Operations Management in Action

Operations Management in Action

Surgeons at BC Children’s Hospital have a way to shorten wait lists and get their patients through the hospital system more efficiently and with less stress.

Drs. Geoff Blair and Jacques Leblanc, head and assistant head of sur- gery, respectively, have created a simulation model not unlike those used by air- lines and the military during wartime, that allows for the movement of large numbers of people.

Using a branch of mathematical analysis known as the “science of deci- sion making,” including game theory and queueing theory, Drs. Blair and Leblanc believed it was possible to simulate and test how changes to patient flow and surgical scheduling would affect throughput, patient waiting times, and budgets, without adversely affecting ongoing operations.

They created a model to simulate the processes from the moment a patient enters BC Children’s for surgery to the time they leave. This includes OR prep time, how long each procedure takes per surgeon, how patients are moved from the ICU to the appropriate unit, available bed space and actual recovery times.

Programmers added random factors provided by hospital staff such as unscheduled emergency cases, unplanned fluctuations in staff levels due to ill- ness, equipment breakdowns, and clerical problems. They also came up with a block schedule analysis tool the hospital could use to test various schedule changes, and assess their impact on future wait times, staffing, and costs.

The team used aliases of more than 30,000 actual patient files from the previous three years to ensure confidentially was maintained during the simu- lation process. They checked the accuracy of the program by running a one-year simulation of patient flow, and the results compared to what actually tran- spired during that period.

The value of the simulation is that it can effectively get past the “what if?” while staff try to come up with ways to deal with growing wait lists and escalating costs, says Dr. Blair.

“In healthcare, we’ve been relying on our puny human minds to try to deal with all these systems of queues and decision-making. And the interaction of everything has to be brought to bear to get a kid at Children’s Hospital who needs surgery into the operating room, with everybody who needs to be there, with all the equipment that needs to be there, on time, at an appropriate time that reflects the child’s medical condition. It’s not unlike the same complexity airlines face when faced with ticketing, boarding passes, security, baggage han- dling, runway availability, 747s unloading their passengers while another 747 is boarding passengers. So why not use the same sort of approach?” he asked.

Chapter 10: Simulation 285

The model has been a boon for block scheduling, he says. OR times have been set since the hospital opened in 1982, but with the aid of the simulation tool, those surgeries with the longest wait lists can be reallocated without caus- ing a logjam.

“The human tendency is to avoid changing, because we don’t know what’s going to happen,” said Blair. “There’s a real inertia in healthcare. And governments are increasingly demanding more proof that we deserve more money, to prove that we are being as efficient as possible. What we can do with this block scheduler [is] simulate what will happen because we have all the data entered in real time—every patient who is waiting is now in the system— and we can then take time away from orthopedics, cardiac surgery and give it to ophthalmology, say, and then we just run it. Within minutes, we can see graphically exactly what will happen to the waiting lists in cardiac surgery, orthopedics, and ophthalmology. It allows us to tweak time in areas that are not in bad shape and not inconvenience anybody; we can predict scientifically, in a valid way, what will happen if we make this change. This removes the fear of change.”

The Simulation Tool, as it’s called, will also be used to better manage outpatient clinics as well as OR time. It’s also going to be applied to better management of beds.

In one striking scenario used by the programmers, Wednesday morning rounds with surgeons and nurses were changed to the afternoon and resulted in an additional 54 surgeries being performed over the course of a year.

“This isn’t a cure for cancer,” said Blair. “But it could go a long way to

‘curing’ some of the major problems that affect us every day in our attempts to

deliver care to our kids here.”

SOURCE: Haley, L. 2005. “Simulation Aims to Speed Patients Through Surgeries.” Medical Post 41 (1): 2–3. Reprinted with permission.

Overview

Simulation is the process of modeling reality to gain a better understanding of the phenomena or system being studied. Simulation allows the user to ask and answer “what if” questions in an environment that is more cost-effective, less dangerous, faster, or more practical than the real world.

Although the simulation techniques outlined in this chapter are com- puter-based mathematic models, simulation does not require mathematic models or computer analysis. A fire drill, for example, is a simulation of what would or could happen in the event of a real fire. The drill is run to address

286 Performance Improvement Tools, Techniques, and Programs

any problems that might arise if there were a real fire, without the danger associated with a real fire. There are many different types of simulations appro- priate for many different contexts.

The report Building a Better Delivery System: A New Engineering/ Health Care Partnership, a joint effort between the National Academy of Engi- neering and the Institute of Medicine (2005), identifies engineering tools and technologies that could be employed to help to overcome the current safety, quality, cost, and access crises faced by the healthcare industry. This report specifically cites systems modeling and simulation as tools that have the power to enable healthcare organizations to improve quality, efficiency, and safety.

This chapter provides an introduction to simulation and the theories underlying it. The major topics covered include:

  • How, where, and why simulation can be used;
  • The simulation process;
  • Monte Carlo simulation;
  • Queueing theor y; and
  • Discrete event simulation.

After completing this chapter, readers should have a basic understand- ing of simulation. This should help them understand how simulation could be used in their organizations to evaluate choices and optimize processes and systems.

Uses of Simulation

Simulation can be used for many different purposes, including performance, proof, discovery, entertainment, training, education, and prediction (Axelrod 2006).

Performance simulation can actually carry out some task for an organ- ization or entity. It is related to artificial intelligence and usually simulates a human behavior. Voice recognition and robotic-assisted surgery are examples of performance simulation.

Simulation can be used to discover new relationships and principles and to provide proof of a theory. For example, Conway’s Game of Life (Berlekamp, Conway, and Guy 2003; Poundstone 1985) is used to discover and prove that simple rules can result in complex behavior. Simulations can also be used to entertain, as with virtual reality video games.

Simulation is often used for education and training purposes. Increas- ingly, simulators are being used to educate healthcare professionals in med- ical concepts and decision making as well as to train them in therapeutic and diagnostic procedures. For example, the mannequin simulator Resusci-Anne has been used for CPR training since the 1960s (Grenvik and Schaefer

2004). Since then, simulations related to medical training and education have become increasingly sophisticated. Training simulations allow users to prac- tice decisions and techniques in a safe environment where an incorrect deci- sion does not have serious consequences.

Predictive simulation can be used to evaluate the design of new products, systems, or procedures as well as to analyze and improve existing products, sys- tems, or procedures. This chapter focuses on predictive simulation—specifically, Monte Carlo simulation and discrete event simulation.

The Simulation Process

Simulation begins with development of a model. Once the model has been built and validated, the output of the simulation is analyzed to address the original question or problem (Figure 10.1).

Model Development

The first step in model development is to define the problem or question to be answered with the simulation. The usefulness of the simulation will be driven by the accuracy of the problem definition.

The next step in developing a simulation model is defining the con- ceptual model. Here, the system is described analytically or mathematically; inputs and outputs are determined, and relationships are defined. Because real-world systems are complex and difficult to represent analytically, assump- tions about the system must be made. A perfect model of the system is sel- dom possible, and approximations appropriate for the study must be made. There is usually a trade-off between model validity and model complexity; all things being equal, however, a simpler model is better.

Once the conceptual model has been defined, information required for the simulation must be collected. Data related to the probability distribu- tions of random variables in the system, data defining the relationships in the simulation, and data related to the output behavior of existing systems are collected. These data will be used in running and validating the simulation.

The final step in model development is actually building the computer model. In the past, this meant coding the software for the model. Today, many commercially available software packages make this step relatively simple.

Model Validation

The validity of a simulation is related to how closely the simulation mirrors reality and answers the question that was asked. Simulations can be devel- oped that are technically correct but do not accurately reflect reality or do not address the intended question or problem. Therefore, assessing the valid- ity of the simulation is an essential, but often difficult, step.

Ideally, the simulation is run and quantitative output data of the sim- ulation are compared to output data from the real system to determine whether they are similar. Alternatively, experts are asked to determine if the design and output of the simulation make sense to them. If the simulation is not deemed valid, the model must be redeveloped.

Simulation and Output Analysis

Here, the simulation model is actually run and output data are collected. If a number of different variables and variable states are of interest, experimental design can be used to determine the specifications of those variables so that the experiments can be optimally run in a timely, cost-effective manner. Ensuring reliable results may require many replications of the simulation. The results of the simulations must be collected, organized, and stored. Finally, the output data must be analyzed to determine the “answer” to the original question or problem.

Process Improvement in Practice

Six Sigma

If the primary goal of a process improvement project is to improve quality (reduce the variability in outcomes), the Six Sigma approach and tools described in Chapter 8 will yield the best results. As discussed previously, Six Sigma uses seven basic tools: fishbone diagrams, check sheets, histograms, Pareto charts, flow charts, scatter plots, and run charts. It also includes statistical process control to provide an ongoing measurement of process output characteristics to ensure quality and enable the identification of a problem situation before an error occurs.

The Six Sigma approach also includes measuring process capability—a measure of whether a process is actually capable of producing the desired output—and benchmarking it against other similar processes in other organi- zations. Quality function deployment is used to match customer requirements (voice of the customer) with process capabilities given that trade-offs must be made. Poka-yoke is employed selectively to mistake-proof parts of a process.