Peri-operative hemodynamic therapy: only large clinical trials can resolve our uncertainty

In this issue of Critical Care, Cecconi and colleagues report the findings of a small trial of goal-directed hemodynamic therapy (GDT) in patients undergoing major orthopedic surgery under regional anesthesia. This is an interesting trial and the first of which we are aware to test the efficacy of GDT in this patient group. The findings suggest that this intervention is associated with improved adverse event rates after surgery. However, in this trial, as in all small trials of GDT, the potential for bias leaves some uncertainty regarding how widely the findings should be implemented. Such limitations may be impossible to completely eliminate from trials of complex interventions, but large multicenter trials may allow us to substantially decrease bias and improve the generalizability of the findings.

Th e peri-operative use of fl uid and inotropes guided by cardiac output monitoring is variously known as optimization, fl ow-guided therapy, and goal-directed therapy (GDT). In this issue of Critical Care, Cecconi and colleagues [1], who are experienced researchers in this fi eld, report the fi ndings of an interesting single-center trial of GDT in patients undergoing hip replacement under regional anaesthesia. Th e fi ndings of this investigation suggest that GDT may be associated with reduced adverse event rates after orthopedic surgery. Th ere is a clear need to improve survival for patients undergoing major surgery [2,3], and in many small trials, GDT has been associated with better clinical outcomes, particularly following high-risk procedures [4][5][6]. Th is benefi cial eff ect appears to be related to improvements in micro vascular fl ow and tissue oxygenation [7]. Th e fi ndings of previous studies have suggested benefi t in patients undergoing proximal femoral fracture repair [8,9], but we are unaware of any previous trials of GDT in elective orthopedic surgery. Consequently, there has been little evidence to inform the use of GDT in this large popu lation of patients. Th ere are distinct challenges associated with the design and conduct of GDT trials, and these challenges are not easily solved. Th us, the fi ndings of this and all previous GDT trials must be interpreted in the context of the design choices that the investigators have made.
GDT is a term that describes a potentially more eff ective and objective method to determine the optimal dose of intravenous fl uid and inotropic therapy through the use of a clinical algorithm. Ordinarily, the clinician will administer fl uid or inotropic therapy (or both) on a subjective basis according to their preferred clinical endpoints. Hence, there is signifi cant variability in clinical practice which, in a small trial, must be replaced by a control group algorithm designed to refl ect best usual care. Failure to use this approach will expose the investigators to the possible accusation of manipulating the trial outcome through poor control group care. In the recent past, even the suggestion of such impropriety has proven damaging for highly respected investigators [10]. In the trial of Cecconi and colleagues [1], the choice of arterial pressure as a control group end-point for fl uid therapy illustrates the dilemma. Fluid challenges are commonly administered to correct hypo tension. However, under regional anaesthesia, arterial hypotension is more likely to result from vasodilatation than hypovolemia. We could therefore argue that usual practice is incorrect in this case and yet more eff ective control group care will diminish any apparent treatment eff ect of GDT. It is easy to criticize control group care in clinical trials of complex interventions though much harder to fi nd a better alternative. In large multicenter trials, the variable nature of usual care is much less likely to lead to an erroneous fi nding. Even then, it would be advisable to recommend some treatment standards to avoid practice misalignment. Less critically, the GDT algorithm must result in

Abstract
In this issue of Critical Care, Cecconi and colleagues report the fi ndings of a small trial of goal-directed hemodynamic therapy (GDT) in patients undergoing major orthopedic surgery under regional anesthesia. This is an interesting trial and the fi rst of which we are aware to test the effi cacy of GDT in this patient group. The fi ndings suggest that this intervention is associated with improved adverse event rates after surgery. However, in this trial, as in all small trials of GDT, the potential for bias leaves some uncertainty regarding how widely the fi ndings should be implemented. Such limitations may be impossible to completely eliminate from trials of complex interventions, but large multicenter trials may allow us to substantially decrease bias and improve the generalizability of the fi ndings.
adequate intervention in the majority of patients while minimizing potential harm. In the trial of Cecconi and colleagues, only 65% of patients achieved the pre determined target for oxygen delivery (DO 2 ). Th is observation likely refl ects a GDT algorithm that incorporates judicious limits to the dose of inotropic therapy. It seems clear that the continued pursuit of hemodynamic goals in patients who do not respond is harmful [11]. Th e adverse eff ects of fl uid and inotropic therapy will be most serious for patients with signifi cant heart disease. In this respect, it would have been helpful to report the timing of cardiovascular complications. Arrhythmias may be triggered by hypovolemia or myocardial ischemia, but the eff ect of study group allocation would have a more transient eff ect on this category of complications than for others such as infection.
Blinding is another important potential source of bias in GDT trials. GDT is a complex intervention that involves human decision making. In this context, it is impossible to fully blind investigators to study group allocation. Small GDT trials are particularly vulnerable to bias, and investigators may fi nd it diffi cult to demonstrate the adequacy of their procedures for assessing nonbinary outcomes such as complications. Th is problem can be minimized through the use of blinded assessors who determine clinical outcomes according to predetermined criteria. In the report by Cecconi and colleagues, these processes could be better described to clearly demonstrate methodological rigor. Th e use of scoring systems such as the postoperative morbidity survey (POMS) may prove useful for assessment of minor morbidity, but once again, large multicenter trials would appear to be the most eff ective way to reduce bias.
Th is interesting trial adds another piece to the jigsaw of optimal peri-operative hemodynamic therapy. Th e apparent effi cacy of this approach during elective orthopedic surgery in self-ventilating patients under regional anesthesia is of particular interest. We agree with the authors that GDT has a potential role in the treatment of such patients. We also agree that further research is required to confi rm their fi ndings. Small clinical trials add to the literature, but the lack of confi rmation in large trials is a signifi cant limitation of the evidence base for GDT. We hope that this interesting investigation will be followed by a robustly designed multicenter trial designed to confi rm or refute the fi ndings.