Statistical analysisAll values are

Statistical analysisAll values are presented as number with percentage for categorical variables and median withinterquartile range for continuous variables. Comparisons between continuous variables weremade using a t-test or Mann–Whitney U test, as appropriate. Categorical data were analysedusing the Chi-square test or Fisher’s exact test, as appropriate. Event-free survival curveswere estimated by the Kaplan-Meier method and compared using the log-rank test. Hazardrates were determined by survival analysis using the Cox proportional hazards model. Inmultivariable models, covariates included those with a p value < 0.2 and those that wereclinically relevant. We selected the number of variables in each multivariable model in orderto have approximately one variable for every ten patients so as to reduce overfitting. All testswere two-tailed, and p < 0.05 was considered statistically significant. SPSS version 20 (IBM,Armonk, New York, USA) was used for statistical analysis.

Statistical analysis
All values are presented as number with percentage for categorical variables and median with
interquartile range for continuous variables. Comparisons between continuous variables were
made using a t-test or Mann–Whitney U test, as appropriate. Categorical data were analysed
using the Chi-square test or Fisher’s exact test, as appropriate. Event-free survival curves
were estimated by the Kaplan-Meier method and compared using the log-rank test. Hazard
rates were determined by survival analysis using the Cox proportional hazards model. In
multivariable models, covariates included those with a p value < 0.2 and those that were
clinically relevant. We selected the number of variables in each multivariable model in order
to have approximately one variable for every ten patients so as to reduce overfitting. All tests
were two-tailed, and p < 0.05 was considered statistically significant. SPSS version 20 (IBM,
Armonk, New York, USA) was used for statistical analysis.