Predicting US and State-Level Cancer Counts for the Current Calendar Year: Part I – Evaluation of Temporal Projection Methods for Mortality

摘要: BACKGROUND: A study was undertaken to evaluate the temporal projection methods that are applied by American Cancer Society predict 4-year-ahead projections. METHODS: Cancer mortality data recorded in each year from 1969 through 2007 for United States overall and state National Center Health Statistics obtained. Based on 2000, 2001, 2002, 2003, Projections were made 4 years ahead estimate expected number of cancer deaths 2004, 2005, 2006, 2007, respectively, state, using 5 methods. These predictive estimates compared observed occurred all cancers combined 47 sites at national level, 21 level. RESULTS: Among models compared, joinpoint regression model with modified Bayesian information criterion selection produced closest actual deaths. Overall, results show has larger error than 3-year-ahead death counts when same method is used. However, new performed better current state-space method. CONCLUSIONS: The Joinpoint smallest considered state. This will be used project starting 2012. 2012; © 2012 Society.