Monitoring change in health care through statistical process control methods

A Nuffield Trust Research report .

The ability to detect real change in the way care is being delivered will be critical over the next few years as the NHS faces probably its greatest financial challenge. Using information in the right way will be especially important if managers and policy-makers are to make the right decisions about the effects that new models of care are having.


The family of approaches known as statistical process control (SPC) have been widely used for monitoring outcomes in industry and have gained acceptance in many health care settings. However, they are less commonly applied to look at population-level changes across organisations – the changes we now see emerging in new models of care. SPC methods can be applied to looking at some of the measures (such as changes in emergency admissions) that are commonly used to monitor major programmes of organisational change, as seen in new models of integrated care or the Vanguards.

This report describes three forms of SPC:

  • Shewhart or Runs charts
  • Variable life-adjusted displays (VLAD)
  • Cumulative sum charts (CUSUM)

The key value of these methods is their ability to distinguish real changes from random noise seen in all measures. This means that they are more likely to detect genuine changes earlier – and, conversely, more likely to detect the absence of change earlier.

The methods described in this research report are relatively uncommon as a means for monitoring change across health systems. However, realising the potential of these methods may not always be plain-sailing and the report describes a number of potential barriers in terms of access to the data, skills and a willingness to use analytical methods that are more complex than normal.