Does Central Venous Oxygen Saturation-directed Fluid Therapy Affect Postoperative Morbidity after Colorectal Surgery?: A Randomized Assessor-blinded Controlled Trial

• ❖ Whether restricted, goal-directed fluid administration improves outcomes is unclear.

• ❖ In patients undergoing elective open bowel surgery in whom a restricted fluid regimen guided by Scvo², using colloid boluses compared with traditional crystalloid fluid administration, was used, there was no differences in the rate of perioperative complications despite significant differences in the quantity of fluid administered.

THE optimal amount of intravenous fluid for surgical patients is still vigorously debated.1,2The rationale behind the traditional liberal fluid regimen has been questioned. Recent studies with a restricted or an individualized, goal-directed fluid administration have shown better outcomes.3–8Because there is no simple way to predict who will respond to extra fluid by increasing cardiac output, a pragmatic approach is to observe the result of an intravenous bolus on a chosen target. Many of the studies on goal-directed fluid therapy have used a target derived from esophageal Doppler ultrasound, mostly stroke volume. This procedure is minimally invasive but requires expensive equipment and extensive training and has not gained widespread use in daily practice.

During anesthesia, when oxygen consumption is low and steady, it is reasonable to assume that central venous oxygen saturation (Scvo²) reflects cardiac output and oxygen supply. Scvo²may therefore be a useful physiologic indicator to guide fluid responsiveness and administration. After cardiac surgery, Scvo²correlated well with cardiac output after administration of fluid boluses.9In high-risk surgical patients, a low perioperative Scvo²was found to be a predictor of postoperative complications,10,11implying the possibility of improving outcome after surgery by treating low values.

Outcome studies for ordinary surgical patients receiving Scvo²-guided fluid therapy have not yet been published, but in high-risk patients, it may have shown a benefit.12In search for a low-cost fluid management strategy useful for daily practice, we conducted a prospective randomized trial in patients undergoing elective open bowel surgery. The aim of the study was to assess whether a restricted fluid regimen guided by Scvo²would reduce the rate of postoperative complications compared with our traditional fluid treatment.

Patients scheduled for elective open colorectal and lower intestinal surgery in a university and a municipal hospital of the same health trust were assessed for eligibility. Exclusion criteria were as follows: age less than 18 yr; inability to give informed consent; serum creatinine above 177 μM; serious stenotic, valvular heart disease; and receive anticoagulant therapy, the latter because large amounts of hydroxyethyl starch (HES) might increase the risk of epidural bleeding.13Patients were included consecutively whenever one of the investigating anesthesiologists was present.

Patients were randomized into two groups, Scvo²-guided fluid therapy (Scvo²group) or traditional fluid therapy (control group), immediately before induction of anesthesia. Randomization was performed by a web-based system developed and administered by the Unit of Applied Clinical Research, Norwegian University of Science and Technology (Trondheim, Norway). It was based on blocks of 8 to 20 patients and stratified for hospitals and for rectal surgery versus  more proximal bowel surgery. The risk of complications was assumed to be higher in rectal surgery.14 

The Scvo²group was treated with a low crystalloid fluid supply and additional colloid boluses guided by Scvo². The control group had an effluent crystalloid volume based on estimated losses and maintenance, hemodynamic variables, and urine output.

The study was approved by the Regional Committee for Medical Research Ethics in Western Norway, University of Bergen, Bergen, Norway (ClinicalTrials.govID: NCT00468793), and written informed consent was obtained from all participants.

In the Scvo²group, a central venous line was inserted in the right internal jugular vein after induction of anesthesia. A blood sample was drawn for Scvo²analysis and processed immediately in a blood-gas analyzer (ABL 700; Radiometer, Brønshøy, Denmark; or Cobas b221; Roche, Basel, Switzerland). Daily error and reliability checks were performed by a trained lab technician in addition to the hourly self-calibration tests. Coefficient of variation for measurement of hemoglobin oxygen saturation is specified to 0.5%. A 75% Scvo²cut-off was chosen based on known physiologic data.10,11,15Further samples were taken and colloid boluses administered according to the flow chart shown in fig. 1. A mean arterial pressure below 60 mmHg would usually precipitate a new Scvo²analysis, but the exact limit was left to the anesthesiologist. After arrival to the recovery area, Scvo²was analyzed within 30 to 60 min, and a colloid bolus was given if saturation was less than 75%. If the patient displayed clinical signs of hypovolemia (i.e. , decline in blood pressure, urine output, and peripheral circulation), a sample was drawn and a bolus of fluid was given if Scvo²was less than 75%. The central venous line was removed the next morning (day 1).

Oral bowel preparation (sodium phosphate [Phosphoral®]; Casen-Fleet, Madrid, Spain) the afternoon before surgery was left to the discretion of the surgeon. Each patient was allowed to drink water until 2 h before surgery and was asked to empty his or her bladder. Every patient was weighed upon arrival in the operating theater. An epidural was inserted in a lower thoracic interspace. The catheter was tested and then maintained with a continuous infusion of bupivacaine 1 mg/ml, fentanyl 0.002 mg/ml, and epinephrine 0.002 mg/ml, usually at 8 ml/h with adjustments, if required, perioperatively and postoperatively. Pulse oximetry saturation (Spo²) was recorded with the patient supine on the operating table. Standard monitoring included continuous electrocardiography, noninvasive arterial pressure, capnography, and pulse oximetry. We used invasive blood pressure only if indicated. Anesthesia was induced with thiopental or propofol, fentanyl, and vecuronium. The patients from the municipal hospital were maintained with propofol and remifentanil at continuous infusion, whereas most patients in the university hospital received isoflurane and fentanyl. Nitrous oxide was not used. The fraction of inspired oxygen was increased to higher than 35% if needed to keep Spo²of at least 95%. Ventilation was adjusted to pco²30 to 37 mmHg. Patients were actively warmed with intravenous fluid warming and forced air, and temperature was maintained above 35.5°C. Low molecular weight heparin was given as thromboprophylaxis. The anesthesiologist decided when to transfuse blood or use a vasopressor. A bolus of ephedrine or metaoxedrine was the first choice; norepinephrine was used for continuous infusion.

All patients received intravenous metronidazole 1,500 mg and doxycycline 400 mg or cephalothin 2 g as antibiotic prophylaxis. This administration implied approximately 800 ml crystalloid. The amount of fluids given on the day of operation (day 0) was summarized at 8:00 am on day 1. Thereafter, intravenous fluid prescription was left to the surgeons, unaware of the patient's allocation.

Those assigned to the Scvo²group treated with a bowel preparation had 500 ml Ringer's acetate (Fresenius-Kabi, Halden, Norway) to compensate for the fluid deficit.16They were maintained on Ringer's acetate 100 ml/h. HES 130/0.4 60 g/l in 0.9% NaCl (Venofundin®; B Braun, Melsungen, Germany) was given as a 3 ml/kg bolus over 10 to 15 min if required according to the algorithm (fig. 1). To restrict crystalloid supply, bleeding was compensated by HES 1:1. Postoperatively, the patients had glucose 5% 80 ml/h and Ringer's acetate 1:1 only to replace losses on stomas or drains. Extra fluid was given if there were clinical signs of hypovolemia and Scvo²was less than 75%. Total HES supply was limited to 50 ml · kg⁻¹· 24 h⁻¹.

The control group was managed according to a traditional fluid scheme,17(i.e. , 1,000 ml Ringer's acetate loading during the first hour in theater followed by approximately 800 ml/h (10–12 ml · kg⁻¹· h⁻¹)). The rate of infusion was increased if urine excretion dropped below 0.5 ml · kg⁻¹· h⁻¹or the arterial pressure declined. The first 500 ml of bleeding was compensated for 1:4 by Ringer's acetate; bleeding of more than 500 ml was compensated for by 1:1 HES. HES was otherwise occasionally given perioperatively or postoperatively when considered necessary by the anesthesiologist as a result of a drop in blood pressure. Postoperatively, 5% glucose 1,000 ml was prescribed until the next morning, and extra fluid was given after clinical signs of hypovolemia.

In the morning on day 1, after a period of at least 5 min without supplemental oxygen, Spo²was recorded by the attending nurse. Patients were weighed on the same electronic scale as before surgery by a physiotherapist or a nurse. If they were unable to stand upright, weighing later in the morning was tried once. A serum creatinine analysis was scheduled for postoperative day 3 and then recorded from the patient's chart.

Within 4 to 6 weeks, all but four patients had an ambulatory follow up in the hospital by the operating surgeon. The other four patients were followed-up with the use of a structured telephone interview by a nurse only. The postoperative status was routinely documented in the patient's chart. The participating surgeon registered postoperative complications from the patient's chart after follow up according to a predetermined and defined list of complications (table 1). These were chosen to have at least a possible connection with fluid management and most of them required interventions and/or treatment. The perioperative record data and the patient's group assignment were not immediately available and thus blinding was assured.

The primary endpoint of the study was whether the patient had at least one postoperative complication within 30 days. Calculation of sample size was based on an anticipated 30% frequency of patients with one or more postoperative complications. To show a 50% reduction with 80% power at 5% significance level (two-sided), 119 patients had to be included in each group. Secondary endpoints were postoperative serum creatinine, Spo², and weight.

Data were analyzed in SPSS version 15.0 (SPSS Inc., Chicago, IL). Categorical outcomes were compared with the chi-square test, continuous variables with a two-sided t  test, or the Mann–Whitney test for independent samples, as appropriate. P  less than 0.05 was considered statistically significant. Odds ratio (OR) is reported as the odds of having an unfavorable outcome in the Scvo²group compared with control.

From April 2007 to May 2009, 293 of 403 possible patients were approached and 241 patients were included in the study, 226 from the university hospital and 15 from the municipal hospital. One-hundred twenty-one patients were allocated to the Scvo²group and 120 to the control group (fig. 2). Five patients who had given consent were not randomized for one of the following reasons: investigating anesthesiologist was not available; rescheduled or postponed surgery; or inability to randomize because of data failure. In the Scvo²group, two patients had the surgical procedure converted after randomization—one to a stoma revision and another to laparoscopic surgery—and thus did not fulfill the inclusion criteria, and three patients had an explorative laparotomy instead of the planned procedure. The protocol had to be violated in two patients (Scvo²group) because of rhabdomyolysis in the legs after long-term exposure to the lithotomy position. They were successfully treated postoperatively with alkalization and 4–5 l noncolloid fluid to prevent renal damage. All patients were analyzed in their allocation groups based on intention to treat. In a few patients, secondary outcome variables were lost because of insufficient documentation (fig. 2).

The groups were comparable in most parameters apart from age and type of surgery (tables 2 and 3). The median ages (range) in the Scvo²and control groups were 58 (21–87) and 63 (25–91) yr, respectively. Colectomy due to inflammatory bowel disease was overrepresented in the Scvo²group, and only 5 patients older than 80 yr were recruited in this group compared with 15 in the control group. Estimated blood loss, number of patients given erythrocyte transfusions, and the duration of surgery did not differ significantly between the groups (table 3). All central venous catheters were placed in the superior vena cava, confirmed by postoperative x-ray. No complications were related to the central venous access.

The Scvo²group had less total fluid on the day of operation than the control group (mean difference, 2622 ml; 95% CI, 2275–2968 ml), whereas the amount of colloid given was higher (table 4). The Scvo²group received 10.1 ml/kg (range, 0–30) HES, the control group 6.4 ml/kg (range, 0–29). The lower total fluid supply in the Scvo²group resulted in 76 ml (95% CI, 23–129 ml) less perioperative urine output. The calculated output per kilogram and hour includes urine in the bladder at the time of catheterization (table 4). The loss on stomas and drains was highly variable but similar in the groups.

The number of patients with one or more complications was equal in both groups (OR = 0.98 [95% CI, 0.6–1.6]) (table 1). There was no difference in the incidence of renal impairment defined as more than 33% increase in serum creatinine (OR = 1.6 [95% CI, 0.6–4.3]). In several patients, creatinine increased weeks after the operation because of large losses on stomas. No patient needed renal replacement therapy. Fewer patients in the Scvo²group had postoperative paralytic ileus (i.e. , 5 vs.  17; OR = 0.26 [95% CI, 0.09–0.73]). There were 12 reoperations in the Scvo²group and 14 in the control group. Five patients were treated in the intensive-care unit, mainly because of respiratory failure (table 1). At follow up within 4 to 6 weeks, the number of patients who reported that their wounds had not healed properly was 30 (of 121) in the Scvo²group versus  35 (of 116) in the control group (OR = 0.76 [95% CI, 0.43–1.35]). Mean length of stay in the hospital was 13.3 days in the Scvo²group versus  12.6 in the control group. Both groups had a median length of stay of 11 days (range, 5–63). No patients died within 30 days.

The fluid load resulted in a mean weight gain of 0.8 ± 1.8 kg (mean ± SD) in the Scvo²group and 2.5 ± 1.6 kg in the control group (P < 0.001) (fig. 3). Excluding the two patients in the Scvo²group treated with forced diuresis (4.2 and 4.1 kg increase, respectively) reduced the mean weight gain to 0.7 ± 1.7 kg. The decline in Spo²on the first postoperative day was 2.0% in the Scvo²group and 2.8% in the control group, with a mean difference of 0.75% (95% CI, −0.07–1.58%), which was not significant (P = 0.07). Postoperative serum creatinine was higher in the Scvo²group than in the control group. When preoperative creatinine was subtracted from creatinine on day 3, the difference between groups was 5.6 μM (95% CI, 0.4–10.8 μM) (P = 0.03) (fig. 3).

In this randomized controlled study of patients undergoing open colorectal and lower intestinal surgery, the rate of complications was the same whether fluid was given according to a traditional regimen or followed a restricted regimen supplemented by Scvo²-guided optimization.

Intervention groups given less fluid have shown a better outcome after gastrointestinal surgery.3,5,6Evaporative losses may be less than formerly anticipated, and the concept of a third space has been questioned.18,19After 800 ml antibiotic fluid, we administered 100 ml/h Ringer's acetate perioperatively in our study group. With a mean 150 min duration of surgery, the sum of crystalloid is similar to a suggested maintenance rate of 6 mg · kg⁻¹· h⁻¹based on echocardiography in open colorectal surgery.19We thus managed to limit the amount of fluid compared with our traditional regimen in accordance with trials on fluid restriction.3,5,6In contrast to these studies, we found no benefit of a low-fluid regimen apart from a possible lower incidence of postoperative paralytic ileus in the group given less fluid, supporting the view that fluid overload might impair gastrointestinal motility.5,20 

The control group had what some would define as a liberal, others an excessive, fluid supply.1,3The volume comprises anticipated evaporation, third space loss, maintenance, compensation for reduced cardiac filling as a result of vasodilation, and positive pressure ventilation plus extra fluid, as recommended when urine output or arterial pressure is low. The positive fluid balance following this strategy was reflected in 2.5-kg postoperative weight gain. A similar increase in weight from 1.9 to 2.8 kg has been reported in the high-volume groups in previous trials comparing high and low fluid supply.5–7 

Bellamy has illustrated the association between the risk of adverse effects and fluid supply as a U-curve in a diagram where the y-axis represents postoperative morbidity and the x-axis the fluid load. Both too little and too much fluid would increase the rate of complications.2Bundgaard-Nielsen et al.  21suggested that the shape of the U-curve may vary because of comorbidities and form a V-shape during critical illness, where titrating fluid to the nadir of the curve would be essential to reduce morbidity. In relatively healthy patients, the curve would look like a horizontal parenthesis with little difference in outcome whether fluid supply was optimal or not. This may explain why we found no difference in complication rate in our study.

The best documented method in recent years for goal-directed fluid therapy is esophageal Doppler ultrasound. Although meta-analyses seem to favor a beneficial effect on morbidity in gastrointestinal surgery,3,4,22most of these studies included a small number of patients, and the primary outcome was length of stay and not postoperative complications. The esophageal Doppler ultrasound method, however, is not without problems comprising cost, operator dependency, and need for training,23and an unresolved time to estimate a reliable stroke volume, especially in cases of arrhythmia.22,24It is inconvenient in awake patients and excluded in upper gastric and esophageal surgery. Other methods for goal-directed therapy are based on the arterial waveform. They are under investigation with only few outcome studies yet presented.25In search of an easily available method to individualize fluid therapy, and to secure that a low volume basal fluid infusion would not restrict cardiac output and oxygen supply perioperatively, we chose Scvo²as the target variable.

Central venous oxygen saturation reflects oxygen extraction mostly in the head and upper part of the body. It cannot automatically replace mixed venous oxygen saturation. The relation between them changes in different pathologic conditions.15,26However, for most elective surgical patients, Scvo²will change in accordance with mixed venous oxygen saturation. Pulmonary artery catheters are seldom indicated in gastrointestinal surgery.27We postulated therefore that Scvo²could be a useful physiologic indicator to guide fluid therapy perioperatively in this group of patients.

We chose 75% Scvo²as the goal, although there is not much Scvo²data available in ordinary surgical patients to support that threshold for fluid supply. Extra fluid was supplied when a colloid bolus caused an increase in Scvo²as a result of an increase in cardiac output. Jenstrup et al.  28administered fluid before anesthesia and gastrointestinal surgery based on this principle and obtained a maximal Scvo²of 71.5%. A 100-ml intravascular volume deficit was suggested to correspond to a 1% decline in Scvo².28,29Donati et al.  12used both fluid and inotropic therapy to obtain an oxygen extraction ratio corresponding to 73% Scvo²during intra-abdominal surgery in high-risk patients. Because the amount of fluid in the groups was similar, the use of dobutamine in the intervention group may have been decisive for the reduction in the rate of organ failure from 30 to 12% in their study.12In high-risk surgical patients, Scvo²lower than 73% predicted postoperative complications, and the authors encouraged prospective trials using Scvo²as a therapeutic goal.10In the sepsis algorithm, 70% was chosen as the limit for Scvo².30The benefit of setting a limit as we did, instead of maximizing the target for the individual patient, is unresolved. Our strategy resulted in 120 ml perioperative colloid (in addition to substitution for the bleeding) and a weight gain of 0.8 kg. Using a lower threshold in line with the referred studies on high-risk patients10–12,30would probably have resulted in even less fluid to the Scvo²group.

Postoperative complications within 30 days have been reported to be a strong predictor of long-time survival.31Hence, this might be a more relevant outcome measure than the frequently used length of stay. A postoperative complication has been defined as “any deviation from the normal postoperative course.”14The lack of standardization can explain why a variety of complications with highly variable frequency is reported in fluid-related studies regarding gastrointestinal surgery.5,6,8This makes an exact interpretation and evaluation of the outcome differences difficult. Minor complications such as postoperative nausea, vomiting, and urinary tract infections were not included in our study. The frequency of complications was nevertheless higher than the anticipated 30% complication rate from the power calculation, rendering it possible to detect at least a 42% decrease in rate of complications with 80% power.

HES is potentially nephrotoxic.13Because the restricted crystalloid supply and a high colloid load could increase the risk of renal damage,32we wanted to register all possible renal dysfunction. There is no consensus on a single criterion to define postoperative renal dysfunction, and serum creatinine may not be an ideal parameter.33Patients treated with the Scvo²regimen had a higher serum creatinine on day 3 than the control group. This may reflect a higher strain on the kidneys, but the clinical importance is unclear. However, we could not demonstrate a higher frequency of renal impairment in the Scvo²group within 30 days.

Our trial has several important limitations. Despite general and epidural anesthesia, muscle relaxants, and temperature control, fluctuations in oxygen uptake, which alters Scvo², may occur perioperatively. Other possible confounders (e.g. , triggers for transfusion or vasopressor therapy) were not strictly defined. A regimen based on a restricted crystalloid supply with a goal-directed fluid administration strategy was compared with our traditional regimen. Thus, blood loss was compensated for differently in the groups. Accordingly, like many other goal-directed studies, more colloid was used in the intervention group.4,8In our study, the mean difference between the groups was 4 ml/kg. Apart from staying mostly in the intravascular compartment causing less edema than crystalloids, HES has been claimed to have other positive effects (e.g. , reducing neutrophil adhesion to the endothelial cell layer).13The clinical importance is unresolved, and this potential benefit was not reflected in a better outcome in the Scvo²group. Neither was the possible advantage of lower mean age in this group. There are no standard criteria for assessment of postoperative complications. We registered in retrospect, not at a fixed time point, a predefined set of complications that occurred before day 30. Although we used data from the patient's chart, collected by the clinical team, and therefore believe we detected all important complications, this approach remains subjective and may not be sufficiently robust for a randomized trial of a clinical intervention.

In conclusion, patients receiving fluid therapy aimed at Scvo²of at least 75% did not achieve a more favorable outcome after gastrointestinal surgery than patients treated with a traditional fluid regimen. It is noteworthy that the volume of intravenous fluid administered was less in the Scvo²group. If goal-directed fluid therapy is a key issue in reducing complications after bowel surgery in ordinary patients, this trial cannot support Scvo²with a target of at least 75% as a recommended strategy for fluid administration. Whether a larger study may detect a smaller reduction in complications, or if it may be better to maximize Scvo²rather than setting an upper limit, remain to be shown.

The authors thank Geir Egil Eide, Ph.D. (Statistician Professor, Centre for Clinical Research, Haukeland University Hospital, Bergen, Norway), for statistical support, and all the doctors and nurses from Haukeland University Hospital and Voss Hospital (Voss, Norway) who helped us carry out the study.