There were no differences in the number of postoperative complications or length of hospital stay when GDFT was guided by PVI instead of esophageal Doppler during major open abdominal surgical procedures.
Previous studies show that both length of stay and incidence of complications were reduced when PVI was used as a part of a colorectal surgery multimodal enhanced recovery protocol in 109 patients, compared to historical controls [
24]. The use of PVI also decreased perioperative lactate levels compared to controls [
6,
25]. Warnakulasuriya et al. [
8] reported no difference in short term outcome when PVI was compared to esophageal Doppler in 40 patients. Based on these reports and our findings, a clinician wishing to pursue GDFT can choose PVI over esophageal Doppler in the large majority of patients undergoing major open abdominal surgery. PVI, in contrast to esophageal Doppler, is not sensitive to interference from diathermy and does not require frequent access to the patient’s head for probe repositioning. Also, PVI can be measured without single-use equipment avoiding the cost for single-use esophageal probes (USD 130 in our setting). However, as illustrated by the request for Doppler data in two PVI patients, the clinician might still want to have access to a reliable method for measuring intraoperative cardiac output to increase the amount of hemodynamic information during unusually complex situations or in cases of known vascular and/or myocardial dysfunction. Data on cardiac output and thus contractility, systemic vascular resistance and oxygen delivery were available in the Doppler group but not incorporated in the treatment protocol, and decisions about inotropic and/or vasoactive support in both groups were left to the clinical judgement of the responsible anesthetist. Despite the more extensive hemodynamic information in the Doppler group, beside more use of the vasopressor phenylephrine in the PVI group, the amount of inotropic support did not differ between the treatment groups.
Other non-invasive methods for dynamic monitoring are available. One meta-analysis reported the offline assessment of respiratory variation in pulse oximetry plethysmographic waveform amplitude (∆POP) for predicting fluid responsiveness to be comparable to that of PVI (AUC 0.89 vs 0.95), but this technique is as yet not available in real time [
26]. Also, pulse pressure variation and stroke volume variation can now be obtained non-invasively using the Nexfin/Clearsight™ or the CNAP™ systems, but their capacity for predicting fluid responsiveness differs between studies [
27‐
30]. The same applies to measurements of the variability of the diameter of the caval vein or other veins during positive pressure ventilation [
31]. Contrary to PVI, the effect on clinical outcome of all these techniques has to the best of our knowledge not been studied.
PVI is a dynamic indicator of fluid responsiveness based on cardiopulmonary interaction while stroke volume optimization is not. As both methods are claimed to be of benefit for fluid management, we argue that these methods, though differing in concept, can be compared with each other in a randomized trial focusing on clinical outcome [
3]. Maintaining a patient below a certain PVI value might appear to be less aggressive then aiming for the plateau of the Starling curve, but this depends on the selected cut-off values. Probably, a higher cut off value for PVI with the same settings (such as tidal volume) would have resulted in less fluid administration and vice versa. We have previously shown that a PVI value of 10.5% was optimal in this material [
9].
Limitations
Limitations of this study include the absence of a control group without either PVI or Doppler to guide volume optimization. Early reports on stroke volume optimization using the esophageal Doppler were favorable [
32]. Since then the technique has been questioned, partly because of negative studies on GDFT involving esophageal Doppler [
33] and other devices [
34,
35], and partly because of questions raised about the Doppler method itself [
36]. Based on our results, it is not possible to tell whether PVI and Doppler both result in similar improvements in outcome, or whether neither method improves outcome when compared to treatment without GDFT. Such a control group was reflected upon at the time of designing the study (2011), but deemed unethical since, based on the evidence available at that time, we felt there was no equipoise about the benefits of GDFT. Equipoise about the use of GDFT is supported by our findings previously reported about the limited ability of both PVI and Doppler to predict fluid responsiveness [
9]. Also protocol compliance, in terms of time during surgery with achieved PVI and SV goals respectively, was not recorded.
The amount of fluid used during the optimizations represents a part of the total fluid given, because of the per protocol complementary colloid fluid administration during the induction of anesthesia and for bleeding, and crystalloid fluid for maintenance and for the correction of preoperative dehydration. This increased the possibility that both groups would receive similar amounts of fluid; however, we did not find a difference between the groups in the amount of colloids used specifically for the optimizations.
We selected a difference of 10% in number of complications at 30 days to determine required sample size. Therefore the study was not powered to detect smaller albeit still significant differences in postoperative outcome. Also, it could be discussed whether a non-inferiority approach would have been more adequate. This would entail a larger required sample size.
The studies which were used to determine sample size scored complications in different ways and reported incidences of complications which are higher than can be expected in current practice using improved surgical and anesthetic techniques. The findings of this study cannot be applied to laparoscopic surgery. The majority of patients, although undergoing major surgery, were classified as ASA 1 or 2. The potential of GDFT to improve postoperative outcome is more pronounced in patients at higher risk of complications [
37], and this could have influenced the possibility to find differences in outcome between the methods. Also, since intraoperative blood pressure is correlated to postoperative outcome, a target MAP should have been specified in the protocol.
Since the reliability of dynamic parameters including PVI increases with increasing tidal volume, a tidal volume of 8 ml kg
− 1 (ideal or actual body weight) or more is often recommended [
38]. However, as large tidal volumes are probably detrimental in surgical patients [
39] we, in line with earlier reports evaluating PVI, chose a tidal volume of 7 ml kg
− 1 ideal body weight [
6,
7,
13]. Whether or not outcome is improved by using larger tidal volumes in surgical patients monitored with dynamic parameters is unknown.