Coalface data

Information's direction of travel needs to change

There’s a question I’m fond of asking NHS information analysts. It’s a question that’s a bit over-simplistic, and it’s a bit ‘binary’, but it’s this: “Which of your customers get the lion’s share of your time and attention? Is it the top-level senior managers in your organisation, or is it the clinicians and service managers who work at the coalface?”

And the answer is always the same. We devote most of our time and attention to serving the information needs of the top-level people. The coalface barely gets a look-in.

Now this if fine if our main job—our raison d’être—is just keeping the board updated on what’s happening, helping them keep tabs on things, enabling them to monitor steady states. But if we are working in healthcare organisations that are constantly trying to change what they do, trying to improve their outcomes, trying to transform their processes, then I think information’s direction of travel needs to change.

I think we need to prioritise the coalface when it comes to providing and presenting data. I don’t mean that we should necessarily do this at the expense of the boardroom. Don’t get me wrong: those top-level decision-makers do need information; it’s just that the coalface needs it more. And I want to try to explain why.

Here’s an analogy from personal finance. As things stand, we’re like a bank that somehow expects its customers to stay in the black but without making bank statements available to them. That really basic thing most of us take for granted, where we can quickly check what date our salary gets paid into our account and what dates the payments (direct debits, standing orders etc) leave the account: we don’t provide that equivalent service to the NHS coalface.

Here’s something Ben Goldacre posted on Twitter eight months ago:

I'm at a bigwigs meeting to discuss "nudge theory" in health. Many of the projects involve data feedback. But the key ingredient here is not the nudge; it's access to data feedback. The NHS is hopeless at procuring tools to give clinicians actionable data on how they're doing.

‘Data feedback for clinicians’ is a box we should all be ticking as a matter of course. I may not be talking here about precisely the same sort of data that Ben Goldacre was talking about in his tweet, but I’ve had enough conversations with clinicians over the years to realise that they hardly ever get data fed back to them on a regular basis that tells them ‘how they’re doing’. Most of my work is acute hospital based, and I know that charge nurses and ward-based doctors are generally not fed even the most basic data on the activity, length of stay and bed occupancy relating to the patients they care for and the wards they work in. We need to change that.

It’s interesting that Ben Goldacre uses the word ‘tools’ in his tweet. He’s saying we don’t just have to think about the data we have to feed back to clinicians; we also need the right tools to do the feeding back of the data. And he’s right, but there’s a risk of us getting too hung up on which tools we should use. Information professionals nearly always think of ‘tools’ as being digital, screen-based, automated things. Apps, dashboards, drop-down menus. But at the risk of sounding like some old-fashioned Luddite, how about we forget about screens for a minute and how about instead—at least to begin with—we just print out the data onto A4 paper. It might be clunky, but it stands a chance of making the data a bit less intimidating than it already is. Let’s get it right on paper first before we start thinking about more sophisticated, digital tools. Oh, and while we’re on the subject of tools, why not combine the sheet of A4 paper with another ‘tool’ that we number-crunchers frequently overlook: the face-to-face conversation. But—whatever tool we decide to use—let’s resolve to get more information in front of coalface clinicians and service managers in a way that has resonance for them.

But back to bank statements for a minute. If I’m arguing here that issuing bank statements is one of the absolute basic core things a bank needs to do if it wants its customers to ‘behave’, I want to stretch this analogy to make another point. I don’t think we any of us ever really get taught how to read a bank statement, or how to make sense of it, or how to act on the information contained in it. But most of us work all of that out pretty quickly. We realise that it’s not just the amount of money that comes into, and goes out of, our bank account each month that matters, but also when the transactions occur. If we are in the dark about the timing of direct debits and standing orders in relation to the timing of when our salary gets paid in, we’ll end up paying a hefty price for our ignorance, so yes, we learn how to read a bank statement pretty quickly.

But here’s the thing: we all find a way of reading bank statements that works for us. And we achieve this because bank statements contain information that is bespoke and relevant and there are costly consequences if we ignore that information. And that’s how I think coalface-facing data should work: we just need to start by identifying one or two key, easy-to-digest indicators that describe what’s happening at local level, and then feed them back once a week, once a month, however often is appropriate. And we need to have short conversations with the coalface staff about that data, too. People will quickly work out how to extract meaning from the indicators, and we’ll also get a sense of what really matters from the questions they ask us about how they want to tweak or re-design the indictors, or from the extra detail they ask us to drill down to.

I love that the word “actionable” is contained in Ben Goldacre’s tweet. The idea that we might give someone on the coalface data every week, every month whatever, and the knowledge or insight that the clinicians get from that data leads them to take action as a result of it. Either individually or as part of a team effort. That’s what can happen very quickly at coalface level: action can follow decisions in a matter of days or even hours; whereas when information is presented and discussed at board level, the same process of turning it into action can take weeks or months.

A final thought. One of the things that bothers me whenever I attend workshops about whole system issues—and this is particularly the case with the emergency care whole system—is that everybody constantly stresses the ‘whole system complexity’ of it all. And it reminds me of when I learnt about labelling theory when I did my undergraduate degree in Sociology all those years ago. If you keep labelling something as complex, then it eventually becomes complex, not necessarily because it is complex, but instead because of the label you’ve applied to it.

I have a hunch that devolved data, local data, coalface data—call it what you will—might work as an antidote to complexity. What if we just need to provide—regularly—each part of the whole system with the right data so that it can quickly see what it’s doing, and what it needs to do in order to help the overall system work better? What if that’s all we need to do?

[9 March 2018]