Vaccinating millions of vulnerable people in the middle of winter requires a safe, efficient and effective process.
It is not safe to have queues of people waiting outside in the freezing cold. It is not safe to have queues of people packed into an indoor waiting area.
It is not safe to have queues full stop.
And let us face it, the NHS is not brilliant at avoiding queues.
My experience is that the commonest cause of queues in health care processes something called the Flaw of Averages.
This is where patients are booked to arrive at an interval equal to the average rate they can be done.
For example, suppose I can complete 15 vaccinations in an hour … that is one every 4 minutes on average … so common sense tells me it that the optimum way to book patients for their jab is one every four minutes. Yes?
Actually, No. That is the perfect design for generating a queue – and the reason is because, in reality, patients don’t arrive exactly on time, and they don’t arrive at exactly one every three minutes, and there will be variation in exactly how long it takes me to do each jab, and unexpected things will happen. In short, there are lots of sources of variation. Some random and some not. And just that variation is enough to generate a predictably unpredictable queue. A chaotic queue.
The Laws of Physics decree it.
So, to illustrate the principles of creating a No Queue design here are some videos of a simulated mass vaccination process.
The process is quite simple – there are three steps that every patient must complete in sequence:
1) Pre-Jab Safety Check – Covid Symptoms + Identity + Clinical Check.
2) The Jab.
3) Post-Jab Safety Check (15 minutes of observation … just-in-case).
And the simplest layout of a sequential process is a linear one with the three steps in sequence.
So, let’s see what happens.
Notice where the queue develops … this tells us that we have a flow design problem. A queue is signpost that points to the cause.
The first step is to create a “balanced load, resilient flow” design.
Hurrah! The upstream queue has disappeared and we finish earlier. The time from starting to finishing is called the makespan and the shorter this is, the more efficient the design.
OK. Let’s scale up and have multiple, parallel, balanced-load lanes running with an upstream FIFO (first-in-first-out) buffer and a round-robin stream allocation policy (the sorting hat in the video). Oh, and can we see some process performance metrics too please.
Good, still no queues. We are making progress. Only problem is our average utilisation is less than 90% and The Accountants won’t be happy with that. Also, the Staff are grumbling that they don’t get rest breaks.
Right, let’s add a Flow Coordinator to help move things along quicker and hit that optimum 100% utilisation target that The Accountants desire.
Oh dear! Adding a Flow Coordinator seems to made queues worse rather than better; and we’ve increased costs so The Accountants will be even less happy. And the Staff are still grumbling because they still don’t get any regular rest breaks. The Flow Coordinator is also grumbling because they are running around like a blue a***d fly. Everyone is complaining now. That was not the intended effect. I wonder what went wrong?
But, to restore peace let’s take out the Flow Coordinator and give the Staff regular rest breaks.
H’mm. We still seem to have queues. Maybe we just have to live with the fact that patients have to queue. So long as The Accountants are happy and the Staff get their breaks then that’s as good as we can expect. Yes?
But … what if … we flex the Flow Coordinator to fill staggered Staff rest breaks and keep the flow moving calmly and smoothly all day without queues?
At last! Everyone is happy. Patients don’t wait. Staff are comfortably busy and also get regular rest breaks. And we actually have the most productive (value for money) design.
This is health care systems engineering (HCSE) in action.
PS. The Flaw of Averages error is a consequence of two widely held and invalid assumptions:
- That time is money. It isn’t. Time costs money but they are not interchangeable.
- That utilisation and efficiency are interchangeable. They aren’t. It is actually often possible to increase efficiency and reduce utilisation at the same time!