Improving Health Care Quality. Cecilia Fernanda Martinez
Additional discussion on variability in the context of quality improvement can be found in Deming (1986) and Hoerl and Snee (2012).
1.4 Quality Improvement Frameworks
Establishing a quality culture requires an ongoing focus on quality throughout an organization, along with a framework and a set of tools for identifying, making, and maintaining improvements. These frameworks serve as a common approach within an organization, enabling project teams to undertake improvement initiatives in a systematic way using a well‐defined series of steps. A key component of quality improvement is data collection and analysis to assess process performance. Thus, statistical tools play a key role. Nonstatistical tools, such as brainstorming and process mapping, also have an important role in quality improvement initiatives, particularly in the early stages. There are a number of different frameworks that can be adopted; we will provide a brief overview of two of the most commonly used frameworks.
1.4.1 Define–Measure–Analyze–Improve–Control (DMAIC)
The DMAIC framework is a systematic approach to quality improvement applied in Six Sigma programs. The American Society for Quality (ASQ 2019), defines Six Sigma as “a method that provides organizations tools to improve the capability of their business processes.” The DMAIC cycle begins with the Define phase where a team is assembled to develop a project charter that describes the process to be improved and the objectives of the initiative. During this phase, requirements and improvement opportunities are elicited from stakeholders. A clear problem statement is a central part of the project charter. In the Measure phase, a process or value stream map is created to provide stakeholders with a common understanding of how the process operates and serves as the basis for generating improvement ideas. Process performance indicators are also identified, such as delay times and errors. In the Analyze step, the process and associated data are examined to discover potential sources of variation or error. During the Improve phase, process changes that will reduce or eliminate sources of error or variation are developed. Once process changes have demonstrated their effectiveness, they are implemented. Finally, the Control phase puts in place monitoring systems, such as control charts, to ensure that the quality improvements are maintained over time.
The TJR project, parts of which are described in Chapters 12–14, employed the DMAIC framework. Statistical tools including process capability analysis, hypothesis tests, box plots, and dot plots were used in each of the various steps of the process. Insights gained from these tools were critical for the identification of the root cause of the unnecessary process delays. Taken together, in the Improve stage, process root cause countermeasures were brainstormed, solutions designed and evaluated, and pilot testing took place to measure the effectiveness of the solution before its full implementation. In the Control stage, the process elapsed time was monitored in order to maintain the improvements.
1.4.2 Plan–Do–Check–Act (PDCA)
The PDCA framework, synonymous with the Plan–Do–Study–Act framework, is frequently applied to develop and test a quality improvement idea. In the Plan phase, a plan is developed to see if a process change idea will yield a desired improvement. This phase includes developing a problem statement and identifying data to collect to evaluate the change. During the Do phase, the change is implemented as specified in the plan, usually on a small scale. The Check phase evaluates the change using data collected in the Do phase. Finally, in the Act phase, a change that demonstrates significant improvement is deployed as appropriate throughout the organization. If the change does not produce the desired effects, it may be modified and retested or discarded.
As an example, nurses in a hospital wanted to reduce the severity of injuries associated with patient falls. They initiated a PDCA cycle to experiment with fall mats placed next to a patient's bed. They developed a plan to acquire and test the fall mats on a single unit. This change reduced the severity of injuries associated with falls and was adopted on a hospital‐wide basis. PDCA initiatives are often conducted sequentially devising, testing, and deploying a series of process changes.
1.4.3 Choosing a Framework
Often, the DMAIC or the PDCA framework is seen as THE framework for quality improvement. While it is good for an organization to have a framework that they typically employ, there should also be a recognition of other frameworks and tools that should be used, depending on the problem to be addressed. The difficulty of process improvement efforts is not the lack of improvement or analysis approaches but matching the right approach to the problem under study. Figure 1.2 provides a matrix for consideration when deciding how to approach a particular type of problem. Typically, process improvement objectives fall into three main categories: (i) reduce process errors, (ii) reduce processing time or waiting times, and (iii) increase utilization of resources. Likewise, there can be three difficulty levels of problems: (i) too easy, problems with known root cause/solutions, (ii) just right, focused problems with nonobvious solutions, and (iii) too difficult, complex, and large problems with unknown root causes most likely coming from different sources. Projects that attempt to solve category three problems are typically known for trying to solve “world hunger.” This type of project should be narrow‐scoped before attempting any improvement effort. Nevertheless, the improvement methodology should match the problem difficulty level and improvement objective. For example, as shown in Figure 1.2, less difficult projects can be approached with Kaizen. Kaizen is a continuous improvement approach that utilizes short, intensive “events” where dedicated teams work to develop and implement incremental improvements. Lean is the term coined by MIT researchers to describe the way Toyota improved their processes by focusing on value‐added activities to identify waste and thus streamline processes (Roos et al. 1991). Thereby, lean works well for projects with less complex problems and when the primary interest is in minimizing time and reducing wasteful activities. For nonobvious solution projects, more analysis is often required; in particular, Six Sigma/DMAIC is well suited for minimizing errors. Lean Six Sigma lies at the intersection of these two process improvement objectives, and for more complex problems, process methodologies that look into the redesign of products, processes, and sustainability of resources are better suited for systemic problems such as design for Six Sigma (DFSS).
Figure 1.2 Framework‐type of problem matrix.
There are other methodologies used when designing new products such as TRIZ, which is a Russian acronym from “Theory of Inventive Problem Solving,” which is based on universal principles of creativity and invention for the design of innovative solutions to design problems (Altshuller 1999). Last, the concept of robustness is also used when solving complex design problems where the objective is to reduce variability in the performance of a product by making improvements in the product design. While these latter approaches originated in the manufacturing sector, these can also be applied to healthcare by focusing on the process or products used necessary for providing patient care. These quality improvement approaches, however, are beyond the scope of this casebook.
1.5 Statistical Tools for Quality Improvement
The use of data and measurement is key to the quality improvement philosophy. Therefore, data collection and analysis tools play an important role in improvement initiatives. The process of applying statistical tools to a quality improvement initiative begins with collecting data that will address the question posed. Generating pertinent and reliable data forms the basis for analysis that guides process changes. The application of formal methodologies in study, experiment, and survey design help assure that the data collected meets