Statistical analysis

Analyses will be performed on an intention-to-treat basis, that all subjects will be analysed in the group which they are assigned to. Per-protocol analyses will also be performed for the primary endpoints. All statistical analyses will be performed with SAS V.9.2 software (SAS Institute, Cary, North Carolina, USA).

Numeric variables will be presented as mean (SD) or median (minimum, maximum; or IQR) and analysed with independent sample t test or Wilcoxon rank sum test. Categorical variables (including the incidence of 7 day delirium after surgery) will be presented as number of cases (percentage) and analysed with Χ² test, continuity correction Χ² test or Fisher exact test. Time-to-event variables (including the duration of 3-year survival after surgery) will be analysed with Kaplan-Meier survival analysis, with difference between groups assessed with log-rank test. A Cox proportional hazard model will be used to calculated hazard ratio (HR) and the 95% CI to estimate the between group difference. Subgroup analyses will be performed according to the locations and the final pathological diagnoses of cancers.

Two-tailed tests will be used in all statistical analyses, and p values of <0.05 will be considered to be of statistical significance (unless otherwise specified).