Admission Control Biases in Hospital Unit Capacity Management: How Occupancy Information Hurdles and Decision Noise Impact Utilization

Providing patients with timely care from the appropriate unit involves both correct clinical evaluation of patient needs and making admission decisions to effectively manage a unit with limited capacity in the face of stochastic patient arrivals and lengths of stay. We study human decision behavior in the latter operations management task. Using behavioral models and controlled experiments in which physicians and MTurk workers manage a simulated hospital unit, we identify cognitive and environmental factors that drive systematic admission decision bias. We report on two main findings. First, seemingly innocuous “occupancy information hurdles” (e.g., having to type a password to view current occupancy) can cause a chain of events that leads physicians to maintain systematically lower unit utilization. Specifically, these hurdles cause physicians to make most admission decisions without checking the current unit occupancy. Then—between the times that they do check—physicians underestimate the number of available beds when occupancy increases from admissions are more salient than occupancy decreases from discharges. Second, decision-related random error or “noise” leads to higher- or lower-than-optimal utilization of hospital units in predictable patterns, depending on the system parameters. We provide evidence that these patterns are due to some settings providing more opportunity for physicians to mistakenly admit patients and other settings that provide more opportunity to mistakenly reject patients. These findings help identify when and why clinicians are likely to make inefficient decisions because of human cognitive limitations and suggest mitigation strategies to help hospital units improve their capacity management. This paper was accepted by Charles Corbett, operations management.