OBJECTIVES:In January 2009,NICE introduced supplementary guidance for end-of-life(EoL) treatments.This guidance allows EoL treatments with an incremental cost-effectiveness ratio(ICER) over acceptable range to be recommended,given that 1)the treatment is indicated for a small patient population(dropped in 2016)2)indication has a short life expectancy(<2 years) 3)the treatment extends life by more than three months.There has been a tremendous increase in the proportion of technology appraisals(TAs) considering EoL guidance.However,in these TAs,EoL decisions were made without sufficiently taking into account the uncertainty surrounding the EoL criteria satisfaction.The aim of this study was 1)to systematically review 10 years of EoL supplementary guidance,and2)to explore how the EoL guidance implementation can be improved.METHODS:All TAs between January-2009&December-2018 were screened for EoL discussions.Data were extracted on the recommendation,EoL decision,company claims,and criteria estimates including the methodology and associated uncertainty.In addition, the EoL decision uncertainty is incorporated in a stylized economic model for an oncology drug (i.e. in Monte-Carlo simulations,£50,000-per-QALY-gained EoL-threshold is used only in the iterations that satisfied the EOL criteria).RESULTS:35% of the TAs included an EoL discussion(n=123).57% of the EoL discussions led to a positive decision.In 5.7% of the positive EoL decisions,a negative recommendation was given,versus in 43.8% of the negative EoL decisions.54.9% of the EoL decisions was made,while at least one criterion was surrounded by considerable uncertainty.The EoL criteria were evaluated/reported in inconsistent and non-transparent ways(e.g. some TAs reported median overall survival(OS) from the trials,whereas some reported modelled mean,in almost 30% of the OS estimates,no specified method was presented).The stylized example demonstrated that incorporating EoL criteria satisfaction uncertainty can decrease the probability of being cost-effective considerably(i.e. from 60% to 40%) and might lead to a negative incremental net benefit for a technology considered cost effective under EoL cost-effectiveness threshold.CONCLUSIONS:New technical-guidance on EoL evidence generation/reporting should be published,and the uncertainty on EoL criteria satisfaction should be included in economic evaluations.