What is the optimal type of fluid to be used for peri-operative fluid optimisation directed by oesophageal Doppler monitoring?


Dr C. Morris
Email: cmorris@doctors.org.uk


The objective of this review was to determine the optimal type or class of intravenous fluid to be used during peri-operative patient optimisation guided by oesophageal Doppler monitoring and to identify future directions for research. We undertook a literature review of patients undergoing major (general, colorectal, orthopaedic and urological) surgery, whose fluid therapy was managed using peri-operative oesophageal Doppler monitoring. We identified 10 studies that included 891 randomised patients. A variety of regimens and types of fluid were used in association with oesophageal Doppler monitoring, including crystalloid, gelatin and hydroxyethyl starch. A wide variety of hydroxyethyl starch preparations were used, including high molecular weight and highly substituted hetastarches, and lower molecular weight tetrastarches. Most studies were of high quality, associated with reduced hospital stay, but underpowered to evaluate other outcomes. In units with established enhanced recovery facilities, the benefits of colloid based on oesophageal Doppler monitoring were not reproduced. There is little evidence to support preferential use of any particular type of fluid during oesophageal Doppler guided optimisation; however, routine use of colloids is associated with significantly higher costs and may increase hospital stay. Furthermore, many of these fluids have not been evaluated in patient populations in whom optimisation is being applied or proposed, and the potential for harm cannot be excluded. Recommendations for future studies are provided, including adequate power for primary end points beyond hospital stay and adequate follow-up, and inclusion of a crystalloid comparison group.

Despite being available for a number of decades, there can be little doubt that peri-operative oesophageal Doppler monitoring has recently become a ‘hot topic’ within anaesthetic and surgical practice and national health policy. There is consistent evidence that patients whose fluid is managed in response to dynamic stroke volume parameters using an oesophageal Doppler monitor have better outcomes than those receiving ‘standard care’ including clinical judgement of fluid requirements and central venous pressure monitoring; in certain types of surgery, use of the oesophageal Doppler monitor has been argued to be a standard of care [1]. It is beyond the scope of this article to consider the effectiveness of the oesophageal Doppler monitor further and this has been reviewed elsewhere [2, 3].

However, it is fundamentally incorrect to expect any monitoring device to improve patient outcomes; rather it is the therapeutic intervention it directs that could realise this potential. For example, a pulse oximeter alone has never saved anyone’s life, but in triggering oxygen administration it certainly has [4]. The significance of this distinction (i.e. monitor vs therapy) and the impact on outcome has been played out previously, e.g. the pulmonary artery catheter [5]. However, the oesophageal Doppler monitor has been the main focus of research interest and the therapy it directs (fluid boluses) has received less scrutiny. This is perhaps surprising considering the controversies within intravenous fluid therapy generally (e.g. crystalloid vs colloid), use of reduced chloride ‘balanced’ solutions, and possible impact of certain types of colloids (e.g. allergic reactions or renal impairment) [6–10]. Furthermore, as national initiatives are suggesting wide application of this technology, the increased costs of certain types of fluids, especially colloids, have significant implications [1, 11]. It is imperative therefore that if clinicians are to realise the best outcomes using care guided by oesophageal Doppler monitoring, then this should be coupled with use of the most effective type of fluid; determining the nature of this type of fluid is the focus of this review. In reviewing the available evidence, recommendations for future research on fluids directed by oesophageal Doppler monitoring are made.


We searched the PubMed, Cochrane and Ovid databases and the internet using the Google search engine. We also searched UK (http://www.ukcrc-ctu.org.uk/Pages/default.aspx) and US (http://www.clinicaltrials.gov/) trial registries to identified ongoing registered studies. We used the terms below alone and in combination:


We undertook hand searches of all reference lists from articles identified by the searches to find additional relevant references.

We excluded comparisons between cardiac output monitoring devices and optimisation using alternative, non-oesophageal Doppler monitor devices (e.g. pulmonary artery catheters), and those not primarily investigating peri-operative optimisation (e.g. comparing types of fluids without oesophageal Doppler monitoring). Studies using oesophageal Doppler monitors outside the peri-operative setting (e.g. for the management of sepsis or critical illness) were not considered. Studies routinely combining fluids and vasoactive infusions or inotropes to optimise patients were excluded, as was work presented in abstract form or at meetings, outside of publication in peer reviewed journals. Where data were incomplete for this analysis, we attempted to contact study authors directly.

Methodological scoring was done using the method of Jadad [12, 13] and power calculations were determined for each study.


The initial search terms identified 137 citations for ‘oesophageal Doppler monitor(ing)’, 285 for ‘fluid optimisation’ and 6119 for ‘goal directed’. After application of the limits ‘adult’, ‘human’, ‘English’, ‘clinical trial’‘PRCT’ or ‘meta-analysis’, these were reduced to 10 citations for ‘oesophageal Doppler monitor(ing)’, nine for ‘fluid optimisation’ and 294 for ‘goal directed’. The largest group (‘goal directed’, 294 citations) was subsequently screened using the terms ‘surgery’, ‘peri-operative’, ‘crystalloid’, ‘colloid’, ‘hydroxyethyl starch’ and ‘gelatin’. Following hand searches of the remaining citations and application of the exclusion criteria described above, 10 studies were included for further analysis. These included 891 randomised patients suffering from a wide variety of conditions (colorectal, orthopaedic, general and urological) and the principal features are summarised in Table 1 [14–23].

Table 1.  Summary of the reviewed studies conducted using oesophageal Doppler guided fluid optimisation.
StudyPopulation (n)Primary outcomeFluids used (Doppler guided)Control groupMean volume of peri-operative colloid; mlMean volume of peri-operative crystalloid; mlComments
  1. *study not formalised but predominant colloid used was Hespan™.

  2. HES, hydroxyethyl starch (where a branded fluid solution was specified this has been stated); D, Doppler guided group; CVP, central venous pressure; S, standard care; EF, left ventricular ejection fraction. ODM oesophageal Doppler monitor.

Mythen et al. [14]Cardiac surgery before cardiopulmonary bypass (60)Gastric intramucosal pHi6% HES (Elohes™, Fresenius Kabi, Runcorn, Cheshire, UK)Standard care900 (D) vs 250 (S)875 (D) vs 1125 (S)ASA grade 3 with EF > 50%. Protocol allowed vasoactive infusion
Sinclair et al. [15]Proximal femoral fracture (40)Time to dischargeHES*Standard care750 (D) vs 0 (S)725 (D) vs 1000 (S) 
Conway et al. [16]Colorectal Surgery (57)Cardiac outputHESClinician discretion28.0 (D) vs 19.4 ml.kg−1 (S) Total fluid 64.6 vs 55.2 ml.kg−1
Gan et al. [17]General, urological and gynaecological surgery (100)Hospital stay6% HESStandard care847 (D) vs 282 (S)4405 (D) vs 4375 (S)Protocol allowed use of vasoactive infusion
Venn et al. [18]Proximal femoral fracture (90)Time to dischargeGelatin (Gelofusine)Standard care or fluid guided by CVP1207 (D) vs 1123 (CVP) vs 448 (S)1120 (D) vs 1156 (CVP) vs 1286 (S) 
Wakeling et al. [19]Colorectal surgery (128)Hospital stayGelatin (Haemaccel or Gelofusine)Same fluids guided by CVP2000 (D) vs 1500 (S)3000 both 
Mckendry et al. [20]Cardiac surgery (174)Hospital stayHES*Standard care1667 vs 1042 colloid353 (D) vs 328 (S)Postoperative regimen also allowing use of inotropes
Noblett et al. [21]Colorectal surgery (108)Hospital stayGelatin (Volplex)Standard care1340 (D) vs 1209 (S)2298 (D) vs 2625 (S) 
Senagore et al. [22]Colorectal surgery (64)Hospital stayHES (Hextend™, Bio Time, Inc, Berkeley, CA) or Hartmann's solutionStandard care3800 crystalloid (D) vs 3300 colloid (D) vs 2850 crystalloid (S)N/AAll patients received baseline crystalloid bolus and maintenance. Established enhanced recovery programme
Futier et al. [23]Variety of intra-abdominal procedures (70)Postoperative complications6% HESComparison liberal vs restrictive crystalloidN/AN/AAll patients managed with ODM, central access. Protocol allowed vasoactive infusion

The methodological quality of the studies was generally high (Table 2). The most common primary outcome measure was length of stay and no study was adequately powered to investigate mortality.

Table 2.  Summary of the methodological qualities of the reviewed studies conducted using oesophageal Doppler guided fluid optimisation.
StudyRandomised?Double blind?Patients tracked?Jadad score*Power calculation?†Comment
  1. *Jadad score maximum = 5 [11]; †power calculation relative to primary outcome (Table 1) LOS, length of stay; HES, hydroxyethyl starch.

Mythen et al. [14]YesYesYes5YesInvestigation of intramucosal pH changes during cardiac surgery including cardiopulmonary bypass
Sinclair et al. [15]YesYesYes5Yes 
Conway et al. [16]YesYesYes3Yes 
Gan et al. [17]YesYesYes5Yes32% gynaecology
37% urological
79% ASA 1–2
Venn et al. [18]YesYesYes4YesPredicted mortality all groups > 11%
Wakeling et al. [19]YesYesYes4Yes 
Mckendry et al. [20]YesNoYes3YesNurse directed care post-surgery in critical care area
Noblett et al. [21]YesYesYes5Yes 
Senagore et al. [22]YesYesYes4YesLaparoscopic colectomy, excess LOS with HES
Futier et al. [23]YesYesYes5YesComplex design investigating restrictive vs non-restrictive crystalloid regimens while responding with HES to Doppler parameters

A variety of intravenous fluids were used in these studies including crystalloid, gelatin-based and a range of hydroxyethyl starch (HES) based fluids. Table 1 shows the mean or median volumes of optimisation fluids used; Doppler guided cohorts received a larger total volume and larger range of administered colloid boluses, reflecting individualised treatment: individual study data have not been presented further.

Although all crystalloid fluid challenges used Hartmann's solution [23], no study sought to compare directly the effect of a balanced solution compared to 0.9% saline, either as a primary crystalloid or as a suspending solution for a colloid. Within the colloid bolus groups, large variability existed in the type of fluid administered; the nature of the fluids used that contain HES is summarised in Table 3. Although the various gelatine-based fluids show fundamentally less variability, they were not standardised and included the proprietary products 4% succinylated Gelofusine™ (Braun Medical Ltd, Sheffield, UK) [19, 20] or 3.5% urea linked Haemaccel™ [19] and 4% succinylated Volplex™ (Beacon Pharmaceuticals Ltd, Tunbridge Wells, Kent, UK) [22].

Table 3.  Summary of the varieties of hydroxyethyl starch (HES) preparations used during oesophageal Doppler guided fluid optimisation in the reviewed studies.
StudyBrand nameConcentration; mlMolecular weight; kDaSubstitutionC2:C6Base solution
  1. NS, not specified; MW, molecular weight; Substitution, refers to the manufacturer’s ratio of glucose residues of the starch molecule carrying an added hydroxyethyl group; C2:C6, ratio of hydroxyethyl groups carried at the C2 position of glucose to that at the C6 position; *the only commercially available preparation meeting this description is Voluven or Volulyte (Fresenius Kabi) with C2:C6 of 9:1.

Mythen et al. [14]Elohes62000.624:10.9% saline
Sinclair et al. [15]Hespan64500.704:10.9% saline
Conway et al. [16]NS
Gan et al. [17]NS6
McKendry et al. [20]Hespan64500.704:10.9% saline
Senagore et al. [22]Hextend66500.704:1Lactate based
Futier et al. [23]NS*61300.40NSNS

Given the variety of regimens employed using different types of intravenous fluids, it is not possible to identify a hierarchy of outcomes related to different fluid types. The available evidence shows that oesophageal Doppler guided optimisation has been successfully performed using a variety of fluids to reduce hospital stay.


When using the oesophageal Doppler monitor to direct individualised peri-operative fluid optimisation, a wide variety of fluid preparations have been used. As emphasised, the primary purpose of this analysis was to determine the optimal type of fluid to be used in conjunction with the oesophageal Doppler monitor during peri-operative optimisation, and we suggest future directions for research; we were not considering the effectiveness of the oesophageal Doppler monitor per se [2, 3]. Although fluid optimisation directed by oesophageal Doppler monitoring is a consistently successful strategy for reducing hospital length of stay, the current studies of such are consistently underpowered to allow definitive conclusions regarding alternative outcomes. Importantly, there are few recommendations that can be made regarding the type of fluid to be used during optimisation.

It is conceivable that the type of fluid does not matter and that using the oesophageal Doppler monitor with any individualised regimen fluid will shorten hospital stay. Indeed, one review of optimisation gave no specific recommendations for fluid types and stated that ‘correct dosage of fluid therapy improves patient outcomes after surgery’ [24]. Although the correct quantity for individual patients is certainly crucial, it is likely that using different types of fluids (the primary therapeutic intervention during fluid optimisation guided by oesophageal Doppler monitoring) will demonstrate significant differences if they become the primary focus of investigation.

Previous work in the critically ill has implicated hyper-oncotic and large molecular weight (≥ 200 kDa) ‘traditional’ HES in accumulation and acute kidney injury (AKI) [9, 10]. What perhaps would traditionally be regarded as ‘small’ rises in plasma creatinine are associated with an excess mortality and long-term effects and this requires closer scrutiny in future trials [25–27]. Furthermore, deficiencies in caring for AKI in the peri-operative period have been highlighted in a recent National Confidential Enquiry [28]. Whether AKI occurs peri-operatively with more ‘modern’ iso-oncotic, smaller starch molecules remains unclear, and a study of fluid therapy guided by oesophageal Doppler monitoring using 6% 130-kDa HES in 0.9% saline solution in patients with proximal femoral fractures is underway [29]. Larger molecular weight hydroxyethyl starches have been used without AKI during oesophageal Doppler monitor optimisation [14, 15, 20, 22], whereas 130-kDa HES has been associated with a trend towards increased AKI [23]; until the effect of these fluids on outcome is addressed in future studies with adequate statistical power, the use of HES will remain controversial. Intravenous gelatin-based fluids (gelatins) are renally filtered but have not been associated with renal failure. They have been implicated in allergic reactions and altered plasma viscosity and are not approved for use by the Food and Drug Administration in the US [30]. Use of a variety of gelatins has reduced hospital stay [18, 19, 21] without evidence of significant adverse effects. Colloids have also been implicated in effects on blood coagulation, although the effects are complex and related to dose and volume [31]. However, otherwise minor effects on coagulation may become significant where haemostasis must be more meticulous (e.g. venous bleeding following prostatectomy) and may potentially undermine the benefits of oesophageal Doppler monitoring.

Conversely, crystalloids have been cited as requiring larger volumes to produce equivalent physiological effects to colloids, and have been implicated in positive balance complications (e.g. oedema of bowel anastamoses) [32]. It should be noted that despite a larger net fluid balance (2.5 vs 1.9 vs 2.1 ml.kg−1.h−1, for control vs oesophageal Doppler guided crystalloid vs oesophageal Doppler guided HES therapies), patients managed using crystalloid therapy without oesophageal Doppler monitoring had the shortest time to discharge [22]. While the equivalent intravascular expansion ratio for crystalloid:colloid has traditionally been taught as 3:1, the SAFE group established it as being 1.4:1 in the largest evaluation to date of crystalloids vs colloids [6]. Although intuitively attractive, whether colloids are able to reduce the total volumes of fluid administered in a clinically relevant fashion (and thus reduce net peri-operative fluid balance) remains to be established [23]. Furthermore, the nature of the base crystalloid (0.9% saline or ‘balanced’ solution) may have a significant clinical impact, although this has been hard to establish [7, 8]. We believe that when the currently available intravenous fluids are compared appropriately in a head-to-head fashion in high risk patients, superiority will become apparent; such a comparison should be actively sought.

There is little doubt that there is a strong drive within the UK health system for clinicians to use fluid optimisation guided by oesophageal Doppler monitoring. Peri-operative (fluid) optimisation in general is a central component of attempts by monitor manufacturers and the NHS to reduce peri-operative hospital stay [33] and a central tenet of the ‘Modernising Care for Patients Undergoing Major Surgery’ initiative from the Improving Surgical Outcomes Group (ISOG) [1, 34]. The closely linked GIFTASUP multidisciplinary group suggested that ‘intra-operative treatment with intravenous fluid to achieve optimal stroke volume should be used where possible as this may reduce postoperative complication rates and duration of hospital stay’ and these recommendations were made to the House of Commons Health Select Committee by the ISOG [35–37]. Quite correctly, the GIFTASUP guidance does not recommend a type of fluid. The NHS Institute for Innovation and Improvement through the Enhanced Recovery Programme emphasises the importance of fluid optimisation and suggests the UK consider seven index procedures for enhanced recovery (Table 4) [11].

Table 4.  Index procedures suggested in the Enhanced Recovery Programme and approximate UK annual caseload (total is ∼270 000).
Surgical procedureUK annual rate
Colectomy and proctectomy [38]20 000
Proximal femoral fracture repair [39]70 000
Elective hip and knee replacement [40]160 000
Radical cystectomy [41]8000
Radical prostatectomy [42]3–4000
Hysterectomy [43]4–5000

Flow guided optimisation is potentially applicable to 270 000 cases per year, with no recommendation possible for the type of fluid to be used [38–43]. Furthermore, in some types of surgery (see Table 4), the benefit of oesophageal Doppler monitoring has never been directly demonstrated (e.g. primary hip or knee arthroplasty) and some procedures have evolved since the original work using the oesophageal Doppler monitor, e.g. the emergence of the laparoscopic colectomy. These procedures, with their associated pneumoperitoneum, may not generate the same data using targeted fluid management (of which oesophageal Doppler monitoring is the most common), as an open procedure [23]. Similarly, the anaesthetic preference for elective joint arthroplasty surgery has become regional anaesthesia, where the feasibility and tolerability of oesophageal Doppler monitoring are low. Extrapolation of historical data to procedures and populations that were never evaluated is notoriously unreliable, with the potential risk of as yet unexplored differences in clinical effects and possible harm as a consequence–especially when one considers the possible impact of the different types of fluid therapy. It is of concern that the study conducted in a centre with an established enhanced recovery programme demonstrated a length of stay of 64.9 h for patients receiving ‘standard’ care, 71.8 h for those receiving crystalloid therapy guided by oesophageal Doppler monitoring and 75.5 h for those receiving HES therapy guided by oesophageal Doppler monitoring; i.e. colloid based care directed by oesophageal Doppler monitoring resulted in a paradoxical increase in patients’ length of stay [22]. Studies of fluid management guided by oesophageal Doppler monitoring that reported shortened hospital stay were conducted during the evolution of the enhanced recovery concept, raising the question of whether such guided fluid therapy exerts a significant effect if enhanced recovery is established.

The work by Futier raises significant concerns [23]. Although all patients were managed using oesophageal Doppler monitoring (triggering boluses of HES), this study primarily compared postoperative complications between patients treated with restrictive (6 ml.kg−1.h−1) or liberal (12 ml.kg−1.h−1) Hartmann's solution. The study was powered for pre-defined postoperative complications rather than patients’ length of stay and it took place in a centre with established enhanced recovery processes. The restrictive crystalloid group was associated with a higher rate of complications and received more colloid (HES 130-kDa) as guided by oesophageal Doppler. The restrictive group also demonstrated a statistically non significant trend towards increased rates of renal failure including the need for dialysis (11% vs 0%); there is a requirement therefore for adequately powered studies examining these organ failures. A valid emerging hypothesis is that liberal administration of crystalloid, perhaps in the absence of oesophageal Doppler monitoring, in the presence of a functioning enhanced recovery programme may achieve comparable or superior results to management using boluses of colloid guided by oesophageal Doppler monitoring [22, 23].

It is beyond the context of this article to consider a complex economic analysis of the oesophageal Doppler monitoring device itself, where any costs are potentially offset by reduced hospital stay or avoidance of alternative monitors (e.g. central venous access), and such an analysis has been undertaken elsewhere [44]. However, the wholesale use of colloids in preference to crystalloids will have significant cost implications, particularly if these were associated with increased adverse events such as AKI. If HES was used routinely, excess annual costs in the UK could exceed £4 million; this would be approximately £1 million in the case of gelatins (Table 5).

Table 5.  Annual fluid costs, assuming typical intra-operative Doppler guided boluses of 2.0 l for 270 000 enhanced recovery cases (prices according to those paid by the Royal Derby Hospitals in 2010).
Type of fluid (price per litre; £)Annual fluid cost; £
Hartmann’s solution (0.80)216 000
Gelatin (6.00)1 620 000
Hydroxyethyl starch (16.00)4 320 000

We conclude that there is an urgent need for research into this area of fluid management and that the number of cases being considered for enhanced recovery programmes and optimisation should make multiple centre studies feasible. Given the breadth of applications being advocated for optimisation, the largely unexplored potential for harm rather than benefit, and resource consumption, we believe that further studies are ethically mandated and clinicians should particularly consider the following areas:

  •  Conducting trials in patients in whom fluid therapy is not guided by oesophageal Doppler monitoring, but in centres that have established enhanced recovery programmes and have achieved low lengths of stay. Reducing confounding variables and the use of standardised care (e.g. peri-operative analgesic regimens and resumption of oral diet) are key components in demonstrating the independent effect of the oesophageal Doppler monitor and the type of fluid it directs.
  •  The specific components of an enhanced recovery programme should be standardised and more explicit. They should be evaluated as separate components for effectiveness to avoid confounding (e.g. not attributing shorter hospital stays to the monitor when aggressive physiotherapy is responsible).
  •  More clinically robust end-points beyond patients' length of stay (e.g. mortality and organ failures such as AKI or respiratory failure) require evaluation, as do the peri-operative sequelae (e.g. nausea and vomiting, time to recovery of bowel function).
  •  To provide adequate statistical power to detect the endpoints described above will inevitably require larger studies, and endpoints specific to the type of fluid should be employed (e.g. AKI, accumulation of molecules and total volumes of fluid administered).
  •  Follow-up should be appropriate to the endpoint of interest (e.g. AKI has a high rate of progressing to chronic renal impairment and later mortality).
  •  Studies should continue to include a crystalloid arm, mindful of more realistic ratios of required fluid volumes (i.e. crystalloid:colloid ratio of 1.4:1)
  •  Comparisons of ‘balanced’ crystalloids, both as the primary intravenous fluid or suspending fluid for a colloid, require robust evaluation.
  •  Evaluations should be specific to the surgery being considered. For example, results from patients undergoing open colectomies should not be extrapolated to those receiving laparoscopic procedures and results from patients with proximal femoral fractures should not be extrapolated to those undergoing elective hip arthroplasties.
  •  ‘Given the paucity of the existing economic evidence base, any further primary research should include an economic evaluation or should provide data suitable for use in an economic model’ [44].

Competing interests

No external funding or competing interests declared.