Purpose: With higher detection and therefore increased prevalence of asymptomatic abdominal aortic aneurysms (AAA), an increasing number of vascular surgeons are making decisions about the timing of surgery. Overall peri-operative mortality is approximately 5% for elective open repair. The growth-related risk of rupture crosses this operative mortality risk when AAA reach 5.5cm. Until this size is reached, guidelines recommend that expansion be monitored by imaging every 3-6 months. Such “watchful waiting” may be anxiety-provoking. The “availability bias” may make recent, emotionally-vivid, anxiety-provoking events -- such as AAA rupture -- influence statistically-optimal choices. We investigated the role of induced anxiety in the decision to “operate” on an expanding AAA analogue among vascular surgeons.

Methods: In a randomized field experiment, vascular surgeons (n = 81) participated in a computer-based, decision-making task in which they chose when to stop inflating an expanding AAA-analogue. While performing the task, participants received updated information concerning size and bursting probabilities. These matched literature-based AAA sizes and burst probabilities, including a constant “opt-out” (i.e. operative) risk of 5%. Participants were randomly assigned to a pre-task demonstration in which the analogue either burst (anxiety induction condition) or did not (control condition). The number of expansions before “opting out” (choosing surgery) or a burst occurred (AAA rupture) were recorded. Additional information collected included pre-manipulation anxiety levels, risk-aversion, uncertainty intolerance, demographics, surgical experience, and game strategy. To account for censoring resulting from bursts, Kaplan-Meier curves and Cox-survival regression on “opting out” were conducted.

Results: On Kaplan-Meier curves, anxiety induction led to an average of 3 fewer expansions versus controls (12 vs. 15) at the median “survival” time. Cox-regression survival analyses demonstrate the anxiety induction increased the likelihood for “opting out” among surgeons at each decision point (HR = 3.32; p < 0.05), even after controlling for age, pre-manipulation anxiety levels, risk attitudes, uncertainty intolerance, and surgical experience. The only other significant predictor of opting out was attention to statistical risk (HR = 3.82; p < 0.05).

Conclusions: Induced anxiety accelerated surgeons' choice for “surgery” for an expanding AAA. Clinically, this corresponds to 9-18 fewer months of “watchful waiting”. There may be an underappreciated role played by emotions like anxiety in medical decision-making among experts, even when evidence-based, statistical guidelines are known.