Abstract
<jats:p>Drug shortages represent persistent supply disruptions in the U.S. pharmaceutical market, threatening patient access and increasing drug costs. Prior research commonly treats shortages as binary events and relies on static designs, limiting insight into how shortage characteristics drive cost escalation. This study uncovers the heterogeneity behind drug shortages and pharmacy acquisition costs of generic non-injectable drugs. FDA drug shortage records with weekly National Average Drug Acquisition Cost (NADAC) prices were fit with fixed-effects models, duration-specific models, and a double machine-learning framework to characterize heterogeneity in shortage-price associations by duration, market structure, and shortage reasons. In the baseline two-way fixed-effects model, active shortage designation alone was not associated with a significant increase in NADAC under clustered standard errors. In duration-specific models, shortages lasting more than four consecutive weeks were associated with approximately 7% higher NADAC, while each additional cumulative shortage week was associated with approximately 0.37% higher NADAC. Estimated CATEs varied widely across drugs in each week. Allocation restrictions, raw material and distribution disruptions, together with a lack of manufacturers, characterized shortages with higher estimated CATEs. These findings support monitoring both shortage persistence and supply-chain mechanisms to mitigate impacts on healthcare systems.</jats:p>