Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Learning networks (LNs) have demonstrated success as a useful model for building a learning health system, envisioned by the National Academy of Medicine as a system in which innovation and continuous improvement are achieved through stakeholder alignment and in which both known best practices and new knowledge generation are embedded in healthcare delivery processes.1 The results-oriented LN model has become increasingly popular in paediatrics, where there is often a lack of evidence-based best practices. Paediatric LNs aim to improve outcomes and generate new knowledge by using an actor-oriented network structure composed of multiple care sites, a group of varied stakeholders (including patients, families, clinicians, researchers and health system leadership) and use of data for improvement, research and innovation.2 Establishing an effective LN requires intentional design to achieve alignment around a common goal, build standard processes and infrastructure that enable collaboration, and create a shared commons for information exchange.3 This network architecture enables LNs to study variation across sites, test ideas to improve outcomes, identify best practices from these ideas and then enhance efficient dissemination of these best practices across sites.3 4 Using this model, several paediatric LNs have reported significant and sustained improvements in outcomes, including decreased incidence of healthcare acquired conditions,5 increased rates of inflammatory bowel disease remission6 7 and reduction in mortality of infants with high-risk congenital heart disease.8
As emphasised by Britto et al, because not all improvement interventions work equally well, LNs must have methods to test ideas to determine which interventions work best.3 However, despite the importance of this observation, the mechanisms of testing or learning that occur within an LN have not been previously described. Here we describe the learning and improvement (L&I) framework, a proposed framework that organises the types of L&I that occur within LNs as described by Britto et al, ranging from insights gained through one-on-one conversations with other LN teams to network-wide collaborative improvement projects.
L&I framework overview
The L&I framework is shown in figure 1. This framework was developed through our collective experience designing and leading multiple LNs (online supplemental file 1), many of which are centrally supported by the James M. Anderson Center for Health Systems Excellence at Cincinnati Children’s Hospital. Based on these experiences and observations, the L&I framework was designed by two coauthors (KEB and SW) and iteratively adapted through discussions and practical applications among the coauthors as well as other leaders in the National Paediatric Cardiology Quality Improvement Collaborative and Paediatric Acute Care Cardiology Collaborative networks. This framework can be used once an LN is formally established, including specification of mission and vision statements, aims and outcomes.3
All successful LN-based improvement begins with generation of an idea that emanates from shared learning and culminates in measured improvement of outcomes. Ideas for improvement projects may be generated from a variety of sources, including published research, successful single-site improvement projects, identification of variation in practice and outcomes, or expert consensus. These ideas are subsequently shared, developed and disseminated through three separate but related pathways, described briefly and summarised in table 1. More detailed information and additional examples are available in online supplemental file 1. A single idea may ultimately impact LN outcomes through one, two or all three of these pathways, with its course determined by a variety of factors, including characteristics of the idea itself, individual sites’ need for improvement, and LN resources and priorities. Activity at the local level is distinguished from activity at the network level in figure 1, and the ratio of activity between the two levels varies across the three pathways.
LNs may use this framework prospectively to organise their improvement work, prioritise resources, catalogue types of learning that result within their communities and potentially measure the value of LN participation for an individual site. Expected outputs from the successful application of this framework include a self-sustaining pipeline of effective improvement projects and maturation of new LN leaders.
To illustrate how the L&I framework can be used to prospectively plan LN efforts, we will share representative examples from the Paediatric Acute Care Cardiology Collaborative (PAC3), an LN founded in 2014 which has deliberately used the framework to organise and prioritise improvement work for children hospitalised with heart disease.9 In online supplemental file 1, we have provided additional examples from two long-established paediatric LNs: (1) the Children’s Hospitals’ Solutions for Patient Safety Network, which works to eliminate serious harm to both staff and patients in children’s hospitals, including conditions such as surgical site infections, catheter-associated central line infections, pressure injuries and falls resulting in injury,5 and (2) the National Pediatric Cardiology Quality Improvement Collaborative, which aims to improve outcomes and quality of life for infants with a high-risk congenital heart defect.8 10 11
Improvement generated through network-wide projects has the most expansive impact on patient outcomes and has therefore been a major focus of the impact of LNs to date. Given the significant resource investment required for a network-wide project, these projects follow a rigorous structured process for design, implementation and sustainment (table 1) and ideally are driven by strong evidence. Because network-wide projects are anticipated to result in a significant change in an outcome of interest to the LN, an idea for a network-wide project may come from review of LN registry data showing variation in outcomes or practices across LN sites, or alternatively, network-wide projects may represent ideas that previously moved through the other L&I framework pathways.
The network-wide project process requires significant quality improvement, data analysis and project management resources from the central LN team including support for topic-specific learning, regular webinars, aggregate and site-level data reporting (typically using statistical process control), and network-wide communications. Network-wide projects should contribute to scientific knowledge and findings are expected to be published in scientific journals.
Example of network-wide project
PAC3’s first network-wide project focused on reducing the duration of postoperative chest tubes, which are standard and ubiquitous after paediatric cardiac surgery. A small team of experts identified potential variation across sites in chest tube management processes and outcomes which was subsequently verified through prospective data collection at nine sites.12 Through transparent discussion and learning from a model site, PAC3 leaders designed an intervention to reduce chest tube duration. Nine sites implemented the intervention and successfully reduced both postoperative chest tube duration and length of stay.13 Following this accomplishment in the testing and learning phases, 10 additional sites joined the network-wide improvement phase and succeeded in lowering chest tube duration. Importantly, the original nine sites sustained improvements in both chest tube duration and length of stay during this phase.14
Incubator projects are defined as self-organising multisite projects, where LN actors propose, plan and execute the project with limited support from LN staff. In other words, incubator projects are independently driven while remaining LN supported. Incubator projects result from promising ideas based on solid theory and often arise from effective single-site projects, which may be successful at other sites during the incubator phase and potentially evolve into network-wide projects. Incubator projects include at least two sites and may represent a specialised subset of a larger network-wide project (table 1). Effective incubator projects may be considered either for informal spread to all LN sites or for a formal network-wide project.
LN support provided to incubator projects may vary across LNs, as well as across projects within a single LN, due to differences in leader skillsets and available LN resources. Leading an incubator project provides emerging local leaders with valuable experience in designing and leading multisite improvement on a smaller scale than the network-wide projects. Incubator project leaders are expected to share their results regularly with LN leadership and ultimately with the entire LN community. Publishing results in scientific journals is strongly encouraged.
Examples of incubator projects
Using the L&I framework prospectively, PAC3 actively solicited incubator projects to advance its improvement agenda, understanding that some but not all may develop into network-wide improvement projects. In February 2019, the PAC3 Quality Committee selected an incubator project aimed at reducing diuretic administration following cardiac surgery inspired by early evidence of success at one centre that wished to test the intervention’s generalisability. The three participating centres have achieved >50% reduction in postoperative total diuretic usage without an increase in adverse events. Based on early reports from this incubator project, other PAC3 centres began testing changes independently, so PAC3 will share this potential best practice through the share and adapt locally pathway to facilitate knowledge sharing and rapid adoption rather than transition it to a network-wide project.
Share and adapt locally activities encompass a wide variety of interactions among clinical teams and clinical care centres within an LN (table 1). Although this type of multidisciplinary learning has not been emphasised in many LN publications reporting improved outcomes, it is described in the LN literature.3 4 15 Given the sheer volume of these interactions, they likely represent a significant portion of the learning that individual sites derive from LN participation and therefore may have broad impacts on patient outcomes. Indeed, two of the hallmark principles of collaborative improvement, both ‘All teach, all learn’ and ‘Share seamlessly’, refer to this type of learning.
The LN architecture includes multiple share and adapt locally activities which are opportunities designed to facilitate connection and sharing between participants. Learning through the share and adapt locally pathway can occur informally during in-person meetings: a site leader might learn about how another site has addressed a common problem over discussion at lunch and then test these ideas locally. Similarly, poster sessions highlighting individual site improvement projects or research findings provide more structured opportunities for connection and sharing. Once a site team learns about an idea through the LN structure, the team then adapts, tests and ultimately implements it locally. An idea that proves to be effective at a local level might then progress onward to an incubator or network-wide improvement project. Conversely, teams who have not participated in a specific incubator or network-wide improvement project can learn from the final products of these formal projects, including LN presentations, manuscripts or change packages shared through various LN communication channels.
Using the L&I framework and positive feedback from participants, PAC3 established a structure to support and facilitate share and adapt locally learning activities in a variety of ways. Like other LNs, PAC3 has an online commons where individuals can pose questions about any topic related to acute care cardiology. Local practice changes have occurred based on these interactions: one site changed its policy on the use of high-flow nasal cannulas on the acute care cardiology unit based on benchmarking other sites’ policies via a question posed on the commons. Similar to other LNs, PAC3 supports a quality improvement poster session at each of its network-wide meetings. As expected and desired, individual site presentations during these sessions have developed into approved incubator projects. Finally, PAC3 leaders prioritise a social gathering at each of its in-person meetings to facilitate the development of relationships, thus building a community of trusted peers to drive further share and adapt locally learning activities.
We have described the L&I framework, which may be a useful tool to organise and plan activities within an LN. Importantly, the framework recognises and celebrates various types of L&I work, all of which can provide significant benefit to participating sites and ultimately improve patient outcomes. The framework provides a structure to efficiently capture and spread organic learning that occurs in diverse and unrelated settings throughout the LN. Cultivation of all three pathways of L&I should theoretically create a pipeline of effective improvement projects as well as experienced LN leaders. Examples from PAC3 illustrate how LNs can prospectively use the framework to plan activities and thereby guide decisions about LN resource allocation, even at an early stage of development. Ultimately, the pipeline of effective improvement projects, as well as experienced LN leaders, is essential for the long-term vitality and sustainability of an LN. While this prospective framework was developed based on the LN model described by Britto et al,3 it is possible that the concepts of different pathways of learning may be generalisable to LNs with other structures.
To achieve their stated aims of improving patient outcomes, LNs need to identify, disseminate and sustain effective interventions across participating sites on an ongoing basis.2 3 Identification of such projects poses a significant challenge even for mature LNs such as the National Pediatric Cardiology Quality Improvement Collaborative and Solutions for Patient Safety. By intentionally cultivating share and adapt locally learning activities and providing support for, and evaluation of, incubator projects, LN leaders can recognise which interventions are sufficiently generalisable and impactful to warrant a network-wide project. PAC3’s experience emphasises the utility of the L&I framework for young LNs to identify, test and disseminate promising interventions early in the LN maturation process. For a mature LN, it may be helpful to use this framework to understand the results of past work, how to optimally navigate present work, and how to plan for future work. For example, the National Pediatric Cardiology Quality Improvement Collaborative has recently recognised the need to establish a formal mechanism for incubator project proposals to better identify and test effective ideas to subsequently promote to network-wide improvement projects. Of note, because the L&I framework represents a new approach to organising LN activities, the interplay between each of the pathways is still being tested and we continue to learn. It is possible that the three pathways will not be as distinct as outlined in this manuscript or that variation will exist between projects. For example, PAC3 and the National Pediatric Cardiology Quality Improvement Collaborative have learnt that the support for an incubator project depends on both the experience of project leaders and the complexity of the improvement effort.
In addition to identifying effective network-wide improvement projects, LNs need to identify and further develop strong leaders to sustain their organisations.2 Typically, these leaders first enter the LN as individual actors who join local site teams. The L&I framework structure provides awareness as to how the LN achieves improvement and enables individual actors to increasingly engage in LN activities as desired, moving along an engagement spectrum as described by Hartley et al.16 For those actors who will become LN leaders, the L&I framework provides a mechanism to develop LN leadership skills through progressive engagement in leading multisite collaborations. Ideally, as they reach higher levels of engagement, these LN leaders are experienced in leading by influence, codesign with patients and families, quality improvement methodology and strategic planning. As such, the L&I framework may be particularly useful for LNs who have identified opportunities for improvement in the domains of leadership and quality improvement using the LN maturity grid.2 For example, LN leaders might include a review of current incubator projects as part of succession planning to identify promising future LN leaders for further development.
The L&I framework is intended to be high-level, requiring customisation by each LN. If an LN chooses to formally adopt this framework, LN leaders must specify available resources to support each path, criteria for selection of projects and project leaders, appropriate venues for reporting progress, etc. The capacity of an individual site to participate in multiple network-wide projects is an important constraint to keep in mind in addition to the LN leaders’ and staff’s capacity for support. For example, PAC3 determined that it can pursue a maximum of two concurrent network-wide projects due to these constraints. Importantly, in order to continue to feed improvement ideas into the project pipeline, LNs must provide ample opportunities for sites to connect, such as through share and adapt locally activities in an All teach, all learn setting.
Another key issue for LN sustainment is continued engagement and participation by clinical sites. As such, the value of membership in an LN must be clear to individual sites, particularly since LN membership requires local investments in addition to annual participation fees. To demonstrate the return on investment for decision makers, individual site team leaders can use the L&I framework to track and organise how ideas from the LN have been applied in their local environment. For example, a PAC3 local site leader could describe how her team participated in the network-wide project to reduce variation in postoperative chest tube duration and length of stay while also benefitting from share and adapt locally activities to develop a new advanced practice provider onboarding curriculum through shared educational materials. Indeed, from an individual site perspective, given the large volume of available ideas, share and adapt locally learnings may provide the greatest return on investment in LN membership. LN leadership should aim to help site leaders recognise these ideas and their potential value to their local site decision makers by intentionally discussing and celebrating all forms of LN L&I. Lastly, sites may also be motivated to continue participation if their faculty and staff have discovered leadership and academic opportunities within the LN.
The L&I framework was developed based on a few paediatric LNs that intentionally learn from each other, which may limit generalisability to other LNs. This limitation may be somewhat mitigated by the fact that the LNs highlighted in this manuscript focus on different clinical aspects, for example, safety, inpatient care and outpatient care. In addition, most LNs are based on an adapted Institute for Healthcare Improvement Breakthrough Series model17 and therefore operate using similar principles. Another limitation is that LN member centres that inform the L&I framework have been US/Canadian academic medical centres which may limit generalisability to other settings.
We have described a proposed L&I framework that can organise and prioritise activities within an LN to establish a pipeline of effective improvement projects and experienced LN leaders, both of which are essential for sustained success and value. This framework provides structure to efficiently capture and spread organic learning that happens in diverse and unrelated settings throughout the network. In addition, the framework provokes and facilitates learning in a focused area of importance to the network. In the spirit of continuous improvement, we will continue to test, learn and adapt the L&I framework. This framework will also benefit from being tested across multiple LNs.
Patient consent for publication
Contributors KEB and SW designed the learning and improvement framework, drawing on ideas from working with each of the coauthors and networks described in the manuscript. KEB drafted the manuscript and is responsible for the overall content as guarantor. All authors contributed ideas to the draft, critically reviewed the manuscript and approved the draft.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.