Methods: We mailed a 4-page questionnaire to 1502 randomly selected U.S. pediatricians. The questionnaire presented a clinical vignette of a previously healthy 1-year-old boy with a fever of 39.1°. To assess anticipated regret, we had participants choose which scenario they would regret more: 1) treating the boy with antibiotics, with subsequent anaphylaxis and complete recovery, or 2) not giving antibiotics, with subsequent pneumococcal sepsis and complete recovery. To assess the availability heuristic, we asked if participants could recall a potentially preventable case of sepsis that resulted in a bad outcome, and/or a case of anaphylaxis caused by medication that resulted in a bad outcome. To measure fear of malpractice, we used Frank's 6-question “Malpractice Fear Scale;” we categorized malpractice fear scores into tertiles labeled “high,” “moderate” and “low.” We conducted chi-square tests to make bivariate comparisons. To determine independent predictors of having more anticipated regret from the sepsis scenario, we conducted multivariate logistic regression controlling for malpractice fear, recalling a bad case of sepsis and/or anaphylaxis, years since medical school graduation, gender, foreign medical school graduate, general pediatrician, and US region of residence.

Results: 653 pediatricians returned completed surveys (49.5% of eligible). The majority (59%) had more anticipated regret from the sepsis scenario, while 41% would regret the anaphylaxis scenario more. About half of participants (55%) recalled a bad case of pneumococcal sepsis and 16% recalled a bad case of anaphylaxis. Having more anticipated regret from the sepsis scenario was more common among participants who recalled a bad case of sepsis (63%) than who did not (54%, p=0.02). Similarly, having more anticipated regret from the anaphylaxis scenario was more common among participants who recalled a bad case of anaphylaxis (50%) than who did not (39%, p=0.046). In adjusted analysis, significant predictors of having more anticipated regret from the sepsis scenario included high malpractice fear (odds ratio[OR]: 1.9, 95% confidence intervals [CI]: 1.2, 2.9) and recall of a bad case of sepsis (OR 1.5; 95% CI: 1.1, 2.2), while recall of a bad case of anaphylaxis significantly lowered the odds (OR: 0.6; 95% CI: 0.4-0.9).

Conclusions: The availability heuristic and high malpractice fear independently predicted pediatricians' anticipated regret.