Presenter: Alexander Slade (University of Illinois at Urbana-Champaign. Dept. of Agr. & Consumer Economics)
Discussant: Reagan BaughmanUniversity of New Hamphire
Diabetes is a growing problem in America, with about one in ten Americans suffering from the disease. Its treatment differs from the treatment of other chronic diseases since lifestyle factors such as diet and exercise play profound roles in its management. Importantly, diabetes can be substantially or completely reversed (i.e., put into remission) through health investment behaviors that stimulate weight loss through physical activity, and by following a low-carbohydrate diet with lower average plasma glucose concentration (G). Despite numerous interventions, the goal of effectively changing initial and long-term health behavior has been elusive. Here, using the RAND-enhanced Health and Retirement Study (HRS), I investigate the determinants of newly diagnosed diabetics’ decisions to initiate health investment behaviors such as exercising, quitting smoking, and drinking less. In addition, possible mediating factors such as risk preferences and subjective probability of longevity are assessed to clarify the link between a new diagnosis of diabetes and subsequent changes in health behavior.
The paper follows Grossman’s (1972) theory of health production in which individuals derive utility from health, face a depreciating health stock (that depreciates at a faster rate with diabetes), and can invest in behavior to maintain or improve their health stock. In this context, individuals maximize their well being with respect to their available resources, and people with a high G invest more in physical activity, quitting smoking, and reducing alcohol consumption.
This study uses the HRS, a longitudinal study of older Americans, and estimates the effects of changes in disease status, risk preferences, and demographic characteristics on body mass index (BMI), smoking, drinking, and participation in frequent physical activity. Changes in behavior across two waves of individual responses are compared by differencing responses. A cross-section regression model is used to estimate the effects on changes in BMI, and a multinomial logit is used to estimate changes in smoking, drinking, and frequent physical activity. To determine the probability of engaging in a health investment behavior conditional on not having engaged in that behavior up to that point, a hazard model is employed. For each time period, a logit model is fit to clarify the determinants of whether individuals engage in the health behavior.
The preliminary analysis indicates that a new diagnosis leads to significant changes in health behaviors in directions that support good health, including an increase in physical activity and reduction in BMI. Changes in health behaviors vary by race and education. Results of this study may extend our knowledge of how specific health shocks serve as information that affects behavior. Further, given that there are many individuals with diabetes who do not know it, these results underscore the importance of faster and more directed diagnosis and treatment for specific populations, which can lead to better long-term outcomes.
The 3rd Biennial Conference of the American Society of Health Economists took place at Cornell University.
Software © 2010 iHEA - International Health Economics Association