Use simulation to learn results in advance
June 1, 1999 10 minutes read
Use simulation to learn results in advance
Predict outcome of changes more accurately
Simulation is a process reengineering tool used to analyze current systems and then conduct "what if" scenarios. It allows managers to see changes represented before they actually put time and expense into altering their current system.
"Clients love this tool because they are able to actually see the effects using computer animation. This is an amazing tool and has an enormous potential in the health care field," says Deborah Benson, a senior consultant with VHA Performance Consulting, based in Charlotte, NC.
The tool allows you to test out ideas in a simulation before spending money and staff to implement them. "Often, people find out the results were not what they expected," says Benson. "At VHA, we love this tool because it helps us to identify concrete recommendations which will give ED managers the biggest bang for the buck."
ED managers at Holmes Regional Medical Center in Melbourne, FL, used simulation models to decrease average length of stay. "We were able to pick and choose the areas we wanted to expend more of our resources on, and predict success more accurately," says John McPherson, MD, FACEP, medical director of the ED.
The simulation can validate the benefits of changes. "It gave us more confidence that we’d have a fruitful investment. So instead of just considering making specific changes, we’re actually investing actual time and resources to bring these services on board," says McPherson. "The tool helped to steer us in making some major resource allocation decisions."
Conversely, poor investments were also determined. "By modeling these potential changes, we could see that some of them would indeed have minimal effect," McPherson explains.
Test changes before you make them
Before using the simulation, the ED managers at Holmes Regional participated in a benchmarking study to set specific goals for improvement. "Nationally, the length of stay in the ED has been under criticism by patients. We would at least like to meet the average of our peers and not be an outlyer," says McPherson.
Once the ED’s goals were set [by the manager], the consultants were asked for recommendations on how they could be achieved. Possibilities included using ultrasound in the ED, adding a trauma service to expedite the stays of the trauma patients, increasing staffing, switching to bedside registration, implementing an abbreviated and secondary triage to enhance flow, and adding an observation and chest pain unit.
The simulation model determined the impact of each of these changes on length of stay. "This enabled us to test the changes in a virtual world’ without actually implementing changes to assess whether these changes and expense would be worthwhile," says McPherson. "Therefore, we have been able to proceed with confidence with our triage enhancement plan, our observation and chest pain unit, and our trauma service."
Data were collected to replicate the ED’s day-to-day operations onto the model. "The data and their accuracy will have the biggest impact on the model’s success," says Jan Hatzel Walker, RN, MSN, CEN, director of emergency services at Forrest General Hospital in Hattiesburg, MS. Architectural drawings of the ED are used, along with processes such as triage, registration, ordering, frequency of tests, and volume.
Once the simulation has all the necessary information, current bottlenecks can be spotted, notes Walker. "Or you can change a process and see how it affects the patient flow," she says. "How much faster would the patient move through the system if you initiated a new protocol to get x-rays done up front? If you initiated a stat lab in the ED and reduced procedure time, what would be the effect on the length of stay?"
At Forrest General, a simulation model was used when the ED was redesigned. "We wanted to be sure we were going to improve the flow of patients and reduce the length of stay with the new space," says Walker.
The factors Forrest General considered were, according to Walker:
• the need for radiology services in the new ED;
• if the increased cost would justify the reduction in time; and
• would the new design be able to stand a 20% increase in volume?
All of this was shown to the Forrest General group via the simulation model.
Data are collected 24 hours a day for seven days, pertaining to all aspects of an ED visit affecting length of stay. "By altering some of these variables by simulation, we could project how it [altered] total length of stay," says McPherson.
The following information was recorded for every patient:
• time of arrival;
• time at triage;
• time to bed;
• time to being seen by a doctor;
• what kind of orders were generated; and
• how many were from protocol and how many from physicians.
"We also input what time registration was done. Because we do all registration at the bedside, that could occur at any time in the process," says Walker. "We also input what time the labs came back and how much time passed from when the physician made the disposition to when the patient was discharge. From that data, they created the model for simulation."
By gathering this data, the ED was able to examine its operations as they currently exist. "That way, we had a better handle on our current performance, so we could tailor our performance improvement plan appropriately," McPherson explains. "This gave us a clear goal, as to how much time we needed to shave off our total length of stay."
The efficiency of the ED’s fast track was also examined. "We were able to assess how well we use those beds, and find out if we were appropriately triaging patients to fast track or not," says McPherson. Staff were interviewed at length about perceived delays, and simulation themes were chosen based on their input.
The simulation recommended the following changes, which were made in Forrest General’s ED:
• implementation of a stat area for the ED in the lab;
• assigning a technician or nurse to accompany physicians into the exam room and do order entry at the bedside;
• dedication of two radiology rooms to ED patient procedures;
• ordering medical records on all patients as an automatic process at registration;
• adding a new triage station that staffed by a float nurse to be put in effect when necessary; and
• adding a chest pain area used for any patients when needed for overflow.
Some of the simulator’s findings were expected, while others were not. "They recommended we implement bedside registration, which we had been planning for already. However, the enhanced triage process was not high on our radar screen, but the clear benefit of that became obvious," says McPherson.
The benefit of increased staffing resources can also be determined by the simulator. "We were trying to decide whether to put another doctor on from noon until 10 p.m.," says Walker. "We found out that putting another doctor on during peak time during our two busiest days would have enough of an impact to take eight minutes off our average length of stay."
Making key changes in processes was also found to have a major effect on length of stay. "By initiating the physician’s order at the bedside, it would reduce the patient’s length of stay by between eight and 11 minutes," says Walker.
The ED also examined how increased volumes of 5%, 10%, 15%, and 20% would affect the overall length of stay, and found there would be unacceptable delays. "Because of that, we went back to the architect and said, It’s been shown when you build a brand new facility that you usually have an 8% to 10% increase in volume. Based on this data, we don’t think that what you’re building us can handle those increases,’" Walker says.
Planned changes had little benefit
At Forrest General, a decision was made not to add a stat laboratory in the ED after the model indicated there would be a limited change in the length of stay. "Therefore we were better able to evaluate the expense versus the overall benefit," says Walker.
First, several assumptions were agreed to by the lab and the ED, pertaining to where the time would be saved if the stat lab were added. "We had to agree on how much time was tied up in getting the specimen to the department and to the person running the test. We assumed there would be no time saved with order entry to collection, because the ED does its own phlebotomy collections here."
It was determined that three to four minutes would be saved for patients with lab tests, says Walker. "In the end when we looked at minutes saved, it wasn’t enough to justify the cost of duplicating all the care in the ED," she explains. "It showed the cost was high for number of minutes saved, and there were other ways to save minutes without that severe cost."
Instead of adding a stat lab in the ED, a compromise was reached. "We realized that we could accomplish the same thing with a minimal cost, by building a stat area in the lab, with a tube system," Walker says. The system would have transported lab samples through a tube straight from the ED to the stat area.
In the same way, the simulation model showed that adding ultrasound capability to Holmes Regional’s ED was not worth the expense. "We looked at the total length of time it took us to get an ultrasound out of the ED, sending the patient to radiology, with the delays inherent in bringing them back and getting a result read by the radiologist, and how much time would be saved by having it done directly in the ED," says McPherson. "We didn’t feel making that change was going to enhance our service enough to justify the cost."
Limitations in the model
At Holmes Regional, the question of whether or not to add a stat lab is still undecided, due to limitations in the simulation model. "Outliers have a huge impact on our length of stay. If 15 out of 115 patients wait an hour or two longer, it can create major bottlenecks," says McPherson. "If we added a stat lab, I could just walk right across the hall to ask what’s going with Mr. Wilson in bed eight, whereas we don’t get good feedback from a distant laboratory."
Although the simulation model showed that the ED wasn’t too far away from the average, it didn’t take into account the outliers. Outliers are EDs that exceed or significantly fall below the average. An example would be an ED that has a much longer or shorter wait time than most.
"The mean alone is not the whole story. In our case, the benchmarks for certain labs was not too far away from our numbers. We had a little longer times in hematocrits and hemoglobins but it wasn’t as great as we thought. However, the outliers can really create an exponential and ripple effect in discharge delays," says McPherson.
These intangibles can’t be measured by the model, notes McPherson. "It’s tough to quantify the outliers with the simulator," he explains. It can’t measure their impact on the overall process and the average used to compare the ED to others. "The mean is a comparable entity, it’s difficult to quantify the impact of outliers on the ED."
As a result, the simulator wasn’t able to evaluate some of the intangible benefits of having an adjacent lab with the personnel part of the ED team, McPherson explains. "Because the major benefits of adding the stat lab would be in limiting some of these outliers," he says. "The simulation model was focused on decreasing the mean benchmark, which wasn’t as valuable in this case."
Consultants who do simulation usually do not have clinical backgrounds, which can be a disadvantage, Walker advises. "You need to be very careful in explaining processes to them," she says. "I found they did not spend enough time going over the model with me. You need to be the one directing how much time is spent reviewing the model."
It’s important to spend time actually looking at the model, Walker notes. "As a non-clinical person, all the consultant sees is the data. They may tell you a certain process would take eight minutes off your length of stay."
However, by looking at the model they’ve given, the applicable information may come from observing the larger picture. Even a portion of the data may help. "But by looking at the model, we can get information by seeing where patients are in the process. For example, how many patients are in x-rays in the evening? If you have seven patients over there, we know that part of the bottleneck comes from only having two rooms running at that point in time," she explains. "They do send you a copy of the model. You can’t readjust it, but can at least take time and look at it."
Validate your intuition
The findings generated by simulation can help you to validate necessary changes to administrators. "There are a lot of things you know intuitively, and you can be saying them over and over, but no one is really listening," says Walker.
Some of the recommendations are common knowledge. "The fact that observation units may decrease length of stay is nothing new. But the simulation certainly helped us to justify the addition of our observation unit, which became very evident. It was quite helpful in our planning with the administration and getting support," says McPherson. "It clearly showed the 24-hour observation unit decreased length of stay, which was one of our major goals. The model adds validity to intuition."
Similarly, the model showed that by initiating the triage enhancement plan, 20 minutes could be cut out of each patient’s ED stay, reports McPherson. "The trauma service was actually implemented before the results of the model came back. The simulation model showed it would have a significant decrease in ED stays of trauma patients, and the reality was fairly consistent with simulation," he says.
The tool’s graphs, timelines, and process descriptions helped administrators to acknowledge specific benefits of changes. "It allowed us all to clearly visualize that if we made certain changes, there wasn’t just a potential gain, but a probable gain," says McPherson.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content