Heuristic Thinking in Patient Care

This paper studies heuristic thinking and cognitive bias using a natural experiment from the field. The setting for the study is a set of acute care hospitals, where we examine the care process and discharge decisions for individual patients. Determining a patient’s suitability for discharge is cognitively taxing, calling for the decision maker to draw on up-to-date clinical expertise and detailed information. We postulate that bounded rationality in decision making can lead the care provider to substitute clinical readiness for discharge, a more cognitively complex attribute, with the observed days spent in the hospital, a more easily accessible heuristic. Identifying the use of the heuristic is challenging, as patient readiness for discharge is often correlated with the time spent in the hospital. Our identification strategy, motivated by regression discontinuity design, exploits a discontinuity associated with the admission process at midnight. On the basis of more than 177,000 patient discharges spanning eight years, we find support for the use of the heuristic: 19.2% (95% confidence interval (CI) [14.6%, 23.8%]) of patients incur an increased length of stay by a day on average as a result of the heuristic. We also find that the reliance on the heuristic is associated with overtreatment; postmidnight patients on average cost $512 (95% CI [$100, $923]) more but without any corresponding improvement in healthcare outcomes. A counterfactual analysis shows that eliminating the heuristic lowers the bed capacity needed to maintain the same patient throughput. For example, for a hospital unit in our study with a bed capacity of 106, eliminating the heuristic would allow the hospital to maintain the same throughput with 1.61 fewer beds on average. These findings have a number of implications for policy formulation and managerial decision making. This paper was accepted by Vishal Gaur, operations management.