Quality improvement (QI) is a continuous process — not merely a one-time effort, but an ongoing pursuit. If that sounds at all discouraging, consider the alternative: If you do not continue your QI efforts, you risk returning to the status quo, where processes are difficult, costly and frustrating. A key part of QI, then, is learning to hold on to whatever gains you have achieved.

The best way to hold past gains is, very simply, to keep moving forward. In fact, the plan-do-study-act (PDSA) cycle is nothing more than a formalized way of saying “keep moving forward.” In the previous article in this series (“A Team Approach to Quality Improvement,” April 1999), the authors walked through the PDSA cycle as they sought to improve the care of patients with diabetes. In this article, we will explore how you can hold your gains by continuing through that cycle (see “Bringing it full circle”).

Reality check

Perhaps the most important key to holding the gains of QI is harnessing basic human behavior. If your new process or system is to succeed in the long run, individuals must want it to succeed. And most individuals will only want it to succeed if it decreases work, increases efficiency, decreases frustration or improves profits. Achieving one or more of these goals will help you reach the ultimate goal of improving the quality of care. If you've created a new process that is cumbersome or costs money, for example, you will have great difficulty maintaining that process, even if following it does result in better patient outcomes. As you work to hold the gains, then, you must continually check to make sure that whatever changes you have made, or plan to make, will satisfy these basic human needs.

Keeping the momentum

The main purpose of the PDSA framework is to create rapid-cycle change. In other words, you want to maintain momentum and enact useful changes as quickly as possible. At this stage of QI, you've already gone through the PDSA cycle at least once to enact your improvement idea, but practical and useful ideas are bound to surface after the fact. You should implement good ideas as soon as they appear and then check their impact.

When studying the impact of a change, harvest as little data as necessary. Often, you need as few as six data points to arrive at a quick check of an improvement. If the data look promising, keep the change and continue to collect more data. If they don't look promising, modify the change or discard it. Whatever you do, keep the momentum going.

Lead from the front

The most difficult thing about QI is making it continuous. This flies in the face of how we want to practice medicine. Few of us have the time or the energy to improve everything all the time; the prospect can be daunting. For this reason, you need a charismatic champion who will keep the QI effort manageable and continue moving it along.

This champion should be the person most willing to conduct small pilot studies, to tolerate small adjustments and to handle small failures. This person should expect, to some degree, an increase in work, a decrease in efficiency and a (slight) decrease in profit while he or she is working out the bugs in the new system.

This champion must be charismatic in order to keep the other physicians and staff members motivated and must be respected in the community, based on his or her competence in the area under study. Most often, this champion is a member of the same profession as the target audience. A physician, for example, will have little luck being the champion of a change that primarily impacts nurses, just as an administrator will have little luck leading physicians. The implication here is that large health care delivery systems must come to rely on the leadership of a more diverse group of professionals than just administrators. To achieve this, physicians and nurses must be given the training necessary to succeed in these new leadership roles.

The power exchange

A key principle of QI is that variance should be minimized in order to test the effects of a change. The way you control variance is to limit the choices within a process. Decreasing personal choice isn't something individuals support easily, however, so you must provide them with an alternate power to wield.

This alternate power is simply the ability to have some say in the ongoing improvement process. For example, if your practice were trying to improve the care of patients with pneumonia, you would say to your physician colleagues, “We need to agree on one antibiotic regimen that we will all use for outpatients with pneumonia.” As a group, then, you would decide which antibiotic regimen to use based on the literature and best practices in your community, and you would stick to it until you gathered enough data to determine whether you chose the best regimen. If you did not choose the best option, you would choose a new regimen based on updated data. In the end, you will have succeeded in reducing variation, and your colleagues will be able to support you because they had the power to choose which regimen to follow and the power to change their minds based on new data.

Encouraging feedback

Each time a physician (or staff member) does not comply with the new process, he or she must document why. The reason behind this is not to track individual compliance rates but to improve the process, through user feedback, to the point where the practice reaches at least 80 percent compliance. Of course, individuals may be reluctant to report their noncompliance for fear of punishment. Be sure, then, that you stress the importance of their feedback to improving the process, and assure them that the data will not be used to punish them.

In any QI project, no one should be identified as an “outlier” until the system stabilizes and you have approximately 80 percent compliance with the new system. Even then, it is important to realize that establishing a stable system is not the end of the QI process. Once again, you must continue moving forward.

Self-correction

An interesting phenomenon occurs when physicians are shown how their practices vary from their colleagues' practices. Everyone shifts toward the mean. Sometimes, an outlier who actually has a better practice style will shift downward, toward mediocrity. Why? We don't want to be different.

If you throw in outcomes data, however, and show how a particular practice style improves outcomes compared with the standard practice of the community, another phenomenon occurs: Physicians will practically stampede one another trying to find out how the better outcomes were achieved. Practices will now shift not to the mean but to the better practice style. Why? We want to do the best we can for our patients.

By seeking out these best practices, you gain new ideas about how to improve your care process and give physicians a model they can aspire to. For example, you could measure the average HbA_1c levels for physicians in your organization and show that data to the physicians, protecting individuals' identities. If everyone can agree that an HbA_1c of less than 7 percent is desirable (based on the literature), you would then look to see whose patients are closest to this goal on average. Next, you would ask the best-performing physicians to volunteer the details of their practice styles, and you would take those details and craft a simple care process model based on this “best practice.” You would then track compliance and alter the process as needed until you achieve 80 percent compliance.

Data should not be used punitively (except in the most extreme cases, and then only after the physician has been given ample opportunity and support to change his or her practice). The vast majority of physicians will change their practices on their own if given data they can use.

Habit holds the gains

Once you change your practice style, you begin to form new habits. This is unavoidable, and it can be a good thing. Actions done habitually require less effort; they become automatic. If the new process becomes habit for the physicians and nurses, you have held the gains.

This is something of a double-edged sword, though. Once set, habits are hard to change if you later discover, as you may well, that further refinements to your new care process are necessary. As the new process becomes habit, you may even be tempted to stop QI. The natural tendency is to want to shake hands and take a much-deserved rest. This is exactly the point at which the improvement process is at its greatest risk of stagnating. Watch out for this and be prepared. As best as you can, acquire the new habit of continuous improvement.

Assignable opportunity

Once you have built a stable system, have created a data-sharing method that does not evoke fear and have made the new process habitual, what should you do next? The answer is to continue collecting data on your new process and to look for assignable variation in the data — in other words, data that are too good or too bad to be due to the process itself. In either case, they are opportunities to improve. When you discover these new opportunities to make the system better, grab them with both hands and begin the cycle again.

Wag the tail

As physicians' practices converge around the “best practice,” you may find that doctors who tend to hide in a tail of the bell-shaped curve are dramatically uncovered. These individuals who practice truly substandard medicine are rare, but they cannot be ignored. You must carefully decide how you will target them.

The pressure you put on them to improve would be much different, however, than the approach a managed care company might take. An insurer might, for example, simply mandate that doctors decrease their lengths of stay by 10 percent or risk being dropped from its provider panel. Following the QI framework, you would take a more constructive approach. You would say to the outliers that you had sought out the best practice in your organization and had shaped a care process model around it. You would show how the other physicians had shifted toward this best practice, as well as how the true outliers, who cannot improve their practice even in the light of good outcomes data, have not. If physicians must be pressured or reprimanded, this is the fairest context in which to do so.

A QI mantra

At its heart, QI is as simple (and as demanding) as this: In everything you do, work to create processes that offer patients optimal care and service. At the same time, work to decrease irritation, decrease work and increase profits. No fancy gimmicks are necessary to gain buy-in or maintain momentum if you can do these things. And no fancy gimmicks can save a new process if it fails to pass this litmus test.

Editor's note: Dr. Giovino wishes to acknowledge Mary Thoesen Coleman, MD, PhD, Scott Endsley, MD, MSc, Suzanne E. Landis, MD, MPH, John E. Rowe, MD, Miriam Schwarz, MPA, and Kathryn Stewart, MD, for their assistance with this article.