Why Hospitals Need Process Simulation
I did my first simulation on an IBM 360, and I input my data using punch cards. For people who don’t have a clue what this means, it was back when dinosaurs roamed the earth, before the new machine had a soul, and before wizards stayed up all hours building the internet. We've come a long way in my four decades of experience in this field. Today, laptops are powerful tools, and simulation programs produce 3D models that allow me to follow people and materials as they move through the simulation, enabling me to see bottlenecks form, long travel times, idle staff, or equipment, among other issues.
In the world of process improvement (Lean Six Sigma, Operational Excellence, etc.), it always amazes me that factories and other industrial operations are big users of simulation to improve operations and facility planning, yet simulation is seldom used in hospitals. Simulation is a powerful, but underutilized tool in Healthcare despite the obvious benefits it could bring.
What do these benefits look like? Let’s start by examining the differences between hospitals and factories. In a factory, we control the inputs with ordering and quality control. The inputs are known, as is their availability (assuming we’re managing our supply chain). The processes are standardized and consistent. If we tighten bolts on a machine, the bolts are in the same place each time and do not vary significantly. This means we can predict with considerable accuracy, and therefore we can control the inputs and process, ensuring consistent outputs.
Now, let’s think about a hospital. Unlike factories, hospitals react to inputs, even if they are scheduled. We’ll examine two quick examples: the Emergency Department (ED) and Surgery. In our emergency department, patients arrive unpredictably, and departments respond accordingly to their arrival. While the volume certainly follows a pattern, we don’t know on any given day at 11 a.m. if we will see six or 16 patients.
Let’s look at Surgery. Most patients are scheduled for Surgery, but we still must allow for emergent or urgent cases. These cases will disrupt the schedule and cause problems unless we’ve planned for these unexpected procedures. Take a simple example: drawing blood. The procedure for drawing blood is standardized. But unlike bolts on a machine that are in the same place and don’t have any variation, patients’ veins do have variation. Drawing blood on a patient with nice, big veins that are easy to find can go pretty fast, while another patient may have collapsed veins that are challenging to tap. In some cases, a patient may look easy, and then pass out. In any given Surgery, two patients having the same procedure may require different materials: stents, grafts, hardware, and/or sutures. In a hospital, most tasks are performed by humans for humans, resulting in natural variations in skill levels, which in turn increase operational uncertainty.
In my previous job, we spent a lot of time simulating ED patient flow and workflow. Why? In many hospitals today, the emergency department (ED) is often considered the front door to the hospital. It is not uncommon for 50% to 60% of the inpatient admissions to come through the ED. So, if patients have a horrible experience in our ED, the next time they’re likely to go to another hospital. There is also some evidence suggesting that patients admitted to an inpatient bed after 3:00 pm tend to experience longer lengths of stay; since hospitals are typically reimbursed on a DRG (Diagnosis-Related Group) or per-case basis, understanding these costs can have a significant impact on revenue projections.
Moreover, seeing patients in a more timely fashion improves the patient experience and reduces avoidable suffering. Improving workflow means that patients spend less time in the emergency department (ED), which enhances the patient experience. It also means we can process more patients through our emergency department (ED) with the same amount of resources, resulting in cost savings, increased revenue, and enhanced capacity.
Understanding how our processes cause bottlenecks is an excellent use of simulation tools. Since our processes are variable, simulation can help us determine over many iterations where the bottlenecks occur in a way that averages will never reveal.
Since patients arrive randomly (though they do follow certain statistical distributions), if we simply use average arrival numbers during any given hour of any given day of the week, we can be very far off on our volume predictions, resulting in a waiting room jammed with patients, staff strained to the breaking point, or long periods of inactivity.
Not only do patients arrive randomly, but their acuity level can also be highly variable. Some patients arrive experiencing life-threatening trauma, others may need urgent but not emergent care, and some may have only minor issues. Thus, assuming average acuity levels can also lead us in a very wrong direction. As Sam Savage, in his book The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty, said, “Plans based on average assumptions are wrong on average.”
In our other example, Surgery, most patients are scheduled ahead of time. Still, we must have contingencies for the emergent patients who come through our ED, inpatient services, or otherwise urgent patients. How do we ensure the capacity to deal with emergent and urgent patients while balancing our costs and resources? (See my previous discussion on balancing the voice of the customer, voice of quality, and voice of the business.) Do we keep an operating room empty, staffed, and ready for random emergencies?
But what if we get two emergent patients or an emergent patient and an urgent patient? Do we operate a second shift to handle urgent patients? Do we use overtime, etc.? There are many ways to address these issues, and simulation can be a valuable tool in this process. We can simulate various scenarios and understand the patient risk, potential resource consumption issues, and explore different strategies to deal with each. This is far more powerful than simply picking a plan based on known averages and hoping it works.
When I think of the value of simulation, I often think of a much simpler example. I worked on a project in a large hospital looking at environmental services (housekeeping)—specifically, discharge cleaning: cleaning rooms previously occupied by discharged patients in preparation for new admissions. The hospital was experiencing backups in its ED, with patients waiting for beds, because the rooms had not yet been cleaned. Examining the scheduled staff and the average discharge rate, staffing levels appeared to be sufficient, and it did not seem we would have a problem. Therefore, we constructed a simple simulation to examine discharges, patient wait times for admission, and staffing.
It was eye-opening. We first learned that the demand for beds was different based on the day of the week and the hour of the day. Most discharges occurred later in the day and into the evening and night, while we were staffed to handle most of our discharges during the day. Second, the demand, matching discharged beds with patients waiting, was more of an issue earlier in the week. We tended to admit surgery patients earlier in the week so we could get them out by the weekend, and that scheduling strategy was bumping up against our need for beds for ED patients. Third, we looked at actual staffing versus scheduled staffing and found many “callouts,” staff calling in sick or otherwise not showing up for scheduled shifts. When this happened, filling the slot was difficult; assuming we could fill it, it sometimes took two to three hours.
The simulation showed us that discharge cleaning spiraled out of control, even if we filled the slot in two to three hours. The demand and capacity for cleaning were running on thin margins, and even a little drop in our capacity would put us so far behind that we would wind up cleaning rooms well past midnight. This resulted in patients experiencing very extended stays in the ED and caused problems for our nursing staff when we admitted patients, especially after 10 p.m.
Simulation helped us understand our challenges in a way that averages or simple math could not, and allowed us to test various potential solutions before implementation. As a result, we adjusted staff schedules, established a dedicated discharge cleaning team, and enhanced our staffing to better buffer callouts. This resulted in happier nursing staff since patients could be admitted earlier in their shift, happier ED staff since they did not have to hold patients waiting for beds, and happier patients since they did not have to spend as long in the ED.
In our book Simulation Solutions: A Practical Guide to Improve Patient Flow & Facility Design in Healthcare Operations, we discuss three situations where simulation is an excellent tool in the following situations:
Volume (Demand) fluctuates in random and difficult-to-predict patterns
Volume for a new facility or patient services needs to be clarified
Effects of volume fluctuations on staffing or facility usage (e.g., exam rooms, operating rooms, etc.) are unclear.
The first and third of these situations are present in many hospital operations.
As biologist Stephen Jay Gould once said, “Our culture encodes a strong bias either to neglect or ignore variation. We tend to focus instead on measures of central tendency [averages, medians], and as a result, we make terrible mistakes, often with considerable practical import.”
Another project I worked on focused on reducing ED waiting times, with the goal of seeing patients within 30 minutes of their arrival— a common goal for many EDs. We used a simulation program (FlexSim) that allowed us to create a lovely 3D model of our waiting room and simulate the ED front-end processes: registration and triage.
The model made it easy to simulate a full waiting room with patients waiting in line for registration and triage as the morning went on. By noon, a second staff member was added to the mix, but the wait times did not decrease until late evening. While we could and did often work this part out with math, the simulation provided a graphic demonstration of how the line was building and the amount of time it took to get things back under control.
However, the greatest value of simulation was that we could try out various scenarios, bringing staff in at 10:00, 10:30, 11:00, and so on, and examining the effect these changes had on wait times. Ultimately, we found a scenario that met our needs. This was far easier than changing a staff member's schedule and trialing scenarios with real people.
Let us recap. Why do hospitals need simulation? The random and sometimes unpredictable nature of demand, the variation in patient need (one patient may be emergent another routine, or two patients with the same diagnosis requiring different treatments because of different underlying health issues), and the variability in process requirements make simulation a perfect tool to both increase the understanding of problems and simplify the process of testing solutions.
Simulation has come a long way since I started working in hospitals in the mid-1970s, especially in the last 10 to 15 years, with the increased power of personal computers and the growing sophistication of object-oriented programs. But we’ve hardly scratched the surface of how simulation can benefit hospital operations and planning.
If you’re curious about simulation in hospital operations and planning, check out our book Simulation Solutions: A Practical Guide to Improve Patient Flow & Facility Design in Healthcare Operations, available from Amazon. It will provide you with a more comprehensive understanding of the technologies and techniques your healthcare organization can adopt to streamline its operations.