Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Industrial Engineering


Mayorga, Maria E

Committee Member

Kurz , Mary Elizabeth

Committee Member

Taaffe , Kevin M

Committee Member

Cho , Byung R


This dissertation presents two applications of discrete-event simulation (DES) to represent clinical processes: (1) a model to quantify the risk of the maternal obese and diabetic intrauterine environment influence on progression to adult obesity and diabetes, and (2) a model to evaluate health and economic outcomes of different smoking cessation strategies. The first application considers the public health impact of the diabetic and obese intrauterine environment's effect on the prevalence of diabetes and obesity across subsequent generations. We first develop a preliminary DES model to investigate and characterize the epidemiology of diabetes during pregnancy and birth outcomes related to maternal obesity and diabetes. Using data from the San Antonio Heart Study (SAHS), the 1980 Census and the NCHS we are able to verify a simplified initial version of our model. Our methodology allows us to quantify the impact of maternal disparities between different racial/ethnic groups on future health disparities at the generational level and to estimate the extent to which intrauterine exposure to diabetes and obesity could be driving these health disparities. The populace of interest in this model is women of child-bearing age.
The preliminary model is next modified to accommodate data and assumptions representing the United States population. We use a mixed-methods approach, incorporating both statistical methods and discrete event simulation, to examine trends in weight-gain over time among white and black women of child-bearing age in the US from 1980 to 2008 using United States Census projections and National Health and Nutrition Examination Survey (NHANES) data. We use BMI as a measure of weight adjusted for height. We establish an underlying population representative of the population prior to the onset of the obesity epidemic. Assessing the rate of change in body mass index (BMI) of the population prior to the obesity epidemic allows us to make 'unadjusted' projections, assuming that subsequent generations carry the same risk as the initial cohort. Unadjusted projections are compared to actual trends in the US population. This comparison allows us to quantify the trends in weight-gain over time. This model is interesting as a first step in understanding the trans-generational impact of obesity during pregnancy at the population level.
The aim of the second application is to understand the impact of different pharmacologic interventions for smoking cessation in achieving long-term abstinence from cigarette smoking is an important health and economic issue. We design and develop a clinically-based DES model to provide predictive estimates of health and economic outcomes associated with different smoking cessation interventions. Interventions assessed included nicotine replacement therapy, oral medications (bupropion and varenicline), and abstinence without pharmacologic assistance. We utilized data from multiple sources to simulate patients' actions and associated responses to different interventions along with co-morbidities associated with smoking. Outcomes of interest included estimates of sustained abstinence from smoking, quality adjusted life years, cost of treatment, and additional health-related costs due to long-term effects of smoking (lung cancer, chronic obstructive pulmonary disease, stroke, coronary heart disease). Understanding the comparative effectiveness and intrinsic value of alternative smoking cessation strategies can improve clinical and patient decision-making and subsequent health and economic outcomes at the population level.
This dissertation contributes to the field of industrial engineering in healthcare. US population-level data structures are not always available in the desired format and there is not one method for managing the data. The key element is to be able to link the mathematical model with the available data. We illustrate various methods (i.e. bootstrap techniques, mixed-effects regression, application of probability distributions) for extracting information from different types of data (i.e. longitudinal data, cross-sectional data, incidence rates) to make population-level predictions. Methods used in cost-effectiveness evaluations (i.e. incremental cost-effectiveness ratio, bootstrap confidence intervals, cost-effectiveness plane) are applied to output measures obtained from the simulation to compare alternative smoking cessation strategies to deduce additional information. While the estimates resulting from the two models are topic-specific, many of the modules created for these studies are generic and can easily be transferred to other disease models. It is believed that these two models will aid decision makers in recognizing the impact that preventative-care initiatives will have, and to evaluate possible alternatives.