Information needed | Information obtained from each study design | Information gained by doing both | Effect of information obtained on RCT protocol | |
---|---|---|---|---|
 | Survey study | Cohort study |  |  |
Overall frequency of steroid use for fluid resistant septic shock | Wide variation in individual physician responses within and between centres | Overall steroid use varied from 35.3% to 71.4% between centres | Despite survey responses to the contrary, one centre had a high rate of empiric corticosteroid administration | Centres with a high rate of steroid administration and centres with ≥50% of respondent physicians stating they often or always administer steroids for septic shock were not included in the RCT |
Use of steroids in patients who received 60Â cm3/kg of fluid and were on two or more vasoactive agents | 95.6% of physicians surveyed stated they would administer steroids to such patients | 50.8% of such patients actually received steroids | The discrepancy between the survey and cohort study findings suggests that physicians believe there may be benefit to corticosteroids but do not always administer them | This discrepancy provided support for the existence of community equipoise, which was needed as justification for the grant to fund the RCT |
Performance of adrenal testing | The majority of physicians stated that they sometimes, often or always perform adrenal testing prior to steroid use | Only 5.1% of patients had adrenal testing performed prior to steroid use | The discrepancy between what physicians say they do versus what they did regarding adrenal testing suggests that physicians may believe that adrenal testing should be performed but rarely do | We identified barriers to conducting adrenal testing including delays in obtaining results, difficulty in interpreting results and the cost of conducting the test. Due to these findings, adrenal testing was not required as part of the protocol |
Physician willingness to randomise patients | 84.3% would be willing to randomise patients on one high dose of vasoactive medication. However, 74.3% would start open-label steroids in patients requiring two high doses of vasoactive medication | It was not possible to determine willingness to randomise but we observed a lower rate of steroid use (50.8%) in the patient group for which 74.3% of respondents said they would administer steroids | The lower rate of actual steroid administration suggests that physicians might be open to randomisation and protocol adherence in the target population but that open-label steroid use would be a significant threat to the feasibility of the study | This finding emphasised the need for a pilot RCT with close monitoring of both the frequency and reported reasons for open-label steroid use |
Dose of hydrocortisone therapy | 65.2% reported using the equivalent of 1Â mg/kg/dose q6h and 26.1% used 2Â mg/kg/dose q6h | 56.3% of patients received 1Â mg/kg/dose q6h and 26.7% received 2Â mg/kg/dose q6h | The majority of physicians used 1Â mg/kg/dose of hydrocortisone with 2Â mg/kg/dose being a common second choice | The paediatric literature varies significantly regarding the dose of hydrocortisone for septic shock. However, given the consistency of the survey and cohort study data, we opted to use an initial bolus of 2Â mg/kg/dose followed by 1Â mg/kg/dose q6h |