Before Autopilot

25 June 2014

I was recently in a meeting with my Chief Executive and Chairman and the CE regaled a story of how pilots used to read crucial information when landing a plane. One could imagine that this would be of a complex nature with multiple variables influencing the planes characteristics. However, what he said was, and I’m no pilot, is that there used to be three sets of cats-eyes on the runway; green, amber and red. Then, depending upon the trajectory of the descent, he would only see one colour, for example if the descent was too shallow he would see the amber colour, if we was descending too fast the red and if the plan was on the perfect trajectory he would see green. The morale to this story, is taking simplicity from complexity. Yes, there are hundreds of different readings taken from different sensors, all showing different numbers and position on gauges. BUT, the red, amber, green (or RAG as we love to call it nowadays) was far more intuitive. The reason? Because it told the pilot what was wrong and how to fix it.

The NHS is awash with data/gauges/dashboards/reports/papers call them what you will, take for example the recent BIG data, but really what is data? ‘We had 500 A&E attendances yesterday’ said the analyst, ‘great, and…….’ said the manager. A number is only part of the story, its only data its what we do with it is important. What the analyst should be really be saying is ‘…which is 15% above predicted levels, which was driven by Ambulance attendances late in the evening by X CCG’. Why should we do this? Because this is taking data and turning into action i.e. here is something interesting and here is where you can focus.

Ideally why should we even focus on a number (well at least to start with)? Some of our reports don’t show numbers at all, they show whether the metric is significant and needs to be worried about. For example, if admissions are within +/- 1 standard deviation it will show as green and as this gets further away from the mean the colours will automatically change. This is used on our daily operational tool looking at metrics on A&E, flow through the emergency floor and further into the Hospital. So that if the issue with breaches is that we started to the day with no beds available in our short stay area we know that this is the area we need to address and we need to keep the conversion rate down, or if there is an increase in our long stay patients do we need to speak to Social Services / Community Providers about inreach into the Hospital?

This gets back to the original idea of the pilot, operational and clinical staff shouldn’t be looking at reports wondering what it is telling them, is 500 attendances a lot? What do we usually have? , the report should show them what is interesting and what they should do about it …………….. and then what if you can automatically alert them to the problem as its arising? No phone calls, no finding out about a problem 4 hours too late. Thats what we do at EKBI, turning data into action. The next step is automated alerting based on a predictive models ; that’s something we’ll talk about in our next blog

Chris Green