Individual patient data comparative analysis of hydroxyethyl starch 130/0.4 v 4% albumin for fluid resuscitation in critically ill patients: statistical analysis plan.

BACKGROUND: Recent randomised controlled trials have compared the effects of albumin and hydroxyethyl starch (HES) v crystalloids on patient-centred outcomes in critically ill patients. The Saline v Albumin Fluid Evaluation (SAFE) trial reported patient-centred outcomes at 28 days in 6933 patients assigned to fluid resuscitation with either 4% albumin or 0.9% saline; the Crystalloid v Hydroxyethyl Starch Trial (CHEST) reported patient-centred outcomes at 28 days in 6644 patients assigned to fluid resuscitation with either 6% HES (130/0.4) or 0.9% saline. As the two trials used a common reference fluid (0.9% saline) and had most trial methods and data collection points harmonised, a comparison of 4% albumin and 6% HES (130/0.4) on patient-centred outcomes at 28 days in critically ill patients using the individual patient data from the two trials is feasible.

OBJECTIVES: To prepublish the statistical analysis plan (SAP) of an individual patient data comparative analysis of the SAFE and CHEST studies.

METHODS: The statistical analyses are described in detail with the appropriate statistical analysis specified.

RESULTS: The planned statistical analyses are described in detail for the SAFE and CHEST individual patient data comparative analysis. We outline the proposed statistical analysis for the baseline characteristics, processes of care and the outcomes. Subgroups are defined with statistical comparisons between the fluid groups and in each subgroup explained.

CONCLUSION: We have developed a SAP for the SAFE/ CHEST individual patient data comparative analysis to increase the internal validity of the study and minimise bias.