Fluid resuscitation for septic shock

Fluid resuscitation in sepsis is viewed as a mainstay of therapy, and in fact many trial authors refer to it using those exact words. The theory of resuscitating septic shock patients with fluid is probably built upon flawed foundationsgiven that preload is usually offered to improve cardiac output which in sepsis is frequently normal or high. However, such theoretical objections are usually disregarded. For one, it is not clear that the current model is in any way deficient - given that at least in Australia we seem to be enjoying a lovely downtrend of sepsis mortality from 35% in 2000 to around 18% in 2012 (Kaukonen et al, 2014). According to ARISE data, we seem to be resuscitating patients with a lot of fluid - the average input at 72 hours seems to be around 7000ml (or around 90ml/kg).

This issue was interrogated in Question 19 from the second paper of 2011. Shamefully, only 29% of the candidates passed this "critically evaluate" question. This beggars the imagination. If you are going to administer 7 litres of something to your patient, then you should probably have some scientific rationale for doing so, and be able to articulate a defense of the practice. The college in fact do want you to defend "early, rapid, and substantial infusion of intravenous fluids"; according to their model answer, fluid resuscitation "is clearly indicated to treat acute hypotensive episodes". The well-informed candidate who takes a contrary stance thereby imperils their mark.

In brief, one may summarise the important points as follows:

Rationale and advantages

  • In sepsis, preload is reduced by third-space losses from leaky capillaries
  • Improve preload = improve cardiac output
  • Recommended by SSG
  • Other forms of shock benefit from fluids
  • 20% albumin may help meet haemodynamic goals faster (ALBIOS) and protect the glycocalyx

Counter-rationale and disadvantages

  • In sepsis, cardiac output is already high
  • Fluid bolus use is based on animal studies
  • No evidence regarding the optimal dose of fluid
  • Other forms of shock are aetiologically different from sepsis
  • Albumin does not improve survival

Classic studies to quote:

  • SAFE: septic group = mortality benefit from albumin (post-hoc, underwpowered sub-group)
  • ALBIOS: clinically insignificant (1-2mmHg) improvement in MAP for the albumin group
  • VASST: positive fluid balance was associated with increased mortality
  • FEAST: increased 48-hour mortality with fluid boluses (but difficult to generalise)
  • SOAP : OR for mortality increased by 1.1 for every 1000ml of positive fluid balance

There is a vast amount of reading material on this; in fact even reading the abovelisted trials is probably beyond the needs of a time-poor candidate. Instead, one may limit oneself to some recent expert opinions. This decreases the risk of forming one's own, should it accidentally be incompatible with the official college views.  The following two articles from the November 2016 issue of Intensive Care Medicine are probably the best short summaries of the available evidence, and offer two contrasted views of this issue.

Rationale for fluid resuscitation in septic shock

  • Septic patients are hypotensive.
  • Hypotension is a major feature of sepsis, and results from both vasodilation and actual fluid loss (specifically, the migration of fluid into the extravascular compartment).
  • Preload decreases, and cardiac output may be affected, decreasing tissue perfusion
  • Thus, fluid resuscitation should improve cardiac output and tissue perfusion.

As an argument from authority, one should also mention the Surviving Sepsis Guidelines, which (on the basis of expert opinion) suggest that fluid resuscitation is a vital part of the first six hours.

Volume of resuscitation fluid

There are two conflicting issues, leading to a cognitive dissonance in sepsis resuscitation. On one hand, the Surviving Sepsis Guidelines recommend liberal fluid resuscitation (30ml/kg) with generous haemodynamic endpoints (MAP over 65, CVP around 8-12). On the other hand, we have trial data and prospective observational findings which associate large-scale fluid therapy with decreased survival. The effect is that ICU staff still end up infusing vast volumes of fluid into their patients while shaking their heads and remarking how bad that is, how harmful, how far divorced from evidence, etc.

How much do we give? In the recently published ARISE trial the total volume pre-randomisation (i.e. given in the Australian ED) was already around 34ml/kg, or around 2.5L (we are talking about the "usual care" arm). When it got to 72 hours, the patients had received 6973 ml (± 4267ml), i.e. from to 2,706 ml to 11,240 ml, or 93.9ml/kg on average. This is on par with PROCESS and PROMISS trials, and various experts have started saying that this fluid therapy is the "international standard of care"

Evidence in support of conservative fluid management

FEAST study is made much of. It was probably the stimulus for Question 19 from the second paper of 2011 (it was published in June of 2011, just in time for the examiners to incoporate it into the late August paper).  The investigators found that fluid boluses significantly increased 48-hour mortality. However, even a casual observer will note that the trial had significant limitations, especially when it comes to its generalizability to the ICU setting:

  • It was set in resource-poor Africa
  • It was a paediatric trial (mean age of 24 months)
  • Most of the patients had malaria (57% of them!)
  • Many (37%) had a haemoglobin under 50g/L
  • Survival was measured at 48 hours, rather than the normal 28 or 90 days.

SOAP study: positive fluid balance was among the strongest predictors for death from sepsis in the ICU. This prospective study enrolled 3.147 adult ICU patients. The odds ratio of mortality in Europe increased by 1.1 for every added litre of cumulative fluid balance in the first 72 hours. One might offer the criticism that sicker and older patients have the greatest amount of fluid infused as their hypotension is the most refractory, but the SOAP investigators made the greatest effort to correct their numbers for APACHE and age, suggesting that fluid balance is an independent predictor.

VASST trial data, when analysed in 2011 by Boyd et al, also revealed that a positive fluid balance was independently associated with mortality. Of even greater concern is the association between mortality and a CVP of 8-12mmHg, which the SSG recommend as one of their early endpoints. The patients with the lowest CVP (under 8) had the lowest mortality rate. The authors conclude that "optimal survival in the VASST study occurred with a positive fluid balance of approximately 3 L at 12 hrs", suggesting that the usual technique of flooding the patients was somehow detrimental.So, where does this flood therapy come from?

CLASSIC study (Hjortrup et al, 2016) was another good attempt to make sense of this issue. The investigators enrolled patients in whom some initial resuscitation had already been carried out, and then randomised them to receive fluids ad libitum or only when they became "severely hypoperfused", i.e. mottled and oliguric. Unfortunately, after enrolling already overloaded patients (pre-randomisation volumes around 4000-500ml) and suffering a massive rate of protocol violations (up to 45%) the study could not achieve a wide enough practice separation between groups. With groups so similar (total input of 12.5L vs 13.7L at five days) it is difficult to take seriously the decreased rates of AKI (37% vs 54%) and death (33% vs 41%). Not that the study was powered to detect such differences anyway. 

Arguments for large-volume fluid resuscitation in sepsis

Turns out, the practice of torrential resuscitation is based almost entirely in animal data and expert opinion. Those experts, in turn, base their opinions on their understanding of physiology and on their experience (filled with fondly remembered success stories of waterlogged septic patients). Hilton and Bellomo weighed in with their opinion in 2012, questioning the value of the saline deluge. Their article is an excellent resource for arguments against wanton abuse of fluid. Key issues raised by them are as follows:

  • Argument from physiology: fluid boluses are necessary to increase cardiac output and improve the delivery of blood to the organs. However:
    • There is no evidence that cardiac output is always decreased in sepsis.
    • There is no evidence that hypotension in sepsis is uniformly harmful
    • Nobody has considered the effect of increasing delivery of toxin-rich blood
    • No data exists to balance the physiological benefits of fluid boluses against the physiological costs of oedema.
  • Arguments from animal experiments: increased survival is demonstrated in fluid-bolused animal models of sepsis. However:
    • We are not mice.
    • The animals frequently receive weird fluid therapy (eg. by subcutaneous infusion)
    • Even animal models disagree as to the magnitude and duration of the effect
    • Many animal models of sepsis contradict the paradigm by featuring increased cardiac output and increased organ blood flow
  • Arguments from analogy: fluid resuscitation works for other forms of critical illness. However:
    • Other forms of shock have very different pathophysiology
    • Some (eg. haemorrhagic shock from penetrating trauma) clearly don't benefit from aggressive fluid resuscitation
    • Acute kidney injury and ARDS don't benefit from it either.

In general, Hilton and Bellomo suggest that 20-25ml/kg should be the maximum, not the minimum (as suggested by the SSG) and that earlier vasopressors should be promoted. A recent opinion piece by John Myburgh also recommends using lower bolus doses, citing the ProCESS investigators who found no difference in mortality between their study groups (of which the EGDT arm on average received 2.8L within the first 6 hours, whereas the "usual care" group received 2.3L).

Choice of resuscitation fluid

Choice of crystalloid for resuscitation of septic shock

Essentially, the question is - should we be strongly critical of the ED tendency to flood these patients with saline, and should we be using Hartmanns or Plasmalyte instead? Or does it even matter? Raghunathan et al (2014) have demonstrated some survival benefit from "balanced" crystalloid in a retrospective cohort. A meta-analysis of similar vintage found the same association, but the confidence of recommendations was rather low, owing to heterogeneity. Overall, a large randomised controlled trial is being called for.

Utility of albumin in septic shock

The studies published to date have somehow served to encourage both those people who like to use albumin, and those who don't.

SAFE study is often quoted in support of the use of albumin in sepsis.

  • The improvement in mortality in the septic subgroup did not reach statistical significance (the group was not powered to detect a subtle treatment effect).
  • However, the study did demonstrate that albumin was as safe as saline; the lack of evidence for harm was enough to encourage people.

ALBIOS trial is the more recent entry into the scene; again albumin did not appear to be associated with any mortalty benefit.

  • Albumin appears to improve mortality of septic shock patients once haemodynamic stability has been achieved, and was associated with some haemodynamic advantages (eg. a higher MAP, a lower positive fluid balance, and a lower heart rate).
  • As some critics have pointed out, those differences were statistically significant but clinically insignificant- the increase in MAP, for instance, was by 1-2mmHg.

EARSS study (thus far unpublished) also could not demonstrate any mortality benefit at 90 days, according to the abstract. The full paper is still awaited after two years; it is expected to be another negative paper for albumin.

Overall, a recent meta-analysis did not find enough evidence to recommend the use of albumin in sepsis.

Utility of blood transfusion in septic shock

The Surviving Sepsis Guidelines had at one stage recommended using PRBCs to pursue the ScvO2 goals of therapy (aiming at a haematrocrit of around 0.30). This was on the basis of older studies among post-bypass cardiothoracic patients.

TRICC study: undifferentiated ICU patients did not benefit from a liberal transfusion strategy; however, the group of septic patients was small (23 and 18 pts with Hb of 70 and 90 respectively) and it was difficult to draw conclusions on the basis of this.

TRISS study: septic shock patients did not seem to benefit from a higher haemoglobin threshold. The SSG people have promised to review their recommendations.


An excellent resource for this topic are the chapters in Oh's Manual dealing with severe sepsis (ch 61, by A Raffaele de Gaudio) and with the immunocompromised host (ch 59, by Steve Wesselingh and Martyn A H French).

An older, yet similarly respectable source is Shoemaker (2005); Chapter 155 (Infections in the immunocompromised patient) by Andrew Githaiga, Magdaline Ndirangu and David L. Paterson covers this topic with great detail.

The Surviving Sepsis Campaign has these published guidelines to peruse.

Vincent JL, Sakr Y, Sprung CL, Ranieri VM, Reinhart K, Gerlach H, Moreno R, Carlet J, Le Gall JR, Payen D; Sepsis Occurrence in Acutely Ill Patients Investigators. Sepsis in European intensive care units: results of the SOAP study. Crit Care Med. 2006 Feb;34(2):344-53.

Maitland K, et al and the FEAST Trial Group. Mortality after Fluid Bolus in African Children with Severe Infection. N Engl J Med. 2011 May 26.

Delaney, Anthony P., Arina Dan, John McCaffrey, and Simon Finfer. "The role of albumin as a resuscitation fluid for patients with sepsis: A systematic review and meta-analysis*." Critical care medicine 39, no. 2 (2011): 386-391.

Raghunathan, Karthik, et al. "Association Between the Choice of IV Crystalloid and In-Hospital Mortality Among Critically Ill Adults With Sepsis." Critical care medicine (2014).

Caironi, Pietro, et al. "Albumin replacement in patients with severe sepsis or septic shock." New England Journal of Medicine 370.15 (2014): 1412-1421.

Dellinger, R. Phillip. "Crystalloids for fluid resuscitation in sepsis: where is the balance?." Annals of internal medicine 161.5 (2014): 372-373.

Rochwerg, Bram, et al. "Fluid resuscitation in sepsis: a systematic review and network meta-analysis." Annals of internal medicine 161.5 (2014): 347-355.

Boyd, John H., et al. "Fluid resuscitation in septic shock: A positive fluid balance and elevated central venous pressure are associated with increased mortality*." Critical care medicine 39.2 (2011): 259-265.

Russell, James A., et al. "Vasopressin versus norepinephrine infusion in patients with septic shock." New England Journal of Medicine 358.9 (2008): 877-887.

Hilton, Andrew K., and Rinaldo Bellomo. "A critique of fluid bolus resuscitation in severe sepsis." Crit Care 16.1 (2012): 302.

Marik, Paul Ellis. "Fluid Responsiveness and Fluid Resuscitation." Evidence-Based Critical Care. Springer International Publishing, 2015. 57-87.

Myburgh, J. A. "Fluid resuscitation in acute medicine: what is the current situation?." Journal of internal medicine 277.1 (2015): 58-68.

Yealy, Donald M., et al. "A randomized trial of protocol-based care for early septic shock." The New England journal of medicine 370.18 (2014): 1683-1693.

Finfer, Simon, et al. "A comparison of albumin and saline for fluid resuscitation in the intensive care unit." N Engl j Med 350.22 (2004): 2247-2256.

Patel, Amit, et al. "Randomised trials of human albumin for adults with sepsis: systematic review and meta-analysis with trial sequential analysis of all-cause mortality." BMJ 349 (2014): g4561.

McEvoy, Rinaldo Bellomo, et al. "The SAFE Study Investigators Impact of albumin compared to saline on organ function and mortality of patients with severe sepsis." Intensive Care Med 37 (2011): 86-96.

Flannery, Alexander H., Sean P. Kane, and Angel O. Coz-Yataco. "A word of caution regarding proposed benefits of albumin from ALBIOS: a dose of healthy skepticism." Crit Care 18 (2014): 509.

Charpentier, J., and J. P. Mira. "Efficacy and tolerance of hyperoncotic albumin administration in septic shock patients: the EARSS study." Intensive care medicine. Vol. 37. 233 SPRING ST, NEW YORK, NY 10013 USA: SPRINGER, 2011. - not yet available!

Hébert, Paul C., et al. "A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care." New England Journal of Medicine340.6 (1999): 409-417.

Holst, Lars B., et al. "Lower versus higher hemoglobin threshold for transfusion in septic shock." New England Journal of Medicine 371.15 (2014): 1381-1391.