[MUSIC] This section will provide a brief introduction to understanding variation in processes. In the next few minutes, I'll review what variation in a process means and what it looks like in quality improvement initiatives. In addition, I will review the types of variation and emphasize why it is important to understand the different types of variation as a means of guiding our actions. In summary, this section will be used to briefly review some of these concepts and give you a feel of their relevance. Healthcare processes contain inherent variation, often a mixture of intended and unintended variation. When variation is there for the benefit of the patients, for example, prioritizing patients with breathing related problems in an emergency department. All working with individual patients to understand and meet their specific needs, then this variation is acceptable and desired. In contrast, an intended variation in quality of care results in poor outcomes for patients. The goal of quality improvement is to eliminate unwanted variation as much as one is able to. This leads us to ask, how can we quantify the level of variation in a process over time? Since all improvement takes place over time, analyzing how measures of care, quality change over time is an important way to understand whether changes are resulting in improvement. But how can we tell when the variation that we see in a measure is down to improvement or just the usual behavior of the process? Let's explore these ideas further. Let's consider a measure of unscheduled health care system performance used in several countries for our performance. The percentage of patients attending the emergency department who are discharged, admitted, or transferred in four hours or less. Assisting delivering good quality care results in all, but a small proportion of patients moving through the emergency department within the target time of four hours. The process of delivering emergency care will vary depending on the circumstances. Suppose that several patients with complex scannings arrive at a particular emergency department within a short space of time, delivering appropriate care for these patients may take longer than normal. And this in turn may mean that patients arriving slightly later have to wait longer to be seen. An X ray machine may break down, requiring servicing. These two may cause patients to wait longer to be diagnosed or treated. The combination of many such factors all acting on the process of delivering emergency care results in variation in four hour performance. Imagine you are part of a team seeking to improve four our performance at a medium to large-sized city hospital. The data analytics team produced this chart showing four our performance for all emergency department attendances in two months, May and August. Does this mean the process of delivering timely emergency care has improved between May and August? What if this is what happened in the other months, all these, all these? Looking at these different possible combinations of these time series, it should be clear that simply knowing the four hour performance in two months is not enough to know whether the process itself has improved. In all three possibilities here, performance was better in August than in May. But without the context of what happened before, during and after, we can't say whether this is part of the usual variation in the process, a temporary dip, a trend, or any other pattern. So, a good way to begin to understand variation in a process over time is to plot key measures of interest as time series like this, with the measurement of interest on the vertical axis, and the time on the horizontal axis. This at least gives you some visual idea of the variation present in the boxes. But it does not give you a actual measure of the amount of variation in the process. Nor does it helps us understand whether the variation is usual for the system or down to actions taken by the improvement team. Clearly in our example, this measure is varying over time. The improvement team wants to know whether the process has improved and if so, whether these changes were maintained. How can they tell statistically whether this happened or not? This is what you will learn next, the tools and techniques of statistical process control, which allows us to answer questions like this one to understand variation. You will use R to perform analysis, creating your own charts, and learning how to interpret and act on the results. You should now have a great appreciation of the concept that all health care process have some degree of variation built into them, intentional or unintentional. And that quality improvement seeks to reduce and ultimately eliminate unwanted variation. You have seen why we need statistical tools to help understand variation. And now let's get started learning about these tools. [MUSIC]
