Sepsis Fluid Resuscitation

For decades, sepsis resuscitation was guided by clinical gestalt and central venous pressure targets, with no high-quality evidence defining the optimal approach. In 2001, Rivers and colleagues published a single-centre RCT demonstrating that early goal-directed therapy (EGDT) — a protocolized bundle targeting CVP, MAP, ScvO2, and haematocrit — reduced absolute mortality by 16% compared to standard care. The study transformed sepsis management overnight, embedding EGDT into the Surviving Sepsis Campaign guidelines and making the 'sepsis bundle' a global standard of care. However, the single-centre design, unusually high control-group mortality, and reliance on invasive monitoring raised questions about generalisability.

Three large multicentre RCTs — ProCESS (USA), ARISE (Australasia), and ProMISe (UK) — simultaneously dismantled the EGDT paradigm. Each trial enrolled patients meeting the same sepsis criteria as Rivers but found no survival benefit from protocolized EGDT compared to usual care. ProCESS randomised 1,341 patients to EGDT, protocol-based standard therapy, or usual care and found identical 60-day mortality across all groups. ARISE confirmed this in 1,600 patients across 51 Australasian sites, and ProMISe replicated the finding in 1,260 UK patients. The consistent null results across three continents demonstrated that the elements of EGDT — particularly ScvO2 monitoring, dobutamine, and transfusion triggers — added complexity without benefit, and that contemporary usual care had already incorporated the truly valuable components: early recognition, antibiotics, and initial fluid resuscitation.

With protocolized targets debunked, attention shifted to fluid type and volume. The SMART trial (2018) compared balanced crystalloids (lactated Ringer's) to normal saline in 15,802 ICU patients and demonstrated a reduction in the composite of death, new renal replacement therapy, or persistent renal dysfunction — effectively ending the reflexive use of 0.9% saline. The BaSICS and PLUS trials provided large-scale confirmatory data, though with more nuanced results showing trends favouring balanced solutions without reaching statistical significance for mortality alone. Most recently, CLOVERS (2023) directly compared restrictive versus liberal fluid strategies in early septic shock, finding no difference in 90-day mortality but signalling that more fluid is not necessarily better. Together, these trials established that balanced crystalloids are preferred and that clinicians should exercise restraint in fluid volumes beyond initial resuscitation.

The Surviving Sepsis Campaign 2021 guidelines reflected this paradigm shift by recommending an initial 30 mL/kg crystalloid bolus within 3 hours for sepsis-induced hypoperfusion (weak recommendation), while acknowledging that ongoing fluid administration should be guided by dynamic measures of fluid responsiveness rather than static targets. The guidelines suggest balanced crystalloids over normal saline (weak recommendation) and recommend against hydroxyethyl starch. The 2021 update notably moved away from mandating the full EGDT protocol, instead emphasising early antibiotics and source control as the highest-priority interventions.

Current practice reflects a remarkable reversal from the Rivers era. Clinicians now prioritise early antibiotics and source control over protocolized haemodynamic targets, use balanced crystalloids as the default fluid, and increasingly adopt a 'fluid-conservative' approach after initial resuscitation. Dynamic assessments of fluid responsiveness (passive leg raise, pulse pressure variation) have replaced static CVP targets. The pendulum has swung from 'flood and squeeze' to 'restrain and reassess.' Ongoing trials are exploring even more restrictive initial fluid volumes, personalised resuscitation guided by point-of-care echocardiography, and the role of early vasopressors to limit fluid administration. The field continues to grapple with the fundamental question: how much fluid is enough, and when does it become harmful?