Risk set sampling in epidemiologic cohort studies

摘要: Recent work has extended the methods for analysis of nested case-control studies to accommodate a broad variety risk set sampling designs. These results have implications design sam- pled epidemiologic cohort studies. We describe model which is natural extension Cox proportional hazards and may be used es- timate parameters from sampled data. illustrate how these techniques solve three diverse prob- lems research. craft manufacturing firm in San Diego County, Cal- ifornia, by families group employees who had died esophagus cancer. Their claim was that occupational exposures carcinogenic agents were blame disease. In order objectively as- sess whether there truly an unusually high rate cancer among any partic- ular substances at related cancer, mortality study undertaken. Inves- tigators compiled basic information such as birth dates, sex, race dates employment 14,067 company records. Other data sources required determine vital status and, if dead, cause death. For study, number deaths various causes then com- pared those expected based on United States cause-specific rates. There 14 cases observed compared 12.27 ex- pected (standardized ratio (SMR) = O/E 1.14, 95% confidence interval 0.62-1.92; Garabrant, Held, Langholz Bernstein, 1988). Although this excess certainly not significantly different 1, generally lower