Is Knowing Half the Battle? Impacts of Risk Information from Obesity Screening

Wednesday, June 15, 2016: 10:55 AM
B26 (Stiteler Hall)

Author(s): Bryant Hyuncheol Kim; Suejin A Lee; Wilfredo Lim

Discussant: Laura Argys

Obesity has been recognized as a serious problem worldwide. It has been highlighted that people who are obese are more likely to develop cardio-cerebrovascular diseases, such as diabetes, hypertension, heart disease, etc. Many policies for controlling and preventing obesity are based on information interventions. These include policies such as mandatory posting of calories on menus (Bollinger et al., 2011), nutritional labeling on packaged foods (Mathios, 2000), and weight report cards in schools (Prina and Royer, 2014). Screening for obesity also takes advantage of the idea that information could affect individual behavior. Since 2011, the Centers for Medicare and Medicaid Services initiated a new obesity screening and counseling benefit in Medicare coverage. In addition, the US Preventive Services Task Force (2012) recommends screening for obesity in adults of all ages, and intensive behavioral interventions for patients with BMI of 30 or higher. However, the direct effects of screening for obesity are not well known.

This paper aims to examine how information on obesity classification from screening affects changes in behaviors and future health outcomes. This paper exploits the National Health Screening Program (NHSP) in Korea, which provides free general health screening every two years to the entire population aged 40 and over. In this program, individuals are not only informed of their exact value of BMI, but also provided with their obesity classification as “normal” if BMI is between 18.5-24.9 and “obese” if BMI is greater than or equal to 25. The NHSP uses lower BMI threshold than the World Health Organization (2000) standard of 30. Using the fact that the obesity classification varies discontinuously at BMI threshold 25, we apply a regression discontinuity design and assess individuals’ responses to obesity classification while controlling for unobservable factors correlated with both the attainment of information and the responses to that information.

We use the sample cohort database provided by the National Health Insurance Services for about 210,000 participants from 2009-2013. The sample cohort database is a 2% random sample of the administrative database, provided for policy and academic research purposes. We assess individuals’ behavior changes, including alcohol and tobacco use, exercise, medical activities, and follow-up examinations. In addition, this paper assesses the question of whether the changes in behaviors yield changes in future health outcomes in the subsequent rounds of screening. Future health is measured by BMI and other measures such as systolic blood pressure, total cholesterol, and blood sugar level, which are important factors to manage in order to lower the risk of complications of obesity.

We find no evidence that the “normal” and “obese” groups show different changes in behaviors, such as drinking, smoking, exercise, and medical activities. However, we find that the probability of taking follow-up examinations is higher for the “obese" group than the “normal” group. We also find a small but statistically significant decrease in the future blood sugar level for the individuals in the “obese” group in both two years and four years after the baseline screening. However, we find no impact on other health measures.