Time-based measures of treatment effect : reassessment of ticagrelor and clopidogrel from the PLATO trial

OBJECTIVE: Treatment effects to binary endpoints using time-to-event data in randomised controlled trials are typically summarised by reporting HRs derived with Cox proportional hazard models. Alternative and complementary methods include summarising the between-treatment differences on the metric time scale, quantifying the effect as delay of the event (DoE). The aim of this study was to reassess data from the PLATO study expressing the effects as the time by which the main outcomes are delayed or hastened due to treatment. METHODS: PLATO was a randomised controlled double-blind multicentre study (n=18,624), conducted between 2006 and 2008, which demonstrated superiority of the antiplatelet treatment ticagrelor over clopidogrel in reducing risk of several cardiovascular events. In the present study, four of the main PLATO outcomes were reassessed by calculating the time by which an event may be delayed due to the treatment. RESULTS: The effects of ticagrelor, as compared with clopidogrel, consisted of a substantial delay of the evaluated outcomes, ranging from 83 to 98 days over 400-day follow-up. The Delay of Events Curves showed that the effects progressively increased over time, and the significant findings were concordant with those presented in the original PLATO study. CONCLUSIONS: This study confirmed evidence of a beneficial effect of ticagrelor over clopidogrel, and provided the magnitude of such effects in terms of delayed event time. Investigating time-to-event data with a percentile approach allows presenting treatment effects from randomised controlled studies as absolute measures of the time by which an event may be delayed due to the treatment. TRIAL REGISTRATION NUMBER: PLATO (www.clinicaltrials.gov; NCT00391872); Results