OBJECTIVES: Within acute coronary syndromes (ACS), the risk of experiencing fatal and non-fatal cardiac events is highest immediately after diagnosis and decreases over time. Visual inspection of survival curves from pivotal ACS trials suggest three potential risk periods. The highest risk (unstable disease) period typically lasts up to 10 days from diagnosis. Patients then become more stable but are still at a high risk of events until approximately 30 days from diagnosis. Beyond 30 days patients are considered stable and at a lower risk of events. Different agents may be best suitable for different periods and may affect different events. The objective of this research is to estimate a model which enables the effects on fatal and non-fatal events following an ACS episode to be disentangled and that distinguishes between periods of disease without accurately knowing how long these periods are. METHODS: A Markov model is estimated which distinguishes between three time periods and between fatal and non-fatal events. A likelihood function is derived as well as a Bayesian procedure to estimate the model parameters. The approach is tested using simulated data. Subsequently, event free survival data and overall survival data comparing ticagrelor with clopidogrel are taken from the Kaplan-Meier curves presented in the publication of the PLATO trial and model parameters are estimated based on these data.  RESULTS: Using simulated data the model mimics the data generating process perfectly and the approach seems quite powerful in distinguishing periods and differences with patient numbers of 500 and above. CONCLUSIONS: When applied to the PLATO study we conclude from the model that  ticagrelor lowers the probability to experience an event in the unstable and stable high risk disease periods.