Contributions of a survey and retrospective cohort study to the planning of a randomised controlled trial of corticosteroids in the treatment of paediatric septic shock

Randomised controlled trials (RCTs) are challenging to conduct in a paediatric critical-care environment. Background work, including surveys and observational studies, is often used to determine disease estimates, sample sizes and design protocols when planning such RCTs. Our objective was to determine the necessity of performing a survey or a retrospective chart review or both when planning an RCT on corticosteroids in the treatment of paediatric septic shock.

We compared information on corticosteroid use for moderate to severe paediatric septic shock obtained from a survey of physician beliefs and stated practices with that obtained from a retrospective cohort study. The survey was conducted between February and March 2012 and the retrospective study included children from birth to 17 years of age admitted from January 2010 to June 2011. The survey and the retrospective study were conducted at four academic tertiary care centres in Canada.

Survey responses from 23 physicians and retrospective data from 81 septic shock patients were included. The survey identified time to discontinuation of vasoactive infusions as the most feasible and clinically important outcome for an RCT on corticosteroids for paediatric septic shock. The retrospective chart review provided means and standard deviations for the suggested primary outcome, from which we could estimate sample sizes and justify the minimal clinically important difference. The survey found that physicians believe that patients with severe septic shock were most likely to benefit from corticosteroid administration but the majority stated they would be unwilling to randomise such patients, suggesting a lack of individual physician equipoise. The combined information from the survey and retrospective study suggested that enrolment of patients with moderate septic shock would be more feasible but that strategies would still have to be implemented to prevent open-label corticosteroid use.

The survey provided valuable information on the choice of primary outcome, target population and physician equipoise. The retrospective study provided estimates of patient numbers, the minimal clinically important difference, evidence for community equipoise and physician practice patterns. Strong consideration should be given to performing both types of studies prior to conducting RCTs in paediatric critical-care environments.