This issue has never come up in the fellowship exam, but merits some attention from the preparing candidate because studies have been published and words have been written on this topic, most notably by local experts who also happen to be examiners. So, one might ask – why have the Myburghs and Bellomos of this world not written any SAQs asking for us to “critically evaluate the use of balanced crystalloid” etc? Well, perhaps because it would be hard work marking such a thing, or because there is equipoise, or because the question does not discriminate safe intensivists from unsafe ones. Either way, no past paper “model answers” exist to help us divine what the college officially thinks on this issue. As such, the topic is probably of little importance to the person who is one week before exam. Its relevance in this section is largely theoretical, to protect trainees from the possibility that they might come across an examiner at a viva who happens to have some strong unbalanced views about saline.
As far as published evidence and opinion go, there is much in the literature. The exam candidate can safely limit their reading to virtually any top Google result for a search string where “saline” and “balanced” figure prominently. Of the saline-bashing, the most violent prison-shower-like attacks happen in articles by Dileep Lobo (see here, here, here here and here, as well as here and a little bit here). The man has strong opinions and has infiltrated various organisations giving rise to anti-saline guidelines (more on that later). Of balanced views, there are many of notable quality, but one which stands out is a paper by Cortez Bonor and Vincent from 2014. The time-poor candidate can limit their reading to this single resource. An updated alternative is the 2016 review by Reddy et al. LITFL also has a great one-stop resource. For a good review of the reasons as to why saline should not be dismissed as toxic pickle brine, Thomas Woodcock's 2014 critique of the NICE fluid guidelines and Chris Palmer's ACEP 2014 editorial are also worth reading.
Rationale and boundaries for this discussion
To be clear, the specific question is whether there is any realistic benefit to the use of balanced crystalloid solutions (such as Plasmalyte or Hartmanns) instead of normal saline for maintenance or resuscitation in the undifferentiated cohort of “critically ill” patients.
Arguments against the use of saline
One can make an argument that saline has physiological effects which are detrimental, and which are exaggerated in the ICU where the volume of resuscitation fluid is greatest. These are discussed in some detail in a recent article by Santi et al (2015). Although the discussion has a somewhat paediatric inclination, the arguments they offer remain interesting, and are summarised below. Wherever possible, the evidence supporting or dismissing the argument is offered.
Normal anion gap metabolic acidosis
Though this is usually well tolerated in the healthy patient (eg. Lobo’s surgical registrars) the ICU population is rather less likely to shoulder it effortlessly. Without reproducing the entire list of consequence of acidosis, it will suffice for us all to agree that it has negative physiological effects on the nearly dead ICU patient. Negative inotropy, impaired response to catecholamines, increased respiratory drive and so forth.
Whether this theoretical concern translates into real-life dead patients is difficult to ascertain. Taken as a homogeneous group, the mortality of all-comer ICU patients is probably unaffected. In undifferentiated ICU patients, Gunnerson et al (2006) were unable to find a strong association between normal anion gap metabolic acidosis and increased mortality (29% dead vs. 26% in the non-acidotic group, vs. 56% for lactic acidosis). However, some patient groups are probably going to be worst affected, for example patients with trauma (Ho et al, 2001) – acidosis is part of the “lethal triad” which is a widely acknowledged Bad Thing. Similarly, Neyra et al (2015) were able to demonstrate increased mortality from hyperchloraemia in critically ill septic patients (with an OR of death around 1.37 associated with a 5mmol/L increase in chloride).
Acute kidney injury
Hyperchloraemia seems to be associated with decreased renal blood flow velocity and decreased renal cortical perfusion among healthy volunteers (Choudhury et al, 2012). In their article, Santi et al theorise that this may arise from afferent arteriolar vasoconstriction in reaction to increased chloride delivery to the macula densa. Surely, ICU patients do not need any extra reasons to have renal impairment. Does this really affect ICU-relevant outcomes? Prospective non-RCT data (eg. Bellomo et al, 2012) has supported the idea that chloride does some sort of renal damage; of the linked study cohort at Austin Hospital, the chloride-enriched group had almost twice the incidence of requiring dialysis (10% vs. 6.3%). Thus far, no association with increased mortality has been found, in spite of the fact that acute kidney injury is traditionally supposed to increase ICU mortality.
Contrary to hypertonic saline (which is supposed to have some sort of anti-inflammatory effects) normal saline is suspected of activating the release of inflammatory mediators. This, however, comes from either animal studies or small scale human data. For instance, Wu et al (2011) reported substantially worsened SIRS variables in patients with pancreatitis who were resuscitated with normal saline. However, this was a study using the now-defunct (and widely derided) SIRS definitions, and they used CRP to measure inflammation (which may not be the best-ever biomarker of such things).
Normal saline is thought to have some sort of negative influence on perioperative haemostasis and blood transfusion requirements. Cortez et al (2014) summarise the data and conclude that probably there is no such effect in the majority of patients. By massaging the data through post-hoc subgroup analysis of other people’s findings, these authors were able to identify a high risk group of surgical patients who had increased transfusion requirements associated with the use of normal saline.
Nausea and vomiting
Though Heidari et al (2011) found an increase in the risk of post-operative nausea and vomiting with normal saline among ninety Pakistani elective surgical patients, no such increase was found by Karaca et al (2006) among sixty Turkish patients undergoing spinal anaesthesia. Is this for real? Who knows. Certainly this has little bearing on the overall experience of a “proper” ICU patient, i.e. somebody who is sedated and ventilated with multiple organ problems. The nausea signal is lost in the mess of the overall unhappy state of critical illness.
The irrelevance of the cost
Operating from the privileged perspective of a first-world ICU, one may easily dismiss arguments regarding the cost of various resuscitation fluids. When a day of ICU stay costs $5000 AUD even before you start throwing levosimendan at people, haggling over the couple of dollars difference between saline and Hartmanns becomes mildly ridiculous. Of course, the situation is very different in resource-poor environments where saline might be the only resuscitation fluid available. In which case no physiological arguments regarding acidosis and AKI are going to deter the clinician from continuing to use this fluid for resuscitation, rendering the cost discussion irrelevant.
Arguments in support of normal saline
The grey-haired intensivist might angrily dismiss the modern infatuation with balanced crystalloid as a fad. In my day, they might growl, there was only the imperative to save lives, and none of this physiological wankery. Indeed, human experience with normal saline has been extensive, and its ongoing popularity cannot be purely attributed to cost advantage or the inaccurate but reassuring use of the word “normal” in its name. The core message of this counterargument is that it would be inappropriate to change established practice unless there is strong evidence that there exists an even better practice.
Inadequacy of anti-saline studies
The list of saline-bashing studies offered above contains virtually no randomised controlled trials. Most of the studies which suggest a disadvantage from the use of saline are observational, either prospective or retrospective. The majority of clinical trials which have yielded strong pro-balanced-crystalloid sentiment have been small in scale and have focused on short term biochemistry rather than “hard” patient-centered outcomes.
Furthermore, the association of increased morbidity and mortality with hyperchloraemia found in prospective and retrospective studies may be due to the association of raised chloride with the volume of administered fluid. That chloride they found did not rise by itself- it took litres and litres of saline. Excessive fluid administration is well known to be associated with worse outcomes (remember SOAP, for instance). Thus, this hyperchloraemia may be merely a marker of over-resuscitation, and the real result of these studies is that excess fluid causes harm.
The benign acidosis
The abovementioned Gunnerson et al (2006) study did not find any evidence of increased mortality among an unsorted population of ICU patients. Some might argue that even in the cohort of critically ill patients normal anion gap metabolic acidosis is benign and self-limiting. Or, rather one might say that the mortality influence from this minor metabolic disturbance is so minor that its significance is lost in the fog of various ICU interventions and complications, many of which have a more profound mortality difference.
Increased efficacy as a resuscitation fluid
Sodium stays in the extracellular fluid, as does chloride. The presence of these both in the resuscitation fluid favours its retention in compartments which are clinically relevant during resuscitation. Ergo, normal saline should be a better choice than a fluid which does not contain as much sodium or as much chloride. If one takes resuscitation seriously as “the thing we do to improve organ perfusion and function”, then normal saline may achieve that goal better than balanced crystalloids.
The very real relevance of cost
At the organisational level (eg. that of a statewide health service) the change from saline to another more expensive crystalloid does indeed impose a significant cost difference, which annually could escalate into the millions. Consider that the local emergency department saw approximately 70,000 patients last year. If even only 10% of them required fluids, and of those only half received more than 2000ml, the total cost in Plasmalyte148 would already be over $200,000 (going by the street price of Baxter fluids in my ‘hood). This dollar could be spent elsewhere.
The adverse effects of balanced crystalloids
All those expensive fluids are themselves not benign. The following observations can be made about them:
- They are not biologically equivalent, and none approach the concentration of human body fluids very closely.
- Hartmann’s is hypotonic, which is counterproductive if you want to avoid oedema.
- Hypotonic fluids, when used in liberal volumes, may give rise to cerebral oedema (i.e. the uncritical use of such fluid can produce a blood-brain osmotic gradient)
- Calcium in Hartmanns can react with blood products and inactivate antibiotics (eg. ceftriaxone)
- Little is known about gluconate (a component of Plasmalyte), which is worrying.
- Acetate has a recognised cardio-depressant effect (Kirkendol et al, 1979)
- Alkalosis arises from the overuse of balanced crystalloid, and this is not a benign process (at leats no more benign than the normal anion gap acidosis due to saline).
Saline versus balanced crystalloid in big randomised trials
Up to 2014, Cortez Bonor and Vincent identifed 28 studies, which were all small and the meta-analysis of which had led to no firm conclusions about anything. Since then a few more developments have taken place, of which a few were run by ANZICS CTG and therefore have exam relevance (as one might run afoul of a primary investigator in the vivas).
This was a double-crossover RCT conducted in 4 ICUs from New Zealand. 2278 patients were enrolled. The hypothesis was that the use of Plasmalyte would be protective against acute kidney injury. It was not. However, of the 1100-something patients in each group, only 38 went on to have dialysis, which begs the question - how sick were they, really? Were these “real” ICU patients? On average, they all received only about 2000ml of resuscitation fluid. Overall in-hospital mortality was around 7-8%, suggesting that the population was perhaps too well to benefit from elegant biochemical manipulation. Indeed, a subgroup analysis which excluded short-stayers and elective post-op patients found a statistically insignificant (but interesting) trend towards increased mortality (19% vs 15%).
SMART (Semler et al, 2018)
This was a huge trial (15,802 patients) from Vanderbilt Medical Centre in Nashville, Tennessee. Again, these were supposedly "critically ill" patients with an expected mortality of around 8%. Like SPLIT, the resuscitation volume was even smaller small - the balanced group got about 1L of balanced crystalloid, and the saline group got about 1L of saline, on average. It is therefore surprising that there was a statistically significant trend towards improved mortality in the balanced group, which amounted an absolute risk reduction of around 0.8%. Unfortunately, instead of good honest mortality, the investigators used a composite endpoint which combined mortality together with dialysis, persistent renal dysfunction, ICU free days, and so on. That makes the primay endpoint result even less plausible. Fortunately, an a priori subgroup of "real" ICU patients with sepsis, the mortality difference was greater (25.2% vs 29.4% in the saline group)
SALT-ED (Self et al, 2018)
This was also a large sing-centre trial which looked at ED patients. 13,347 patients were enrolled and on average receoved about 1L of one fluid or the other. Again, there was no mortality or hospital stay related difference between the groups, probably owing to the trivial fluid dose they received. Like SMART, these guys used the "MAKE30" composite endpoint (Major Adverse Kidney Events within 30 days) and found a 1.1% difference in this outcome, which achieved statistical signficance but which was still underwhelming.
PLUS (Finfer, Bellomo et al)
At the time of writing in 2017, this was still in its “set up phase”, and at the time of the later revision in 2020 it was still collecting data, but wrapping up. The idea is that ultimately this CTG megatrial will compare Plasmalyte with saline in a gargantuan cohort of 8800 patients, looking mainly at patients who need "real" resuscitaton, and not this panzy 2000ml volume. "Proper" ICU patients are boggy as hell, we say, and it is this group who will benefit most from the choice of a balanced crystalloid. With Finfer and Bellomo at the helm, one can expect it to appear in NEJM (if not on a platinum tablet launched into interstellar space).
Support for balanced crystalloid in society guidelines
Apart from NICE, the various societies do not seem to have specific guidelines. The unfortunately abbreviated British GIFTASUP and the even more unfortunate 2014 GIFTAHo update favoured the use of lactate or acetate-balanced solutions for everything except for replacement of chloride-rich nasogastric losses (but that was virtually guaranteed with the presence of Lobo among the authors). These recommendations were not met with uniform agreement. There are both opponents (Woodcock, 2014) and defenders (Stroud Nolan and Soni, 2014).
“Own Practice” statement
Generally, the examiners like to see an answer to “what would you do in this situation?” as a means of determining whether the candidate is likely to have well-formed opinions (rather than just memorised trial results). However, a good exam result is favoured by a well-formed opinion which falls in line with the examiner’s opinion. Locally, there appears to be a pro-balanced sentiment. Glassford et al published a 2016 retrospective study of “fluid ecology” in Australia and New Zealand demonstrating that balanced crystalloid use is on the rise (“widespread bi-national preference” is how they described it).
Thus, for exam purposes, basic statements can be made:
- The aim of ICU management at its most fundamental level is to normalise the physiological homeostasis of the patient.
- The use of balanced crystalloids is as valid as the use of saline when it is used to achieve this aim. Acidotic patients should receive “alkalinising” balanced crystalloid, and alkalotic hypochloraemic patients should receive saline.
- In patients with unimpaired acid-base balance, the fluid choice should aim to maintain that balance. Thus, these patients should be resuscitated with balanced crystalloid. This is not because there might be some sort mortality or renal perfusion benefit. Rather, the pursuit of physiological normality is the goal. This separates a “reasonable” fluid choice from the ideal fluid choice.
- The mindless use of any specific fluid choice will result in adverse effects, regardless of which fluid one chooses. The user of resuscitation fluid must be familiar with the physiological consequences of administering normal saline (or Hartmanns, or Plasmalyte), as it would be insanely irresponsible to infuse a patient with 8000ml of a substance which one does not fully understand.