Beruflich Dokumente
Kultur Dokumente
Keywords:
Abstract This article reviews if a restrictive fluid management policy reduces the complication rate if
Elective surgical procedures;
compared to liberal fluid management policy during elective surgery. The PubMed database was
Fluid therapy;
explored by 2 independent researchers. We used the following search terms: “Blood transfusion
Review, systematic
(MESH); transfusion need; fluid therapy (MESH); permissive hypotension; fluid management;
resuscitation; restrictive fluid management; liberal fluid management; elective surgery; damage control
resuscitation; surgical procedures, operative (MESH); wounds (MESH); injuries (MESH); surgery;
trauma patients.” A secondary search in the Medline, EMBASE, Web of Science, and Cochrane library
revealed no additional results. We selected randomized controlled trials performed during elective
surgeries. Patients were randomly assigned to a restrictive fluid management policy or to a liberal fluid
management policy during elective surgery. The patient characteristics and the type of surgery varied.
All but 3 studies reported American Society of Anaesthesiologists groups 1 to 3 as the inclusion
criterion. The primary outcome of interest is total number of patients with a complication and the
complication rate. Secondary outcome measures are infection rate, transfusion need, postoperative
rebleeding, hospital stay, and renal function. In total, 1397 patients were analyzed (693 restrictive
protocol, 704 liberal protocol). Meta-analysis showed that in the restrictive group as compared with
the liberal group, fewer patients experienced a complication (relative risk [RR], 0.65; 95% confidence
☆
Funding: No funding was received.
☆☆
Conflict of interest: None.
⁎ Correspondence: Pim B.B. Schol, MD, Department of Obstetrics and Gynaecology, Maastricht University Medical Centre, PO 5800, 6202 AZ, Maastricht,
The Netherlands. Tel.: +31 43 3876700, +31 6 45124220; fax: +31 43 3874765.
E-mail addresses: pbbschol@hotmail.com (P.B.B. Schol), i.terink@student.maastrichtuniversity.nl (I.M. Terink), marcus.lance@mumc.nl (M.D. Lancé),
hcj.scheepers@mumc.nl (H.C.J. Scheepers).
http://dx.doi.org/10.1016/j.jclinane.2016.07.010
0952-8180/© 2016 Elsevier Inc. All rights reserved.
Liberal or restrictive fluid management in elective surgery 27
interval [CI], 0.55-0.78). The total complication rate (RR, 0.57; 95% CI, 0.52-0.64), risk of infection
(RR, 0.62; 95% CI, 0.48-0.79), and transfusion rate (RR, 0.81; 95% CI, 0.66-0.99) were also lower.
The postoperative rebleeding did not differ in both groups: RR, 0.76 (95% CI, 0.28-2.06). We conclude
that compared with a liberal fluid policy, a restrictive fluid policy in elective surgery results in a 35%
reduction in patients with a complication and should be advised as the preferred fluid management
policy.
© 2016 Elsevier Inc. All rights reserved.
1. Introduction 2. Methods
Although fluid therapy is a cornerstone in current surgical In this systematic review, the PRISMA statement for
practice, no consensus on the optimal perioperative fluid man- reporting reviews was applied [11]. The PubMed database
agement exists, and the existing trials are contradictory. was explored by 2 independent researchers to identify appro-
Since Shires in 1961, a liberal transfusion practice is advo- priate articles. We used the following search terms: “Blood
cated [1]. Today's textbook management is approximately 20 transfusion (MESH); transfusion need; fluid therapy (MESH);
mL/kg per hour fluid transfusion (crystalloids and colloids) to permissive hypotension; fluid management; resuscitation; re-
account for fasting, third space, and urine losses [2]. On top of strictive fluid management; liberal fluid management; elective
the standard management, blood loss will be compensated 3 to surgery; damage control resuscitation; surgical procedures, op-
4 times the actual loss [2]. Although a more liberal fluid man- erative (MESH); wounds (MESH); injuries (MESH); surgery;
agement is practiced widely, it has never been properly evalu- trauma patients.” A secondary search in the Medline,
ated [3]. Excessive fluid therapy is associated with negative EMBASE, Web of Science, and Cochrane library revealed
outcomes, even in healthy patients (American Society of An- no additional results.
esthesiologists [ASA] 1) [2,4-7]. One important side effect of Studies had to meet the following criteria to be included:
the liberal approach is volume overload which may cause re- (1) a randomized controlled trial, (2) a population that was ad-
duced pulmonary function, postoperative reduced gut motility, mitted for any kind of elective surgery, and (3) a comparison
and reduced subcutaneous oxygen tension [5,7]. More fluid of restrictive and liberal fluid management with complication
puts a greater demand on the cardiac and urinary systems pre- rate and/or hospital stay as outcome measurements. No restric-
disposing for cardiac morbidity and urinary retention [5]. In tions were set with regard to age, ethnicity, or sex. Articles
addition, the crystalloids and colloids transfused interfere with were excluded if a goal-directed approach of fluid manage-
coagulation due to dilution, acidosis, or faster clot disintegra- ment or if an additional anesthesia was used in either of the
tion [5,8-10]. groups (eg, restrictive policy with epidural compared to stan-
Recently, more restrictive perioperative fluid management dard care without an epidural anesthesia) was used. Screening
policies have been studied in randomized controlled trials was done on title and abstract; if this provided insufficient in-
challenging the liberal practice. Despite avoiding an overload- formation, the full text was read. Inclusion and exclusion were
ing effect, restrictive fluid management and its potential hypo- done independently by 2 researchers. Disagreement about in-
volemic state are associated with impaired cardiac output. This clusion or exclusion was resolved through discussion, and a
results in inadequate oxygenation putting the organs at risk for third researcher was decisive if needed. The reference lists of
ischemia, infarction, and organ failure [2]. With all strategies included articles were screened for additional articles.
having their own risks, the most important goal is to achieve The following data were extracted and summarized: (1)
an optimized state with a normovolemic patient. number of participants and type of surgery, (2) intervention
In this systematic review, we will evaluate a liberal vs a re- protocol, (3) outcome measures, and (4) results. The required
strictive policy intraoperatively in general elective surgery. data were available in all selected articles.
The primary outcome of interest is total number of patients One researcher (IMT) performed the quality assessment
with a complication and the complication rate (defined as the (Appendix 1), according to the CONSORT guideline for
total number of complications given in the trials per group). reviewing randomized controlled trials [12]. All items were
The secondary outcome measures are hospital stay, infection scored and given the following codes:
rate (the total of peritonitis, sepsis, wound infection, pneumo-
nia, urinary tract infection, and wound abscess), postoperative + (1 point) good, clearly described and taken into
bleedings (defined as the total number of postoperative bleed- account;
ings that occurred requiring transfusion and surgical treat- +/− (half a point) moderately well described, not entirely
ment), transfusion need, and renal function. clear;
28 P.B.B. Schol et al.
− (zero points) bad, not described or not taken into A meta-analysis was performed on the primary and second-
account; ary outcomes with Review Manager version 5.2. The statistic
N/A (not considered in the final judgment) when the item used was risk ratio (RR). As the included studies are heteroge-
was not applicable to the study. neous regarding surgical types, a random-effects model (with
Liberal or restrictive fluid management in elective surgery 29
Mantel-Haenszel method) was chosen. To calculate with ad- included any elective surgery with an American College of
verse events in percentages, we divided the total number of pa- Cardiology risk score grerater than 3 [20]. Criteria for exclu-
tients with any complication in a group by the total nummber sion were similar in all studies. Inclusion and exclusion criteria
of patients in that group. are shown in Table 1. The intraoperative fluid management in
9 of 12 studies consisted of Ringer's lactate [15-21,23,24].
Other used solutions were buffered glucose 2.5% [13], Hart-
man solution [22], and normal saline 0.9% [14]. All solutions
3. Results used were crystalloids, in 5 studies with addition of colloids
(hydroxyethyl starch) [14,15,17,18,24]. In each individual
3.1. Study selection study, the restrictive group received no more than the
liberal group. Overall, the liberal group received more fluids
The 2 individual PubMed searches resulted in 2330 and compared with the restrictive group (mean 4048 mL [2928-
1692 articles. After screening on title and abstract and remov- 5775 mL] vs mean 2019 mL [997.5-3517 mL] respectively;
ing duplicates, 4003 articles were excluded. Eighteen articles see Figure 1). Total amount of complications was available
remained, 1 of which was not accessible in full text. Two addi- in all studies with the execption of 2 studies [16,19]. Other rel-
tional articles that met the eligibility criteria were found by evant outcome measures were hospital stay and renal function.
screening the reference lists. A total of 19 full-text articles have An overview of study characteristics is given in Table 2.
been read. Six were excluded as they did not meet the eligibil-
ity criteria: goal-directed fluid management (n = 2), the ab- 3.3. Results of individual studies and meta-analysis
sence of a liberal group (n = 1), the absence of a restrictive
group (n = 1), unclear intervention protocol (n = 2), and com- A statistically significant difference in total number of com-
plication rate or hospital stay was not outcome measures (n = plications within 30 days was found in 5 studies. All but 1 of
1). A total of 12 randomized controlled trials were included in these 5 studies showed more complications in the liberal group
the systematic review [13-24]. in comparison to the restrictive group [14,20,22,23]; the other
one showed more complications in the restrictive group [17].
3.2. Study characteristics One study showed the infection rates between both groups.
These results were not statistically significantly different
All subjects (1397 patients were analyzed [693 restrictive [19]. An overview of results per trial is given in Table 3.
protocol, 704 liberal protocol]) were scheduled for elective The total number of complications was subdivided in dif-
surgery. The patient characteristics and the type of surgery ferent categories. Percentages of bleeding, wound infection,
were heterogeneous, varying from age older than 65 years pneumonia, sepsis, cystitis, and peritonitis were extracted as
and morbidly obese patients to cancer patients. All but 3 stud- shown in Table 3. For the exact definitions used per outcome
ies reported ASA groups 1 to 3 as inclusion criterion. One in- per study, see Appendix 2. Percentages of bleeding, sepsis,
cluded all patients with an abdominal aortic aneurysm repair and peritonitis did not differ between the 2 groups. Wound
[22]; one did not select on specific ASA critiria but excluded infection, pneumonia, and cystitis were more common in the
on age, weight, and additional diseases [16]. The third study liberal group.
30
Table 2 Study characteristics
Study Participants Age R Age L Male/ Male/ Fluid management restrictive Fluid management liberal Outcome measures
(n) in years in years female R female L group during operation group during operation
Abraham-Nordling 161 68 (59-77) 69 (62-79) 43/36 45/37 Buffered glucose 2%-5% IV 2 Buffered glucose 2%-5% IV 2 Primary: postoperative hospital
et al 2012 [13] mL/h/kg mL/h/kg and RL 5 mL/h/kg stay
Secondary: complications
within 30 d
Brandstrup et al 141 64 (42-90) 69 (41-88) 33/36 37/35 No replacement for third space Normal 0.9% saline; first hour 7 Primary: complications within
2003 [14] loss, 500 mL of glucose 5% in mL/h/kg, second and third hour 30 d
water less oral fluid intake 5 mL/h/kg, then 3 mL/h/kg. Secondary: death and adverse
during fast 500 mL normal saline 0.9% in- effects including postoperative
Blood loss replaced with 6% dependent or oral intake during hypotensive episodes and renal
HAES 1:1 fast. function impairment
Blood loss up to 500 mL: 1000-
1500 mL normal saline; Blood
loss N500 mL additional HAES
6%
Gao et al 2012 [15] 179 72 (65-89) 73 (65-87) 54/39 49/37 RL; first hour 7 mL/h/kg, then 5 RL; 12 mL/h/kg Primary: complications within
mL/h/kg Blood loss replaced with 6% 30 d
Blood loss replaced with 6% HAES 1:1 Secondary: death and adverse
HAES 1:1 effects
Holte et al 2004 48 34 (21-65) 37.5 (23-63) 3/21 5/19 15 mL/kg RL 40 mL/kg RL Primary: pulmonary function,
(laparoscopic) exercise capacity,stress
[16] responses (aldosteron,
antidiuretic hormone,
angiotension II, atrial
natriuretic peptide and renin),
balance function
Secondary: pain, nausea,
vomiting, hospital stay, and
recovery.
Holte et al 2007 32 73.5 (56-87) 76.5 (53-93) 6/10 9/7 RL: first hour 7 mL/h/kg, then 5 RL: 18 mL/h/kg and Voluven 7 Time to discharge,
(colonic surgery) mL/h/kg and Voluven 7 mg/kg mg/kg readmissions within 30 d,
[17] complications within 30 d
Holte et al 2007 48 71.5 (58-80) 71.5 (55-83) 13/11 10/14 RL: 10 mL/h/kg and Voluven 7 RL: 30 mL/h/kg and Voluven 7 Time to discharge,
(knee arthro- mL/kg mL/kg readmissions within 30 d,
plasty) [18] complications within 30 d
Kabon et al 2005 253 53 (39-67) 52 (38-66) 60/64 65/64 RL: 8-10 mL/h/kg RL bolus 10 mL/kg before Primary: surgical wound
31
32
Table 3 Results of individual studies: complications within 30 days
Study Total Total Total Total P Bleeding Bleeding Wound Wound Pneumonia Pneumonia Sepsis Sepsis Cystitis Cystitis Peritonitis Peritonitis
participants participants complications complications R, n (%) L, n (%) infection infection R, n (%) L, n (%) R, n L, n R, n L, n R, n (%) L, n (%)
R (n) L (n) R, n (% a) L, n (% a) R, n (%) L, n (%) (%) (%) (%) (%)
Abraham- 79 82 50 (39) 47 (57) 0.079 0 (0) 2 (2.4) 10 (12.7) 11 (13.4) 0 (0) 1 (1.2) 1 (1.3) 4 (4.9) – – 1 (1.3) 1 (1.2)
Nordling
et al 2012
[13]
Brandstrup 69 72 26 (33) 83 (51) 0.013 1 (1.4) 5 (6.9) 9 (14.5) 18 (26.4) 3 (4.3) 9 (12.5) 0 (0) 4 (5.6) 1 (1.4) 5 (6.9) 1 (1.4) 0 (0)
et al 2003
[14]
Gao et al 93 86 46 (33) 84 (45) 0.079 2 (2.2) 1 (1.2) 12 (13.8) 22 (25.6) 7 (7.5) 15 (17.4) 2 (2.2) 1 (1.2) 2 (2.2) 3 (3.5) 1 (1.1) 0 (0)
2012 [15]
Holte et al 24 24 – – – – – – – – – – – – – – –
2004 (lap-
aroscopic)
[16]
Holte et al 16 16 18 (37.5) 1 (6) 0.03 1 (6.3) 0 (0) 1 (6.3) 0 (0) 2 (12.5) 1 (6.3) – – – – – –
2007
(colonic
surgery)
[17]
Holte et al 24 24 1 (4) 3 (12.5) – – – – – – – – – – – – –
2007 (knee
arthro-
plasty) [18]
Kabon et al 124 129 – – – – – 14 (11.3) 11 (8.5) – – – – – – – –
2005 [19]
Lobo et al 45 43 9 (20) 24 (42) 0.046 – – – – – – – – – – 0 (0) 2 (4.6)
2011 [20]
Matot et al 52 55 7 (13) 10 (18) 0.60 2 (3.8) 2 (3.6) 1 (1.9) 0 (0) – – – – – – – –
2012 [21]
McArdle 10 11 1 (10) 14 (64 ) 0.024 – – 0 (0) 0 (0) 0 (0) 4 (36.4) 0 (0) 1 (7.1) – – – –
et al 2009
[22]
Nisanevich 77 75 17 (17) 32 (31) 0.046 0 (0) 0 (0) 7 (9.1) 11 (14.7) 3 (3.9) 5 (6.7) 0 (0) 1 (1.3) – – 2 (2.6) 3 (4)
et al 2005
[23]
2. Total complicationrate
4. Blood transfusion
5. Post-operative rebleeding
Data concerning hospital stay were available in 7 studies surgery showed in both subgroup analyses a significant reduc-
[13,16-19,22,23]: McArdle et al [22] and Nisanevich et al tion in total patients with a complication which favors the re-
[23] found that the length of hospital stay was significantly strictive approach (data not shown).
lower in the restrictive group; 9 vs 18 days (P = .010;
McArdle et al) and 8 vs 9 days (P = .01; Nisanevich et al),
whereas Abraham-Nordling et al [13], Holte et al (knee arthro-
plasty) [18], Holte et al [17] (colonic surgery), and Kabon et al 4. Discussion
[19] found no difference. Holte et al [16] (laparoscopic) found
that hospital stay was longer in the restrictive group. In the re- Fluid management during surgery is been discussed for
strictive group, 15 of 23 patients could be discharched the many years, yet no consensus exists on the optimal course of
same day of surgery compared with 21 of 22 patients in the lib- action. The British consensus advocates an optimal stroke vol-
eral group (P b .03). No data are available on the length of ume guided fluid therapy (goal-directed therapy [GDT]) for
stay of the remaining patients. orthopedic and intraabdominal surgeries but is not directive
Renal function data were available in 3 studies [14,21,22]; as to what this stroke volume should be and the volume of
however, different methods were used to assess renal function. the suggested bolus therapies is authority based [25]. No
McArdle et al [22] measured the urinary albumin/creatinine ra- guideline is available of the ASA and European Society of An-
tio and found a significantly higher value (suggesting impaired aesthesiology. The Enhanced Recovery After Surgery society
renal endothelial function) in the liberal group. Brandstrup advocates to avoid water and salt overloading, and intraopera-
et al [14] reported a significantly lower serum creatinine in the tive GDT could be helpful to achieve this [26].
liberal group upon arrival in the recovery room. There was no GDT individualizes the amount of fluid given to a patient with
difference found in the subsequent days. In addition, Matot stroke volume as the directive measurement. However, the anes-
et al [21] found that there was no significant difference be- thesia itself can induce a hypotensive state, reduced urine output,
tween the restrictive and liberal groups and that mean creati- and reduced heart rate without the patient being hypovolemic.
nine serum concentrations were within the reference range at Therefore, GDT can still lead to fluid overloading [7,27-30].
all times. Furthermore, Matot et al measured low urine outputs Our current analysis advocates for a restrictive approach.
in the majority of the patients in both groups without any sta- We showed that restrictive fluid management decreases 30-
tistical difference between the liberal and restrictive groups. day complications after elective surgery. As a secondary out-
Figure 2 shows the results of meta-analysis of the come measure, we detect fewer infections in the restrictive
primary and secondary outcome measures: total patients with group. In addition, transfusion need is significantly lower in re-
a complication; the complication rate; and secondary outcome strictive groups, although the blood loss did not significantly
measures transfusion need, postoperative rebleeding, and cu- differ between both groups.
mulative infection rate. Brandstrup et al [14] report in their study higher rates of anas-
The total amount of patients with a complication is signifi- tomotic leakage and more infections in the liberal group. Both
cantly higher in the liberal approach: RR of 0.65 (95% confi- may lead to sepsis and wound healing problems. This is in line
dence interval [CI], 0.55-0.78). In addition, the total with McArdle et al [22], Peng et al [24], and Nisanevich et al
complication rate is significantly lower in the restrictive policy [23] who argue that tissue edema due to the liberal fluid regi-
group compared with the liberal policy group: RR of 0.57 men might be responsible [22-24]. Gastrointestinal edema re-
(95% CI, 0.52-0.64). A higher risk of infection is found, and sults in gastrointestinal dysfunction and, therefore, an increased
more transfusions are needed in the liberal policy group: RRinf risk of anastomotic dehiscence [31]. This is supported by the re-
of 0.62 (95% CI, 0.48-0.79) and RRtrans of 0.81 (95% CI, 0.66- sults of Peng et al who demonstrate the increased amount of extra-
0.99), although the postoperative rebleeding did not differ in cellular fluids in the liberal group on the first 2 postoperative days.
both groups: RR of 0.76 (95% CI, 0.28-2.06). Cellular swelling impairs intracellular signaling mechanisms re-
No meta-analysis could be performed with regard to the sponsible for adequate (immune) responses [32].
secondary outcome measure renal function because too little Gao et al [15] shows that fluid overload might promote in-
data were available for valid results. fection due to an altered immune system. The authors demon-
Not all data could be included in the meta-analysis of total strate a higher CD4+/CD8+ ratio which suggest a better
complication rate as the total amount of complications per preserved immune response. They argue that lymphocyte sig-
group in some studies exceeded the total amount of patients naling is impaired due to cell swelling in the liberal approach.
in the group [14,17,20,22]. The investigation of Holte et al in colonic surgery [17] could
There was no difference in the amount of complications per not show more anastomotic leakage or infections.
patient in both groups (data not shown). The restrictive ap- An increase of cardiovascular events could be caused by
proach results in 35% fewer patients with a complication in fluid overload which stresses the circulatory system. More-
our analysis. A subgroup analysis did not show any difference over, fluid overloading contributes to pulmonary dysfunction
between studies using crystalloids only and studies using a com- resulting from oedema. The edema and the potentially result-
bination of crystalloids and colloids (data not shown.) A sub- ing hypoxia may give rise to respiratory failure and to pulmo-
goup analysis for abdominal surgery only and nonabdominal nary infection [5,33].
Liberal or restrictive fluid management in elective surgery 35
Although the mean amount of blood loss in the restrictive fluid Lobo et al, Holte et al (laparoscopic, knee, and colonic),
management group does not differ from the mean amount of Nisanevich et al, and Abraham-Nordling et al described blinding
blood loss in the liberal fluid management group (mean 343 mL of surgeons for the intervention, contrary to the other studies.
[0-1146 mL] vs 372 mL [0-1100 mL], respectively), the distribu- However, surgeons are not the primary guardians of fluid man-
tion of transfusion rate favors the restrictive group. This might be agement; fluid management is the responsibility of the anaesthe-
explained by a higher degree of haemodilution. In other words, the siologist. Anaesthesiologists are not further involved in the study
hemodilution seems responsible for unnecessary blood transfu- procedure and patient care thereafter limiting the risk of informa-
sion, which itself is responsible for increased mortality. The tion bias. No study mentions blinding of patients for the allocated
immunomodulation caused by blood transfusion may contribute intervention. Because all studies are randomized controlled trials,
to wound healing distress and infection, perhaps explaining the confounding is not likely to occur. Randomization sequence was
higher infection rate in the liberal fluid therapy group. generated by a computer in all studies, and allocation conceal-
Finally fluid overload may interfere with coagulation. Crys- ment was done by opaque, sealed envelopes. The overall bias
talloids have shown to promote a hypercoagulant state, possibly of the individual studies is considered to be low.
predisposing to thromboembolic events [5,34,35]. The exact We are not the first to analyze this subject. Although other
mechanism remains unclear, but it may be due to dilution of an- reviews exist, we do contribute to the subject with our review
ticoagulants such as antitrombin III and protein C [34]. If the di- as more recent studies have been added. The most known re-
lution is more pronounced, the combined effects result in a views on the subject are Corcoran et al [39], Boland et al
coagulopathy which might promote bleeding [36]. Hydro- [40], and Bundgaard-Nielsen et al [41]. Corcoran et al com-
xyethyl starches (HES) are known to interfere with platelet func- pare goal-directed, liberal, and restrictive regimes in which
tion, von Willebrand factor, and factor VIII and protein C they conclude a goal-directed regime is superior to a liberal re-
coagulation cascade promoting a hypocoagulant state in even gime. They do not comment on liberal vs restrictive regime
small quantities [37,38]. In this review, the HES effects are not compared to each other. Since Corcoran et al and
pronounced as only 5 [17,18] used HES preparations. Between Bundgaard-Nielsen et al, the studies of Gao et al, Abraham-
Holte et al (colonic surgery) [17] and Holte et al (knee ahtro- Nordling et al, Matot et al, and Peng et al have been published
plasty) [18], the amount given per policy did not differ between [13,15,21,24]. Therefore, we were able to peform a meta-
the study groups. Gao et al [15], Lobo et al [20], and Peng et al analysis in contrast to Bundgaard-Nielsen et al [41]. Boland
[24] gave significant different amounts of colloids to each study et al [40] restrict itself to solely abdominal surgery. Three ran-
group. In all 5 studies, there was no difference in hypercoagulant domized controlled trials were included in Boland et al and
events or hypocoagulant events. Corcoran et al which we did not include: Vermeulen et al
[42], Gonzalez et al [43], and Mackay et al [44]. These 3 trials
studied the effects of postoperative fluid management; there-
fore, they do not meet our inclusion criteria. The latest review
5. Limitations available is Eng et al [45] which subspecializes in colorectal
surgery and pancreatic surgery. Eng et al also include
The risk of bias is fixed at a low level because we only in- Doppler-guided therapy in their review. By doing so, they do
cluded randomized controlled trials. However, there are some not research the true effect of a liberal vs a restrictive fluid
limitations to be discussed. The heterogeneity of the included management intraoperatively. The pancreatic surgery studies
studies might be a limitation. Particularly, the broad variety are all retrospective studies. One is a randomized controlled
of participants and types of surgery can possibly influence trial analyzing the effects of different crystalloids used instead
the results. Most of the studies are single center which can of the different amount administered.
add to the heterogeneity. In contrast, the shown advantage of
restrictive fluid management in all these different patient
groups induces generalizability to the world's population, im- 6. Conclusions
proving the external validity of the results. The statistical anal-
ysis in the meta-analysis is low for heterogeneity suggesting
Restrictive fluid management policy in comparison with a
the results are valid for the overall population.
liberal fluid management policy during elective surgery gener-
In addition, the risk of bias in individual studies needs to be
ally led to fewer complications within 30 days after the proce-
addressed. The results of quality assessment are shown in
dure, a lower infection rate, and a lower need for blood
Appendix 1. Overall, the quality of Brandstrup et al [14] and
transfusion. We therefore advocate a more restrictive filling
Peng et al [24] are best and worst appraised, respectively. It
policy in elective surgery.
has to be taken into account that not every item has the same
weight. Peng et al and Holte et al (knee and colonic) clearly de-
scribe how and when participants were selected. In all other
studies, the selection procedures are less clearly described. An- Acknowledgments
other potential threat for overall validity is information bias.
Blinded assessment of outcome measuresis done in all studies. There are no acknowledgments.
36
Appendix 1. Quality assessment
2. Elevated temperature
and positive culture
Holte et al 2004 n/a n/a n/a n/a n/a
(laparoscopic) [16]
Holte et al 2007 (colonic Wound requiring Temperature N38°C, clin- n/a n/a n/a
surgery) [17] drainage ical signs, positive x-ray
Holte et al 2007 (knee n/a n/a n/a n/a n/a
arthroplasty) [18]
Kabon et al 2005 [19] Purulent exudate and a n/a n/a n/a n/a
positive culture. 1992
revision if CDC criteria
for a period of 15 days.
ASEPSIS system
Lobo et al 2011 [20] CDC criteria for CDC criteria for Medical guidelines of CDC criteria for CDC criteria
infections infections American College of infections for infections
Chest Physicians AND
Society of Critical Care
Matot et al 2012 [21] Pus and positive Radiographic findings Bacterial infection plus Positive urinary Requiring
culture (new infiltrate) plus 2 of 2 clinical signs (hypo/ culture with clinical surgery
the following: hyperthermia, symptoms (dysuria,
temperature N38°C, tachycardia, tachypnea, frequency, fever) or
leucocytosis, positive leucocytosis or leucocytosis or urinary
sputum culture leucopenia analysis with bacterial
count N100,000.
McArdle et al 2009 [22] Infection requiring 2 of 3 criteria: temperature Clinical signs and n/a
drainage N38°C, clinical signs, temperature N38°C or
positive x-ray. b36°C
Nisanevich et al 2005 Pus and positive Radiographic findings Positive urinary
Bacterial infection plus Requiring
[23] culture (new infiltrate) plus 2 of 2 clinical signs (hypo/culture with clinical surgery
the following: hyperthermia, symptoms (dysuria,
temperature N38°C, tachycardia, tachypnea,frequency, fever) or
leucocytosis, positive leucocytosis or leucocytosis or urinary
sputum culture leucopenia) analysis with bacterial
count N100,000.
Peng et al 2013 [24] Surgical removal of 1. Elevated temperature Positive blood culture Elevated temperature, Reoperation
pus and positive and radiographic with or without DIC or dysuria, and positive
culture findings multiorgan culture
dysfunction.
OR
2. Elevated temperature
and positive culture
38 P.B.B. Schol et al.
[42] Vermeulen H, Hofland J, Legemate DA, Ubbink DT. Intravenous fluid [44] MacKay G, Fearon K, McConnachie A, Serpell MG, Molloy RG,
restriction after major abdominal surgery: a randomized blinded clinical O'Dwyer PJ. Randomized clinical trial of the effect of postoperative in-
trial. Trials 2009;10:50. travenous fluid restriction on recovery after elective colorectal surgery.
[43] Gonzalez-Fajardo JA, Mengibar L, Brizuela JA, Castrodeza J, Vaquero- Br J Surg 2006;93:1469-74.
Puerta C. Effect of postoperative restrictive fluid therapy in the recovery [45] Eng OS, Melstrom LG, Carpizo DR. The relationship of perioperative
of patients with abdominal vascular surgery. Eur J Vasc Endovasc Surg fluid administration to outcomes in colorectal and pancreatic surgery: a
2009;37:538-43. review of the literature. J Surg Oncol 2015;111:472-7.