Translational Epidemiology: An Application of Lung Cancer Prevention Strategies in USA and Mexico

Abstract

Lung cancer is the number one cause of cancer-related deaths in the US and worldwide. This is an aggressive disease with few treatment options if discovered at a late stage, as it often is. Therefore, reducing the burden of lung cancer should focus on primary and secondary prevention strategies: tobacco control and lung cancer screening. However, to implement effective interventions, we need to be familiar with a variety of methods and engage with different perspectives and stakeholders. Through the lens of translational epidemiology, this dissertation presents three analyses applied to different contexts to address the lung cancer burden. First, we describe a study using mixed methods to design and evaluate a web-tool for lung cancer screening informed decision-making in Metro Detroit. Through participatory design workshops, we identify design features that may be salient for other developers of web-based decision aids for both lung cancer screening and other health interventions. Afterwards, we conduct a quasi-experimental study with before-after surveys to test a modified version of a web-based decision aid, shouldiscreen.com, in an African American community in Detroit. We find that using the decision aid moderately improves lung cancer screening knowledge and concordance of individual preference for screening with clinical guidelines. Use of the tool also decreases decisional conflict with regards to lung cancer screening. We contact these survey participants after six months over the phone and ask if they had taken steps to see a clinician and enquire about lung cancer screening. A subset of the survey takers also participate in a focus group to give more in-depth feedback about their experience using shouldiscreen.com, which we use to triangulate our findings in the participatory design workshops. This aim demonstrates various design and implementation challenges for web-based lung cancer screening decision aids, particularly when considering the needs of low-resource communities with limited access to the internet and individuals with low levels of literacy. Second, we present an age-period-cohort analysis of smoking patterns in Mexico using nationally representative cross-sectional data from 1987 to 2016. Specifically,we estimate trends in smoking prevalence, initiation and cessation by birth cohort and sex. The results show that while smoking prevalence and initiation have decreased,progress has slowed and even reversed for younger birth cohorts. Moreover, the analysis reveals that there has been a shift in smoking patterns from daily to non-daily smoking in recent years. This shift may have implications in how tobacco control policies would be implemented in future. Third, we develop a computational model of smoking prevalence for Mexico, which accounts for the current patterns of use uncovered in Aim 2, and explicitly tracks daily and non-daily smokers. The model suggests that if smoking initiation and cessation remain at current levels, there will be around 18 million male and 4 million female smokers in 2050. These projections serve as a basis for future assessments of ongoing and new tobacco control policies.