Critique of the Surviving Sepsis guidelines

Among the CICM examiners' most favourite punching bags, none are most prominently featured in the exam than the Surviving Sepsis guidelines. Trainees are usually invited to "critically evaluate" the guidelines - which means, describe their rationale, explain the pros and cons and give evidence to support their "own practice" statement. Because this has long been as staple of the Fellowship Exam, to score less than 7/10 is inexcusable for anybody who's done any amount of preparation for this test, and all future versions of these questions should be viewed as easy marks by the savvy candidates.

• Question 1 from the first paper of 2014 required the candidates to generate a coherent ANZICS-themed response to the SSG and it was generally answered poorly.
• Question 19 from the second paper of 2006 is analogous, but refers to the now-defunct 2003 version of the guidelines, which had even more problems.
• The most recent guidelines have been interrogated in Question 12 from the second paper of 2015.
• Given that in April 2018 the SSG updated their guidelines again (Levy et al, 2018), one could very reasonably expect to see an SAQ about this in the 2019 papers (allowing for the typical 18-month lag).

If one were too junior to generate their own opinion of the 2012 guidelines, where would one turn? Literature comes to the rescue with prefabricated opinions from experts.

• A brilliantly informative flame war developed in Chest (2013):
  • Paul Marik and the ANZICS team trashed the bundle packages in Chest (2013).
  • Dellinger and Townsend of the SSG offered a rebuttal.
  • Marik et al rebutted their rebuttal.
• Marik also had a go at the earlier guidelines package in 2011 (Annals of Intensive Care), and discussed the controversies in the management of sepsis current to 2014.
• Vo and Kahn critique the GRADE system in Critical Care (2013).
• The SSG authors offered their response to the ProCESS and ARISE trials.
•  Lena Napolitano (2018) has summarised the relevant changes to the guidelines between 2016 and 2018
• IDSA (Infectious Diseases Society of America) explaining why they didn't endorse the 2016 version of the SSG (Gilbert et al, 2018)

More recent entries, relevant to the new 2018 SSG:

• PulmCCM critique (harsh)
• PulmCrit critique (harsher)
• Pepper et al (2018), examining the evidence on which the SSG (and subsequent US performance measures) were based (spoiler: "few trials, poor-quality and confounded studies")

One could do worse than regurgitating these expert opinions in the exam. For one, some of the quoted experts might be the actual CICM examiners. To render easier the process of retention and disgorgement, these opinions are distilled in the discussion below.

In brief summary:

Now, in some detail:

Advantages of the SSG

High quality of the presented package

• Excellent literature search and summary.
• A good source of literature references
• Well presented and easily accessible website.
• Produced by an international panel of experts reviewing and grading the evidence.
• GRADE system makes it easier for clinicians to assess the strength of a recommendation.

Evidence in defence of the guidelines

• There is no evidence they do any harm (eg. there has been no evidence of harm from ScvO₂ monitoring)
• Even in Australia (where ANZICS did not endorse the 2008 guidelines) patient management seems to be carried out with some awareness of the SSG protocols (eg. the "usual care" group in the ARISE study were resuscitated rapidly according to well-accepted endpoints and received their antibiotics early, ... but somehow that was not "goal-directed therapy").
• Failure to standardise care may result in an inefficient variation in care.
• Bundled interventions may work synergistically and the whole package may generate greater benefit than the sum of its parts.
• The recommendation for early antibiotics meets high standards for inclusion in protocols.

Criticisms of the SSG

Arguments against bundled care in general:

• Marik et al: "Systematic promotion of evidence illustrates the power of bundles to magnify ineffective therapies."
• All-or-none bundles may promote harmful therapies together with helpful ones. People end applying evidence-based treatment together with unsupported treatment. Remember Xigris, propagated by the SSG despite abundant evidence against its use.
• There is no evidence that "bundle synergy" exists.
• The quality of care might end up being measured according to bundle compliance, which is dangerously divorsed from actual "quality".

Objections on the basis of methodology:

• Many of the recommendations are not based in high-level evidence
• None of the recommendations are based in Level 1 evidence.
• The GRADE system is subjective.

Objections to the guidelines themselves:

• The guidelines digress significantly from sepsis, and creep across into other areas of ICU management.
• Controversial recommendations are made, which may not be supported by evidence:
  • There is no survival benefit from routine ScvO₂ monitoring (SSG authors agree)
  • There is no survival benefit from routine dobutamine use.
• The emphasis on EGDT is not supported by strong evidence.

Objections to the evidence offered in support of widespread implementation:

• All the trials demonstrating a benefit from SSG implementation were "before-and-after" trials.
  • These are chronically subject to publication bias, patient selection bias, temporal bias, and the Hawthorne effect.
• All the trials tested the whole bundle, but frequently only one specific component was independently associated with a treatment effect (eg. early administration of antibiotics in the Edusepsis study from Spain)
• The offered studies demonstrate a treatment effect in spite of poor adherence to the bundle (again Edusepsis - as an example mentioned by Marik et al - 5% decrease in mortality was attributed to the six-hour bundle, but the sites only complied with it 10% of the time).

Empirical evidence against the use of the guidelines:

• Procolised care seems to have no benefit - yes, perhaps the "usual care" group in the ARISE trial all seemed to get some sort of early therapy, directed by certain resuscitation goals, but the investigators confirmed that none of the sites followed any sort of specific resuscitation protocol.
• Sure, the guidelines seem to be associated with an improvement in mortality, but in Australia where they are not followed,
  • the mortality already began to improve before the first guidelines were published
  • the mortality from sepsis remains better in SSG-free Australia than in countries where the SSG bundles are routinely implemented.
• The ProCESS,  ARISE and ProMISE studies did not demonstrate any survival benefit from protocolised care (see a detailed breakdown of these trials in the chapter on early goal-directed therapy for sepsis)