Using Cometing Risk Model for Death and Multi-State Stochastic Process for the Behaviour of Betel-Quid Chewing and Smoking to Estimate the Potential Years of Life Lost Due to Betel Quid and to Estiamte the Impact of Qutiing Betel-Quid Chewing

Betel-quid chewing is a well-recognized risk factor for the incidence of oral premalignancy and oral cancer. Moreover, the higher hazard ratios of betel-quid chewing on non-oral cancers and other chronic diseases have been recovered in recent years. The elucidation of the effect, and maybe the dose-response relationship, of betel-quid chewing for competing causes of death is of important role for public health. However, the majority of studies assumed independent censorship for different death events. Such assumption may cause biased estimate of the variance of varying hazards to death, and lead to incorrect statistical inference. Therefore, it is necessary to develop a non-homogeneous competing risk stochastic model to capture the correlation between different hazards and the fact that hazards increasing with age. It has been demonstrated that the habits of betel quid chewing and smoking are highly correlated, for both their use and their quittance. To understand the process of starting to use betel-quid and smoking and the process to quit the habits can help the public health workers to plan more efficient education programs. Collectively, we propose a three-year project to analyze the community-based cohort with longitudinal follow-up. We aim to achieve the following objectives: Year 1: (2012/08 ~ 2013/07) (1) To analyzed the all causes survival by status of betel-quid chewing with a community-based cohort. (2) To develop a non-homogeneous competing risk model to evaluate the effect of betel-quid chewing for death from oral cancer, other sites of cancer, and major chronic diseases Year 2: (2013/08 ~ 2014/07) (1) To develop a multi-state process model to estimate the co-compliance of betel-quid and smoking, the incidence rate of betel quid chewing and smoking, and the quitting rate of both habits. (2) To use a random effect model to correct the measurement error for the starting age and quitting age of betel-quid chewing and smoking Year 3 (2014/08 ~ 2015/07) (1) To evaluate the efficacy of quitting program for betel-quid and smoking based on parameters estimated in previous two years. (2) To estimate the potential years of life lost due to betel-quid chewing