Science of Improvement

The fundamental concept of improvement is that improvement requires change. If a system is not changed, it can only be expected that the system will continue to achieve the same results. In the words of Paul Batalden, “Every system is perfectly designed to achieve exactly the results that it achieves.” Within this phrase is embedded the central idea underlying modern health care improvement: performance is a characteristic of a system. Therefore, in order to achieve a different level of performance, it is essential to make changes to that system in ways that permit it to produce better results. Poorly designed systems lend themselves to inefficiency and poor quality. QI approaches identify unnecessary, redundant, or missing parts of processes and attempt to improve quality by clarifying and/or simplifying procedures.

An equally important concept is that while improvement requires change, not every change is an improvement. Because not every change is necessarily an improvement, changes must be tested and studied to determine whether the change improves the quality of care. Effective change takes into account how parts of a system are coordinated and link together, rather than focusing on just one part. For example, changes in staff skills and knowledge through training will only yield improvement to the extent that the lack of training was the major cause of poor performance of the system. If problems in processes are not also addressed, then even trained staff will not be able to accomplish their work to the best of their abilities. This concept remains true for other areas besides training, such as measurement, inspection, telling someone what to do, and the investment of resources. Actions in these areas will not create the desired outcome to improve the quality of care unless the overall process of care delivery is also improved.

In health care improvement, teams test changes by applying an improvement or change model. Many improvement models exist; the common feature underlying all improvement models is that an intervention is introduced, and one or more indicators are monitored to see the intervention’s effect on the desired outcome or output. If the intervention yields the desired improvement, it is then instituted as part of the new work process and ramped up to other providers in the organization. If it does not, it is either modified or discarded.

An improvement model that has been used extensively in health care improvement in USAID-assisted countries is that developed by Associates in Process Improvement and is described in The Improvement Guide (Langley et al. 1996). The Model for Improvement incorporates the Shewhart Cycle for Learning and Improvement, otherwise known as the Plan-Do-Study-Act (PDSA) Cycle. In this model, a change believed likely to yield improvement is proposed. However, whether it will yield an improvement or not is a hypothesis that needs to be proved or disproved. A plan is developed for testing the change, the plan is implemented, and the effect of that test is studied to see whether the change did in fact yield the improvement expected. What action is taken next is based on the result of the test.