Shan Huang defends her PhD thesis at the Department of Economics

Candidate:

Shan Huang, Department of Economics, University of Copenhagen

Title:

Essays on the economics of health care provision

Supervisor:

Hannes Ullrich, Associate Professor, Department of Economics, University of Copenhagen

Assessment Committee:

Summary:

This thesis comprises four self-contained chapters. Each chapter builds on empirical evidence from microdata in order to address a novel research question from the field of health economics. The chapters explore topics related to the behavior of healthcare providers, the value of data, and the sustainability of health insurance systems. By collectively investigating these topics, this thesis aims to advance our understanding of efficient healthcare provision.

Chapter 1: 'Provider effects in antibiotic prescribing: Evidence from physician exits', investigates how the practice styles of primary care providers contribute to antibiotic consumption. Policymakers strive to reduce antibiotic consumption, a major contributor to bacterial resistance. However, lower prescribing levels may jeopardize patient health. We explore whether targeting physicians' prescribing styles can effectively reduce antibiotic use without compromising patient health. To differentiate provider effects from patient effects, we leverage physician exits from primary care in Denmark, where physicians and patients are exogenously separated due to clinic closures, physician moves, or retirements. Our findings reveal that practice styles account for 49% of antibiotic prescribing differences between providers, implying that standardizing practice styles could reduce differences by half. However, the effect magnitude is even greater (83%) when restricting to second-line antibiotic use. Moreover, higher prescribing intensities do not indicate greater efficiency at the margin but are instead associated with more treatment failures and prescriptions without diagnostic tests, and no reduction in the risk of patient hospitalization for infection. This study is coauthored with Hannes Ullrich.

Chapter 2: 'Assessing the value of data for prediction policies: The case of antibiotic prescribing', explores the benefits of linking separate administrative data to improve targeted health care provision in the context of antibiotic prescribing. We treat antibiotic prescribing as a prediction policy problem and employ a machine learning algorithm trained to identify bacterial urinary tract infections in primary care. We measure improvements in prediction quality and antibiotic targeting by evaluating the algorithm's predictions against physician decisions. We find that including simple demographic information leads to considerable improvements in prediction quality, measured by the area under the operating curve (AUC). Additionally, including patients' rich medical histories improves AUC only modestly but reduces antibiotic use substantially. Our findings highlight the importance of assessing data needs based on policy objectives and not solely on prediction quality. This chapter is coauthored with Michael Ribers and Hannes Ullrich.

Chapter 3: 'The role of physician altruism in the physician-industry relationship: Evidence from combining experimental and observational data', examines the link between altruism, a professional norm among health care providers, and physicians' ties to the pharmaceutical industry. We match data on altruistic preferences for 280 physicians in the US, obtained from a revealed preference economic experiment, with information on their receipt of monetary and in-kind transfers from pharmaceutical firms, along with claims data on their drug prescriptions. We find that physicians with less altruistic preferences receive industry transfers with a monetary value that is, on average, 112.55% higher than those with strong altruistic preferences, indicating selective targeting by pharmaceutical firms. Moreover, we observe that a positive association between industry transfers and higher overall drug costs or brand prescribing rates is primarily driven by physicians with less altruistic preferences. Our study highlights the role of professional norms in physicians' clinical behavior. This chapter is coauthored with Anirban Basu and Jing Li.

Chapter 4: 'The effect of a ban on gender-based pricing on risk selection in the German health insurance market', examines the financial sustainability of health insurance systems by studying the impact of the unisex mandate -- a regulatory change that banned gender-based pricing -- on risk segmentation between public and private health insurance in Germany. While public health insurance companies cannot set prices based on expected health expenses, private health insurance companies consider health risk, including gender, as pricing factors. The unisex mandate made private health insurance relatively more appealing to women and less appealing to men. Based on survey data from the German socio-economic panel, we provide empirical evidence that the unisex mandate substantially increased women's switching from public to private insurance. The effect varies across employment groups and aligns with different financial incentives. The changes induced by the unisex mandate worsen the private sector's risk pool and improve the public sector's risk pool, as women have on average higher health expenses than men. Our study highlights how regulatory changes can impact risk segmentation in the health insurance market. This chapter is coauthored with Martin Salm.

An electronic copy of the thesis can be requested here: lema@econ.ku.dk