9)
2.07 (1.56; 2.76)
χ2 test. b Linear trend for χ2 test.
1.00 (ref)
<0.001
ACCEPTED MANUSCRIPT 424
Table 1. Clinical variables distribution according to patients’ outcome (cont.). Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
≤ 3 days
91 (64.0)
51 (35.9)
> 3 days
29 (35.4)
53 (64.6)
None
78 (83.9)
15 (16.1)
<3
39 (39.4)
60 (60.6)
≥3
3 (9.4)
29 (90.6)
RR (CI 95%)
p value
1.00 (ref)
<0.001
a
AKI duration
M AN U
Maximum number of organ failure
RI PT
b
1.00 (ref)
3.76 (2.30;6.13)
5.62 (3.49;9.05)
None
70 (89.7)
8 (10.3)
<3
46(46.9)
52 (53.1)
5.17 (2.61;10.24)
≥3
4 (8.3)
44 (91.7)
8.94 (4.61;17.33)
Hemoglobin level (minimum)
86 (60.6)
TE D
≥9mg/dl <9mg/dl
Sepsis
56 (39.4)
34 (41.5)
48 (58.5)
86 (79.6)
22 (29.6)
EP
Yes
1.00 (ref)
1.00 (ref)
1.00 (ref)
82 (70.7)
Sepsis
23 (45.1)
28 (54.9)
Severe sepsis
11 (16.9)
54 (83.1)
1.51 (1.15;1.96)
No
34 (20.4)
82 (70.7)
3.47 (2.35;5.13)
No
94 (72.3)
36 (27.7)
Yes
26 (27.6)
68 (72.3)
AC C
Type of sepsis
<0.001
c
0.008
1.48 (1.13;1.95)
34 (20.4)
No
c
<0.001
SC
Number of organ failure
1.80 (1.37;2.36)
<0.001
3.47 (2.35;5.13)
1.00 (ref)
0.002
Use of vasoactive drugs
425
a
χ2 test. b at AKI diagnosis. c Linear trend for χ2 test.
1.00 (ref) 2.61 (1.93;3.54)
< 0.001
ACCEPTED MANUSCRIPT 426
Table 1. Clinical variables distribution according to patients’ outcome (cont.). Variable
Survivors n=120
Non-survivors n=104
n (%)
n (%)
No
88 (81.5)
20 (18.5)
Yes
32 (27.6)
84 (72.4)
No
118 (57.2)
88 (42.7)
Yes
2 (11.1)
16 (88.9)
RR (CI 95%)
p value
1.00 (ref)
<0.001
a
Mechanic ventilation
1.00 (ref)
2.08 (1.66;2.61)
TE D
M AN U
SC
χ2 test.
EP
a
AC C
427
RI PT
Dialysis
3.91 (2.59;5.9)
< 0.001
ACCEPTED MANUSCRIPT 428
Table 2. Fluid balance volume in milliliters (median and interquartile range), before, during
429
acute kidney injury and total, according to patients’ outcome.
FB before AKId
Non-survivors
0 (0;1064.5)
0 (0; 5340)
a
Survivorsb
pc
0 (0;0)
< 0.001
4135 (550.5;10204.5) 6906.5 (3016;16725) 1557.5 (97;6408) < 0.001
Total FBf
4483 (571.5;11051.5) 7565.5 (3243;19551) 1557.5 (55;7125) < 0.001
RI PT
FB during AKIe
a
Non-survivors: n = 37, for FB before AKI, and n = 104, for FB during AKI and total FB. Survivors: n = 47, for FB before AKI, and n = 120, for FB during AKI and total FB. c p value for median comparison between non-survivor and survivor’s fluid balance (MannWhitney test). d FB before AKI: fluid balance since ICU admission until AKI diagnosis. N = 84 (140 patients had AKI in the 1st day, and for this reason, they do not have this FB). e FB during AKI: fluid balance since AKI diagnosis until renal recovery, ICU discharge or death. N = 224 f Total FB total: fluid balance since ICU admission until ICU discharge or death. N = 22
EP
TE D
M AN U
SC
b
AC C
430 431 432 433 434 435 436 437 438
All
ACCEPTED MANUSCRIPT 439
Table 3. Resistance and reactance values, adjusted for height (R/H and Xc/H,
440
respectively), according to patient´s outcome. R/H
Xc/H
(mean ± standard deviation) (mean ± standard deviation) 20.5 ± 7.5a
Non-survivors
220.5 ± 74.6
15.8 ± 7.7
TE D
M AN U
SC
p < 0.05
EP
a
RI PT
246.6 ± 69.6a
AC C
441
Survivors
ACCEPTED MANUSCRIPT 442
Table 4. P values of the comparison of resistance mean values adjusted for height
443
(R/H) of patients, grouped according to the timing of acute kidney injury (AKI)
444
diagnosis and outcome.
Group 2b (R/H = 243.5)
0.05
Group 3c (R/H = 233.8)
Group 4d (R/H = 195.8) <0.001 <0.001 0.007
Group 1: Survivors with AKI in 1st day of Intensive Care Unit (n = 73). bGroup 2: Survivors with AKI after 1st day of Intensive Care Unit (n = 47). cGroup 3: Non-survivors with AKI in 1st day of Intensive Care Unit (n = 67). dGrupo 4: Non-survivors with AKI after 1st day of Intensive Care Unit (n = 37).
EP
TE D
M AN U
SC
a
AC C
445 446 447 448
Group 3c (R/H = 233.8) 0.12
RI PT
Group 1a (R/H = 249.7)
Group 2b (R/H = 243.5) 0.72
ACCEPTED MANUSCRIPT 449
Figure 1
EP AC C
451
TE D
M AN U
SC
RI PT
450
ACCEPTED MANUSCRIPT 452
Figure 2
M AN U
SC
RI PT
453
454
AC C
EP
TE D
455