Integrated Care Systems – Finding hope: scaling up

I am continuing to think about the new Integrated Care Systems developing across health and care in England following their statutory go-live on 1st July, with many challenges facing them common to many large and complex systems, whatever the sector. In the last 1-minute read, I explored how leadership in complex systems needs to and can develop a shared sense of belonging, particularly drawing from learning from working with the Suffolk and North East Essex Integrated Care System. Today, I am thinking about one of the other challenges: how to do things effectively at scale.

The Voltage Drop

Professor John List is chief economist at Walmart and a professor at the University of Chicago. His recent book, “The Voltage Effect: How to Make Good Ideas Great and Great Ideas Scale,” explores how some ideas that initially seem so promising fail to work on a bigger scale. He calls the disappointment of the gap between promising pilots and full-scale rollouts the ‘voltage drop’. I am sure we can all think of examples – local pilots that seem to work well and energise people at improving hospital discharge, tackling educational attainment, or engaging staff. Yet when spread, the spark is lost, and the benefits have gone missing. He believes there are 5 causes of this and proposes how they can be overcome, drawing from Implementation Science.

Is that a real impact: False positives?

We think a positive result in a small experiment or pilot process is something. As we know from clinical research, you need large enough numbers to demonstrate real effect, not just randomness.

So, how do we guard against false positives in change in complex systems? Prof List recommends doing at least 3 independent replications or tests before rollout.

Representing our population?

We may have found an intervention we think works – an online interactive mental health tool. But is it scalable? What if the population we developed and tested it with is atypical, i.e., it was engaged young people with high levels of digital skills and good access to online services? Whereas the rollout population lacks the skills, access, or motivation?

So, make sure your test groups at smaller scales reflect the larger population you aim to reach. Work with a random sample of people. Just because the service may not scale to everyone does not mean it won’t be useful for the segment that can access it – we will need different solutions for the others. A recent NHS example, I think, was the development of pilot diagnostic cancer hubs in urban centres: the model didn’t fit rural areas. A different solution was needed to deliver the same outcomes.

Representing our situation?

A key question is whether the core driver of any success will scale. Is it something unique or specific? That’s also why he says truly great restaurants rarely become chains. Can you really scale culinary genius? I think of some of the great consultancy projects we have had over the years – was it our ‘amazing insights or the unique combination of people and circumstances? How do we know?

However, if your core drivers can remain constant as your scale grows, then you can scale. Automated digital technologies are a good example of this – for example, the basics of Instagram—take a picture, post it, others view it—works as well for one billion of us who use it today as it did for the first 100 when it launched. WHO preoperative checklists might be a good example of scalable simplicity in health and care.

We didn’t see that coming: Unintended consequences.

When designing ideas and initiatives early on, we must anticipate unintended consequences and negative knock-ons. How can we design positive knock-ons? For example, a well-evidenced positive know-on is that all patients in research active organisations benefit, not just those directly involved in research.

Do we know what will happen in our local system if we reduce or increase wait times, or increase the numbers of beds or open alternative access routes? Simulation using models which explore system dynamics and allow for behaviour changes, can be really useful for anticipating consequences – unintended or not. Complex system behaviours are counter intuitive at times.

What does everyone else do? The supply side

When we think scaling, we also need to consider what will happen to the workforce, leadership, costs, and benefits. How will they change? What are upfront investment, infrastructure developments and ongoing operations? The Public Sector tends to focus on the benefit profile of the rollout programme, whereas the private sector usually focuses on costs – and whether there are economies or diseconomies of scale. This means we need to be wary of copying across sectors without checking what we are most interested in.

So, Prof List suggests putting all the messiness of real-world challenges into the original experiments and testing. I worked for a number of years on med tech adoption in the NHS – so often, the apparent research-based costs and benefits in a pure ‘research setting did not match the real-world evaluation and true clinical utility. At times, however, they were unexpectedly better – a positive unintended consequence!

I hope this is some useful food for thought as you think of all those pilots running across your system – and how they might be scaled up to deliver benefits and drive equity.