Introduction to the studies of cohort in epidemiología and to the analysis of survival

The cohort design is a major methodological tool in epidemiologicaland medical research. This article introducesthe basic concepts and statistical methods supporting thisdesign. After describing the fundamental ideas of prospectiveand retrospective cohort designs and of fi xed and variablecohort designs, this article focuses on the basic tools used invariable cohort designs. These tools are part of the so calledsurvival analysis. The concepts of censoring, life table, survivalfunction, the Kaplan-Meier estimator and the cumulativehazard function are then described. Statistical methods forcomparing two survival functions such as the Cochran-Mantel-Haenszel and log-rank tests and the log-rank stratifi edtest are also explained. Finally, the alternative person-yearsmethod is introduced. The concepts and methods are illustratedwith a retrospective cohort study of the age of onsetof daily smoking in patients with schizophrenia