| dc.description.abstract | Hospital congestion is a pervasive problem that causes care delays, frustration for patients and their families, and stressed staff. This could potentially reduce the quality of care and ruin a hospital’s reputation. Hospital congestion also affects new patient admission, mainly through the emergency department (ED). ED overcrowding, which has been a challenge for years, is primarily caused by patients waiting in ED for being admitted to the inpatient unit (IU). The main reason for the delay in ED patient transfer to IU is inpatient bed unavailability, which can also contribute to canceling elective surgeries and rejecting patient admission to intensive care units (ICU). The situation worsens with a pandemic virus outbreak, which boosts demand for ED and ICU beds. Thus, improving access to IU beds helps smooth the patient flow not only from the ED but also from other upstream units such as ICU and post-anesthesia care unit. One efficient way to release IU beds is to improve the discharge process and minimize non-medical inpatient days. This dissertation studies improving hospital discharge in both operational and strategic levels. Discharge planning on the day of discharge is necessary to ensure effective performance. Discharge delay reduces patient satisfaction and increases hospital congestion and length of stay. Patient satisfaction is impacted by adherence to patient preferred discharge time. Preferences arise from many factors including waiting for family, avoiding rush hours, or waiting to feel better. Flow congestion manifests in patient boarding, and length of stay is extended if discharge delay incurs extra overnight stay. These factors are often in conflict, thus, good hospital performance can only be achieved through careful balancing. In the first part of this dissertation, discharge planning problem is formulated as a two-stage stochastic program with uncertainty in discharge processing and bed request times. The objective minimizes a combination of discharge lateness, patient boarding, and deviation from preferred discharge times. Patient boarding is integrated by aligning bed requests with bed releases. The model is solved for different instances generated using data from a large hospital in Texas. Stochastic decomposition is compared with the deterministic equivalent and the L-shaped algorithm. A shortest expected processing time heuristic is also investigated. Computational experiments indicate that stochastic decomposition outperforms the L-shaped algorithm and the heuristic, with a significantly shorter computational time and small deviation from optimal. The L-shaped method solves only small problems within the allotted time budget. Simulation experiments demonstrate that the developed modeling approach improves discharge lateness and patient boarding compared to current practice. In addition to patients being discharged to home, some wait for a transfer to the next level of care. These patients may experience several days of non-medical stay in IU until the hospital finds a post-acute care facility that fits their needs. The second part of this dissertation studies the feasibility of creating a “post-discharge-unit” (PDU) for patients, who are medically ready for discharge but are being delayed for some reason, to improve access to valuable IU beds. We use a multistage stochastic program to address PDU capacity planning and cost-effectiveness issues. The random variable is the number of bed requests from upstream units, including the ED, ICU, direct admissions, etc. Our model takes the impact of PDU on upstream patient flow, e.g., ED congestion and hospital admission into account. We use the stochastic dual dynamic programming algorithm to solve the model. An extensive numerical analysis is carried out using the data from a large hospital in Texas. An analysis of the impact of a variety of parameters, including PDU’s fixed and operational costs, and length of ALC stays, on PDU capacity and cost savings is performed. The results show that a PDU is cost-efficient and improves access to IU beds significantly, even when the ALC population is small, which is counter-intuitive. Another important finding is that PDU size in hospitals with a larger ALC population is more sensitive to increasing the PDU fixed and operational costs. In other words, the PDU size decreases faster when ALC population is larger. | en |
