Scared to Death: Information Avoidance and Diagnostic Testing

Americans' use of preventive care is only half the recommended level. In fact, greater utilization of preventive care has become a national health policy objective. Basic economic theory suggests that a reduction in the price of preventive care will increase the amount demanded. Recent U.S. reforms embodied in the Affordable Care Act (ACA) reflect this insight by requiring that insurance companies impose no consumer cost-sharing on approved preventive services. However, previous economic studies suggest that price is not the only important factor that impacts the demand for preventive services. In addition, empirical evidence suggests that some people are health information avoidant, which means that they prefer not knowing information about their health even when the screening tests are free and very accurate.

To explain this puzzle, this study embodies insights from the economics theoretical literature to incorporate health anxiety, which represents the stress or disutility associated with the anticipation of positive (i.e., bad outcome) test results, as an additional potential cost of having a test in an individual's forward-looking, dynamic decisionmaking process. With data from the Health and Retirement Study (HRS) survey and its linked biomarker data, I evaluate the roles of many contributors to the observed type-2 diabetes screening behavior. The mechanisms nested in the model that explain why someone who is at risk for diabetes, for example, would not have a blood sugar test are monetary and time costs of doctor visits, the marginal effectiveness of medical and non-medical inputs for controlling blood sugar levels, an incorrect perception of health, life expectancy, and health anxiety.

To account for the simultaneity of many decisions, the dynamic selection into different health states, and unobserved heterogeneity that may be correlated across behaviors and outcomes, I jointly estimate an individual’s decisions (about doctor visits, exercise, smoking, and excessive drinking), stochastic outcomes (blood sugar tests and hospitalization), and health production functions (body mass accumulation, blood sugar evolution, death, and subjective health and longevity expectation transitions) derived from her optimization problem. Blood sugar testing is modeled as a stochastic outcome that occurs at a doctor visit in order to capture the role physicians play as agents for patients. Individuals base visit and health behaviors decisions on subjective beliefs about their health. Individuals' pessimism levels serve as proxies for health anxiety based on the theory and after controlling for many other mechanisms in the model. The equations are correlated contemporaneously and over time through unobserved heterogeneity that is modeled as random effects with discrete distributions.

Estimation results suggest that monetary costs, time costs, health and longevity expectations, and health anxiety are all important contributors to an individual's diabetes screening behavior. Specifically, a health anxious individual is more likely to avoid a diabetes screening test during a doctor visit as well as through reducing the number of doctor visits. Individuals' health-related behaviors also respond to health information associated with the screening tests. Different policy experiments aiming to improve individuals' diabetes screening behaviors and subsequent health outcomes are also explored in the paper.