The RAMER REVIEWS: The ARISE FLUIDS Trial
Written By: Michelle Qu, DO; Edited By: Brian Smith, DO, MA, MMSc-Med
What is the study question?
In adult patients presenting to the emergency department with early septic shock, there are usually two approaches. Compared to more liberal fluid volume resuscitation and later vasopressors, does the approach of restricted fluid volume and early vasopressors increase the number of days alive and out of the hospital at day 90?
Surviving Sepsis Campaign recommends at least 30 mL/kg of IV fluid within the first 3 hours of sepsis-induced hypotension but this is based on limited evidence and therefore, is a weak recommendation.
What study design is used to answer the question?
It was an investigator-initiated, multicenter, open-label, parallel-group, pragmatic randomized controlled trial conducted at 51 sites across Australia, New Zealand, and Ireland
What is the exposure or intervention?
Two contrasting hemodynamic resuscitation strategies delivered for a minimum of 6 hours and up to 24 hours in a critical care area (ED or ICU). IV fluid resuscitation ceased after randomization and vasopressor therapy commenced immediately in the vasopressor group. Maintenance fluids for this group were discouraged. The fluids group had an initial bolus of up to 1,000 mL was infused within 60 minutes after randomization. Additional 500 mL boluses were given for persistent hypotension or hypoperfusion. Administration of 30 mL/kg within 3 hours of ED presentation was recommended unless contraindicated. Vasopressors were initiated only if the MAP target was not achieved after the clinician determined the patient was volume-replete or not fluid-responsive.
What is the outcome (or disease) and how is it measured (or defined)?
The primary outcome was no different between the two groups. The median number of days alive and out of the hospital at day 90 was 76 days in both the vasopressor group and the fluids group. All secondary outcomes appeared similar between the two groups. The 90 day mortality was 16.4% in the vasopressor group vs the 14.4% in fluids group. 28 day mortality was 12.9% in the vasopressor group vs 10% in the fluids group. The vasopressor group showed faster cessation of vasopressors and shorter ICU stay. The most notable safety finding was a significant difference in pulmonary edema: 0.6% in the vasopressor group vs. 5.0% in the fluids group. Rates of invasive ventilation (15.4% vs. 15.1%) and renal replacement therapy (6.2% vs. 5.2%) were similar between the two groups.
How were participants selected or recruited for the study?
Participants were recruited from 51 emergency departments across Australia, New Zealand, and Ireland, with enrollment occurring under a waiver of prior consent (opt-out or consent-to-continue model) given the emergency nature of septic shock
Who was included in the study?
Adults with clinically suspected infection were eligible if they met all of the following: systolic blood pressure <90 mm Hg or MAP <65 mm Hg despite receiving at least 1,000 mL of fluid in bolus doses (including prehospital fluids), lactate >2.0 mmol/L, IV antibiotics agent already initiated. These criteria align with the Sepsis-3 definition of septic shock (hypotension requiring vasopressors plus lactate >2 mmol/L), ensuring enrollment of a clinically severe and well-defined population
Who was excluded from the study?
The following groups of people were excluded from the study: If the person received >2,000 mL of IV fluids prior to randomization, if there was >6 hours elapsed since ED presentation. The cap of 2,000 mL of fluid prior to the study was critical to this study because it ensured that the patients that were enrolled in this study did not already meet the guideline recommended 30mL/kg. This helped the study generate evidence about initial fluid resuscitation volumes that are recommended with limited evidence. Other criteria for exclusion included barriers to the intervention, including care limitations, need for immediate surgery, or clinician-determined unsuitability.
If randomization occurred at what point did it occur? (only in experimental design)
Randomization occurred within 6 hours of presentation in the ED.
What do the authors say about power/sample size of the study? Is the power adequate to answer the study question?
The study initially enrolled 1000 patients into the study with a power of 90% based on the original ARISE study in 2014. 37 patients did not give consent, which resulted in a total of 963 patients. However, the trial was underpowered to detect clinically important differences in mortality the confidence intervals for 90-day mortality are wide enough to include both meaningful harm and benefit.
Are the study groups’ characteristics comparable at baseline?
The groups were comparable including demographics, illness severity, hemodynamics, infection sources, and prerandomization treatments.
What statistical tests/methods are used in the analysis of the results? Is the “intention to treat” principle maintained (if the study is a clinical trial)?
The intention to treat principle was maintained. The primary outcome (days alive and out of hospital at day 90) was analyzed using quantile mixed-effects regression with treatment group as a fixed effect and site as a random effect.
What are the results of the study?
The primary outcome showed no difference between the vasopressor and fluids strategies: the median days alive and out of hospital at day 90 was 76 days in both groups. Among patients receiving vasopressors, the vasopressor group had earlier cessation and earlier ICU discharge, though these are considered process outcomes influenced by group assignment. Adverse events were similar except for pulmonary edema occurred significantly more often in the fluids group (5.0% vs. 0.6%)
Are the results statistically significant?
The primary outcome was not statistically significant, the P value was 1.00, indicating no difference between the vasopressor and fluids groups in days alive and out of hospital at day 90. The value of 1.00 for the primary outcome is a definitive null result; the trial was adequately powered (90% power to detect a 4.5-day difference) and achieved intervention separation, yet found no benefit. It strongly suggests that the two resuscitation strategies produce equivalent clinical outcomes. The only outcome that reached statistical significance was the safety outcome of pulmonary edema: 0.6% in the vasopressor group vs. 5.0% in the fluids group.
Are the results clinically significant?
The trial shows clinical equivalence between the two resuscitation strategies. The only clinically significant finding — reduced pulmonary edema with restricted fluids — may inform practice by supporting a more cautious approach to fluid administration in patients at risk for fluid overload, even though it does not change the overall conclusion of equivalent efficacy
What are the biases/limitations of the study that could cause you to question the author’s conclusions?
The trial was unblinded, which could influence co-interventions, monitoring intensity, and discharge decisions. intervention was restricted to the first 24 hours and did not control for fluid or vasopressor management during the remainder of hospitalization. If clinicians in both groups ended up doing the same treatment after the initial 24 hours, the initial separation may have been insufficient to produce lasting outcome differences. The study population had a 90-day mortality of ~15%, meaning that the population was moderately ill. Compared to CLOVERS trial, these patients were slightly more sick but still not high mortality in general.
Which limitations or biases are unique to the study design (see below), and discuss how these potential problems are addressed by the authors.
The authors argued that post-intervention decisions occurred later under the care of clinicians who were "generally unaware of the treatment assignment," minimizing bias on the primary outcome. The authors chose an objective primary outcome which is less susceptible to subjective bias.
What is the overall internal validity of this article. Did the authors rule out chance, bias, and confounding as explanations for their findings?
The authors adequately addressed chance and confounding. The trial was powered at 90% to detect a 7-day difference in the primary outcome. However, it was underpowered for mortality. Confounding was well-controlled through randomization. Characteristics were well-balanced across all major variables such as demographics, illness severity (APACHE II, SOFA, lactate), hemodynamics, infection source, and prerandomization fluid volumes were nearly identical between groups. There was some biases that were not adequately addressed in this article such as performance bias. However, selection bias was mitigated using computer generated randomization.
What is the overall external validity or generalizability of this study. Would these results change your practice? How do these results fit into what we already know about this subject? (Understand that to achieve internal validity discussed above researchers often sacrifice external validity)
Given the findings of this study, I would still practice similarly. The patient’s fluid status pre-resuscitation will likely be the defining characteristic that helps me determine wheter to start fluids or vasopressors first for sepsis induced hypotension. Early resuscition with either method is not deemed more superior than the other, I would continue the similar resuscitation method I had before. I would trial 30mL/kg unless the patient is fluid overloaded to avoid pulmonary edema.
How does this fit our current knowledge? (refer to current practice, previous beliefs, and/or previous studies)
Very similar findings in the relevant studies of CLOVERS and CLASSICs. However, these patients are slightly more sick than CLOVERS study but definitely less sick than CLASSICS. This is also a study that is focused on the early resuscitation in managing septic shock.
Potential limitations/biases according to study design:
Cross-Sectional
No way to establish temporal relationship exposure and outcome measured at the same time.
Selection Bias
Self-Reporting Bias
Response Bias
Case-Control
Selection Bias (of cases and controls)
Recall Bias
Questionable Temporal Relationship (decrease exposure before outcome)
Cohort
Selection Bias
Loss to Follow up
Change in habits over time
Experimental Design
Selection Bias
Loss to Follow up
Improper or biased randomization procedures
Inadequate blinding of participants and investigators to exposure/treatment