Confirmation Bias Affects Estimation of Blood Loss and Amniotic Fluid Volume: A Randomized Simulation-Based Trial

This study was aimed to determine if confirmation bias affects diagnoses in obstetrics, specifically estimation of blood loss and amniotic fluid volume.

We performed a randomized simulation-based trial. Participants went through the following three consecutive scenarios: (1) the first involved estimating the volume of blood (actually a blood-like substance) in a container at the simulation model's perineum. The actual volume was either 500 or 1,500 mL. Participants were told it was blood seen after a vaginal delivery. One group was told that the "patient" was normotensive, the other was told that the "patient" was hypotensive. (2) The second scenario involved estimation of amniotic fluid from an ultrasound picture of four quadrants, with one group told that the patient was normotensive and the other group told that the patient had chronic hypertension. (3) The third scenario was a "negative image" of the first (i.e., if they had been randomized to the 500 mL/normotensive in scenario one, then they would be presented with the 1,500 mL/hypotensive). They also filled a survey including demographics and tolerance of ambiguity and confirmation bias scales.

From April 2018 through May 2018, a convenience sample of 85 providers was recruited. Participants were more likely to overestimate blood loss when they were told that the patient was hypotensive (p = 0.024), in comparison to when they were told the patient had normal blood pressure. They were also less likely to estimate the amniotic fluid as normal when they were told that the patient was hypertensive (p = 0.032).

Confirmation bias affects estimates of blood loss and amniotic fluid.