Hemodynamic Responses to Crystalloid and Colloid Fluid Boluses during Noncardiac Surgery

We present a planned subanalysis of a large multicenter randomized trial that compared goal-directed administration of hydroxyethyl starch and lactated Ringer’s solution on a composite of cardiac, pulmonary, gastrointestinal, renal, infections, and coagulation complications. There were no statistically significant or clinically meaningful differences in postoperative minor or major complications with either fluid.⁶  The original trial was conducted after approval of the institutional review board at three participating institutions. The trial was registered at ClinicalTrials.gov NCT01195883 (registered on September 6, 2010; principal investigator: Andrea Kurz) and EudraCT 2005-004602-86. Written informed consent was obtained from all patients before randomization.

The original trial enrolled adults less than 80 yr old who had elective moderate-to-high-risk abdominal surgeries expected to last at least 2 h. We excluded patients with an American Society of Anesthesiologists physical score higher than III, a body mass index greater than 35 kg · m^–2, compromised kidney function (defined as a creatinine clearance less than 30 ml · min^–1), an estimated cardiac ejection fraction lower than 35%, severe chronic obstructive pulmonary disease, documented coagulopathies, and esophageal or aortic abnormalities.

Patients were given 5 to 7 ml · kg^–1 of lactated Ringer’s solution during induction of anesthesia, which was followed by a continuous infusion of 3 to 5 ml · kg^–1 · h^–1 of a balanced crystalloid solution, normalized to ideal body weight, throughout surgery. The patients were randomized to groups receiving goal-directed fluid bolus administration of 250 ml with either balanced crystalloid (lactated Ringer’s solution) or colloid (hydroxyethyl starch 6%) solutions.

Fluid boluses were administered in response to stroke volume and corrected aortic flow times in accordance with a previously established algorithm.¹⁰  Specifically, we gave a bolus of the designated fluid when the corrected flow time was less than 350 ms. Boluses were repeated as necessary until no further increase in stroke volume was observed. Vasopressors were given as necessary for hypotension resistant to fluid administration. The corrected flow time and the cardiac index were typically measured at 10-min intervals, before and after fluid boluses, and more often if clinically indicated. The corrected flow time is inversely related to afterload and resistance. The optimized cardiac index is the clinical outcome of interest that resulted from goal-directed fluid management. We provide a detailed description of our study procedure in the Supplemental Digital Content (http://links.lww.com/ALN/C738, eAppendix 1).

The time weighted average of the cardiac index was estimated as the average cardiac index for a patient during the surgery while assuming a straight-line relationship between consecutive cardiac index measurements:

where cardiac index_k is the kth cardiac index measurement, t_k is the time of the kth measurement, T is the total time from the first to last observation, and WA is the weighted average. Cardiac index measurements less than 0.8 l · min^–1 · m^–2 or greater than 8 l · min^–1 · m^–2 were excluded as presumed recording errors.

One of our secondary outcomes was the generalized average real variability (ARV) of MAP, which was computed as:

where T is the number of minutes from the first blood pressure measurement to the last, and BP_k is the kth blood pressure measurement. MAP values exceeding 250 mmHg were assumed to be recording errors and excluded.

The statistical analysis plan was completed before accessing data for this analysis. At the Cleveland Clinic (Cleveland, Ohio), hemodynamic data were recorded electronically. The automatic data transfer into our database was lacking for some patients; therefore, no hemodynamic data were available in approximately 7.9% of the patients. Our analysis was therefore restricted to paired corrected flow time and MAP measurements with time stamps.

In the original trial, the randomized groups were well balanced on most of the demographic factors collected. The absolute standardized difference for smoking status was 0.12, and all others had absolute standardized differences less than 0.1. Our current primary and secondary analyses required us to remove patients with missing outcome variables, potentially altering the baseline balance. Our strategy to control confounding in all analyses was as follows: (1) if none of the baseline factors had an absolute standardized difference greater than 0.1, no adjustments would be made; (2) if only a few baseline factors were imbalanced, they would be included in our models; or (3) otherwise we had planned to develop a propensity score model and use inverse probability of treatment weighting when assessing the treatment effect on outcome variables. Propensity modeling proved unnecessary.

We assessed the treatment effect on mean time-weighted average of the cardiac index using a two-sample t test either with or without inverse weighting by the propensity score, as indicated. If that outcome was not normally distributed, we would attempt a transformation. If still not normally distributed, our plan was to use a Wilcoxon rank-sum test. If only a few variables were imbalanced, the treatment effect was assessed in a multivariable linear model while adjusting for the imbalanced variables.

We repeated the primary analysis using imputed values for the time-weighted average of the cardiac index. Conservatively, patients with missing outcomes were assigned to the overall (for combined groups) 75th percentile time-weighted average of the cardiac index if they were in the crystalloid group and the 25th percentile if they were in the colloid group. This imputation and analysis were repeated using the overall largest and smallest observed values.

As an additional investigation into possible effects of missing data, we created univariate logistic models predicting the presence of a missing time-weighted average of the cardiac index as a function of the baseline variables. We then repeated the primary analysis adjusting for all baseline variables that were associated with missing outcomes at the α = 0.05 level.

In addition to analyzing the time-weighted average of the cardiac index, we modeled the intraoperative cardiac index as a time series. A repeated-measures model with an autoregressive correlation structure was fit to adjust for within-patient correlation. Missing outcomes were not imputed for this analysis.

We assessed the immediate effect of a bolus by computing the change in the cardiac index from before to after bolus administration. Therefore, we measured the averaged difference between the cardiac index measured before a bolus was given and of the next available cardiac index measured after the end of a bolus. Responses to boluses of each fluid were compared with a repeated-measures linear model with unstructured or autoregressive correlation to account for correlation within a patient’s repeated measurements.

To evaluate the time that elapsed between the end of one bolus and the beginning of the next, we conducted a time-to-event analysis. The last bolus for each patient was considered right censored at the end of surgery. A Cox proportional hazards frailty model was created to assess differences in time until the next bolus between groups. The frailty model considers patients as a random effect to account for correlation in the repeated measurements within subjects. We also assessed the interaction between the sequential (i.e., first, second, third, …) bolus number and the treatment group. Therefore, we categorized the number of boluses into rough quartiles and analyzed the effect of colloids versus crystalloids on the duration on the effect of a bolus for each bolus number grouping. We also compared colloids versus crystalloids on the number of boluses using the Wilcoxon rank-sum test and Hodges–Lehman estimator of median difference.

We planned to compare differences in the average real variability of MAP with each fluid using a two-sample independent t test if the groups were balanced on baseline characteristics, a linear model if only a few factors were imbalanced, and a weighted t test if propensity score methods were required. Statistical software from SAS (USA) and the R programming language were used for all analyses. A significance level of α = 0.05 was used for all hypotheses. The statistical analysis plan can be seen in the Supplemental Digital Content (http://links.lww.com/ALN/C738, eAppendix 2).

In the primary analysis, 973 patients (480 colloids and 493 crystalloids) were available. For this analysis, we had 90% power at the 0.05 significance level to detect a Cohen’s d = 0.21 for the time-weighted average cardiac index using a two-sample t test. A previous study by Szabó et al.¹¹  found the SD of the cardiac index to be 0.71, which translates to 90% power to detect a difference in the cardiac index of 0.15 l · min^–1 · m^–2.

Clinically meaningful differences were not defined a priori. Post hoc, we defined a difference in the cardiac index of at least 0.5 l · min^–1 · m^–2 as clinical meaningful. The study by Noblett et al.¹²  showed that goal-directed fluid management significantly reduced hospital stay and decreased morbidity in patients undergoing elective colorectal resection (cardiac index, 3.8 ± 1.3 l · min^–1 · m^–2 in the goal-directed group and 3.2 ± 1.2 l · min^–1 · m^–2 in the standard or care group).

Among 1,057 patients in the original trial, 7.9% had no time-stamped cardiac index measurements (43 belonged to the colloid group and 41 were from the crystalloid group). There were thus 973 patients available for the primary analysis. Their baseline characteristics and intraoperative factors are given in table 1. Our restriction on the range of cardiac index measurements between 0.8 and 8 l · min^–1 · m^–2 and MAP less than 250 mmHg removed less than 0.5% of measurements and no patients from analysis.

Of all the confounding factors, the absolute standardized difference was 0.11 for smoking status and 0.10 for preoperative hemoglobin. Consequently, we adjusted for both in our primary analysis and in secondary analysis 3. Four patients had missing hemoglobin measurements, and one had an unreasonably high value. These patients were assigned the median hemoglobin value for the entire sample. The subset of patients with at least one bolus administration was only imbalanced on smoking status (absolute standardized difference equal to 0.11), and thus this was the only variable adjusted for in secondary analyses 1 and 2.

Our primary analysis used a linear model with adjustments for smoking status and preoperative hemoglobin to assess the mean difference in time-weighted average of the cardiac index between patients randomized to crystalloid- or colloid- guided fluid management. The time-weighted average of the cardiac index was slightly greater in patients given colloid, only by 0.20 (95% CI, 0.11 to 0.29) l · min^–1 · m^–2 (fig. 1; table 2).

We conducted sensitivity analyses on the time-weighted average of the cardiac index using imputed missing values at the quartiles, at the extremes, and with adjustments for variables associated with missing outcomes. We also modeled the cardiac index as a time series (instead of using time-weighted average for each patient) using a repeated-measures model with an autoregressive correlation structure. All of our sensitivity analyses except imputation at the extremes were consistent with the primary results. Imputation at the extremes yielded a significant but opposite result (table 2). This “opposite” result was due to this analysis being very conservative, assigning a “worst” value (43 patients in the colloid and 41 patients in the crystalloid had no cardiac index measurements available).

We excluded 13.7% patients who received no fluid boluses throughout surgery from two secondary analyses: the immediate effect of each bolus administration on the cardiac index and the duration of action of each bolus administration. We therefore included 912 patients in these analyses.

The cardiac index immediately after boluses increased only very slightly more after colloid than crystalloid, with the difference in l · min^–1 · m^–2 (95% CI) being only 0.09 (0.06 to 0.12; P < 0.001). The estimated mean (SE) increase of the cardiac index after a bolus was 0.37 (0.02) l · min^–1 · m^–2 for the crystalloid group and 0.46 (0.02) l · min^–1 · m^–2 for the colloid group.

In the Cox proportional hazards frailty model, the hazard for needing additional boluses in the colloid group was lower: 0.60 (0.55 to 0.66; P < 0.001; table 3). The data exhibited minor deviations from the proportional hazard assumption when examined graphically (fig. 2). In addition, the median [quartiles] number of boluses per patient was 4 [2, 6] for colloids and 6 [3, 8] for crystalloids, with a median difference (95% CI) of –1.5 (–2 to –1; P < 0.001; Wilcoxon rank sum test). Furthermore, we found a significant interaction between the fluid group and the number of boluses (hazard ratio [95% CI], 0.97 [0.94 to 0.99]; P = 0.019), indicating that with each additional bolus, the likelihood to need a further bolus was lower in patients receiving colloids than those receiving crystalloids. Patients given colloids also had a significantly lower hazard at each bolus count category. When pairwise comparisons were made, the effect of colloids was found to be different only between the first and third quartiles (P = 0.004; fig. 3).

For our third secondary analysis, we excluded 86 patients because no MAP values were available. Therefore, 971 were analyzed. We found no significant difference in average real variability of MAP, with the difference being only –0.03 (–0.07 to 0.02; P = 0.229; table 3).

Intraoperative hemodynamics and fluid management have a substantial impact on postoperative outcomes.^13,14  Excessive or insufficient fluid administration during surgery is associated with increased morbidity and mortality.¹⁵ 

The administration of a colloid-based goal-directed fluid regimen during surgery did not result in a clinically meaningful higher time-weighted average of the cardiac index as compared to the administration of a crystalloid based goal-directed fluid regimen in patients undergoing moderate- to high-risk abdominal surgery. We found that the cardiac index increased significantly more immediately after a colloid bolus administration as compared to a crystalloid bolus administration. Furthermore, the need for an additional fluid bolus after the administration of a colloid bolus was significantly lower as compared to when a crystalloid fluid bolus was administered. There was no significant effect in time-weighted average MAP, as well as in average real variability MAP.

Evidence that the administration of goal-directed colloid solutions reveals clinical negligible effects on hemodynamic parameters is important and might raise the question of whether it is the blood pressure we should treat perioperatively or whether we should rather follow flow-based parameters. We saw a statistically significant but clinically unimportant difference in the flow-driven parameters and no difference in blood pressure. Thus, it probably is not surprising that we, in our main trial, did not see a difference in our primary outcome, which was a composite of major postoperative complications.⁶  However, we cannot forget that our main trial was a comparison between crystalloids and colloids, both of which were administered guided by esophageal Doppler to a specific goal. Our current study also evaluated whether this goal is reached faster with colloids as supposed to crystalloids.

Colloidal solutions are used during surgery with the aim to increase cardiac output and further to reduce the amount of intraoperative fluids needed to provide hemodynamic stability. Hydroxyethyl starch, for example, has a high molecular weight, which makes it more difficult to cross the glycocalyx barrier, and therefore it is more likely to stay in the intravascular space.^7,9  Although some trials, for example our main trial, found no differences between crystalloids and colloids in regards to hemodynamic parameters and on postoperative outcome,^6,16,17  colloids are still commonly used by many clinicians.

The study of Yates et al.,¹⁶  which included 202 patients undergoing colorectal surgery, evaluated the effect of goal-directed crystalloid versus hydroxyethyl starch bolus administration on postoperative gastrointestinal function. Yates et al.¹⁶  found no significant difference in recovery of gastrointestinal function, and furthermore there was no difference in stroke volume and stroke volume variation between the groups. The FLASH trial (Fluid Loading in Abdominal Surgery: Saline versus Hydroxyethyl Starch), is the most recent study and included 775 patients. High-risk patients undergoing major abdominal surgery were included to evaluate the effect of goal-directed hydroxyethyl starch versus saline administration on a composite of postoperative complications and death.¹⁷  Consistent with our original trial, they also found that the intraoperative amount of fluids was significantly lower in the colloid group as compared to the crystalloid group.^6,17  Furthermore, intraoperative stroke volume index was significantly higher in the colloid group as compared to the crystalloid group. Nevertheless, similar to our main trial, they found no significant differences in their composite outcome, which included acute kidney injury, acute respiratory failure, acute heart failure, major sepsis complication, unplanned reoperations, and death between both groups.

In the FLASH trial, hemodynamic parameters increased significantly when hydroxyethyl starch was administered; however, the effect was clinically negligible. In detail, mean stroke volume index was only 4 ml · m^–2 higher in the colloid group (47 ± 13 ml · m^–2) as compared to the crystalloid group (43 ± 13 ml · m^–2).¹⁷  Similarly, in our study, the significant difference in time-weighted average of the cardiac index was only 0.20 l · min^–1 · m^–2. This means that the time-weighted average in the cardiac index was only 200 ml higher in the colloid group as compared to the crystalloid group. In this context, the net effect of intraoperative goal-directed hydroxyethyl starch administration on the cardiac index is too small to have a relevant impact on clinical outcome. This might be one explanation for the negative results of the FLASH trial and of our main trial.^6,17  There was no significant difference between goal-directed colloid and crystalloid fluid administration on postoperative morbidity and mortality.^6,18 

The OPTIMIZE trial (optimization of perioperative cardiovascular management to improve surgical outcome) included 734 high-risk patients undergoing major gastrointestinal surgery. Patients were randomized to receive intravenous fluid and inotropes guided by cardiac output or to receive intravenous fluid according to clinical standard of care.¹⁸  The type of intraoperative fluids were not specified and were at the discretion of the attending physician. Patients in the cardiac output–guided group received significantly larger amounts of colloids as compared to the standard of care group (1,250 ml in the goal-directed group vs. 500 ml in the standard of care group). Interestingly, there was no significant effect of cardiac output guided fluid administration on the composite of major postoperative complications and 30-day mortality. However, the authors indicated that goal-directed fluid therapy was associated with lower complication rates. Furthermore, they suggested that cardiac output-guided fluid therapy led to a more individualized approach, achieving the correct dose of fluids.¹⁹  It therefore seemed reasonable that the administration of fluids during surgery in a goal- directed way might be more important, whereas the type of fluid administered during surgery plays a much smaller role than initially assumed. This might be especially true in patients with significant cardiovascular comorbidities having moderate to major surgery.²⁰ 

Our study has some limitations. Due to associations between missing primary outcomes and baseline characteristics, the generalizability of these results may be slightly different than for the original trial. The results of our sensitivity analyses adjusting for these variables were consistent with the primary analysis results, so it is unlikely that the results are biased due to missing data.

Because we performed goal-directed fluid management in both study groups, a difference in the cardiac index might have been unlikely. Nevertheless, we administered fluid boluses in response to corrected flow time. Corrected flow time is inversely related to afterload/resistance. A low corrected flow time is most commonly an indicator for low preload and therefore an indicator for fluid administration. However, similar corrected flow time values between both groups did not necessarily mean that there were no differences in the cardiac index. Because organ perfusion and oxygen delivery is dependent on the cardiac index, we assumed that differences in the intraoperative cardiac index between the colloid and crystalloid group might be more important and clinically meaningful than differences in corrected flow time values. Therefore, we used the cardiac index as our primary outcome.

We used boluses of 250 ml for our intraoperative goal- directed fluid management. This is a larger volume as compared to the amount of fluid bolus volume of more recent trials performing goal-directed fluid therapy.²¹  Our study was conducted during a time when 250 ml was used as standard of care for goal-directed fluid algorithms. Specifically, we used the previously published fluid management algorithm by Gan et al.,¹⁰  who showed that esophagus guided fluid bolus administration of 250 ml significantly reduced the length of hospital stay in patients undergoing major abdominal surgery.

Although it is becoming more evident that the perioperative administration of starch solutions is not superior as compared to the administration of balanced crystalloid solutions, they are still commonly used during surgery. As has been shown in the FLASH trial and in our main trial, there was no difference in intraoperative and postoperative major and minor complications within 30 days after surgery.^6,17  Nevertheless, it is still believed that at least the hemodynamic effects of hydroxyethyl starch solutions are superior to crystalloid solutions. Our substudy showed that the measured differences in hemodynamic parameters are clinically negligible. This might explain why we did not see an effect in our main trial. Taking this into account and further that hydroxyethyl starch solutions are more expensive and still carry the risk of serious side effects, they should not be administered routinely in patients undergoing moderate abdominal surgery.

Overall, this is the largest study investigating the hemodynamic effects of goal-directed colloid administration versus goal-directed crystalloid administration during surgery. We were able to show differences in pharmacokinetics between crystalloids colloids. We found that the duration of action and the intravascular stay was significantly longer in the colloid group, which resulted in fewer boluses needed in the colloid group. Furthermore, the hazard for a further bolus was significantly lower in the colloid group. Although colloids remain longer in the intravascular space and reduce the amount of needed volume to maintain hemodynamic stability, there is still no effect on morbidity and mortality.^6,17 

In summary, we evaluated hemodynamic data from a large trial showing that goal-directed colloid administration is not superior to goal directed crystalloid administration in regard to a composite of major complications. Although we found a statistically significant difference in time-weighted average of the cardiac index in our study, the effect was clinically negligible and does not support the administration of hydroxyethyl starch in this patient population.

Supported in part by Fresenius Kabi (Bad Homburg, Germany). Deltex Medical (Chichester, United Kingdom) provided the esophageal Doppler monitors and disposables.

Dr. Sessler has indicated financial relationships with Edwards Lifesciences (Irvine, California); Sensifree (Cupertino, California); and Perceptive Medical (Newport Beach, California), advisory board. Dr. Maheshwari has indicated a financial relationship with Edwards Lifesciences. The other authors declare no competing interests.

Full protocol available at: ak@or.org. Raw data available at: ak@or.org.

Cleveland Clinic, Cleveland, Ohio

Kamal Maheshwari, M.D.

Michael Kot, M.D.

Tatyana Kopyeva, M.D.

Andrea Kurz, M.D.

Edward J. Mascha, Ph.D.

Amanda Naylor, M.A.

tAtila Podolyak, M.D.

Daniel I. Sessler, M.D.

Dongsheng Yang, M.S.

Sven Halvorson, M.S.

Medical University of Vienna, Vienna, Austria

Edith Fleischmann, M.D.

Barbara Kabon, M.D.

Christian Reiterer, M.D.

Oliver M. Zotti, B.S.

Mina Obradovic, M.D.

Florian Luf, M.D.

Jakob Muehlbacher, M.D.

Samir Sljivic, M.D.

Ahmed Bayoumi, M.D.

Corinna Marschalek, M.D.

Klaus Eredics, M.D.

Alexander Taschner, M.D.