Defensive behavior in healthcare: the role of organizational factors and perceived litigation risk
摘要
Defensive medicine is usually framed as a legal or insurance problem, but it is also an operational one: it wastes hospital capacity, creates bottlenecks, and misallocates resources. Given its psychological nature and operational consequences, we position this research within the Behavioral Operations Management literature, which mainly treats behavioral attitudes as antecedents of operations decisions. We distinguish our study by examining how operational, not only legal, conditions relate to defensive clinical decisions, an interaction between meso-level constraints (such as congestion) and macro-level legal threats for which quantitative evidence remains scarce. This paper experimentally investigates how hospital overcrowding and physician overtime interact with litigation risk to influence defensive overtreatment. In a within-subjects online experiment, 56 physicians responded to 504 clinical scenarios in which perceived litigation risk, overcrowding, and overtime were manipulated. Defensive overtreatment, operationalized as deviation from clinical guidelines, is the dependent variable; perceived litigation risk is the primary independent variable; and overcrowding and overtime act both as direct factors and as moderators of the litigation-risk effect. We estimate ordered logistic regression with physician-clustered standard errors. The findings reveal a significant interaction: overcrowding reduces overtreatment in isolation (consistent with a load-adaptive response) but increases defensive behavior when combined with high perceived litigation risk. Physicians’ personality traits (Neuroticism and Agreeableness) and lack of previous emergency experience are significantly correlated with overtreatment. These results suggest directions for healthcare operations: risk-stratified patient pathways may help decouple operational pressure from legal stressors, and the personality correlations point to a potential role for decision-support tools.