More and more potentially curative interventions become available in oncology. Thus, a growing proportion of cancer patients might indeed be cured of their disease. For reimbursement decisions, it is common practice to estimate long-term overall survival based on trial data and for instance parametric survival curves. These standard parametric methods, however, do not explicitly take into account that patients might be cured of the disease. A mixture cure model does that by assuming that a study population is a mixture of cured and uncured patients. We assess the pros and cons of implementing mixture cure models in a Markov framework, compared to other parametric approaches.