Over the course of the past
couple of months I have been doing a lot of work in the healthcare field,
helping hospital improve the throughput of patient. I’ve done work in various hospital departments
including Emergency Departments, Radiology Departments, Out Patient Clinics,
etc. The common denominator in all of
these departments was the length of time it was taking to process and treat
incoming patients. All of these departments
were experiencing excessively long patient waiting times to receive their
needed treatment. In my posting today I
want to address these long wait times and discuss why I believe they occur with
such regularity.
The best way to demonstrate
this phenomena is to look at two identical processes such as an eye clinic. In both clinics, as depicted in the graphic below, patients enter into Department 1, are
processed and treated, then sent to Department 2 where they receive more
treatment and finally on to Department 3 where they receive the final part of
their treatment before being released.
Both clinics have the same resources and capabilities and both clinics
have the same arrival and departure rates of 30 patients per hour.
Obviously what is
drastically different is the total amount of patients within each clinic’s
system. Clinic 1 has a total WIP level
of 60 patients within all three departments while Clinic 2 has a total WIP
level of 20 patients and includes the total number of patients being seen or
treated as well as the patients waiting in front of each department. We’ll come back to some of the reasons why
the WIP levels might be different in the two clinics, but for now let’s explore
the impact of these WIP differences between Clinic 1 and Clinic 2.
The bottom line here is that
high levels of WIP (in this case, patients) will always lead to long lead
times. Actually the average response
time of the system is proportional to the level of WIP within the system. So automatically we know that Clinic 1’s
response time will be longer than Clinic 2’s.
In fact, John Little, in his work on queuing theory, gave as a way to
calculate the average response time.
Little explained that the average response time of a process is equal to
the amount of WIP divided by the system throughput rate or RT = WIP/T. The response time for Clinic 1 is therefore
60 patients divided by 30 patients per hour or 2 hours. Clinic 2’s response time is 20 patients
divided by 30 patients per hour or 0.666 hours or 40 minutes. Two hours versus forty minutes….in which
clinic would you like to receive treatment?
It should be clear to
everyone that most of the time spent through these clinics, but more especially
for Clinic 1, is waiting. There have
been many studies that have demonstrated that the actual work times or value-added
times are typically less than 10 percent of the total response time. So in Clinic 1 the actual work time is
probably on the order of 12 minutes (i.e. 120 minutes x 10 % = 12 minutes).
Earlier I said we’d explore
some of the causes of excess WIP in Clinic 1 versus Clinic 2. One of the major causes of excess WIP is not
controlling the input rate. In other
words, a lack of synchronization into, through and out of any process will
create excessive amounts of WIP. In comparing
Clinic 1 to Clinic 2, one of the major reasons for Clinic 1’s WIP problem is
that there was no scheduling system in place.
That is, Clinic 1 did not have a scheduling process in place so waves of
patients entered the clinic. Clinic 2,
on the other hand, required patients to call ahead and schedule a time for
their treatment. In doing this, Clinic 2
avoided the “large batches” of patients coming to their clinic. Clinic 2 did take “walk-ins,” but the
walk-ins were told that they would have to be worked-in when time slots became
available.
Another potential reason for
Clinic 1’s excessive WIP was that they did not “pull” patients into their
process. Clinic 2 employed either a
Kanban System or Drum Buffer Rope to synchronize the flow of patients. Clinic 2 recognized the existence of a
constraint and never let the constraint sit idle. Clinic 2 did things like staggering lunches
and breaks so that the constraint was always working. In our diagram, Department 2 was the
constraint, so patients would experience excessive wait times in front of this
department.
The most negative effect of
the excessive WIP in this healthcare scenario was unhappy patients which caused
the patient satisfaction metric to decrease significantly. Controlling the amount of WIP within any
process is paramount to keeping patient lead times to a minimum.
Bob Sproull
2 comments:
Thank you and great post, Bob.
I'd like to add that there's another factor that you touched on briefly that starts to take its toll on a system like this. It's what I call the "Livestock" factor. Most truckers will tell you that carrying livestock can be quite a bit more challenging than regular stationary payloads. They tend to have a mind of their own and move around a lot. So, the same goes for people needing care who are in the queue rather than some part or assembly or some form that needs to be signed. They tend to have a mind of their own.
It reminds me of 'waiting on the flight line' versus 'circling the airport' when your plane arrives before the gate is available. Seems the passengers are a lot more patient in the air than they are once they've reached the ground. However, it costs the Airline a lot more in fuel to do so. But, dealing with a bothered and frustrated passenger (OR patient for that matter) can be quite stressful for the service providers. Being stressed out can provide even more constraints to speedy quality service.
We Americans can get pretty "im-patient".
Excellent points Dennis and thanks for contributing. I like the term Livestock Factor!! Again, thanks for your comments.
Post a Comment