PS3-07: A Microsimulation Model of the Natural History of Smoking in the US

Abstract

Background: Smoking and other forms of tobacco use is the largest preventable cause of morbidity and mortality in the US. An understanding of the natural history of this behavior can help in improving smoking cessation efforts. Aim: To develop a microsimulation model to depict the natural history of smoking in the US using data from the National Health Interview Survey (NHIS) and a structured literature review. Methods: Data from NHIS and the literature were used to model smoking prevalence by year of age over the lifetime of a birth cohort of US teenagers. The model accounted for age at initiation, quit attempts by age, success of quit attempts, and relapse. Smoking states in the model were “never smoker,” “current smoker,” “recent quitter,” and “former smoker.” Demographic variables such as age, gender and race/ethnicity were also analyzed to understand variations in smoking behavior by individual characteristics. Utilization of clinical interventions to promote cessation was built-in to the model in order to create two scenarios for use in evaluating the benefit of interventions to reduce smoking. In these two scenarios, lifetime smoking patterns in the absence and presence of clinical intervention were estimated. Use of these scenarios facilitated the comparison of smoking patterns with and without clinical intervention. The second scenario was also used to validate the model against observed prevalence rates and will facilitate future analyses of smoking patterns with selected community interventions incremental to current use of clinical interventions. Results: Model estimate of overall smoking prevalence of approximately 20% (for ages 18–65 years) was comparable to the available national estimates for the US. The model also successfully reproduced the smoking prevalence by age, gender and race/ethnicity, thus allowing for future analyses of various interventions at the clinic and at the community level. The model predicts a reduction of smoking prevalence by 4% at age 65, due to clinical interventions. Conclusions: Microsimulation modeling provides a useful framework for understanding smoking prevalence by individual variables such as age, gender and race/ethnicity. The current model will be expanded to include the impact of community interventions targeted at smoking reduction/cessation.