Smoking and Mental Illness Co-morbidity: Implications for Mortality Outcomes and Treatment Interventions

Abstract

Smoking and mental illness are leading contributors to mortality and morbidity in the U.S. They are also significantly associated with each other, as people with mental illness smoke at much higher rates compared to the general population. While it is known that people with mental illness have shorter life expectancies than people without, the extent to which this is associated with their increased smoking had previously been unknown. In Chapter 2, I use survival analysis and the National Health Interview Surveys 1997-2009 mortality follow-up data to construct lifetables by smoking and mental health status. I find that at age 40, never smokers with SPD lose approximately 5.3 years of life expectancy compared to current smokers with SPD who lose 14.9. Thus smoking is a primary driver of differences in life expectancy by mental health status.

In Chapter 3, I use simulation methods to adjust for recall error in national surveys and produce revised estimates of lifetime prevalence of major depression (MD) in the U.S. I develop a simple compartmental model of MD calibrated to data from the National Survey on Drug Use and Health (NSDUH). I show that over 40% of adults with lifetime MD underreport their histories of depression. After adjusting for recall error, 28.7% and 16.0% of women and men have lifetime MD compared to 15.6% and 9.5% when relying on self-report.

In Chapter 4, I develop the first model to evaluate the joint impact of smoking and mental health in the U.S. I calibrate a system dynamics model of smoking and depression comorbidity to NSDUH data 2005-2015 and show that the smoking prevalence disparity by depressive status is projected to widen over time. From 2016 to 2050, women and men with MD are expected to become increasingly more likely to smoke compared to their never depressed counterparts. Adults with current MD represent 6.7% of the adult population, but more than 640,000 smoking-attributable deaths are projected to occur in this group. While the proportion of smoking-attributable deaths among adults with depression is projected to decrease with time, even by 2050 parity with those who have no history of MD would not be achieved.

In Chapter 5, I evaluate the health gains associated with smoking cessation interventions that target patients with depression. I simulate the effects of widespread access to cessation treatment and increased utilization of mental health services among smokers with depression. Under a highly optimistic scenario that assumes all patients with depression receive cessation medication from their mental health professionals in 2018, less than 31,400 premature deaths would be avoided by 2050. This represents only about 5% of the nearly 600,000 smoking-attributable deaths that are projected to occur among adults with depression during the same period. Increases in the level of mental health service utilization would offer some additional but marginal health gains. While cessation interventions in mental health settings would reduce the smoking disparity by depression status, the mortality benefits associated with doing so are modest and should be pursued alongside more aggressive tobacco control strategies.

This dissertation advances knowledge about the impact of smoking and mental illness comorbidities on population health. It furthermore demonstrates the potential for systems science approaches to inform the epidemiology of behavioral health conditions, assess changes to tobacco use disparities over time, and evaluate the long-term effects of interventions.