Many possibilities and areas of application for simulation exist that can be pursued in a clinical setting.
Simulation for skills training, risk management, and workflow analysis
First, simulation-based skills trainers are increasingly used. Indeed, in healthcare and in the military and EMS domains, mannequins are commercially available that can simulate patients in a variety of health conditions with partial or full replication of physiological systems — circulatory, urinary, nervous, musculo-skeletal. These simulators bleed, drool, breathe, and speak, and are invaluable as trainers for the more inexperienced clinical staff to develop new skills at no risk to the patient, and for advanced practitioners to refresh theirs. The use of realistic castings to simulate injuries (moulage) is also increasing. Attached to mannequins, they can provide invaluable training and reduce shock for medics eventually deployed to the battlefield.
Second, modeling and simulation of scenarios, as in probabilistic outcomes from one or more decisions — think desired increase in revenue, reduction in waste, or elimination of harm events as a consequence of a certain investment in technology or a change in policy — can be carried out via Monte Carlo methods. These algorithms are best applied when there is significant uncertainty in the inputs to the model being studied, and one wants to gauge risk. As one lets the inputs range over a series of values and runs the model over and over again, the answers or outputs take the form of a range of values, each associated with a higher or lower likelihood of occurrence.
Third, patient flows can be simulated and studied from admission to discharge with a view to optimizing the design of facilities, streamlining operations, or improving care. This is not restricted to inpatients at hospitals, and can be done for outpatients at physician practices as well. Discrete-event simulation (DES) uses state-based modeling, with events triggering transitions between states. The DES of patient flow at a hypothetical ambulatory clinic or physician practice shown in the video clip below for illustration was developed on a PC using a commercial modeling and simulation tool, and a screen capture of the running simulation was done with another tool.
As the simulation runs, you can see the operational blocks, representing activities at the clinic during a five hour time frame, light up in various colors — green indicates an action occurs, red indicates blockage waiting for a resource — as patients move through the clinic. The changing numbers on top of each block indicate the patients waiting or being served at each station. The model is laid out in horizontal swimlanes — I will explain swimlanes in a coming post — that group the specific activities corresponding to patient intake, treatment, and discharge respectively — activities include registration, taking a patient’s vitals, drawing blood, walking the patient between three exam rooms, imparting education, doing the check-out, etc. Note that the animation speed is irrelevant, and is typically varied for ease of visualization. The simulation produces a number of interesting numerical statistics in terms of wait times, service times, queue lengths at various service points, and so on, quite useful for analysis and eventual redesign.
Building such a model allows us to investigate and vary many factors impacting performance measures — the latter include length of visit, wait times, amount of patient movement or transport within a facility, and others — which are likely proxies for patient satisfaction.
These factors (inputs to the model) include:
- alternatives among physical layouts
- number of exam rooms
- location of testing devices
- single or multiple roles for staff
- proportion of appointments vs. walk-ins
- scheduling of consecutive appointments as a function of their complexity (patient mix)
- hours of operation and staffing levels
- (re)sequencing and parallelization of activities within the workflow of the clinic.
Thinking of wellness as a continuum may point one in the direction of moving certain setup activities to the time between visits to help make the eventual encounter a better experience for patients. Dead times that cannot be reduced while a physician is with another patient can also be better utlized by imparting education to patients, for example.
Such an analysis can be conducted simulating an operating time frame that covers a single day, a week, or a month in a matter of minutes, once a model has been developed. The probability distributions behind the operational blocks simulated in the model include memoryless and random (Poisson) for arrival times of walk-ins, exponential for service times (registration, blood draw), Erlang for sequences of service steps, and so on.
In consulting for a client physician along the lines outlined above, I was able to show that patient scheduling could be improved. The AMA categorizes patient ailments, going upwards from 1 to 10 as to increasing severity and thus physician time and decision-making complexity. At the lower levels, we find BP checks, immunizations, sore throats, phlebotomies, and headaches. Higher up are diabetes, flu, and asthma. Considered quite complex are complications from chronic ailments and new patients — something which may not be immediately obvious. Given that many physician practices allocate equal time for appointments, consecutive scheduling of new patients may not be the best thing to do from the point of view of minimizing patient wait time, shortening queues, and not running over at the end of the day. Simulations were run with different and improved scheduling schemes preventing this specific situation, which translated eventually into real-world improvement and greater physician and patient satisfaction at the practice.
Clearly, operations can be studied in this fashion in a variety of clinical contexts, from overall patient flow to specific processes such as twice-daily meds administration, and from facilities sizing, layout, and equipment to internal warehousing and distribution of surgical supplies among hospital sites. The power of simulation to both gain an understanding as to what is going on in a complex setting and to put decision-making on a firmer, data-driven basis should not be underestimated.