Enhancing the Impact of Lung Cancer Screening: Assessment of the Performance of Joint Smoking Cessation and Screening Interventions and Personalized Screening Scheduling Using Microsimulation Modeling

Abstract

Lung cancer is the deadliest cancer in the United States. Low-dose computed tomography for lung cancer screening has proven effective in reducing lung cancer mortality and is thus recommended for ever smokers with a considerable smoking history. This dissertation investigated two strategies to refine lung cancer screening (LCS) processes: smoking cessation intervention in the context of LCS and optimal screening schedules for LCS. I utilized a microsimulation modeling approach to quantify the benefits, harms, and costs of various strategies. I considered the screening eligibility criteria under the 2013 US Preventive Services Task Force guidelines: smokers between ages 55 and 80, smoked for at least 30 pack years and former smokers quit within 15 years. First, I extended the University of Michigan Lung Cancer Natural History and Screening (MichiganLung) model, an established microsimulation model, to compare the effects on mortality of a hypothetical one-time cessation intervention at the first annual screening vs. annual screening alone. I tested the sensitivity of results to different assumptions about screening uptake and cessation efficacy. Across all assumptions, adding a smoking cessation intervention to screening reduced lung cancer mortality and overall deaths compared to screening alone. Our results show that smoking cessation interventions would clearly enhance the net benefits of LCS programs. Second, in collaboration with colleagues from the National Cancer Institute Smoking Cessation at Lung Examination Consortium, we conducted a cost-effectiveness analysis for cessation interventions at the first screen plus annual screening using the MichiganLung model. We considered five cessation interventions, including pharmacotherapy only, or pharmacotherapy with web-based, telephone, individual, or group counseling. Cost-effective cessation strategies included pharmacotherapy with web-based, telephone, or individual counseling. All smoking cessation interventions delivered with LCS were likely to reduce lung cancer mortality and result in life-years gained at reasonable costs. The choice of cessation intervention for screening clinics should be guided by practical concerns such as staff training and availability. Third, although annual LCS is currently recommended, a less intensive schedule may be preferable for low-risk individuals. I utilized a risk-threshold method to determine optimal screening schedules based on individual lung cancer risk, past screening results and other risk factors. Using the MichiganLung model, I compared lung cancer outcomes from adaptive screening schedules to regular (non-adaptive) triennial, biennial, and annual screenings. Adaptive screening schedules had a better benefit-to-harm ratio and were more efficient than regular screening schedules. Individual lung cancer risk and preferences play an important role in the performance of LCS. These findings support the adoption of patient-centered decision-making processes and individualized LCS strategies. Finally, I evaluated the cost-effectiveness of adaptive and regular (non-adaptive) schedules for LCS using the MichiganLung model results. I identified 9 dominant strategies, with 8 being adaptive schedules while 1 being annual screening. Compared with no screening scenario, all strategies had a cost to QALY ratio under $50,000. Compared incrementally, seven out of the eight dominant adaptive schedules were cost-effective under the $100,000 willingness-to-pay threshold, whereas annual screening had an incremental cost to QALY ratio over $120,000. Hence, under a fixed budget healthcare system, adaptive schedules may provide better “value for the money.” Overall, this dissertation identified two strategies that could enhance the impact of LCS by maximizing the net benefits and cost-effectiveness. It furthermore demonstrates the potential for mathematical modeling to translate risk estimates and other epidemiological data into clinically meaningful recommendations.