Background: Effective health care requires a delivery system that is fully coordinated. To achieve this requires structured communication between physicians and nurses yet an increasingly recognized barrier to efficient care is the geographic dispersion of these primary team members. Utilization of systems engineering and simulation modeling can provide a data-driven alternative to evaluate these complex healthcare issues leading to improved patient care. The purpose of our analysis was to apply systems engineering techniques to geographically match patients with their Internal Medicine (IM) housestaff teams.
Methods: Using electronic health records (EHR) we obtained admission, bed assignment, and discharge data for IM patients for 2013. We developed a current state model using SIMIO simulation software that was then validated. Building on the current state model, we constructed our first experimental model to evaluate if the average patient volume could be accommodated by only three nursing units. Building on the first experimental model, a second experimental model was constructed with the inclusion of the following variables: 1) level of patient acuity and 2) bed type. A node algorithm was created to admit patients to the units with the lowest census, as well as randomly selecting teams to ensure patients were distributed equally. Methods and analyses utilized for the first experimental model were replicated for the second.
Results: The first experimental model resulted in an average team census of 11.6 (raw EHR data= 12.0) with an average of 71% utilization of patient load capacity for the unit (EHR derived est. = 72%) and 72% team utilization of patient load capacity. For the second experimental model, analysis revealed two types of patients – those requiring step-down and those requiring general beds. The 3 units used in our first experimental model had a total of 46 general and 35 step-down beds. The proportion of patients admitted to these two types of beds was derived from the EHR and determined to be 76.5% to 23.5%, respectively. The second experimental model was simulated with these parameters. The results showed a general bed utilization of 96% compared to the step-down utilization of 39%. This mismatch increased the average length of stay for general medicine patients to 171 hours as compared to 127 hours for step-down patients.
Conclusions: As exemplified by our analysis, the principles of systems engineering can provide us with a data-driven method with which to evaluate complex hospital processes, which may aid critical decision-making ultimately influencing the ability to provide equitable and quality care. For instance, based on 2013 admissions data and our first model, we were able to successfully establish the feasibility of admitting all housestaff patients to only 3 units instead of the current practice of 17 units. Additionally through the second experimental model we were able to highlight a resource and demand misalignment that exists on those 3 units.
To cite this abstract:Mishra V, Masters H, Phillips A, Storch R, Tu S. You Can’t Improve What You Don’t Measure: A Systems Engingeering Approach to Developing Geographically-Matched Patient Provider Teams. Abstract published at Hospital Medicine 2015, March 29-April 1, National Harbor, Md. Journal of Hospital Medicine. 2015; 10 (suppl 2). https://www.shmabstracts.com/abstract/you-cant-improve-what-you-dont-measure-a-systems-engingeering-approach-to-developing-geographically-matched-patient-provider-teams/. Accessed June 18, 2019.