Drive through services may not sound innovative, or much like a care system. However, the drive through clinic that opened in Birmingham that at the end of October 2020 really was, and here’s why.
Big systems – disaster relief, fresh water, home delivery – are central to our existence but demand our most sophisticated skills. You may, however, have experienced the basic idea when you bought a new kitchen. You guessed what you wanted; the sales assistant mapped it onto your floorplan; it didn’t fit, so you changed your choices; then you realise that two of your favourite tasks (ironing and emptying the dishwasher) couldn’t be done at the same time, so you replanned. Eventually someone pressed a button and you were off!
How to think systems
Two features of any system design are prototyping and iteration. You can’t know from the start exactly what you need for any complex system. So instead, you make an informed guess, and test the prototype. Usually, you have to answer two questions:
- Can it do what I think it ought to do?
- Under what conditions will it fail and what unexpected outcomes might it deliver?
The spiral method, developed by software engineers embraces both these elements and a simplified version is shown in figure 1. Time follows the arrows clockwise around an expanding curve, from concept to commissioned service.
Figure 1 A simplified spiral process relying on a series of ever more sophisticated prototypes
You may have encountered PDSA (plan, do, study, act), a commonplace improvement method. While it can be a useful tool, it cannot accommodate wholesale redesign because you cannot test a new service until you have ‘done’ it, and once you have built your new service you are largely committed to it.
It is possible to map a version of PDSA onto figure 1, but in its own terms, figure 1 sets out 4 phases through which a design cycles. Although the same questions are revisited with each sweep, the level of detail and confidence in the design increases until it is ready to launch.
What is a protype?
The activity in the green sector usually starts with the simplest prototypes, perhaps a sketch or a spreadsheet model. To address increased levels of detail, scale models, and especially computer models, eventually a mock-up might be needed to explore specific operational issues. Over the process, one needs to examine: financial flows; information flows; the logistics of getting patients and staff together; the flows of supplies, drugs and blood to the right places just when they are needed.
Sometimes a drawing is enough to prototype a concept by establishing whether connections can be made. An example from the process is shown in figure 2, which provides a logical model of the flows of people and the flows of information. A particular strength of diagrams such as this – especially if posted on a wall, where stakeholders can stand around and discuss them – is to test the idea against a range of requirements and perspectives. If there is a pen handy for people to comment or amend the drawing, so much the better.
Here, stakeholders are not just frontline staff who run the clinic, but the GPs who would be part of the referral chain, and the A&E departments to which patients might be referred, or from which patients might be redirected to the clinic.
However, in this case, it was important also to understand whether the dependencies and timings would work out: for instance, can the system allow every patient to book an appointment in real-time? Clearly, a static drawing cannot fully explore these timing issues, nor can it identify the dynamics that will lead to glitches that throw up empty slots or double booking – or patients arriving with their notes.
Computer modelling was heavily used in these cases, building numerical descriptions of arrivals, departures and consultation times would not lead to overflow, waiting or queues backing into the street.
Figure 2 Prototyping on paper to follow the physical flows of people and the invisible flows of information needed to make a drive through clinical work.
How to build a system
The easiest way of thinking about the design process is by appealing to a basic SPD (study plan do) cycle as shown in Figure 3, where the ‘do’ is first as models and then as the physical implementation. The other complication here, is that because several physical clinics were built and operated, the earlier ones served as the measurement testbed for later ones, alongside models.
In total, 6 clinics were built and operated:
(a) Temporary Rapid Assessment Centre at the NEC (Apr-Jul 2020)
(b) Temporary urgent treatment centre at Slade Road, Erdington (Oct 2020–Oct 2021)
(c) Temporary urgent treatment centre at Hob Moor Road, Bordesley Green (Nov 2020–Jul 2021)
(d) Permanent urgent treatment centre at Erdington Stockland Green (Oct 2020–present)
(e) Permanent urgent treatment centre at Hurst Lane, Castle Bromwich (Nov 2020–present)
(f) Permanent urgent treatment centre at Bourne Road in Aston, (Nov 2021–present)
The temporary structures used tents and portacabins while the permanent structures were based in repurposed warehouses clinical health centre space, or in the case of the last, a warehouse.
Figure 3 A simplified schema for the development of the first clinic, as seen through a study, plan do (SPD) cycle.
Using a simplified SPD cycle as shown in figure 3, the first and final clinics ((a) and (f) above) are relatively easy to follow: a concept was advanced, one or more prototypes were built using computer models, these were studied for viability, capacity and to establish activity dependencies and durations. Through this process, it was established that the design had the capacity to deliver, it would be possible to for people to make their own bookings and to arrive and leave wait-free! The concept was fine-tuned and then implemented. In both cases, as planned.
In fact, there was more data for the finally commissioned drive through clinic (f), because earlier clinics could also be used as test beds to generate measurements of cycle times of patients arriving and leaving, and to capture throughput information.
Clearly, if you want to get up to 500 patients a day through the clinic, you need not only good modelling and good measurement, but to base your modelling on measurements from systems as similar to the one you are working on as possible. This was achieved, by using earlier clinics as test beds, from which to extract, cycle times for typical patient case mixes, in order to inform the next stage of design (figure 4).
Figure 4 Process of designing and commissioning a series of drive to and drive through clinics
The future is springing out of the margins
Not many NHS projects are commissioned in a few weeks, deliver precisely what they promise and can do so sustainably and at scale. One day, all services will be designed as systems and will work in this way.
Until then, good systems are springing up in unlikely places. Given half a chance, they will behave like the sapling in the sidewalk and establish themselves in their own right.
Simon Dodds
Simon Dodds, MA, MS, FRCS, studied medicine and digital systems engineering. His is a vascular surgeon at University Hospitals Birmingham NHS Foundation Trust
Alan MacDonald
Alan MacDonald, BSc (Hons) studied Biomedical Science and is a data analyst with the Badger Group who works on drive through clinics across Birmingham
Fay Wilson
Fay Wilson MBChB, FRCGP, trained in Birmingham and has practiced there as a GP there since 1985.