An improvement program should ideally over time sustain various elements, including its activities, community-level partnerships, organizational practices, benefits to its clients, and the salience of the program’s core issue. These are called “sustainability outcomes” by Scheirer and Dearing [31], and reflect the various ways that a program can continue to achieve its intended effects. However, this highlights the question of how a program can position itself to best ensure that these sustainability outcomes can be realized. It has been proposed that sustainability itself is the small set of organizational and contextual factors that build the capacity for maintaining a program over time. That is, sustainability is the ability to maintain programming and its benefits over time despite underlying factors that act to undermine and extinguish the effects of the intervention.

Despite its importance for outcomes, sustainability has received relatively little research attention. Findings and insights regarding improvement and implementation are generally documented and learned through reports of successful improvement projects as presented at conferences and seminars and via published articles in research, policy and practice journals. Because the majority of these present short-term impacts only, evidence regarding long-term maintenance of practice change and improvements is limited. Anecdotal evidence, however, suggests that long-term maintenance is rare and difficult to achieve and nearly impossible to document using traditional evaluation methods. The ultimate value of improvement and implementation science requires sustained, ongoing improvements (and measurements) rather than short-term benefits, and thus requires greater attention to the study of, and support for, institutionalization and maintenance of practice change. Recent publications in the implementation science literature advocate greater attention to sustainability and offer frameworks and guidance for studying and achieving sustainability, and represent important contributions to the growing recognition and need for sustainability research, practice and success [32, 33].

Closely related to sustainability challenges are questions of scale-up and spread. The heterogeneity of practice settings and quality problems limits the direct applicability and likely broad effectiveness of improvement and implementation strategies found to be useful in one or a small number of settings. Implementation researchers who recognize the rarity of spontaneous diffusion and widespread adoption of innovations in medical care and health care delivery often neglect to recognize that implementation strategies shown to be effective in one set of sites are unlikely to spontaneously diffuse to other sites, and that their suitability and likely effectiveness may be limited even if some natural spread occurs [34]. Research on scale-up and spread barriers, processes and strategies is limited in the same manner as research on maintenance and sustainability, and represents another “new frontier” for the fields of improvement and implementation sciences as they endeavor to better support more effective and evidence-driven public health policy and practice goals for quality improvement [35].

Consistent with its foundations in clinical research and clinical research stakeholders’ preference for rigorous, study designs prioritizing internal validity (e.g., randomized controlled trials), many implementation scientists similarly prefer experimental, interventional approaches to evaluate implementation strategies and understand implementation barriers and processes [36]. Although these maximize internal validity and meet standard scientific standards for rigor in study design, they entail considerable compromise in reduced external validity, policy and practice relevance [37, 38]. Investigator-initiated and directed implementation and improvement projects often require special arrangements and time-limited support and resources, including many that are not easily sustained nor replicated following the conclusion of the project. These additional resources and support preclude a valid conclusion that the specific practice change strategy under study is responsible for observed improvements in quality, rather than the special (and non-sustainable) support and resources provided. Observational studies of “natural experiments” and other naturally occurring implementation and improvement processes permit study of implementation and improvement processes and strategies without the complicating addition of artificial constraining circumstances and support. Increased interest in observational study designs that minimize threats to internal validity [39] will facilitate implementation research that balances internal and external validity and is better able to generate relevant and sustainable policy and practice insights.

The theories, tools and methods of implementation science prioritize implementation interventions and strategies and favor summative evaluation research designs and methods to evaluate their effectiveness and impacts. This focus on questions of whether improvement occurred emulates research approaches optimized to evaluate clinical interventions such as drugs and devices, but is less appropriate in situations characterized by high levels of heterogeneity and adaptability. These features of quality problems and solutions are implicitly recognized by improvement science methods and require a focus on the formative processes and mechanisms of improvement–formative evaluation–in addition to a focus on impacts and outcomes. Rapid-cycle improvement approaches are able to accommodate local variations in quality problems and causes, differences in contextual factors such as organizational resources and policies, and other factors that contribute to significant variations in improvement outcomes across sites and over time.