Achieving the best health care outcomes is constrained by systems' ability to reliably deliver evidence-based care, every time, to the people who need it. Systems include not only the design of care models and care delivery processes, but also supporting structures and the human interactions in between. These human interactions at every level put improving healthcare squarely in the arena of complex, adaptive systems.

The field of improvement has evolved to address the complex, systemic nature of health care challenges, particularly through the use of adaptive, iterative testing and implementation of changes which empower teams to use data in real-time to test changes which address systems issues and lead to improved outcomes. These teams engage all the key stakeholders involved in the outcome of interest, and can include various cadres of healthcare workers as well as patients, families and communities. For several decades the application of modern improvement methods has evolved from working to improve simple processes in facilities, for example to reduce waiting times, to clinical improvements like reducing the incidence of hospital acquired infections, to more significant improvements in healthcare, like decreases in mortality, and achieving such results at scale.

Given the adaptive, iterative nature of improvement, the method of learning about the improvement should not undermine these very features which make improvement activities successful. For the most part, improvement methods have relied on time series charts, with or without statistical process control. Other designs include: stepped wedge trials, time series analyses with or without control groups and case reports. The use of randomized controlled trials, the gold standard for clinical research, has been limited in the field of improvement, with the fixed conceptualization of the improvement intervention, donor constraints and ethical issues regarding randomization, particularly for those in low- and middle-income country settings, often cited. Nuanced review of these methods and matching them to the aim of the inquiry is needed, as well as efforts to develop the epistemology of the field of improvement.

A key question being asked by those in the field is how we know whether the results achieved can be attributed to the intervention conducted. This question underlies one line of inquiry, however, there are many others which would contribute to our learning about improvement. For each, the choice of methods depends on the aim of the inquiry. For example, increasing knowledge on the contextual factors which underlie the "how" or "why" a successful result was achieved is a very different line of inquiry, requiring different methods, than asking to what degree a result is attributable to an improvement intervention. These are legitimate and welcome questions, as it indicates the field has evolved and grown, and it now poses the next challenge in the evolution of the science of improvement.

The value of such a result will be to help in the design, implementation and evaluation of improvement to increase the validity of the conclusions and the attribution of the results to the activities implemented. This in turn allows us to understand which activities under which conditions are most effective at achieving sustained results in health outcomes.