Fresh Trial

Find the original article here!

Fluid Response Evaluation in Sepsis Hypotension and Shock - CHEST (chestnet.org)

Summary:

The “Fresh Trial” is multicenter trial including 13 hospitals in the UK and the US. It asked if Fluid Responsiveness (FR) assessment improved outcomes when resuscitating patients with sepsis or septic shock. They followed a resuscitation algorithm based on FR for the intervention group and standard care for the control group. FR assessment was done with surface electrode bioreactance to measure SV and change in SV after a passive leg raising maneuver (PLR). If there was > 10% change in SV the patient was considered FR and got fluid. If FR was not present, pressors were added. This is a complex trial with a lot of validity issues, discussed below. Might be better for more advanced learners to dissect.

PICOTT:

Population: 5500 patients screened. 150 pts enrolled into 2 groups. 2:1 randomization to intervention group.

Intervention: Resuscitation based on FR assessment

Comparison: Standard care

Outcomes: Fluid given at 72hrs or ICU discharge, Secondary outcomes mechanical ventilation initiation, renal replacement therapy, worsening creatinine, mortality.

Type of Question: Diagnostic test used as a therapeutic intervention

Type of Study: RCT

Interpretation:

This multicenter trial reports benefit from using FR assessment for resuscitation. They showed a benefit in fluid given at 72 hrs. or ICU discharge. They showed a benefit in the secondary outcomes of RRT and MV initiation.

It is very hard to do a big RCT and unfortunately, in this case I cannot assess if the benefit is true or not. Benefits seem overstated. Let's look at the details!

They decided to randomize 2:1 to the intervention – if I had to speculate, likely because they thought the intervention was clearly more beneficial. Because the trial only analyzed 124 patient and there were only 41 on the control group, this has a high potential for random error. A few random events might give you a positive result. This is clearly seen by the fact that gender distribution was completely imbalanced. The only other potential reason is a problem with allocation concealment, however, one of the authors reported to me that it was done well and as reported.

The FR assessment was done with surface electrode bioreactance, which in the literature has a percentage error of approximately 47% and a standard deviation of -2/+2 L/min in Cardiac Output.

The Primary outcome is a composite outcome. It has a similar physiopathology, but different importance in the components and different frequency. The result is positive only because of the difference in the fluid given (not important), but not in the ICU discharge (more relevant). Additionally, we could argue that fluid given is a result of the intervention being done, and not an outcome. If only 50% of the patients are FR, only 50% of the patients will receive fluid. Self-fulfilling outcome. 

There was no blinding, which would be impossible to do in the clinicians and patients, as they need to follow different algorithms. This leads to the risk of performance bias, in which assessments or co-interventions are biased because we know which group the patient belongs to.

To their credit, they did Intention to treat, but they reported modified intention to treat, which will over-estimate benefit.

The really important outcomes, which are mechanical ventilation (MV), renal replacement therapy (RRT) and mortality were secondary, and they did not enroll the 210 patients they needed. They stopped the trial early after they concluded they had a positive primary outcome at 120 patients. The renal replacement therapy in particular was positive comparing 4 to 7 events only. This is very susceptible to random chance alone and that result would have a very high fragility index. Of note, decisions to initiate RRT or MV have a certain degree of subjectivity and could have been affected unconsciously by lack of blinding.

Regarding the external validity, I am concerned about randomizing only 2.7% of patients enrolled. This will affect generalizability. The primary outcomes were not outcomes that patients will care about or will affect them.

Overall Risk of bias: High

Context: 

Fluid responsiveness has been gaining enthusiasm as a consistent way to assess and treat patients. It would be widely and frequently applicable to patients with sepsis, septic shock, but also to shock in general. Shock remains one of the most common ICU diagnoses. There is a lack of evidence and multiple tools to assess it. Huge knowledge gap.

Teaching points:

·         Composite outcomes.

·         Diagnostic Interventions (Therapeutic intervention).

·         2:1 randomization.

·         Blinding,

·         Intention to treat.

·         Stopping trials early.

Verdict:

Not Settled - Very likely to change in the future