Friday, December 16, 2011

Improving Patient Flow: The Concepts of Natural and Artificial Variability in Demand

Those of us old enough to remember “I Love Lucy” remember the Candy Factory episode where Lucy and Ethel take a job at a candy factory and attempt to wrap candies as they progress down the belt. When the speed is slow enough they manage to keep up with the work, their boss orders that the belt be increased in speed and the demand is too much for Lucy and Ethel to keep up with. This episode is a humorous way to demonstrate how varying the demand affects a system. In healthcare, the mismatch between the supply of caregivers and the demand for their services is the major cause of waiting and delays in care.

It’s not a secret that services like the Emergency Department get overwhelmed at times of peak demand. In the ED, we never know precisely how many patients will arrive in a given day. Our operating rooms are very busy on some days and not so busy on others. As a practicing physician, I used to think that peaks and valleys in demand were just a fact of nature. You just had to “suck it up” if you were busy and then you were owed the break you got when business was slower. I now know that things are not quite so simple. 

Deeper study of the peaks and valleys in patient volumes shows that there are actually two kinds of variability affecting patient demand. 

The variability due to the amount of illness in the community at any moment in time is called natural variability. For example, we know we get more patients coming to the Emergency Department in the winter with fevers and colds and coughs than we do in the summer, and we know we get more patients coming in with trauma and outdoor related injuries in the summer than we do in the winter. Based on historical data, trends, etc. we know approximately how many patients we are going to get every day of the week in the ED. Although we don’t know the exact number, the variability from day to day is due to natural causes beyond our control. Natural variability cannot be controlled; we just have to manage it. We generally need to have a bit more capacity then we will use on average to be able to deal with the peaks in demand. The mathematical science called Queuing Theory helps us staff efficiently when demand is varying naturally.

The second source of variation is called artificial variability. In artificial variability, the cause is man-made. Like in the candy store episode, the manager was controlling the speed of the conveyor belt. The concept of batching is the opposite of continuous flow. Generally individuals or subsystems batch things to reduce their own personal inefficiency. What they generally do not realize is that batching generally makes the larger system inefficient. 

When I was a pediatric resident we knew that if we ordered a lead test on a child, we wouldn’t get it back until later in the week because the lab waited to get a number of blood specimens before running the test in an effort to save money and run them all on one day. The lab director was trying to do the right thing but was not really thinking about the waits and delays for pediatricians or the rework required to get the results later and deal with the patient on another day. Physicians and nurses generally batch discharge work until later in the day because they sense that they need to deal with new admissions and the work of seeing sicker patients first. After all, the patient who is ready to go home is safe, so it feels like batching discharge work is a good thing. However when discharges all happen late in the day, admissions backup at other places, like the ED and the PACU waiting for the beds of the patients who are to be discharged. The largest single source of batching between days at GBMC is the elective surgery schedule. On some weekdays we do many more elective surgeries than on other weekdays. You can imagine how difficult it is to staff Unit 48 or Unit 58 efficiently if they are going to get 10 new patients on Tuesday and 3 or 4 on Wednesday. 

What we know now is that artificial variability is actually a BIGGER source of day to day variability in census and the stress caused by peaks in demand than natural variability is.  We also know now that healthcare workers who are overwhelmed are much more likely to make errors and potentially put patients at risk or much more likely to not be able to get to provide all of the care to each individual if they’ve got too many patients. 

So the goal is to reduce artificial variability caused by batching in order to reduce stress on our providers and make it safer for our patients. We also know that we can actually serve more patients in a more cost-effective way if we smooth the demand and try to get to continuous flow.   Once we have eliminated as much of the artificial variability as we can, we can then predict the natural variability and staff our hospital effectively. 

We’ve engaged some consultants from the Institute for Healthcare Optimization to help us do this work.  We are now in the midst of an initiative to smooth the elective surgery schedule, to keep the patients safer, to reduce waits and delays, to make the surgeons work more predictable and to make it easier to do even more surgeries. Under the leadership of Dr. Jack Flowers, our Chairman of Surgery and Dr. Lewis Hogge, our most recent block time changes have helped to begin the smoothing of patients by destination unit, Unit 48 or 58. 

Another part of our initiative to use science to manage flow and eliminate artificial variability is with our medicine service, under the guidance of Drs. Neal Friedlander, Paul Foster and Fred Chan, who have begun an initiative to smooth the discharge time of their patients to eliminate late afternoon batching. Their new mantra will be to discharge the patient as soon as he or she is clinically ready.  Most of us believe it’s unethical to send patients home in the middle of the night; we should have a relatively continuous flow of discharges from about 8 a.m. to 8 p.m.  Right now, we don’t.  We have a peak of discharges from 4 p.m. – 8 p.m. and we have very few patients going home before that, mostly because the system now has staff batching their work rather than trying to get to continuous flow. As they redesign the sequence of people’s work, or do discharges as a team, they will be helping to maximize the efficiency of the entire system. 

Interestingly, Israeli economist Eliyahu M. Goldratt examined this issue in his 1984 management-oriented novel titled The Goal, which focused on constraints and bottlenecks and how to alleviate them, and applications of these concepts in industry.  For example, Goldratt looked at the issue of why Japanese automakers were doing better than western automakers in producing vehicles, and noted that American manufacturers were focused on issues such as how often a machine was used rather than focusing on what the end goal was, how much product was coming off the assembly line. Required reading at Harvard Business School that was originally published nearly three decades ago and republished twice since, this book was a seminal work in American industrial thought that we can benefit from in healthcare today. Dr. Goldratt followed this novel with his book, The Theory of Constraints, another must-read for students of patient flow.

If it was your daughter, you wouldn’t want her waiting for long hours in the ED for a bed on the unit. We used to take care of this by always having a lot of excess capacity. Now no one can afford this so we have to use science to make the system work better. 

Do you have any thoughts on how we can improve patient flow and eliminate batching at GBMC? Please share your thoughts below.


  1. We also have to make it apparent to the patients that we have a discharge time. Most of the time I am discharging patients late it is because they are waiting for rides home. They know we will wait for their family get off of work. We need to make a policy or posting a time in their room to make them more likely to go home early

  2. While smoothing is great for patient flow in the hospital, not batching has some negative quality issues on a physician level. Standardized work clearly leads to less errors and a better product. Forcing surgeons do multiple different procedures on the same day, instead of repeating the same procedure, can lead to errors and certainly complicates the amounts and types of equipment needed for an individual operating room. Consideration needs to be taken for all the issues. The goal is the best care for each patient.
    That said, done properly, smoothing can make the care better, more pleasant, and more efficient.

  3. One team in the first Frontline Leadership program researched and presented a proposal for a discharge waiting area. Perhaps it's time for this to be considered seriously. There is no need for a patient who is stable enough for discharge to sit in a bed all day until their ride comes, or because they want to eat one more meal. I was once a patient in another area hospital. When my discharge papers were given, it was clear that my stay was over, and I definitely did not receive much attention from the staff until I left. Discharged patients who "linger" also affect staffing decisions. We have to bring in nurses and techs for these people who are in fact discharged and no longer in need of our care; indeed, discharge should invalidate all previous orders so no more meds, treatments, etc., should be given.

  4. Thanks for the comments!

    I have been studying discharge time for some years. Everyone remembers the case who is waiting for a ride home. But when we have studied this, for every patient waiting for a ride, there are 8-10 going to the nursing station asking "Can I go home yet?". We need to take care of this bigger problem first.

    We need to be clever in our approach to those who are truly waiting hours for a ride after we have completed the discharge work. Since they are generally few, we might even be able to pay for their ride home (if it is safe). Every hospital that I know of that built a discharge lounge closed it for lack fo use. What percentage of patients on your unit truly have ride home as the rate limiting step?

    I agree with the comment that the best care for the patient has to be the goal. (Better health, better care, lower cost, more joy). I am not aware that anyone is "forcing" surgeons to do "multiple different procedures on the same day". The surgeon determines what procedures are done.


Thank you for taking time to read "A Healthy Dialogue" and for commenting on the blog. Comments are an important part of the public dialogue and help facilitate conversation. All comments are reviewed before posting to ensure posts are not off-topic, do not violate patient confidentiality, and are civil. Differing opinions are welcome as long as the tone is respectful.