GIFTAHo; an improvement on GIFTASuP? New NICE guidelines on intravenous fluids

Remove far from me vanity and lies: give me neither poverty nor riches; feed me with food convenient for me —Proverbs 30.8 King James Bible (Pure Cambridge 'Authorized Version'; http://www.kingjamesbibleonline.org/Proverbs-30-8/) Five years ago, an editorial in this journal carried a withering attack on the unpublished British Consensus Guidelines on Intravenous Fluid Therapy in Adult Surgical Patients (GIFTASuP), condemning them as: "an object lesson in the perils of writing guidelines… evidence-based scores used inappropriately allow[ing] opinion to masquerade as fact… an illusion of knowledge and certainty where little exists" 1. Sub-optimal fluid therapy is costing patients their lives and causing tens of thousands of patients to suffer complications every year, according to the Chairman of the Guideline Development Group (GDG) that has produced the National Institute of Health and Care Excellence's (NICE's) new Guidelines on Intravenous Fluid Therapy in Adults in Hospital (GIFTAHo) 2. He has therefore declared GIFTAHo to be amongst the most important guidances NICE has ever issued 3. But has it identified from evidence, rather than opinion, the optimal prescribing principles that will deliver improved patient outcomes? In this editorial, I suggest not. The NICE process is clearly defined, and has been designed to squeeze cold facts out of hard published data. The technical team formulates questions about the intervention under consideration, the outcomes of interest and the comparisons that have been made. The GDG then refines, validates and answers these questions based on the distilled evidence. In two chapters, The principles of fluid prescribing and Assessment and monitoring of patients receiving intravenous fluid therapy, it was not considered possible to perform clinical evidence reviews. In these chapters, the GDG drafted recommendations as expert consensus opinion, citing a promotional booklet commissioned from two GDG members by the pharmaceutical manufacturer B. Braun 4: "If the wording and sentiment of the text is similar, it can be attributed to the GDG members' direct involvement in both pieces of work." (full guidance footnote b, p 48). There is listed a substantial cohort of hospitalised adults to whom this Guidance must not be applied: pregnant patients; patients with severe renal or hepatic disease; those with diabetes, burns or traumatic brain injury; and those needing neurosurgery. The full version of GIFTAHo gives a comprehensive overview of the clinical and economic evidence base, and in most instances draws reasonable conclusions from them. However, I argue that some of the recommendations do not logically flow from the evidence. A controversial theme that runs through GIFTAHo is the 'Hartmann's versus saline' division and its slogans: 'normal' saline is abnormal, or a problem not a solution, while Hartmann's solution is 'balanced' and physiological 5. It is important not to be consumed by irrational prejudice in either direction; the rational prescriber needs both solutions to be available in the fluid cupboard, and to be able to use them according to the patient's individual requirement. Below, I precis and comment on selected questions, the relevant evidence, and the consequent recommendations (numbers refer to the full guidance reference/short guidance reference). What is the incidence and clinical significance of hyperchloraemia and hypochloraemia? Recommendation 13/1.2.5: Measure serum chloride if at any time the patient is given a solution containing chloride more than 120 mmol.l −1 Comment: GIFTAHo raises concerns that isotonic sodium chloride causes significant reductions in renal blood flow and glomerular filtration, hyperchloraemic acidosis, gastrointestinal mucosal acidosis and ileus. Based on very low-quality evidence, the GDG suggests that provision of intravenous fluids containing < 120 mmol.l−1 chloride is associated with lower mortality and morbidity than those containing > 120 mmol.l−1. On the other hand, the evidence could not demonstrate whether hyperchloraemia was worse than hypochloraemia and it was not possible to determine whether abnormal serum chloride in either direction was a complication of fluid therapy rather than a symptom of the underlying disease. There could, therefore, be no direct recommendation to prefer a balanced solution, but instead this curious one is invented, likely to frighten inexperienced prescribers away from isotonic saline even though it was the standard resuscitation fluid in the world's largest randomised controlled trials 6, 7 and is really very safe in the volumes we are considering in this guideline for non-experts 8, 9. A more logical recommendation, if chloraemia was a serious concern for the GDG, would be to monitor both sodium and chloride daily and consider the strong ion difference in all patients receiving fluid therapy. A rational approach would then be to use sodium chloride 0.9% whenever an isotonic solution is indicated, with isotonic sodium bicarbonate 1.26% occasionally substituted if it is considered necessary to widen the strong ion difference. Such a recommendation would also make redundant the proposed research question: "Are balanced solutions superior to sodium chloride 0.9% for the resuscitation of patients with acute hypovolaemic shock?" (section 7.5.2 2). Sodium bicarbonate solutions are not mentioned in GIFTAHo; also notably absent is critical discussion of the clinical effects of pathophysiological substitute anions such as lactate, and unmeasured substitute anions including acetate and malate. A selection of bicarbonate-balanced polyionic isotonic solutions are available in hospitals; designed to be used as filtrate replacement solutions in continuous renal replacement therapy, they could nonetheless be prescribed when higher volume resuscitation or replacement therapy is needed. What is the most clinically and cost effective intravenous fluid for fluid resuscitation…? [and]…timing and rate of administration? Recommendation 16/1.3.1: Give sodium-based crystalloids (Na 130–154 mmol.l −1 ) in 500-ml boluses delivered over less than 15 minutes Comment: GIFTASuP's first recommendation back in 2009 dismissed isotonic saline from routine use and advocated 'balanced' polyionic solutions, claiming a high evidence level of 1b. The GIFTAHo GDG of 2013, however, found no controlled trials comparing isotonic saline with a balanced solution. Hyponatraemia as a complication of fluid therapy was not chosen to be a review question and is therefore given little consideration. GIFTAHo curiously defines hyponatraemia as plasma sodium concentration < 130 mmol.l−1, while most laboratories regard 135 mmol.l−1 as the lower limit of the normal range and one expert commentary warns not to infuse hypotonic solutions while the plasma sodium is less than 138 mmol.l−1 9. Hyponatraemic encephalopathy and neurogenic pulmonary oedema get no mention; they may be rare but must be anticipated and prevented in at-risk patients. In Box 1, I list some cautions for the use of hypotonic intravenous infusions. Consideration of tonicity and intracellular fluid volume could have led to a recommendation for the use of fully isotonic solutions for resuscitation, particularly in patients at risk of intracranial hypertension, and post-surgical maintenance therapy. No evidence is offered to support the suggested dose of 500 ml in less than 15 minutes for all resuscitation situations. For the normotensive but tachycardic and oliguric patient with concentrated, low-sodium urine, 250 ml delivered over an hour would be a safe rate. For the non-shocked but hypotensive patient, 250 ml in less than 10 minutes is about right. For the shocked patient, 30 ml.kg−1 warm fluids must be given as rapidly as possible, monitoring the vital signs until the patient responds. The importance of heat flux during aggressive fluid therapy was outwith the remit of GIFTAHo. Recommendation 17/1.3.2: Do not use tetrastarch for resuscitation Comment: colloids are another dimension to the chronic fluid debate that GIFTAHo could have helped illuminate. GIFTASuP recommended either colloid or crystalloid for resuscitation, but warned incorrectly that five times more volume was needed if crystalloid was chosen. The three-dimensional representation of the Twigley-Hillman fluid compartments diagram that appeared in GIFTASuP is reproduced in GIFTAHo's full document, and the misconception that red cell dilution data indicate resuscitative efficacy is perpetuated, as GIFTAHo explains that "in theory, colloids that are iso-oncotic with plasma should expand blood volume by the volume infused…" (section 5.1.4 2). The fact that colloid-induced red cell dilution is less than expected from the Twigley-Hillman paradigm 10 is not primarily a consequence of leaky capillaries. The clinical evidence, as identified by the NICE process, is that: (i) there is no advantage or disadvantage of gelatin vs tetrastarch in terms of mortality or the volume of fluid required for resuscitation; (ii) there is no advantage or disadvantage of gelatin vs either Ringer's lactate or sodium chloride 0.9% in terms of mortality; and (iii) starches have disadvantage over crystalloids for resuscitation. A fourth fact, not considered by GIFTAHo, is that gelatins may trigger sometimes fatal anaphylactic reactions 11. Current theory runs contrary to the original logic of Bayliss and Starling's biophysical colloid therapy for wound shock during the Great War. The Michel-Weinbaum interpretation of the effects of steady-state Starling forces explains that most capillaries only filter fluid to the interstitium, reabsorption being at best limited and transient 12. Contrary to traditional teaching as exemplified in GIFTAHo, adding colloid molecules to an isotonic resuscitation solution is not expected to enhance resuscitation or to absorb extravascular oedema fluid while capillary hydrostatic pressure is low 13, 14. In equipotent resuscitation doses, colloid-containing solutions cause disproportionate anaemia and hypoalbuminaemia compared with a colloid-free equivalent 15, 16. Moreover, all infused semi-synthetic colloid molecules or their breakdown products eventually circulate to the interstitium, where their effects are largely unexplored and unknown 16. On that evidence, recommendation 17 looks too narrow and should recommend avoidance of any plasma substitute unless future colloid vs crystalloid trials establish superiority. The publication of a study suggested at 7.5.4 came too late for consideration, and shows that peri-operative stroke volume optimisation with colloid is not superior to using crystalloid 16. A firm stand would be consistent with the Cochrane recommendations and give confidence to physicians that it is no longer good practice to offer patients plasma substitutes 17. GIFTAHo's reluctance to recommend against plasma substitutes is striking; it says that plasma substitutes ought to work, and that lack of evidence demonstrating this is caused by leaky capillaries. In section 9.1.1.1 (prefaced by "beyond the scope of this guidance"), the GDG opines that "Colloids are not generally used (in patients with replacement needs) unless their deficits are such that they need urgent resuscitation". Perhaps this is intended to be consistent with the evidence-free opinion of the European Medicines Agency (EMA) that hydroxyethyl starch solutions should only be used for the treatment of hypovolaemia due to acute blood loss when crystalloids alone are not considered sufficient 18. Though the Medicines and Healthcare products Regulatory Authority (MHRA) issued a Class 2 recall on starch products in the UK in 2013 19, their return to the Market could be imminent. Another unsupported and potentially harmful opinion in GIFTAHo is that "Concentrated (20-25%) sodium poor albumin could also be valuable in fluid redistribution problems" (sections 5.1.4 and 9.1.2.1). Recommendation 18/1.3.3: Consider human albumin solution 4–5% only for resuscitation in patients with severe sepsis. Comment: this rather ambiguous recommendation, which does not feature in the resuscitation algorithm, is problematic because albumin is expensive and supply is limited, and its indications (and the appropriate dose) are unknown. Does GIFTAHo advocate 'albumin 4–5% only' for resuscitation from severe sepsis or albumin 4–5% 'only for resuscitation from severe sepsis'? Severe sepsis is addressed in the replacement and redistribution algorithm and requires expert help. The NHS cannot afford widespread adoption of this recommendation by non-experts. For routine maintenance: what is the most clinically and cost effective intravenous fluid…? What is the clinical and cost effectiveness of different volumes? What is the most clinically and cost effective timing and rate of administration? Recommendations 19-23/1.4.1-5: Prescribe glucose and fifth-normal saline solution with potassium chloride 0.3% at 25 to 30 ml.kg−1.day−1 with adjustments for obesity, frailty, renal impairment, cardiac failure or malnutrition. Infuse the planned daily amount during daylight hours if possible. Comment: physiology predicts there will be very little, if any, excretion of electrolyte-free water in many hospitalised patients, because the stress hormone vasopressin makes the distal parts of the nephron permeable to water 9. Many of the drugs given to adults in hospital either stimulate the further release of vasopressin or increase the sensitivity of vasopressin receptors 20. Disorders of volume and tonicity are therefore the most common serious problems associated with intravenous fluid therapy 21. It is reported that fifth-normal saline has largely been removed from some hospitals because of multiple episodes of fluid and electrolyte complications 22. An editorial comment warned that "there can be no justification for administering hypotonic fluids in the perioperative setting" 9. A critical appraisal of indications for isotonic maintenance therapy for postoperative adult patients, paralleling the debate engaging paediatric practitioners for many years, would have been welcomed. The research recommendation (8.6.5) to investigate the optimal sodium content of hypotonic maintenance fluids in medical and surgical patients is not an adequate response. I have noted above that the EMA now advises that hydroxyethyl starch solutions may be used for the treatment of hypovolaemia due to acute blood loss when crystalloids alone are not considered sufficient 18, though the MHRA has yet to reverse the suspension of their marketing authorisations in the UK. In response to this regulatory chaos, the GIFTAHo GDG has, at the last minute, modified its research recommendations from 'acute shock' to 'acute hypovolaemic shock'. This distinction has been proposed to preserve the marketing authorisation for starches, which are now accepted to be harmful in patients fulfilling the diagnostic criteria for sepsis, but there is no reason to believe the capillary injury in non-septic shock is very different 23, 24. This last-minute change has serious consequences for the clinical applicability of future research. When faced with a shocked patient, it is often not possible to categorise the cause as septic or non-septic with any confidence. I would go further and suggest that the distinction is spurious; the capillary injury and the harm that plasma substitutes could inflict are the same. It is disappointing that a GDG of NICE must conform to MHRA or EMA decisions. So to my verdict. It is clear to me that a guideline on intravenous fluid therapy in hospitalised adult patients is far too broad a subject to be informed by a restricted number of questions, most of which have never been directly addressed by clinical investigation. One could even ask whether the exercise should have been attempted at all. Many of the published clinical trials were commercially sponsored and the tendency for such studies' results to favour the sponsors' products is well described 25, 26. Polyionic 'balanced' or 'physiological' salt solutions and plasma substitutes are competing for our attention at premium prices, often on the basis of selective data from animal studies, healthy volunteer studies, and appeals to reason based on a flawed understanding of the relevant physiology. Fluid gelatins are being promoted as 'better volume expanders' to prescribers who are unconvinced that crystalloid resuscitation is preferable, while starches are poised for return to the market for treatment of non-septic hypovolaemia. We therefore look to NICE for impartial, clear and unambiguous advice on whether we must buy them, or whether our patients and the service as a whole would be better served by using the standard generic solutions. I regret that GIFTAHo misses an opportunity to bring order to the chaotic arena of fluid and electrolyte therapy, and leaves the door open for patients to be exposed to plasma substitutes in the absence of safety data. Non-inferiority of expensive and dangerous plasma substitutes over their crystalloid equivalent is not enough reason to tolerate their continued prescription. The taxpayer should not have to pay for more expensive solutions until their safety, necessity and superiority over cheaper alternatives are established. I am also concerned that there will be an increase in the use of hypotonic solutions for resuscitation and post-surgical maintenance, and acceptance of mild hyponatraemia that may increase the number of at-risk patients exposed to harm. There is also a danger of patient harm from the proliferation of guidelines that are based on very poor evidence and that undermine the application of informed clinical reasoning 27-29. Box 2 highlights the points that I feel should underpin future guideline development. Alarmingly, plans are already afoot to transform GIFTAHo into a Standard of Care. I am sorry to have to recommend that anaesthetists oppose the adoption of these particular guidelines into their hospital policies. However, that would place the onus on us to develop and maintain local fluid therapy practices and policies that are appropriate to the clinical service, responsive to the requirements of each individual patient, and recognise the need to accept informed clinical reasoning by a suitably qualified person. How hard can it be 30? No external funding declared. TW is a member of the AAGBI Council. These views represent the author's opinions.