Centre for Excellence in Operations (CEO)

Staffing Optimization

Detailed report     

Final presentation

User's Guide

At the start of 2011, we worked closely with Alberta Health Services (AHS) EMS Dispatch, AHS is one of the provinces largest medical services providers. The EMS Dispatch unit is responsible for delivering high quality dispatch services throughout the province?s EMS system.  This includes three core functions: i) emergency ground ambulance evaluation and dispatch, ii) inter-facility patient transfer scheduling and coordination, and, iii) air ambulance flight dispatch and coordination. Before we introduced our model, AHS ran their Central Communications Centre (CCC) scheduling based on a simple staffing model that was introduced early on in the existence of its operations. The current model was adopted to forecast the expected workload throughout a typical week; however the calculations were not overly robust and cannot account for the high variability of demand that a typical EMS centre would see. These high fluctuations had a negative impact on staffing utilization thus contributes a less than optimal staffing scheme. Program managers required the ability to properly plan shift schedules as well as to understand the impacts of alternative workflows. It was our job to create a model that attends to the needs of the managers and include the optimal staffing schedule in every situation and provide a user friendly interface that the managers can use in the future to alter the staffing level as CCC takes on extra demand from the other communities.

Our project strategy was to evaluate the service environment by evaluating data and mapping the process flow of the EMS dispatch centre. We would then use this analysis and understanding to create model inputs and develop a comprehensive staffing solution for the call evaluators at the dispatch centre. Using the provided telephony and computer-aided dispatch data, we found varying call demand depending on the time of the day and the day of the week. We accounted for these seasonality?s in our model. We also found that weekends had a significantly larger call demand caused by the processing of more inter-facility transfer calls. We analyzed hundreds of thousands of data entries and used these to formulate the inputs of our staffing model. We also made assumptions based on the nature of the dispatch centre, which is a queueing system.  The two main inputs obtained from our data were the arrival rate (demand) and service rate. Once we finalized the numbers for each hour of the day and day of the week we assumed an M/M/s queueing system. We used the M/M/s queueing assumptions to solve for the minimum number of server (call evaluators) required for each hour of the day and each day of the week. The queueing model had a threshold time and service level of at least 95 percent of the calls answered within five seconds, we also performed sensitivity analysis around these constraints. After calculating the minimum servers needed, we input these values into a binary schedule model that produced useful staff schedules for each day of the week. The results were also stress-tested on ARENA to confirm our solution. We studied and forecasted upcoming demand to make our model useful when the CCC takes on extra call demand from other service areas in the province. Overall, our model is able to save staffing costs by staffing a feasible amount of call evaluators in the centre that adhere to union and company policies. The results are reduced staffing and planning costs without sacrificing service requirements required in a critical health care business. Our model is robust to future conditions at the EMS dispatch centre and will ultimately provide useful intelligence when staffing this facility and others within Alberta Health Services.