Why Healthcare Needs Simulation Modeling For Complex Decision Making
Have you ever been confronted with a complex decision full of many variables and unknowns? Have you found yourself combing through report after report, but just unable to comprehend all of the data you are looking at and how it all connects to one another? Do you wish you could play out a number of scenarios side-by-side to see what your best choice would be and avoid buyer’s remorse? Well guess what, you certainly can with an analytical modeling tool called discrete event simulation (DES), and here are four reasons why you should be using it in your decision making:
- Simulation Accounts For Randomness in a System
The traditional use of spreadsheets runs into a number of issues. For one, spreadsheets fall victim to being static by typically using averages, and anyone who has worked in the healthcare environment knows that healthcare is far from average and is full of many complex, random processes. These processes range from registering patients when they arrive in clinic, to rooming patients, to taking labs, to performing complex surgical procedures. Each of these are all dependent on one another and can compound as a patient moves through the healthcare system leading to downstream impacts.
To get a better idea of how this randomness impacts a system, imagine a traditional bank involving customers arriving in some random order throughout the day, who then seek help from a teller, and then they leave. In this process, people come in at a rate of say 12 people per hour (or every 5 minutes), and then wait in line for a teller who, on average takes 5 minutes to process that individual. Since customers come in every 5 minutes and then take 5 minutes to get their needs addressed, we should never have anyone wait in line right?
Well if we take into account random probability distributions (here we will use what is known in statistics as an exponential distribution), and run this process 24 hours a day for the next 365 days, the average number of people in line by the end of the year would be about 200, with an average time in in the bank of about 15 hours, 180 times longer than expected [1]! Just imagine the problematic issues of modeling a system that involves multiple of these random processes in a healthcare environment with a static tool. Simulation can overcome this gap by accounting for this variability.
2. Visualize The Process In Real Time
Static spreadsheets don’t always tell a compelling story for all stakeholders involved in a complex decision. What if you’re planning on completely redesigning a rooming process, changing staffing ratios, or redesigning operating room processes? Wouldn’t it be helpful to be able to visualize the the current process and the proposed process in real time, side-by-side, with your stakeholders? This is something that simulation programs can provide that traditional spreadsheet analysis cannot.
An example of this can be seen below where patients are entering an ED, getting triaged by a nurse, sent to the waiting room, and await a room to be available. From here, they get processed by a nurse, then a doctor, and then leave to their next destination.
3. Test Multiple Scenarios Side By Side, Risk Free
Have you ever gone into an improvement project, identified deliverables, and then after implementation saw no changes in your key performance metrics despite completely redesigning your processes?
With simulation we have the ability to build out an entire system and run multiple scenarios lasting anywhere from a few days, to a few decades so you can evaluate whether hiring one new physician or building an entirely new facility would improve your current process or not, completely risk free.
4. Help Frame Your Problem Correctly
The process of collecting data for each step in the current system forces you and your stakeholders to truly understand the current process, and sometimes by doing this, your problem may present itself in the form of inconsistencies or gaps that you were completely unaware of.
That being said, wondering how other organizations have successfully used this tool in you organization? Here are a few examples:
Emergency Department Throughput are you running into issues with throughput in the emergency department? Perhaps you’re currently faced with the decision of adding on additional physicians, nursing staff, building more rooms, or rearranging the locations of lab/imaging equipment so it is closer to the ED? Baptist Health of South Florida was able to use simulation to optimize staffing, reduce the length of stay, and decrease door to provider time by 46% [2]
Capital and Human Resource Utilization are you getting put in a position where someone is asking for a new, MRI machine to improve access to high end imaging? Well with simulation modeling we can model current utilization, and change variables such as the number of support staff or shorten schedule blocks around it to see if we need more machines, or potentially we just need to rearrange existing resources to improve utilization and increase access. With the costs of adding a new MRI machine exceeding $500,000 plus ongoing maintenance, you can avoid a buyer’s remorse by comparing numerous scenarios prior to spending any money.
Patient Access in Primary Care healthcare organizations are increasingly being pressured to improve primary care access in order to meet patient demand with a decreasing supply of physicians. With simulation modeling, we can model out the entire course of events a patient may have in their typical primary care visit and compare scenarios evaluating questions such as: what if we changed schedules from 30 to 20 minutes? What if we added additional support staff (e.g. CMAs)? What if we were to change protocols that would allow RNs to take care of specific issues would this allow more patients to come in for in person visits with their provider? The Portuguese National Health Service (NHS) did just this by designing new family health units and saw a 45% decrease in wait times to see a clinician, accompanied by a 5% increase in overall revenue [3].
Operating Room Utilization with margins getting smaller and smaller, and OR costs hovering around $55 per minute, there is an increased need to maximize the time an OR is being utilized. With simulation modeling, we can model out the entire process from when a patient arrives at the hospital, through the end when they either leave (in the case of same day surgery) or when they are transferred to a floor in the hospital (in the case of inpatient procedures) to detect potential inefficiencies. For example, Memorial Health was able to map out existing processes prior to an expansion, identified the inefficiency existing in a poor elevator (as opposed to the people in an OR), and were able to advocate for a better transportation system in the new building. This led to an average decrease in wait time of 30 minutes ($1650 dollars per patient using the stat above), and ensured that the new lay out could accommodate demand for the next 20 years [4].
When healthcare environments are continuously being pressured to make decisions about complex systems, day in and day out, these decisions can be facilitated by use of discrete event simulation. When used correctly this tool can help to increase resource utilization, decreasing wait times for patients, and increase overall satisfaction by staff providing care to patients through optimized processes and help organizations maximize their profits in a time when they need to further bend the cost curve.
[1] Data collected using an exponential distribution with lambda equal to 5 over the course of 5 replications. No warm-up periods used in data collection.
[2]https://healthcare.flexsim.com/case-studies/better-decisions-emergency-department/
[3] https://bmchealthservres.biomedcentral.com/articles/10.1186/1472-6963-11-274#Sec8
[4]https://www.simul8.com/case-studies/SIMUL8-Memorial-Health-System-case-study.pdf
This is a part of a multi-part series on simulation modeling which will cover applications of simulation and key components involved in developing your first simulation model.
