Hospitals collect large amounts of data related to safety and quality performance. However deriving inferences from these data is not straightforward. One difficulty arises when distinguishing signals from noise; i.e. data arising from special-cause rather than common-cause (chance) variation. Control charts are learning tools that enable decision-makers’ to easily make these distinctions (Carey, 2003).
The present project has at least two goals: (1) Describe how often control charts are already used in by NHS Trusts by means of their publically available board papers and (2) intervene in a trust to promote the use of control charts to improve safety and quality performance there. The paragraphs below describe the advantages of control charts.
In addition to displaying the same information in basic line and bar charts, control charts also include reference lines set two or three standard deviations from the central tendency. Data falling between the lines are likely the result of common-cause and indicate that the process is in-control, while data falling outside the lines are likely the result of special-cause and indicate that the process is out-of-control.
By distinguishing these variations, control charts suggest courses of action to improve total quality performance. Special-cause variations are typically worth investigating to identify and eradicate whatever caused the unsatisfactory performance (or more fully implement whatever caused of the exemplary performance). In contrast, common-cause variations are typically not worth further investigating, because there is nothing special about them to discover. Rather than investigating particular data, to improve performance here one must change the process from which all common-cause variations are derived.
Control charts have proven useful tools to improve variables in emergency medicine; for example, diagnostic test turnaround times, the percentage of patients leaving without being seen and door-to-needle time' in acute myocardial infarction (Thor, 2007).
Surprisingly, even though control charts can support quality improvement initiatives they are underused in NHS trusts. From a random sample of 30 board papers, only 15 contained any control charts. Those that did contain control charts, mostly contained very few (<5). Where control charts were used, such use was often limited to infection or mortality. So it is expected that NHS trust need a helpful nudge to start using control charts more often and for different types of safety and quality performance measures.
In the coming year we are working with NHS trusts to identify and eradicate the barriers to (and promote facilitators of) control chart use, by means of a COM-B model. In our case, a survey derived from this model assesses hospital decision-makers’ capability, opportunity and motivations to use control charts. Our analyses suggest social opportunities are the biggest barrier to control chart use (i.e., people don’t feel socially supported using control charts). In the coming months we will experimentally intervene on a hospital selected performance measure to not only improve that performance measure but also promote control charts’ use in general.