Fig. 1

Results of the univariate and multivariable analysis conducted within the training dataset. The black squares and solid horizontal lines represent the estimate and the associated 95% confidence interval of the proposed prognostic model within a univariate Cox model. The black circles and dashed horizontal lines represent the estimate and the associated 95% confidence interval of the proposed prognostic model within a multivariable penalised Cox model, following the combination of coefficients and bootstrap standard errors across ten imputed dataset using Rubin’s rules. If there are no multivariable results for a variable, then this was penalised to zero during the model building process and not included in the final model