up

Interpreting the Dashboards

Statistical Process Control Charts

On the Patient Safety Dashboardthe first chart is called a Statistical Process Control (SPC) chart. We plot, for a given trust and metric, monthly percentage of patients values over a 2 year period. We also plot the average for the selected Trust, along with average values for Trusts in Yorkshire & Humber and the whole of England. A key part of the charts are the Control Limits (purple lines). These are derived using principles of Statistical Process Control (SPC) charts.

SPC charts have been used for decades in the manufacturing industry as a way of determining normal (or "common cause”) variation vs. special cause variation. A lot of processes exhibit common cause variation. For example, if we were to look at the proportion of patients who are male in a hospital month by month, we wouldn't expect to get exactly the same proportion every month. However, if 1 month there was a significant spike (or a drop) in the percentage, it could indicate a special cause variation. SPC charts use mathematical rules to determine when a process is exhibiting common cause or special cause variation.

The control limits on the charts show the values beyond which the metric would seem to be exhibiting special cause variation. Therefore, if the control limits suggested an upper limit of ~ 55% of patients being male and the value for that month is 56% there is likely to be a reason for this - statistically there is a 99% probability that this isn't “by chance”. SPC charts are used to help determine when it is worth investigating the reasons behind a specific data point.

There are several rules which can be applied to SPC charts. For example, if there are 8 points in a row above (or below) the average, we would say there has been a step change, and we would calculate a new average and control limts from the beginning of this period:

Other rules also exist, such a 6 points in a row increasing (or decreasing), and too many points "near" to the control limits. We will introduce the option of seeing the step changes in the Patient Safety Dashboard in a future version.

Funnel Charts

SPC charts are useful for comparing activity over time, however when comparing data between Trusts a Funnel Chart is more useful. The chart you can see at the bottom chart in the Patient Safety Dashboard is a Funnel chart. Here we plot the number of patients per month on the x-axis, vs, the value (in this example % of patients who are male) on the y-axis. We then calculate curved control limits, using a similar rationale to SPC charts. If a Trust is outside the control limits it means we are 99% sure that they are different to a typical trust. This could be for a variety of reasons; for example if we were looking at % of patients who had a fall, where a Trust had a value below the lower control limit it could be because they don't treat many elderly patients who are more likely to fall - not necessarily that they are better at preventing falls.

On the chart shown above, the control limits are closer to the average for trusts with a higher numbers of patients. This is because as we get more patients we become surer about the reliability of the metric, so the control limits become narrower. Above, the trust plotted in green at 60% is within the control limits, whereas the trust shown in red at 59% is outside the control limits as there are a higher number of patients here.

Preventing Falls

We are using SPC principles to assess the impact of our work to prevent falls

Michael Rooney

michael.rooney@yhahsn.nhs.uk

01274 383965

@MichaelRooneyUK