Background

The U.S. Food and Drug Administration (FDA) directly impacts public health by approving drugs and communicating regulatory information. However, traditional drug approval policies have heavily relied on quantitative analyses based on clinical trials and pharmacological data. While effective in verifying the safety and efficacy of new drugs, this approach has limitations in predicting and controlling complex public behaviors in real-world healthcare settings. Issues such as patients discontinuing medication due to concerns about side effects, or prescribed opioid analgesics entering abuse networks and causing social disasters, cannot be fully explained by simple pharmacological data. Therefore, there is a growing call to incorporate decision-making mechanisms into regulatory science to enhance the effectiveness of drug regulation. In 2017, the U.S. National Academy of Medicine (NAM) formally recommended the adoption of a multidisciplinary systems modeling approach for responding to national public health crises.

Key Findings

This paper, published in the Proceedings of the National Academy of Sciences (PNAS), systematically reports on the actual implementation of behavioral and decision sciences within the FDA's drug regulatory mission. Sara L. Eggers, former Chief of Decision Support and Analytics at the FDA, along with Tamar Krishnamurti, Professor of Medicine at the University of Pittsburgh, and Baruch Fischhoff, Professor at Carnegie Mellon University, present four key pillars through which behavioral science has improved the quality of the FDA's policy decisions.

The first is the 'Benefit-Risk Framework,' which has become a standard in the new drug approval process. This is a visual tool that helps reviewers consistently evaluate data derived from clinical trial data and potential risks in a coherent framework. The second is the 'Decision Support Service,' which provides real-time assistance for high-risk regulatory decisions. This service, composed of internal experts, provides analytical reports that incorporate behavioral science theories in complex drug regulation situations, thereby enhancing the objectivity of the regulations.

The third is the 'Patient-Focused Drug Development (PFDD)' initiative, which quantifies and incorporates patients' actual experiences and preferences. This initiative collects data on patients' pain levels and factors that reduce their quality of life through surveys, which are then included in the evaluation criteria, addressing aspects often overlooked in traditional clinical trials.

The fourth is the dynamic systems model 'FDA SOURCE,' created to simulate the opioid crisis in the United States. This model simulates the distribution of prescribed opioids, addiction rates, limitations of treatment facilities, relapse patterns, and overdose mortality rates in a computer environment (in silico) for the U.S. population aged 12 and over. It precisely models feedback structures, such as changes in patients' risk perception and social transmission effects, to help predict the impact of specific policies when implemented. Recognizing its outstanding scientific value, the model was awarded the 'Jay Wright Forrester Award,' the highest honor of the System Dynamics Society, in 2025.

Significance and Prospects

This research demonstrates that pharmaceutical regulatory science should expand beyond traditional analytical categories to incorporate the prediction of human psychology and behavior through the integration of social sciences. Even if a drug has excellent biological mechanisms, regulatory policies will be ineffective if they cannot predict users' uncertain behavior patterns. This multidisciplinary predictive simulation modeling is expected to serve as a benchmark for designing various public health policies related to public behavior, such as controlling opioids, responding to emerging infectious diseases, and increasing vaccine coverage.

However, to fully integrate qualitative indicators and simulation data from behavioral science into actual legal regulatory guidelines, further coordination with policymakers is required. In addition, to increase the reliability of simulation results, a monitoring system should be continuously operated to validate and update the real-time patient data and socio-structural indicators used in the model.