Submitted by Aaron Leppin
Decisions on whether to initiate statin therapy for cardiovascular risk reduction should be based on individual patient risk and occur in the context of a shared decision making (SDM) conversation. The Statin Choice Conversation Aid is a web-based tool that incorporates patient variables to calculate and present an individual-level risk. It has been shown in multiple randomized trials to facilitate SDM when used in the clinical encounter.
Despite being freely available and well accepted by patients and clinicians, the Statin Choice tool had not been institutionally adopted and integrated into the clinical work flow at any site prior to 2014. This lack of implementation was and is representative of many SDM interventions which, in routine settings, are often not prioritized. The reasons for this are complex but, at least at some level, result from the competing priorities healthcare systems must address and the often-fixed resources they have to do this work. In this context, it stands to reason that health systems and other settings will be more likely to undertake the work of implementing SDM when it is understood clearly to be low. Unfortunately, in most cases, the work of implementing any individual SDM intervention is poorly understood at the outset. The most effective and efficient strategies for facilitating implementation are often even more ambiguous.
In this study, we sought to address these foundational problems by both characterizing the work of implementing the Statin Choice tool and identifying the most useful strategies for doing this work. Specifically, we recruited 3 health systems in the Mayo Clinic Care Network and carefully observed and tracked their efforts to integrate the tool into their EHR and into routine use across all of primary care over an 18-month period.
We used Normalization Process Theory, an implementation theory that organizes the types of work required to embed new practices, to describe the implementation process at each site. We collected multiple types of data from many sources to track the success (or outcomes) of implementation as well. By carefully examining the things teams did (e.g. the strategies they used) to do the work of implementation and the results of this effort (e.g. the outcomes the work achieved), we were able to identify the most useful strategies for making SDM implementation happen. We were also able to gain a clear understanding of the types and amount of work that would be required.
With this knowledge, we were able to develop a multi-component toolkit that could be provided to other settings to support implementation of the tool. As part of this toolkit, we were also able to provide a brief organizational readiness and context assessment. More clearly, because we had observed the implementation process, we were able to provide an assessment that would guide clinical stakeholders in thinking about the specific things they would need to be able to do (e.g. integrate into the record, train clinicians), the ways in which these things can be done (e.g. workflow examples, training methods), and whether the provided toolkit resources (e.g. EHR code language, implementation team manuals, educational templates) was sufficient support to justify going forward.
Importantly, our study identified several strategies that were judged to be of low value in facilitating implementation. This knowledge was critical to the development of the toolkit and to stakeholders as it allowed us to avoid inclusion of things that will only cause more work for clinical teams with little to no benefit.
The conceptual advancements of our research to the field of implementation science include (1) a theoretical connection between the work that stakeholders do to implement SDM and the outcomes this achieves and (2) an appreciation of the need to develop useful toolkits that can support clinical settings in understanding and doing the work of implementation.
It is not our impression, however, that the toolkit we developed will be necessarily appropriate for other SDM interventions. Rather, we believe our research should be used as a template that can be replicated by other teams in other settings and for other interventions.
The full paper was published in BMC Health Services Research and can be found here. This study was made possible by a CTSA Grant (UL1 TR000135) from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH).