Identifying the Bottleneck Unit : Impact of Congestion Spillover in Hospital Inpatient Unit Network

Because a hospital is an interconnected, interdependent network of care units, allocating resources — beds, nurses, and improvement initiatives — to one unit to reduce its congestion may have spillover effects on other units. As such, to identify the bottleneck unit and improve access to care, hospital administrators need to take into account potential spillover effects. We use patient data from a hospital collected over five years to empirically examine whether and how much congestion propagates through the network of inpatient units. Our estimation result suggests that the magnitude of the congestion spillover is substantial. For example, increasing one inpatient unit's utilization by 10 percentage points today can increase its neighboring inpatient unit's utilization by up to 2.2 percentage points tomorrow. Using counterfactual analyses, we estimate the effects of two potential interventions in the study hospital: 1) decreasing the bottleneck unit's utilization by 10 percentage points can free up 4.9 beds in the hospital and 2) adding a bed to the bottleneck unit can free up 2.5 beds in the hospital. These effects are more than twice in magnitude compared to what one can achieve by naively choosing which hospital unit to make changes to. Importantly, we find that the magnitude of both effects is correlated with the size and network centrality of the targeted unit, but not correlated with the average utilization of the targeted unit. In other words, our results suggest that unit size and network structure are key in identifying the bottleneck unit of the hospital inpatient network