Poster Paper:
What Passed Is Past? the Role of Recent Adverse Events in Physician Treatment Decisions
*Names in bold indicate Presenter
Methods: First, I explore how an obstetrician’s treatment choice is influenced by adverse outcomes of prior patients. I use econometric techniques on inpatient EHR data (2015-2018) from a large hospital to estimate the within-physician effect of experiencing an adverse outcome in a patient on the likelihood of switching delivery modes on the next patient. I then examine whether this switching behavior affects quality of care and patient outcomes, both in the short- and long-run. Second, by simulating the hypothetical behavior of a “rational” obstetrician with data from published meta-analyses, I explore whether observed obstetrician switching is consistent with a model of Bayesian physician learning.
Results: An obstetrician is more likely to switch delivery modes following an adverse patient outcome, even after accounting for patient heterogeneity and physician practice style. Immediately after an adverse event, a physician is more likely to: i) switch to the inappropriate delivery mode, ii) increase treatment intensity and iii) experience worse patient outcomes. Over time, however, frequent switching is correlated with significant improvements in the worst-performing physicians. I conclude with evidence from the simulation model that this behavior cannot be a purely Bayesian response, and may indicate the presence of bias in physician decision-making.
Conclusion: Policies should aim to strengthen objectivity in clinical decision-making to avoid biases in subjective judgment, thereby enhancing quality of care and patient outcomes.
