Data analysis plan

The primary outcome and all secondary outcomes will be compared between the intervention and standard of care groups in an intention-to-treat population, with all tests two-sided and with the level of significance set at 5%, after the primary outcome has been collected in the last subject. There is no plan for any interim analysis. We will perform exploratory analyses in pre-specified subgroups of patients stratified according to APACHE II, and the length of intervention. We will also perform unadjusted analysis of odds ratio of being functionally independent (defined as ability to walk, use a telephone, self-care, use the toilet and groom) at 6 months after ICU admission in patients in the intervention and standard of care groups. We will perform adjustments on the disease severity (APACHE II score), admission diagnosis, baseline functional status and age. Missing data for primary outcome will be dealt by reporting both worst-case and per-protocol results; no imputation will be used.