Antibiotic Resistance: Socio-Economic Determinants and the Role of Information and Salience in Treatment Choice (ABRSEIST)

Antibiotics have contributed to a tremendous increase in human well-being, saving many millions of lives. However, antibiotics become obsolete the more they are used as selection pressure promotes the development of resistant bacteria. The World Health Organization has proclaimed antibiotic resistance as a major global threat to public health. At the outset of the project, 700,000 deaths per year were due to untreatable infections. To win the battle against antibiotic resistance, new policies affecting the supply and demand of existing and new drugs must be designed.

The project set out to identify and evaluate feasible and effective demand-side policy interventions targeting the relevant decision makers: physicians and patients. ABRSEIST made use of a broad econometric toolset to identify mechanisms linking antibiotic resistance and consumption, exploiting a unique combination of physician-patient-level antibiotic resistance, treatment, and socio-economic data. Using machine learning methods, causal inference, theory-driven structural econometric analysis, and a survey experiment in the field, it provided rigorous evidence on effective intervention designs. This research improved our understanding of how prescribing, resistance, and the effect of antibiotic use on resistance, are distributed in the general population, which has important implications for the design of targeted interventions. It then estimated a structural model of general practitioners' diagnostic and treatment decision-making under uncertainty, allowing counterfactual analysis of the effects of AI adoption on treatment outcomes. The large-scale and structural econometric analyses allowed flexible identification of physician heterogeneity and was complemented by a nation-wide survey experiment to identify behavioral biases in diagnosis and treatment decisions that interact with AI-based information provision. Then project has contributed to the economic toolset for analyzing the effects of AI in healthcare and highlighted ways to improve rational use of antibiotics with a focus on primary care.