Estimating survival in newly diagnosed cancer patients: use of computer simulations to evaluate performances of different approaches in a wide range of scenarios.

Many empirical studies have proven the usefulness of period analysis in providing more up-to-date estimates of cancer patient survival than cohort-based methods. The aim of this paper is to provide a non-empirical evaluation of several survival approaches over a comprehensive range of scenarios using computer simulations. The simulation model included the following input parameters: number of annual patients, length of survival calculation, number of years of diagnosis, prognosis of cancer, follow-up period, and two additional parameters for modeling assumption on incidence and survival trends. The current study also introduced alternative cohort- and period-based approaches in addition to more traditional methods. Simulations showed that the choice of an appropriate survival approach is strongly dependent on the given scenario: period analysis was effective only for a limited number of circumstances, while alternative approaches appeared to be suitable for more realistic situations, when the follow-up period is different from the incidence period. The results of simulations could be useful for a quick identification of the most appropriate approach when estimating up-to-date cancer survival rates.