Optimal Allocation of Intensive Care Unit Nurses to Patient-At-Risk-Team

Abstract

Medical studies report an association between critical care nursing and patient mortality in an Intensive Care Unit (ICU). Increasing the nurse-to-patient ratio in hospitals is not feasible, mainly due to an ever-growing nursing shortage and limited hospital budget. However, an alternative approach could be reducing the ICU demands through preventing unnecessary and unplanned admissions from wards. This is the primary goal of a nurse-led Patient-AT-Risk-Team (PART) capable of identifying deteriorating patients in wards who could be provided with the required services outside of ICU. Specifically, PART is a nurse-led outreach team with the following aims: First, to avert ICU admission by identifying patients who are deteriorating; second, to safely discharge from ICU by following up patients discharged to the ward; and, finally, to prevent admissions of patients who are either not in critical condition despite showing signs or can be treated in the ward. Constructing such a costly team though is often justifiable when it is staffed with the same ICU nurses, but, lowering the level of ICU nursing could lead to worse patient outcomes such as mortality. Thus, we aim to investigate what nurse allocation policy between PART and ICU would result in best possible outcomes for both patients and hospitals. This thesis is comprised of two distinct but supplementary research papers. Paper I provides econometric models to estimate the impact of the occupancy level of critical care nurses in both PART and ICU on patient hospital length of stay. Paper II proposes queueing and simulation models to obtain the optimal nurse allocation policy between PART and ICU aimed at minimising the ICU mortality rate. The proposed models were validated at Middlemore Hospital using a 12-month history (1 July 2015 to 30 June 2016) of 8,576 visits of the PART to 2,662 patients. The findings of paper I indicate that when the ICU nurses are busy, length of stay increases by 3% (nearly 125 patient-days), and the high utilisation rate of the PART nurses increases the length of stay by 2% (almost 70 patient-days). The results would seem to suggest roughly $300,000 annual saving by allocating a new nurse to PART per shift. Simulation results in paper II suggest that establishing PART at hospitals even with one ICU nurse might reduce ICU deaths significantly (almost 35%). Lastly, the evidence from the "what-if" analysis implies that, in an ICU with 18 beds, the configuration of 3 PART and 15 ICU nurses minimises the ICU mortality rate.