Rob Findlay and Mike Davidge on how your journey to work demonstrates the difference between ‘common cause’ and ‘special cause’

There have been an increasing number of articles and blogs on the use of control charts within the NHS, and they often refer to two different types of variation – “common cause” and “special cause”. But what are these apparently similar-sounding things, and why is the distinction so important?  

Your predictable journey to work

Let’s take an everyday example – your regular journey to work.

If we asked you how long it took, you would give us a pretty good estimate. But it wouldn’t be a single number, like “29 minutes”. Instead it would be a range, like “usually between 20 and 45 minutes.”

How did you estimate it? Well, you have done that journey dozens if not hundreds of times, and know there are lots of factors at play: how many traffic lights are on red; how many cars are ahead of you at that tricky right turn; how easily you find a parking space and many more.

That’s if you drive, and if you use public transport or cycle then other factors will affect your journey in a similar way.

All those different factors combine to give that range of 20 to 45 minutes. You know it’s not worth trying to work out why it took 5 minutes longer today. It takes as long as it takes and you plan accordingly.

So if you have to be at work for 9am, you won’t leave home at 8.30. Although on average you’ll have enough time, leaving then would make you late too often. So you might leave at 8.15 instead, based on your knowledge of the unpredictable process that is your normal journey to work.

’Common cause’ or ‘special cause’?

What we have just described is “common cause” variation – a common set of factors that are part of the process we are measuring. Once we understand this variation, we know what to expect in future under normal circumstances.

Which brings us to the second type of variation – special cause variation – which happens when circumstances are not normal.

There might be an accident on the motorway, or a train strike. Now your journey to work will take considerably longer than “normal”. But this time we can point to the reason why.

And if you have advanced warning of delays on your route, you might employ a Plan B to mitigate the effect, such as leaving at 7am. But you only use Plan B if you need to – you won’t leave home at 7am every day just in case there is an accident.

Variations in the hospital

At work things are trickier, because we need to make decisions about measures that we have less personal experience of. Here we need control charts, which were specifically designed to distinguish between the two types of variation.

We’ll use two examples – from actual trust performance reports – to show how traditional reporting could lead you astray.

In the first, a trust wants Podiatry activity to be within 5 per cent of plan each month. They have fallen outside that range at 94%. This is flagged as “red” just so you don’t miss it.

Podiatry kp is

The temptation would be to go looking for the reason for this poor performance, but a brief look at a control chart tells a different story.

Podiatry graph

The green line shows the average contacts per month over the year. The red lines are the upper and lower control limits and represent the range of common cause variation in this process. Every month falls within these limits, indicating that none deserve special attention.

So although March 2016 is below the plan, it is not “special”. An investigation will not find a root cause because there are none to be found. The danger is that, having gone looking, we might think we have found something and react accordingly – but this will not improve things. If we want to change the common cause variation, we instead need to change the way the service is set up.

In the second example, a trust is monitoring compliance with the Patient Safety Thermometer, a composite measure of four key patient harms. Everything looks rosy, doesn’t it?

Safety thermometer kp is

They are above the national target and improved on the previous month. And this isn’t a flash in the pan – Q2 performance is also above target. Faced with discussing 55 other KPIs, they move on…

Harm free care graph

But if they had studied the control chart, it would have given cause for concern. The green line shows that on average 93.5 per cent of patients report none of the four harms each month. There are no special causes in the chart so this is a stable system showing only common cause variation.

Now look at the lower control limit – it’s well below target at under 90 per cent. So this process can be expected to breach the target in future as part of the usual “common cause” variation.

Armed with this information, the management team should find out why the process is so variable, and act on those sources of variation by redesigning the process.

Mike Davidge is a director with NHS Elect and Rob Findlay is director of Gooroo Ltd.