Toward a More Nuanced Approach to the Early Administration of Intravenous Fluids in Patients With Sepsis

After decades of research and dozens of failed trials, the mainstay of the treatment of sepsis remains early administration of antibiotics and intravenous fluid resuscitation. The importance of early antibiotics in severe bacterial infection is biologically plausible and supported by numerous studies demonstrating a strong association with sepsis mortality. On the other hand, the evidence supporting the association of early intravenous fluid resuscitation with mortality in patients with sepsis is much more equivocal. Elucidating the precise role of intravenous fluids is critical in light of the high burden of sepsis as well as recent policy initiatives, such as the Centers for Medicare & Medicaid Services’ Early Management Bundle, Severe Sepsis/Septic Shock (SEP-1) measure, that mandate early aggressive intravenous fluid resuscitation.

The concept of intravenous fluid therapy for sepsis is based in part on the pathophysiological characteristics of sepsis that induce changes in vascular permeability and decrease fluid retention in the vascular compartment.¹ Much of the clinical evidence supporting the use of intravenous fluid therapy in sepsis stems from the original early goal-directed therapy trial, in which early liberal intravenous fluid therapy and optimization of specific hemodynamic targets led to reduced mortality in patients with septic shock.² This study, and the Surviving Sepsis Campaign that followed, catalyzed the implementation of sepsis treatment bundles in hospitals around the world that led to administration of larger volumes of intravenous fluids during the ensuing decade.³ Concurrently, sepsis-related mortality declined, but it is difficult to attribute this decline to greater intravenous fluid resuscitation, because numerous diagnostic, treatment, and organ support factors have also been changed in the last decade.

The contribution of intravenous fluid resuscitation is difficult to disentangle from the other interventions contained within sepsis treatment bundles, particularly early initiation of antibiotics. A recent Bayesian analysis of 37 studies that included approximately 20 000 patients suggested that the mortality benefit in the early goal-directed therapy trials was explained solely by earlier administration of appropriate antibiotics, rather than intravenous fluids or any of the protocol’s hemodynamic targets.⁴ Another recent analysis of nearly 50 000 patients with sepsis who were treated with mandated care in New York state hospitals found a strong association in mortality with delays in antibiotic therapy but not completion of the initial 30-mL/kg fluid bolus.⁵ A multicenter analysis of the effect of compliance with SEP-1 suggested that failing on the 3-hour antibiotic measure was associated with higher mortality, while failing the treatment bundle on any other component (including the 30-mL/kg fluid requirement) was not.⁶

Furthermore, the liberal intravenous fluid paradigm has recently been challenged by observational studies and randomized clinical trials suggesting poorer outcomes associated with a positive fluid balance in patients with sepsis.^7,8 In addition to endothelial leak, vasoplegia is a central feature of the pathophysiological characteristics of sepsis and may predominate in patients who are not responsive to intravenous fluids.¹ Furthermore, preexisting or sepsis-induced cardiac and lung dysfunction is common, and tissue edema from volume overload can impair blood flow and drainage and contribute to organ dysfunction.

In light of these uncertainties around the role of early intravenous fluid resuscitation in sepsis, the study by Lane and colleagues⁹ is a welcome contribution. The investigators examined the association between intravenous fluids administered by paramedics and mortality in 1871 patients with sepsis who were transported by emergency medical services to hospitals in Alberta, Canada. Sepsis was defined by the combination of infectious diagnoses in the emergency department, admission to the hospital (or death within the emergency department), and organ dysfunction as evidenced by hypoxemia, hypotension, abnormal Glasgow Coma Scale score, or a sepsis-related organ dysfunction diagnosis or procedure code. The overall hospital mortality of the sepsis cohort was 28.2%.

More than half of the patients with sepsis received intravenous fluids from paramedics, at a median volume of 400 mL. The investigators controlled for multiple potential confounders, including baseline characteristics, initial physiological measures, documented symptoms, physical examination findings, paramedics’ suspicion of sepsis in the patient, total prehospital time, and transportation priority, and found that the association of intravenous fluids with mortality depended on the patients’ initial systolic blood pressure. At the median initial systolic blood pressure of 125 mm Hg, intravenous fluids had no association with mortality or may have been harmful. However, for a typical patient with an initial systolic blood pressure of 100 mm Hg, administration of intravenous fluids was associated with significantly reduced mortality (odds ratio, 0.73). A propensity-matched analysis yielded similar results, lending support to the primary analysis.

The finding that intravenous fluids are potentially helpful only in patients with sepsis who initially have hypotension is important in light of the fact that sepsis is a very heterogeneous syndrome. Patients exhibiting other types of sepsis-induced organ dysfunction, such as altered mental status, hypoxemia, acute kidney injury, transaminitis, or coagulopathy, are likely to have differing pathophysiological characteristics than those with hypotension or shock. This study supports the intuitive notion that intravenous fluid treatment strategies should differ according to the sepsis phenotype.

One important limitation of the study by Lane and colleagues,⁹ however, is the potential for unmeasured confounding in the association between early administration of intravenous fluids in patients with hypotension and improved mortality. In particular, patients with hypotension in this study had a shorter time to physician assessment in the emergency department, suggesting that illness severity was a major determinant of rapid treatment. Because antibiotics were not given by paramedics, this time to physician assessment almost certainly affected the time to receipt of antibiotics; as with other observational studies, this finding raises the question whether the intravenous fluids or timely antibiotics were the primary drivers of the improvement in survival outcomes. Another limitation is the possibility of misclassification from the use of diagnosis codes for infection and organ dysfunction to identify the cohort with sepsis, which could have led to the enrollment of patients without sepsis.

In addition, while the study suggests that some intravenous fluids are better than none in patients with sepsis who have initial hypotension, it does not speak to perhaps the most pressing question: what is the right amount of intravenous fluids? We cannot draw conclusions about the appropriateness of the 30-mL/kg fluid strategy recommended by the Surviving Sepsis Campaign and mandated by SEP-1, because the paramedics in the study administered only a median of 400 mL of intravenous fluids. The study does, however, indirectly support the focus on the intravenous fluid resuscitation bundle for patients with sepsis who have initial hypotension. Future investigations will hopefully shed further light on the precise balance between intravenous fluids and vasopressors; a randomized clinical trial of liberal vs restrictive intravenous fluids for septic shock is currently under way.¹⁰ We are now finally evolving from the simplistic paradigm of liberal intravenous fluids for all patients with sepsis toward a more nuanced understanding of optimal targets and strategies for early intravenous fluid resuscitation.

Published: December 14, 2018. doi:10.1001/jamanetworkopen.2018.5844

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Rhee C et al. JAMA Network Open.

Corresponding Author: Andre C. Kalil, MD, MPH, Division of Infectious Diseases, Department of Medicine, University of Nebraska Medical Center, 985400 Nebraska Medical Center, Omaha, NE 68198 (akalil@unmc.edu).

Conflict of Interest Disclosures: None reported.

Funding/Support: Dr Rhee was supported by grant K08HS025008 from the Agency for Healthcare Research and Quality.

Role of the Sponsor: The funding source had no role in preparation, review, or approval of the manuscript or decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.