Gravar-mail: Adjusting for multiple prognostic factors in the analysis of randomised trials