The pragmatic, rapid, and iterative dissemination & implementation (PRIDI) cycle: Adapting to the dynamic nature of public health emergencies

Background Public health emergencies – such as the 2020 COVID19 pandemic –accelerate the need for both evidence generation and rapid dissemination and implementation (D&I) of evidence where it is most needed. In this paper, we reflect on how D&I frameworks and methods can be pragmatic (i.e., relevant to real-world context) tools for rapid and iterative planning, implementation, evaluation, and dissemination of evidence to address public health emergencies. The Pragmatic, Rapid, and Iterative D&I (PRIDI) Cycle: The PRIDI Cycle is based on a “double-loop” learning process, reflecting the iterative and adaptive D&I, along with iterative re-consideration of goals and priorities, interventions and corresponding D&I strategies, and needs and capacities of individuals and contexts. Stakeholder engagement is essential- which itself is an evolving activity. The results of iterative evaluations should be communicated with local implementers and stakeholders through customized feedbacks. Conclusion Even when the health system priority is provision of the best care to the individuals in need, and scientists are focused on development of effective diagnostic and therapeutic technologies, planning for D&I is critical. Without a flexible and adapting process of D&I, which is responsive to emerging evidence generation cycles, and is closely connected to stakeholders and target users through engagement and feedback, the interventions to mitigate public health emergencies – such as the COVID19 pandemic - will have limited reach and impact on populations that would most benefit. The PRIDI cycle is intended to provide a pragmatic approach to support planning for D&I throughout the evidence generation process.

There is usually redundancy and parallelism within systems, which positively and negatively affects the implementation of evidence-based processes and practices. On the positive side, we can learn from the experience of other hospitals who deal with similar situations and challenges (e.g. in allocating ventilators and ICU beds). (23) On the negative side, redundancy and parallelism and lack of communication may result in confusion, con icts, dilution of resources, and burn-out, and lack of monitoring and evaluation of what practices are both feasible and have impact.
Additional complexities that need to be addressed include the striking racial/ethnic inequities that have been apparent with respect to COVID19 morbidity and mortality, related in part to embedded systems that create and reinforce structural and interpersonal forms of discrimination and racism. (12,24) To align both the science of D&I with the practice of D&I in real-world settings, it is important to explicate how health systems can apply D&I frameworks and methods rapidly, effectively, equitably, and with few resources to guide local adoption of evidence-based interventions or emerging best practices/protocols (informed by the best available evidence at the time). In this paper, we re ect on widely adopted D&I frameworks and tools and how they should be adapted to address dynamic trajectories of emergency situations.
The Pragmatic, Rapid, And Iterative D&i(pridi) Cycle Figure 1 shows the PRIDI model for D&I. It depicts the dynamic connection between the cyclical process of executing and evaluating D&I (center), the intervention and strategies (left side), the multi-level nature of the context (upper side), and goals and outcomes of D&I (right side). Consistent with recent emphasis on iterative and pragmatic nature of D&I, (25,26) its journey is not a linear process, particularly in the uid and dynamic contexts of emergencies.
This cyclical process of Assess > Plan > Do > Evaluate > Report should be done quickly and iteratively as an intervention and strategies to support it are rolled out. (27) As shown in Fig. 1, we incorporated this cyclical process to the center of the PRIDI model. The cycle of D&I itself activates lateral cycles (shown by the cyclical relation between the middle circle and three arms), which involve revisiting the mental models, goals and outcomes, interventions and D&I strategies, and individuals and contexts, through the course of D&I cycles. It resembles a double-loop learning model. (28,29) This is particularly critical in emergencies, where traditional mental models may not t the emerging problems and contexts. If we apply Plan > Do > Study > Act (PDSA) cycles using existing models i.e. single loop learning, we might fail to learn from the higher order feedback loops that requires more than incremental improvements in e ciency and time.
Second order learning might inform entirely different approaches based on different assumptions and different mental models.
To the extent possible, monitoring and evaluation should be prioritized, and results should be regularly communicated with stakeholders, and meaningfully and consistently incorporated in any re-design or planned adaptations/modi cations within the system. (2) If an intervention or a D&I strategy is not working, it should be modi ed or abandoned (de-implemented) in a timely manner. Evaluations and monitoring may include information that changes the nature of the evidence supporting the effectiveness of the intervention itself or strategies to support its use (see cyclical path from the implementation to effectiveness).
The engagement of stakeholders within these dynamic contexts is critical throughout this process to understand what is working or not and why, where inequities are emerging, and the feasibility and acceptability of the programs and practices. The double loop nature of the process also has implications for engagement of diverse stakeholders in the context of psychological safety where people feel free to express contrarian views thus fostering opportunities to challenge conventional assumptions. For example, suppose we assume that African Americans by virtue of higher SARS-CoV-2 infection rates and worse outcomes will have higher demand for COVID-19 testing and vaccines in development. This would be a reasonable assumption from which we could develop cyclical PDSAs for messaging. This assumption would suggest that nding ways to promote awareness in the African American community regarding where to get tested and where to receive the future vaccine will reduce disparities in infection. Yet, if the African American community were at the table and divergent views were encouraged based on recognition of second order learning, members might express reservations about COVID-19 testing including risk for family separation, forced quarantine with pay, and greater stigma. Similarly, members might voice deep skepticism towards receiving future vaccines including mistrust of government statements, concerns about being guinea pigs for a vaccine that has been rushed to market, and/or concerns about the vaccine containing virus. This second order learning might suggest a fundamentally different approach rather than incremental changes in content, dose or frequency in messages.
In Table 1, we summarized the suggested aspects of D&I activity that should be collected, discussed, and re-evaluated at each round.  (30), the interventions and strategies to facilitate their dissemination, adoption, and use are the key elements of D&I efforts, that are shown in the left side of Fig. 1: The intervention (e.g. evidence-based practice, policy, program, treatment) to be disseminated and implemented (e.g. establishing a call center for followup of patients in home-isolation, setting up an online meeting model for grand rounds, an online screening tool for self-assessment of symptoms, best practices for mental health screening among COVID19 patients/healthcare workers, safety protocols and policies for birthing mothers) (e.g. the 'what') D&I strategies involve the processes, approaches, or interventions that facilitate and enhance the proactive dissemination and implementation of the interventions. Examples include tailored email/online communication for the self-assessment platform, literacy appropriate instructional packages for patients about the COVID19 call center, staff training/education to learn about the new work ow, motivational incentives to enhance staff participation in grand rounds. See Powell et al. (2015) for full ERIC taxonomy of implementation strategies. (31) Interventions generally include a core (the essence or function of the intervention that is responsible for its impact), and an adaptable periphery (that could be modi ed to adapt to various contexts and situations). (32) Ignoring the distinction of these two components may result in rigid interventions that are not exible to survive varying and unprecedented contextual variations and barriers, or that are too complex or costly to be implemented. As such, it is important that a exible approach is taken during the design of D&I activities, and the local implementers are trusted to adapt the intervention to t into their own local contexts, resources, needs, and policies. Consequently, we added Adaptation as an important phase in PRIDI Cycle (Center of Fig. 1).
D&I adaptation models may be useful to help guide planned adaptations (e.g. ADAPT-ITT) (33), to help balance considerations of t and delity. Ideally, the core component of the intervention should be de ned, dynamically updated (as changes are made over time), and communicated; relevant data could be collected through iterative evaluations to understand the impact of both the core elements of the program, any planned adaptations made, as well as evolutions of the program made across its life course. (34) For example, preventive health messages delivered through health organizations such as the Centers for Disease Control and Prevention (CDC) (35) and local and state health authorities typically target broad audiences and are not always adapted to the needs, values, or expectations of vulnerable individuals and communities. The messages may not address the limited behavioral control of the target audience (e.g. in practicing social distancing or staying at home), may not include information about local services and resources, and may not be adapted to the literacy levels of individuals who may be at greatest risk for COVID-19. (36,37) For example, an individual living in a dense, multi-generational household may have di culty following isolation and physical distancing guidelines or lack digital technology to access functional health literacy resources. (38,39) Communities of color, including Black Americans who have experienced striking COVID-19 inequities, are more likely to be exposed to multiple layers of structural racism, including living in buildings and neighborhoods that are more crowded and have poorer infrastructures, irrespective of income, and may feel unsafe using face masks in public. Asian or Asian Americans may face stigma related to the disease due to misinformation about its origins and spread. An individual living in unstable economic conditions who needs access to employment may not be able to self-isolate for the recommended period of time while symptomatic. Therefore, standard messaging should be adapted to the needs, expectations, and capacities of diverse sub-groups and populations to be able to educate or motivate and improve the understanding of COVID-19 and both individual and community responses to it.

Goals & Outcomes
Evaluation is not a one-time post-intervention process in D&I; it is an iterative, ongoing process that can enhance and inform the evolvability of evidence-based interventions and strategies, including their design, adaptation, re nement, and delivery throughout the process of D&I. Consequently, intended goals and outcomes of D&I should ideally be incorporated from the beginning (right side of Fig. 1). In emergency planning, the value of learning from continuous evaluation is even more essential, as the path forward can be more uncertain, the interventions are more experimental and their evidence-base evolving, and the clinical situation and healthcare contexts can change quickly. As such, it may be useful for decision-makers to have a compass to guide them as to whether they are moving in the right direction or need to re-assess and re-design and challenge existing models that might not t with such a dynamic context. RE-AIM provides a systematic conceptual framework to guide the planning, adaptation, and evaluation of the D&I activities, programs, practices or policies. (25,40) An intervention should: Reach the target populations equitably (Did we reach the those who needed the intervention or would bene t the most from it?); be Effective (Did the intervention achieve its goals and impact on health behaviors/outcomes?); be widely Adopted (Did the settings and stakeholders/decision-makers adopt the intervention?); be Implemented (Did the target users or implementers actually use it as it was intended? How was it adapted?); and be Maintained/Sustained (Did the target users continue using it over time and did it continue to have long-term impact?). Importantly, in light of dynamic contexts, (41) RE-AIM can be iteratively applied to track these D&I indicators to help document where inequities and challenges in each of these areas are arising and to inform re nements of adaptations to respond to changing system challenges (e.g. costs, resources), population needs/values, and evolving evidence.

Individuals And Contexts
The upper side of Figure 1 shows the multi-layer and complex nature of contextual factors and their role in determining the success or failure of D&I efforts. It is critical to consciously consider the complexity of personal, inter-personal, organizational, social, economic, policy, community, and cultural contexts at the design phase, and across the continuous process of re-evaluation and adaptations throughout implementation phases. A seemingly useful intervention may fail to implement, since patients may nd it irrelevant to their needs and characteristics, or may face certain nancial and structural/logistical barriers to access and use it, or may not trust the source of the intervention; staff or administrators may nd it burdensome (since many staff who are running these programs are delivering them in addition to their normal workload, they may be overwhelmed or have many competing demands under limited resources); at the organizational level, infrastructure needed to deliver the program may have geographical, demographic, and structural limitations. External environment factors, such as policies, economic challenges, and cultures and social norms are also rapidly changing. For example, adherence to long-term physical/social distancing may vary based on demographics and cultural backgrounds; (43) Country-level and state-level disease mitigation policies may affect the implementation and sustainment of interventions; (44) and wider economic impact of the lockdowns and current mitigation strategies may affect the effectiveness and sustainment implementation of those mitigation strategies (45) (through activation of feedback loops). Many of these barriers are di cult to overcome in emergency situations; however, having the tools to address them may facilitate development of innovative alternative solutions and enhance the reach and impact of evidence-based solutions, particularly with an eye towards health equity.

Stakeholder engagement:
It may seem like an inappropriate time to engage stakeholders in the context of emergency situations. However, even brief engagement of stakeholders has immense bene ts that make it worth considering, at the design phase and through the cyclical process of re-evaluation and re-design. (46) Stakeholders that are actively involved and engaged in the processes of dissemination and implementation may: (47) Feel more invested to help disseminate, implement, and sustain it Are prepared cognitively and operationally and are more committed to execute it May identify setting or culturally speci c barriers that may have been have missed Provide real time feedback on whether strategies are working and inform important re nements or adaptations of interventions and strategies Enhance relevance and t, and may propose innovative solutions that haven't been considered Stakeholder engagement may be applied at different degrees along the spectrum of implementation, depending on the availability of time and resources, and the nature of the intervention and D&I strategies. (48) Even at its lowest degrees (i.e. information provision and consultation) it can facilitate preparedness and elicitation of some feedback that may be critical in the success of D&I efforts. Given its key importance in informing the process of D&I, we showed 'stakeholder engagement' as a circle surrounding all phases of D&I cycle in PRIDI model (Fig. 1).

Leadership
All mentioned processes are only possible under the context of strong organizational commitment, (49) as well as transformational (inspiring and motivating) and transactional (providing contingent rewards) leadership, (50) that have shown to predict implementation success. (51) Organizational leaders can help maximize the t between all aspects of D&I activities (30), make and effectively communicate strategic decisions, and are nimble and ready to change course midway if the iterative evaluations suggest the need for modi cation of goals and strategies. A successful crisis leader (52) should be well-versed with the subject matter (e.g. public health) or consult team members with expertise in the speci c area; should make evidence-based and timely decisions, while constantly collecting data from the environment; should inspire trust and con dence; and should feel responsible for the safety and welfare of the team members. In emergency situations it is very likely that multiple groups try independently to develop solutions, which may result in fragmented efforts and confusion. The leader should develop an effective project management structure as well as an atmosphere in which teams and individuals have means and feel safe to express criticisms and suggest alternative solutions.

Conclusions
In this paper, we re ected on the cyclical model of 'Assess > Plan > Do > Evaluate > Report' (27), RE-AIM/PRISM framework (30), recent advancement of RE-AIM to incorporate equity (25), and to inform rapid implementation (26), and proposed PRIDI model that takes the dynamic nature of problems, interventions, evidence, contexts, and stakeholders into account. D&I in the context of emergency should be a continuous and iterative process. RE-AIM provides a framework for the evaluation of D&I activities, that includes Reach, Effectiveness, Adoption, Implementation, Maintenance. Recent extensions of this model can also inform more explicit consideration of understanding and addressing health equity and equitable implementation over time and in dynamic contexts. (25) Interventions are disseminated and implemented in complex and multi-layer contexts. Overlooking these complexities will hamper the success of the adoption, use, and impact of the intervention.
The cyclical process of D&I informs double loop learning processes, that may result in revisiting mental models, goals and outcomes, interventions and D&I strategies, and individuals and contexts. The results of cyclical evaluations should also be communicated with local implementers and stakeholders through customized, and actionable feedback. Stakeholder engagement is a key solution to understand and address contextual variations and barriers. It is a continuum ranging from informing the stakeholders to co-ownership, and will be critical to addressing some of the striking racial/ethnic and setting inequities evidence for COVID19, including redistribution of decision-making and resources with the community. The iterative process also accommodates for emergent evidence-generation and potential revisions in the evidence base of the interventions that are being implemented.
Even though the health system priority at this moment is provision of best care to the individuals in need and development of effective diagnostic and therapeutic technologies, (2) exible and prospective planning for D&I is also critical. (2, 53) Without planning and tailoring, with the input and partnership of local stakeholders, D&I strategies will never reach target populations that would most bene t from it; will only be accessed and used by socio-demographic groups that face fewer structural barriers to care (hence deepening the equity gap); and will not sustain as intended. While limited organizational readiness and lack of time and resources are challenges to effective D&I plans, emergency response interventions may fail to meet their objectives and waste limited resources if critical D&I considerations are ignored. Development of infrastructures, organizational cultures, trainings, and establishment of processes towards a Rapid-Learning Health System (54,55) and incorporation of D&I as its key component (55) will prepare healthcare systems and organizations to effectively respond to future emergencies. D&I methods and frameworks also need to adapt to the dynamic trajectories and complexity of emergency situations. This paper calls for dynamic and adaptive D&I models that are responsive to rapid and unpredictable nature of emergencies through rapid and iterative cycles, continuous engagement of stakeholders, and incorporating the evolution of goals, interventions, and contexts. We also call for more dynamic and two-way translational dialogue between D&I and evidence generation research.