ICU Admission Control: An Empirical Study of Capacity Allocation and its Implication on Patient Outcomes

This work examines admission control of the hospital Intensive Care Unit (ICU) which provides care for the most critically ill patients. We focus on how congestion can impact ICU admission decisions and ultimately patient outcomes and hospital costs. We first develop a stylized model for ICU admission that captures key trade-offs in allocating beds to patients with heterogeneous medical needs and stochastic arrival patterns. The insights gained from this model are then used to build an econometric framework to explain how ICU admission decisions are made in practice. In addition, we show that this framework provides valid instrumental variables for identifying the effect of endogenous ICU admission decisions on patient outcomes, where the endogeneity is caused by unobserved patient severity factors affecting both admission decisions and patient outcomes. We estimate these models using patient data from an integrated healthcare delivery system with over 200,000 hospitalizations. We show that busy ICUs are associated with lower chance of admission –a 55% decrease on average–which can further lead to significant health and financial implications. In turn, providing ICU care can improve patient outcomes substantially: hospital length of stay decreases by 1.3 days and the likelihood of readmissions drop by 4.3%. We estimate that if congestion did not create ICU access issues, this hospital network could save over $2.7 million per year. By factoring in the estimated benefit of ICU care determined by our econometric model into our analytic model, we are also able to compare the performance of probable admission policies to that of the optimal policy. We show that relatively simple admission policies that account for ICU congestion can have nearly optimal performance.