Oesophageal Doppler monitoring, doubt and equipoise: evidence based medicine means change

Authors


Until recently, and for most of the last 10 years, I was a routine user – and perhaps even an advocate [1] – of oesophageal Doppler monitoring (ODM) for major surgery. While there are challenges inherent in using ODM, including sometimes obtaining inconsistent or uncertain readings, it is biologically plausible and somewhat evidence based that ODM might at least reduce hospital stay. Extrapolation would suggest that patients would physiologically fare better with ODM-guided therapy, and now ODM is firmly embedded within many ‘enhanced recovery after surgery’ (ERAS) programmes, with National Institute for Health and Clinical Excellence (NICE) approval, and attracts generous tariff rewards if it is used in patients [2, 3]. Like many clinicians, I suspended many doubts over the quality of the evidence and used the ODM, at least while awaiting further studies. However, I have stopped routinely using the ODM for the colorectal cases that make up my anaesthesia casemix, despite the fact that I am ‘incentivised’ (i.e. my hospital is generously paid) to do so and NICE strongly suggest I should. Were we were wrong over the last 10 years in using ODM, or am I wrong now to abandon it?

The process of NICE approval for ODM has been presented by key individuals involved [2, 4]. I believe that this description is of a large group (NICE) pragmatically reviewing an incomplete evidence base in an attempt to benefit patients. The UK is not alone in approving ODM [5]. The merits or otherwise of ERAS have been discussed in this journal recently and for anaesthetists, using ODM to guide individualised fluid therapy is a large part of the anaesthetic contribution to peri-operative ERAS [6]. However, the benefit of ODM (typically within an ERAS programme) has reached the point of established fact for many enthusiasts, with no alternative viewpoint available, and approval by NICE somewhat perpetuates this outcome regardless of the fact that intense controversy remains around both the place of ODM and aspects of NICE's decision making [7, 8].

Incorporating and translating evidence based medicine into clinical practice is challenging and, with hindsight, we all get it wrong on occasions. There are numerous examples of such a sea change in medicine and one of my enduring memories in training was seeing the Cochrane Collaboration ‘headline’ suggesting that albumin may kill patients [9] evolve with emerging evidence to the possibility of its being non-harmful or even possibly beneficial [10-12]. Many clinicians ‘got it wrong’ with activated protein C, steroids in sepsis and head injury, colloids, peri-operative beta blockade and routine epidural analgesia, to name but a few. Clinicians’ desire to improve care [13], compounded by publication bias, means it is perhaps inevitable that one might encounter examples of definitive negative studies that ultimately curb enthusiasm for ineffective therapies following smaller (and erroneously positive) initial studies; often the ‘truth’ around a therapy depends upon how far along this path we have progressed [14-16]. Recent experience with activated protein C in sepsis is a case in point: the initial PROWESS paper suggested impressive benefit but was essentially a post-hoc subgroup analysis [17]. Despite multiple post-hoc analyses [18], enthusiastic voices and clinical experiences [19], positive post-marketing and registry findings, and endorsement by specialist groups and governments, the concerns never went away and negative studies and debate continued [20, 21]. Clinicians were required to embrace doubt regardless of personal hunches and consider the difficult second study. PROWESS-SHOCK was terminated early through lack of efficacy and activated protein C was withdrawn [22]. Hot on the heels of this, the Surviving Sepsis Campaign has overturned the majority of sepsis management guidelines including use of activated protein C; evidence based medicine has rightly won but things were a lot more exciting in ICUs 10 years ago [12]! One way or another, many of us got it badly wrong; hunches were not borne out by research and history will record another sepsis treatment failure and a victory for science.

One of the most concerning responses to the activated protein C story is that some individuals maintain they knew all along that ‘it never worked’. How could this be? Medicine should progress by hypothesis, testing and refinement. PROWESS and subsequent data made a compelling case [17], and yet rigidly holding the view that it was ineffective pre-PROWESS-SHOCK demonstrates the conflict between evidence and belief that dogs the application of evidence based medicine. An inability to embrace doubt makes for dangerous practice and tends to be accompanied by an inability to change when more definitive data do become available. When a sizeable proportion of your colleagues highlight an area of doubt you should at least listen! Failure to do so leads the way to delusion and while faith can form an excellent foundation for one's life it is perhaps less well suited to medical decision making. We should all have deep suspicion and concerns around individuals with unshakable, non-negotiable, fundamentalist ‘hunches’ that cannot acknowledge an alternative interpretation, rather like ‘knowing’ that activated protein C never worked.

Equipoise ultimately means having sufficient uncertainty around the benefit or otherwise of an intervention to be comfortable to see patients randomised in a clinical trial. Equipoise is rarely an absolute and evolves over time; for example one of the largest obstacles to studying the pulmonary artery catheter (PAC) was the unwillingness of clinicians to manage sick patients without what had become a ‘standard of care’. We should be grateful to the nagging doubters challenging the PAC as established practice; reminding oneself “I could be wrong…” is central to practising evidence based medicine [23]. The perfect trial has yet to be run and ‘experts’ who fail to maintain equipoise for most of what they believe should be treated with suspicion.

Of course, equipoise is partly based upon the quality of the supporting evidence. For the ODM, I would suggest this supporting evidence does little to reject equipoise. Based around a largely non-clinical surrogate endpoint (length of stay), with inconsistent benefits on other complications, the evidence base is familiar to most readers and has been presented elsewhere [2, 24-28]. However individuals choose to interpret the available evidence, as a specialty we are still wrestling with the true meaning of the same data [4, 28-34]. Furthermore, formal cost analysis has not been undertaken to support NICE's suggestion of saving > £1100 per patient, which has got medical managers in quite a flutter [35, 36]. There is considerable pressure on clinicians to use the ODM almost disregarding efficacy; if used on every conceivably ‘appropriate’ patient we are told the generous 2.5% tariff refund through Commissioning for Quality and Innovation (CQUIN) payments would be offset by care savings of approximately £850 million nationally [37]. How confident can we really be of these figures when patients are undergoing surgery that was never part of the original research, e.g. laparoscopic/gynaecological/urological [38]? I would suggest it is unwise to extrapolate to this degree and it is equally conceivable that a colossal bill is being run up.

While it is true that therapy (e.g. individualised fluid) might improve patient outcomes, no monitoring device can, and how clinicians respond to the oesophageal Doppler monitor and with what therapy has perhaps received less scrutiny. The majority of protocols dictate colloid boluses in the form of hydroxyethyl starch (HES) or gelatins; large outcome studies on the efficacy and safety of gelatins have never been undertaken and it is of note that in the USA, the Food and Drug Administration never approved their use. Similarly, HES has been aggressively marketed and for some years my local HES sales representative was a frequent and persistent visitor [39] who has been absent since the recent, 7000-patient CHEST study in which a modern, low-molecular weight HES (Voluven™) showed a non-significant trend to increased mortality, and increased renal failure and dialysis [40]. What actually happens to patients if HES increases renal replacement therapy (relative risk 1.21), while conversely increasing global oxygen delivery indices reduces renal failure by 0.71 [41]? One previous study utilising pulse contour analysis and gelatin therapy did improve outcomes; the key point must be to demonstrate both the monitor and specific therapy (i.e. the exact class of fluid) are effective in combination and safe [42]. The ODM, as a monitor, encourages the use of expensive, non-superior HES therapy and possibly therapy that harms; this at least needs further study and more extensive comparison with crystalloid [43]. There has been little evidence of significant harm through an ODM-based strategy, either directly (e.g. oesophageal injury) or indirectly (e.g. acute kidney injury) with HES [43-45], although this has not been rigorously sought and is probably under-reported. Within an established ERAS programme employing laparoscopic surgery, HES therapy actually increased length of stay relative to crystalloid [46].

Science never stops and NICE approval has been followed by the large study by Challand et al., casting further doubt on the benefit of ODM-guided fluids [47]. More recently and fundamentally, the concept of goal-directed ODM-guided fluid therapy has not been shown superior to a strategy of neutral balance [48]. Perhaps it's time to ask the unpalatable question that provided a ‘sensible’ anaesthetic is given, could our interventions in theatre actually have little effect? There remains little to reject the conclusion that 10 ml.kg−1.h−1 crystalloid during surgery or neutral fluid balance may be as effective as ODM-directed colloid therapy in a modern UK hospital [47, 48]. Furthermore, to be a valid physiological construct the results should be replicated with other monitoring devices and it is fanciful to suggest the ODM is the only device that may be capable of detecting ‘volume responsiveness’ [42]. Indeed, by glossing over the physiological basis of fluid loading we risk losing something far greater. Is it really conceivable that a structure as complex as the heart ‘moves along’ a single line of the Frank-Starling curve as fluids are given or taken away? By not putting forward hypotheses for testing, scientific progress ceases; for example, could fluid loading reflect acute haemodilution and altered blood viscosity rather than ‘haemodynamic optimisation’? Can we continue to follow ODM protocols largely developed for open laparotomy, during laparoscopic surgery and pneumoperitoneum? How is it that with little evidence to favour neuraxial anaesthesia for fixation of fractured neck of femur, the vast majority of clinicians avoid general anaesthesia [49]? Why do some clinicians disregard evidence from this population and proceed with a spinal and no ODM, yet the same clinician on a colorectal list might use the ODM as a ‘standard’ [50, 51]?

So what should happen with the ODM now? It is probably premature to say that ODM-directed fluid therapy is ineffective, but the available evidence is hardly definitive. My interpretation of the evidence base is a shift towards lack of benefit and possibly harm, and this is why I don't use the ODM anymore. I have little ‘evidence’ to challenge its advocates’ contention that ODM works for their patients in their hands, but the onus should be on them to demonstrate that. What is essential is that science prevails and research for answers is not stifled by extreme and unfounded views from enthusiasts and condemners. I personally felt the strength of recommendation from NICE was excessive, considering the available evidence, and think there is a pressing need to re-formulate the NICE recommendations and for further high-calibre research into ODM-guided fluid therapy.

However, what we should encourage as individuals and as a specialty is widespread doubt and equipoise for most of what we do and this extends far beyond the ODM. If we cannot embrace doubt, the inevitable consequence is that we cannot change or engage in scientific debate, and ultimately, we will apply ineffective or dangerous therapies to patients. Retaining and managing doubt, while having an opinion, is a fundamentally different position to taking a fanatical view on a given topic. In demanding ‘evidence based’ decisions in medicine we must recognise that it is impossible to generate sufficient evidence of sufficient calibre to match the pace of questions being posed. Consequently, it is inevitable that we proceed with a variously incomplete evidence base supported by opinions and experience – what's important is that we retain the capacity to change direction. I have that nagging feeling as I write this that I may be wrong and in five years time I might be using the ODM again; who knows, I may even be giving septic patients epidural analgesia and steroids, all guided by right heart catheterisation (again)! Recognising these doubts should be seen as a strength and not a weakness; I see more of a concern in unfounded beliefs, incentivised use and sanctions for non-users, as if somehow we have definitively established the ODM in managing high-risk cases and excluded the possibility of harm.

Competing interests

No external funding and no competing interests declared.

Ancillary