|
|
|
Learning Objectives: New modeling method for assessing tobacco use behavior change at the population level
Abstract: Problems: How to assess tobacco behavior changes at population level presents a challenge.
Methods: We tested the Probabilistic Discrete Event System Model (PDES) in solving this problem. Data were derived from the 2000 National Household Survey on Drug Abuse and respondents 12 years of age and above were categorized into six mutually exclusive groups: (1) never-smokers (42.01%), (2) initiators (3.17%), (3) experimenters (15.19%), (4) self stoppers (31.92%), (5) addicted smokers (7.30%), and (6) quitters (0.41%). A PDES was established to model the behavior. Probabilities were estimated for each of the following seven events describing smoking progression (1) from never-smoker to initiator (event a); (2) from initiator to experimenter (event g), (3) from initiator to self-stopper (event b), (4) from experimenter to addicted smoker (event d), (5) from experimenter to self-stopper (event l), (6) from addicted smoker to quitter (event m); and (7) from quitter back to addicted smoker (event r) (Figure). The model was solved with a serious of constructed Markov Chains.
Results: The estimated annual probabilities were 7.6%for a , 84.5% for g, 15.5% for b, 85.1% for d, 14.9% for l, 5.5% for m and 95% for r. Simulation analysis demonstrated that 10% reduction in a leads to 5.7% increase in never-smokers and 4.85% decline in addicted smokers.
Conclusions: We concluded that the PDES can be used to estimate progression probabilities with cross-sectional data. 2) Results from such analysis are informative for tobacco control planning at the macro level.
See more of Data Collection and Analysis
See more of Evaluation and Surveillance
See more of The 2005 National Conference on Tobacco or Health (May 4-6, 2005)