Almost any short visit to your nearest hospital will leave you with the notion that something here can be done better. Waiting times abound even though many people seem to be doing nothing. Patients spend hours in a waiting room to get into the Emergency room. Patients spend hours waiting in the ER to get a hospital bed. Patients spend hours in bed waiting to see a physician. Once the patient is ready to go home, they spend hours waiting to be discharged.
The full set of possible causes for these delays is too long to fully list but will include, staff shortages, supply shortages, random times between arrivals, random activity times, rework and readmission, coordination across multiple units within and outside of the hospital, and misaligned incentives for the agents involved in the process.
With this level of variety of causes and problems to deal with it may seem hopeless to even speak about a standardized process to make things better. However, we do have a couple of powerful tools that can help – namely decomposition and heuristics. Decomposition refers to breaking a process down into smaller, discrete pieces. Heuristics refers to simple policies or “rules of thumb” that can be brought to bear when looking to improve each piece. With this as the philosophical foundation we developed a 6-step process, or “recipe” to make these processes better.
Our Six Step approach
- Process Description
- Data Collection
- DES Modeling
- Metrics of Interest
- Proposed Changes
- Predict Impact
Details of each of the steps involved can be found in Dada and Chambers 2019. A full exposition of the approach combined with discussion of some of the roughly 50 projects we have completed using this approach are available in Chambers, Dada, Williams 2022. Consequently, we provide only a thumbnail account of the approach here.
Step 1 refers to developing a process description with enough detail to ensure that all parties involved are on the same page and have a common understanding of the subject at hand. This is typically done with a Process Flow Diagram as the central element. This depiction will include process steps, resources involved, and information about activity times.
Step 2 is the old-fashioned observation and recording of actual flows, activities, and times. We emphasize that pulling this data from most IT systems that we have encountered does not work. Such data sets are doomed to contain enough inaccuracies to make any uses of the data highly problematic. There is simply no substitute for direct observation, and anyone who tells you is it not needed, is selling something.
Step 3 is Discrete Event Simulation (DES). DES modeling is a central part of our process. This is key because standard models from a static analysis or queueing theory make too many assumptions and lead to over-confidence in predictions. Mathematical jargon allow us to sound authoritative, but such appearances are often delusions of grandeur.
Step 4 is identifying metrics of interest. Novel, or at least setting-specific metrics often have to be created for the problem at hand. Often standard metrics such as Cycle Times, or Waiting Times are sufficient. However, it is equally likely that the specific problem or symptom you want to measure is specialized to your unit or population.
In Step 5 we combine our data, modeling, and metrics of focus to devise experiments that can be done within the DES model. Based on these findings, we can propose changes to make the system run better. This may involve changes in resource levels, sequencing of steps, standardization of activities, or a host of setting specific ideas.
Finally, given all of the work from steps 1-5 we can predict the impact of whatever change we are recommending. With this in hand we move no to implementation.
Take Away’s
The bottom line is that these processes are very hard to understand and to fix. However, rather than throwing up our hands in surrender, we develop a consistent approach which we have used repeatedly to get positive results. Many of these efforts are documented in the publications, and notes included on our Documents page. Additional discussion is also available in our text at, Amazon.com.