Duration of Fluid Boluses in Septic Shock: Fragile Results and Unanswered Questions
First, the decision made by the Institutional Ethics Committee to stop the trial at interim analysis is questionable. Stopping a trial early is ethically justified only if there is “very strong statistical evidence of a treatment difference with a very small p value” (2). Commonly used p values for early stopping are in the region of p value of less than 0.001 (Haybittle-Peto boundary) (2). Only one child at 6 hours and two at 24 hours would have needed a different outcome to derive a p value of greater than 0.05 (Fragility index of 1 at 6 hr and 2 at 24 hr) (3). Obtaining p values of less than 0.05 with low fragility index results are, in our opinion, not strong statistical evidence for this trial (1) to have been stopped, especially as mortality showed a conflicting trend. Low-powered studies with small sample sizes are often subject to the so called “winner’s curse,” resulting in erroneously inflated effect size (4).
Second, probability of death (Pediatric Index of Mortality [PIM]-2) was unusually high at 74%. However, data provided do not reflect a population with greater than 70% probability of death as per PIM-2 (mean base deficit = 8, mean PaO2 = 77 mm Hg, FIO2 and high-risk diagnosis not stated). In our practice, children with physiologic derangement severe enough for PIM-2 to indicate greater than 70% probability of death are more likely to be intubated and ventilated. Therefore, we question the use of “need for intubation and ventilation” as a primary outcome measure in children with septic shock. As described in recent guidelines, “mechanical ventilation can eliminate work of breathing, improve oxygenation and organ perfusion, all of which are typically compromised in the septic child” (5).
Third, the authors state “we used similar volumes in both groups (of up to 60 mL/kg) to adhere to the guidelines as well as to ensure similar volumes in both groups so that the results are comparable.” Although ensuring comparability is important, we question the administration of fluid boluses for comparability rather than clinical need. It is also puzzling that despite a median time of 36 hours to achieve shock reversal, very few patients received any additional fluid boluses in the first 6 hours (median = 0). Clarification about unusually high inotrope scores would also be useful.
The authors must be congratulated on being able to achieve an impressive time to administer first dose of antibiotics (mean = 15 min).
This trial (1) raises important questions that need to be studied further but, due to the reasons above, appears to be a missed opportunity to answer questions about the optimal duration of fluid boluses in children with septic shock.