How does 'pull' work in a hospital?

The 'pull' principle is probably the most important logistical innovation of Toyota. Because it uses a different way of thinking it already requires some effort to understand the principle in a factory. But how can you apply it in a hospital?

How does 'pull' work in a factory of Toyota?
The traditional way to manufacture is in stock. Production is based on demand forecasts, or sometimes just as much as possible, then we'll try to sell it. Because there are many different versions of a product (e.g. different colors and options for a car) a lot of supply is needed to prevent that you can't produce enough of each version of your poduct. This is part of the reason why so much money is tied up in the stock in factories. This is also one of the reasons that led Henry Ford to his classical quote: 'You can order any color T-Ford, as long as it is black'. The less versions of your product, the less stock you need to produce it.

In the nineteen fifties Toyota understood that the future lies in more variations of cars. At the same time Toyota was in deep financial problems and spend a lot of money on their stock. For both reasons they wanted their production processes to become flexible, so that customers can get their preference fast, with minimal stock. Inspired by American supermarkets where the shelves were filled with the same quantaties that were just 'pulled' out by the customers, Taiichi Ohno translated that 'pull' principle to the Toyota factory. An example how that works:

Imagine you make door handles. Pete next to you puts the handles in the doors. Between you and Pete is a table with three drawings of door handles. When you finish one, you put it on the drawing. Pete takes them from the table. Now imagine that Pete is falling behind because the doors don't reach him anymore. He can't install your handles. Soon the three drawings are filled with three handles. You can't produce any handles anymore, because you can't put them anywhere. You stand still untill Pete starts again to take one away (or better yet, you help the team that builds the doors). Thát is the essence of 'pull': the step after you gives you a signal that you can do your part. Because the whole factory works like this, there are no stocks. The system simply makes this impossible. Also, not one car leaves the factory that is not ordered by a customer. Because 'pull' starts with the order of a customer.

It took Toyota fifteen years before the deceptively simple 'pull' principle was applied in one complete factory. Now, 35 years later, all car manufacturers and many other companies try to work according the 'pull' principle. Between delivery of the 20.000 different parts from many suppliers and the assembly of each part in one Toyota now is on average four (!) hours. They manage this with their 'pull' system. 

How does 'pull' work in a hospital?
A hospital can't produce in stock (let's put twenty CT scans in stock?), so the problem of large supplies is unknown to hospitals. Is the 'pull' principle relevant in a hospital?

We don't have any stock, but we do have a lot of waiting in between steps in care processes. The throughput time from first contact until diagnosis often takes weeks if not months. Despite that the addition of the time spend on each step often is less then a day or only several days. From the perspective of a patient our care processes are a stop-and-go experience. The cause lies in the 'push' way of organizing the characterizes hospital flow.

Essentially 'pull' is:

respond to demand

When care is 'pull' organized the demand determines where care providers spend their time on. In hospitals we currently determine where time is spent on long before care is demanded (= 'push' logistics). Some examples of 'push' logistics:

The number of sessions in an outpatient clinic depends on the availability of specialists and perhaps of supporting staff. It is not based on current demand.

Sessions are programmed with fixed appointments, for example six new patients and twelve follow-up appointments. These programms are not based on actual demand, but on the preferences of the care deliverers. The numbers of each type of appointment will not respond to actual demand in a given week, resulting in unnecessary waiting times.

Supporting staff of outpatient clinics are scheduled two months in advance, when the number of sessions is not yet definite. Often sessions will be canceled or added afterwards. It will be difficult to reschedule the staff (and they will not like it). This results in overstaffing and shortage of staff. Also, the canceling of sessions is not based on less demand, but absence of care providers.

The number of OR (operating theaters) for a specialism per week is determined every year with a fixed planning for each week (on Mondays General Surgery has three OR's, Orthopedics two etc.). This is done before any patient is known. Why? In the words of professor Frits van Merode: the division of OR capacity is not a planning instrument, but a 'peace treaty'.

These are all symptoms of 'push' logistics: demand has to comply to our scheduling of capacity, instead that we respond to demand (by the way, that's why my phd research is about the 'reactive capacity of hospitals').

What would 'pull' logistics look like in a hospital? An example:

If in January more patient request an appointment for a Gynecologist, he or she takes more time to see patients and gets more staff support for the extra sessions. Later he or she performs more operations for these patients and gets more OR capacity to do so.

OR capacity is not divided anymore per specialism, but on actual demand. OR capacity is given to individual patients.

This requires a large flexibility from the system. It also requires trust from the care providers in the system 'I'll receive OR capacity for my patients when they need it'.

How can you recognize the consequences of a 'push' system?

General practitioners call to arrange priority for their patients ('push')

This lab result needs priority ('push')

The surgeon needs to call the radiologist to make sure his patient gets a MRI slot sooner ('push')

Or more in general:

'I push todays demand to tomorrow, because today I'm full" ('push')

Instead of:

'I do todays work today' ('pull', as proposed by general practitioner Mark Murray in his 'Adanced Access' concept)

Push and pull in a hospital
 I don't believe that a hospital can be organized completely 'pull'. There are two major reasons why 'push' is for some part inevitable:

1. Uncertainty what the value stream will be. When I order a car the factory knows exactly the specifications before the car is assembled and all steps required to build it. In care processes next steps are often determined during the process (for example: only after diagnosis treatment can be determined). Better yet, often steps change during the process because of complications or because a patient changes his or her mind.
2. Natural variability in care processes. Installing a door handle can be repeated each time in the same time. The same operation will take more time with this patient then the other because one body is not the same as the other.

Because of both factors a hospital cannot create complete work load leveling ('heijunka'). To organize care processes completely 'pull' unacceptable levels of overcapacity are needed. That's why a mix of 'push' and 'pull' will be required. However, it is evident that a lot more 'pull' is needed. For the benefit of patient flow, but also for the benefit of care providers. 'Push' systems lead to strong tensions (who gets how much capacity? Why do my patients need to wait for your department?). 'Push' systems tend to stimulate people to look after their own interest, even at the cost of the common interest.

'Pull' is not better then 'push'. They are two logistical principles that both can be useful, depending on the situation. The interest is to create 'flow': no unnecessary waiting. And that every care provider can spend as much time as possible on providing care (instead of pushing patients through the system or to claim as much capacity as possible). The interest is to have patients experience care as one, fluent process.

How to start?
A good starting point is the planning processes and policies (e.g. scheduling staff, scheduling sessions, programming sessions, OR capacity planning etc.). Use these questions:

• do we base our decisions on actual demand, or on internal (non-demand related) criteria?
• which information do we use concerning actual demand? If there is none: how can we get it?
• does this policy change increase or decrease the flexibility to respond to actual demand?

Finally, we should improve the flow of care processes in respons to the problems that are experienced on the work floor ('pull').