ANP 654 DQ 1/2

1578185 - McGraw-Hill Professional ©

CHAPTER 21 Principles and Models of Quality

Improvement: Plan-Do-Study-Act

Emmanuel S. King, MD, FHM

Jennifer S. Myers, MD, FHM

INTRODUCTION Achieving better health outcomes for patients and populations requires a focus on continuous quality improvement (QI). While physicians pride themselves on being subject matter experts in their focused area of medical practice, such knowledge alone is insufficient to produce fundamental changes in the delivery of health care. Physicians who practice in complex hospital and health care systems must acquire another kind of knowledge in order to develop and execute change.

W. Edwards Deming, an American statistician and professor who is widely credited with improvement in manufacturing in the United States and Japan, has described this knowledge as a “system of profound knowledge” (Figure 21-1). This knowledge is composed of the following items: appreciation for a system, understanding variation, building knowledge, and the human side of change. These concepts are just beginning to be taught to health care professionals and are essential for anyone who wishes to improve the health care delivery system.

1578185 - McGraw-Hill Professional ©

Figure 21-1 Deming’s System of Profound Knowledge. (Reproduced, with permission, from Langley GJ, et al. The Improvement Guide: A Practical Approach to Enhancing Organization Performance, 2nd ed. San Francisco, CA: Jossey-Bass; 2009.)

All hospitalists have witnessed changes that did not result in fundamental improvements within their hospital systems: the computerized order set that was successfully implemented but never revised based on prescribers’ feedback, the paper checklist for medication reconciliation that never gets filled out, or the new rounding system that worked for the first few weeks but then failed to become a standard part of practice due to physician variation or lack of commitment. These are all examples of first- order changes—changes that ultimately returned the system to the normal level of performance. In quality improvement work, individuals must strive for second-order changes, which are changes that truly alter the system and result in a higher level of system performance. Such changes impact how work is done, produce visible, positive differences in results relative to historical norms, and have a lasting impact. Although the model for improvement described below may seem simple, it is actually quite demanding when used properly; and the process is essential to both learning and ultimately changing complex systems.

PLAN-DO-STUDY-ACT AS A TOOL FOR QUALITY IMPROVEMENT

The Plan-Do-Study-Act (PDSA) model is a commonly used method in quality improvement. Shewart and Deming described the model many years ago when they studied quality in other industries. This model first appeared in health care when Berwick described how the tools could be applied using an iterative approach to change. Using a “test-and-learn approach” in which a hypothesis is tested, retested, and refined, the PDSA cycle allows for controlled change experiments on a small scale before expansion to a larger system. The four repetitive steps of PDSA—plan, do, study, and act—are carried out until fundamental improvement, which can be exponentially larger than the original hypothesis, takes place (Figure 21-2).

1578185 - McGraw-Hill Professional ©

Figure 21-2 The Plan-Do-Study-Act Cycle. (Reproduced, with permission, from Langley GJ, et al. The Improvement Guide: A Practical Approach to Enhancing Organization Performance, 2nd ed. San Francisco, CA: Jossey-Bass; 2009.)

PRACTICE POINT

Use a “test-and-learn approach” to solve quality problems. The PDSA—plan, do, study, and act—framework is one popular model to organize your approach to quality improvement work.

PLAN

During the Plan phase, the team generates broad questions, hypotheses, and a data collection plan. It is critically important during this period to define expectations and assign tasks and accountability to every team member. In the planning phase of the PDSA cycle, it is prudent to invest significant time and develop a well-framed question by reviewing related research and local projects and defining meaningful process and outcome measurements. Broad questions at the outset of a PDSA cycle can include “What are we trying to accomplish?” and “What changes can we make that will result in an

1578185 - McGraw-Hill Professional ©

improvement?” The ideal data collection tool answers the question: “How will we know that a change is an improvement?” It is also helpful for the team to generate predictions of the answers to questions early on. This aids in framing the plan more completely, to uncover underlying assumptions or biases before any testing, and to enhance learning in the Study phase by providing a baseline point of comparison.

Teams new to QI frequently will struggle with the question, “How do we measure improvement?” Defining discrete process measures is a good starting point when using PDSA. Process measures are used to assess whether the cycle is being carried out as planned. This is in contrast to outcome measures, which are used to track success or failure and focus on the specific outcome that the team is trying to achieve.

DO

The Do phase in PDSA is a period of active implementation. It involves feedback on the new process from end users and rigorous data collection. An overarching goal of this phase is to capture and document not only compliance with the new process, but also deviations, defects, or barriers in the process. There are always aspects of quality improvement projects that do not go as planned, and flexibility and open-mindedness are critical to maximize learning from improvement. The quality of the Do phase is intimately related to the quality of the Plan phase. A pitfall for many novice QI teams is to give in to the temptation to jump straight to implementing change without spending a significant amount of time planning. A poorly conceptualized improvement plan, an absence of a sound data collection model, or unclear accountabilities can have adverse effects on the implementation or “do” phase of a new initiative.

STUDY

Analysis of available process and outcome metrics and a qualitative appraisal of the process are the key activities in the Study phase. Time should be set aside to perform a critical review of the data collected and compare it to historical data (when available) and baseline predictions. Close attention should be paid to possible defects in any element of the process, including the data collection plan. If such issues are uncovered, the team may need to revise the initial data collection tools and overall plan. Thoughtful review of all trials, even those that were clearly unsuccessful based on metrics, is a critical and valuable process for the team. In fact, the “failures” in a PDSA cycle can yield unanticipated and improved directions. As the Study phase progresses, time should be spent considering if a follow-up PDSA cycle is planned and exactly what elements to include in that cycle.

ACT

The final component in a PDSA cycle is Act. The team should convene for a feedback and action planning session. Frontline workers in the system that is being changed should be included for honest input. A team approach rather than a “top-down” approach facilitates an open review of successes and failures. An action plan that encompasses lessons learned in the first three steps should then be put into motion. During this stage decisions are made about repeating certain test cycles after improvements are made or “spinning off” new test cycles based on the original one.

RAPID CYCLE, CONTINUOUS, AND SEQUENTIAL PDSA

1578185 - McGraw-Hill Professional ©

In its most basic form, the PDSA model described above can be applied to change a single process. However, teams in health care often confront problems that require multiple changes, in parallel or succession, in order for improvement to happen. Caution is advised when initiating several PDSA cycles simultaneously, especially if there are significantly different data collection plans or if the team is inexperienced in QI methods. An alternative is a sequential PDSA model in which one PDSA cycle feeds into the next. This approach, in which teams continually change and refine their processes based on data evaluation and feedback, is called “continuous quality improvement.” Experienced QI teams strive to utilize this approach. Rapid cycle PDSA is a continuous QI process that lends itself well to projects that are focused on relatively small-scale changes. It is typically used by seasoned QI teams who are familiar with the PDSA model and who wish to implement rapid change.

AN EXAMPLE OF PDSA IN ACTION To illustrate the PDSA model for improvement, a real QI project is presented here from start to finish. A hospitalist group sought to implement a new discharge planning toolkit aimed at improving transitions in care through risk assessment at the time of hospital admission. A QI team was formed with representatives from health care professionals involved in the discharge planning process. While their ultimate goal was to reduce unplanned readmissions, their first team goal involved creating a new process to coordinate and request risk-specific interventions from other teams (eg, nurse educators, pharmacists, a nurse for postdischarge follow-up phone calls) for patients deemed “high risk for hospital re-admission or transition in care problems” by a screening tool. In preparation for the project, the QI team also performed a stakeholder analysis, which is a tool that QI teams can use to identify all of the individuals and groups with a “stake” in the process being discussed.

CYCLE 1

PLAN

The initial PDSA cycle involved piloting a readmission risk screening tool. A weekly meeting was convened that included representative users of the tool and assigned specific responsibilities and tasks with due dates to each team member. At baseline, the biggest barriers to overcome were the perception that the new tool was extra work, introducing a paper-based tool in a largely electronic health care environment, and lack of a tight infrastructure tying the requests to existing risk-specific interventions. Based on these concerns, the team reduced the number of interventions on the initial tool. A data collection plan was started and included both quantitative process metrics (eg, compliance rates with the tool, frequency of risk factors identified on the tool) and qualitative data from the users of the tool.

DO

The new tool was piloted for 2 weeks, during which time data was collected and feedback was solicited from the frontline team.

STUDY

1578185 - McGraw-Hill Professional ©

After 2 weeks, the data showed that overall compliance with the tool was moderately high, but that two risk factors, health literacy and depression, had unexpectedly low percentages. On further inquiry, members of the team admitted that when they performed the risk screen, they paused on those two questions and frequently left them blank, concerned that it might take too much time during the admission process and frustrate new users of the tool.

ACT

The team decided to make another edit to the tool before the second PDSA cycle. The health literacy and depression screening questions were removed based on feedback, with a plan to reintroduce them when the tool was more embedded in the hospital admission workflow.

CYCLE 2 PLAN

The second PDSA cycle focused on follow-up data collection with the health literacy and depression screening questions removed from the tool, to test the theory that this would improve compliance. The data collection plan was to track overall compliance with the tool for a 2-week period. In order to isolate any improvement as a result of this one small change, no other changes were made during this time.

DO

The new version of the tool was implemented.

STUDY

Compliance rates significantly increased from moderately high to very high, and the risk factor screening data remained unchanged. Qualitative feedback from frontline users was that the risk screening process was more streamlined and acceptable.

ACT

A brief but successful cycle 2 ended with a plan to add an intervention checklist to the tool in the next phase.

CYCLE 3

PLAN

The goal of cycle 3 was to associate risk-specific interventions (education, follow-up phone calls, and social work interventions) with a patient’s individual risk factor profile. To meet this goal, the team implemented a new version of the tool that included the risk- specific intervention requests and tracked request type and volume. A 2-week cycle was planned with continued weekly meetings during this time.

DO

The team implemented a tool that allowed for interventions to be requested at the time of risk factor screening.

1578185 - McGraw-Hill Professional ©

STUDY

After 2 weeks, the data showed stable high compliance with the form, stable risk factor data, but very low utilization of intervention requests. At feedback meetings, frontline users stated that at the time of admission, they were not ready to place a request for an intervention. They felt that intervention requests should be discussed in a multidisciplinary team on a follow-up hospital day when more information was available.

ACT

Discharge planners on the team suggested that the intervention request process be integrated into daily discharge planning rounds, during which the entire patient care team (physician, nurse practitioner, registered nurse, discharge planners, patient service representative, and social worker) discussed each patient on the service. A nurse practitioner and patient service representative drafted paper forms that could be used to communicate requests for each of the interventions to the appropriate personnel and to document completion of the task. The next phase would trial this new process.

CYCLE 4 PLAN

Cycle 4 was focused on implementing and studying the new discharge rounds process to request risk-specific interventions. The frequency of intervention requests in each category was added to the existing process metrics. Since this was a more substantial change than before and involved more than just one team of frontline users, a 4-week cycle duration was chosen.

DO

Clinicians continued to screen patients using the risk screening tool, intervention request forms were kept on hand during discharge rounds, and the patient service representative and discharge planners prompted the teams to request interventions based on patient risk factors. The requests were forwarded to the appropriate personnel (registered nurse, pharmacist, nurse educator), who then documented completion of the intervention on the form.

STUDY

Compliance rates and risk factor data remained steady, but there was a significant increase in intervention requests in all categories. However, documentation of completion of the intervention was low. It was determined that the documentation requirements were unfamiliar to the intervention teams, which was an oversight.

ACT

For the next cycle further improvements in documentation of intervention completion and a reintroduction of the health literacy and depression screening questions was planned.

LESSONS LEARNED

This example illustrates several important points for successful use of the PDSA model. First, the engagement and involvement of the end users of new QI tools and processes is

1578185 - McGraw-Hill Professional ©

critical to the success of any improvement project. These users are experts in the process who often know what should be tested next, and perfect champions when changes are disseminated on a larger scale. While it may be impossible to address or fix every problem that they identify, hearing their input, implementing changes based on their suggestions, and giving praise for their involvement and patience is an important skill for leaders of QI. Second, flexibility and creative thinking, skills that are used frequently in clinical care, are also essential in QI. In the case study, several barriers were identified such as: concerns about paper forms, perception that certain risk factors would halt the risk screening process, and lack of infrastructure around the systematic documentation of interventions. As these barriers became apparent, the team remained flexible and changed a part of the new process without compromising the integrity and team goals of the project.

PRACTICE POINT Critical to the success of any quality improvement project:

Performing a stakeholder analysis to help identify all individuals and teams that have a “stake” in the quality problem that is being addressed.

Engagement and involvement of the “end” users of new QI tools and processes. Small tests of change that include data collection followed by data analysis and decisions on how to proceed.

ALTERNATIVE MODELS OF QUALITY IMPROVEMENT

In addition to the PDSA model described above, there are other frameworks that have been used to design and execute quality improvement projects. Adopting one specific framework (as opposed to adopting several) allows an organization to learn a common language and approach to improvement. Six Sigma and Lean are two common frameworks that will be briefly described.

Six Sigma was developed by Motorola in the mid-1980s and is focused on reducing variations in a process. Six Sigma is a popular performance improvement methodology which uses a five-phase approach to problem solving, called DMAIC (Define, Measure, Analyze, Improve, and Control). This framework guides users to define their QI goals, measure the current process, analyze root causes of the quality problem, improve the process on the basis of the previous steps, and finally control the process to ensure that variances are corrected before they result in defects and the new process becomes standard work.

Lean manufacturing (or just “Lean”) was adapted from the Toyota Production Systems and is focused on continuously reducing waste in operations and enhancing the value proposition to customers. The Lean approach is based on a few key principles: defining the problem from the customer perspective, identifying the activities required to provide the customer with a product or service, producing the products or services only when needed by customers, and pursuing perfection in the process.

CONCLUSION

1578185 - McGraw-Hill Professional ©

Plan-Do-Study-Act has remained a fundamental tool for continuous quality improvement. Once comfortable applying this iterative app