Proactively tailoring implementation: the case of shared decision-making for lung cancer screening across the VA New England Healthcare Network

• Shared Decision-Making for Lung Cancer Screening presents a unique implementation challenge; Lung Cancer Screening is the only preventive service for which the Centers for Medicare and Medicaid requires a conversation about benefits and harms for reimbursement.

• Traditional approaches to developing implementation plans may fall short in implementing Shared Decision-Making for Lung Cancer Screening, given the complexity of eliciting and responding to patient values during time-limited appointments.

• Tailored implementation plans improve the fit of implementation efforts to specific contexts. We applied the Framework for Reporting Adaptations and Modifications to Evidence-based Implementation Strategies (FRAME-IS) to proactively guide tailoring of our implementation plan.

Lung cancer is the leading cause of cancer death among both the general population and Veterans [1, 2]. In randomized trials, lung cancer screening (LCS) with low-dose chest computed tomography (CT) has been shown to reduce lung cancer mortality by 20% [3, 4]. LCS carries the potential benefit of saving lives due to early detection and treatment of lung cancer, but there are also potential harms. Harms include distress related to screen-detected findings, radiation exposure, invasive procedures that may cause physical complications, and overdiagnosis [5, 6]. Consequently, the United States Preventive Services Task Force (USPSTF), the American Thoracic Society (ATS), and the Department of Veterans Affairs (VA), among others, recommend Shared Decision-Making (SDM) for LCS [5‐12]. Further, the Centers for Medicare and Medicaid Services (CMS) requires a documented SDM encounter for LCS reimbursement [13, 14]. SDM for LCS is notable; it is the first time CMS has instituted an SDM requirement for a preventive healthcare service [14, 15]. SDM for LCS may confer an additional benefit in supporting adherence to annual screening, as has been shown in other cancer screening contexts [16] though further study is warranted in this area [17].

SDM is a process in which patients and clinicians work together to make medical decisions that align with the patient’s preferences, values, and goals, as well as the clinical evidence [18]. SDM involves information sharing, deliberation, and the resultant decision [19]. USPSTF recommends the conversation include information about the benefits, limitations, and harms of LCS [20].

Guideline-concordant SDM for LCS does not routinely occur in practice [21‐23]. Barriers to SDM for LCS exist at multiple levels of influence. Thus, we selected the socio-ecological model (Fig. 1) to stratify barriers [24]. Innovative approaches tailored to the local context are needed to achieve implementation of SDM for LCS in practice [17]. Our project, Whole Health Approach to Implementing Shared Decision Making for Lung Cancer Screening (WISDOM LCS), aims to increase guideline-concordant LCS, namely, the frequency and quality of SDM for LCS, in the VA New England Healthcare System over the five-year grant period. Notably, VA is transforming to a patient-centered, Whole Health (WH) approach to care, that includes centering clinical care around Veterans’ values and life goals [25]. With WISDOM LCS, we seek to leverage VA’s transformation to the WH approach to implement SDM. WISDOM LCS is organized around the socio-ecological model and includes intervention components at the organization, clinician, clinical encounter, and patient levels to address these multi-level barriers. To create our original implementation plan in the grant-writing stage, we conducted a brief literature review to identify activities that were effective in other settings to overcome known barriers to SDM identified in the literature. We used this information and drew from the “Expert Recommendations for Implementing Change” (ERIC) compilation of implementation strategies to formulate our original implementation plan [26].

As tailoring is important for implementation strategies to be effective, we sought to proactively tailor our implementation plan [27‐29]. As an early step in WISDOM LCS, prior to implementation, we conducted a formative evaluation to understand barriers to SDM for LCS across VA New England. We used the Framework for Reporting Adaptations and Modifications to Evidence-based Implementation Strategies (FRAME-IS), a tool to document implementation plan modifications [30]. Implementation science recognizes the importance of local context to the uptake of evidence-based practices. Much work has been devoted to how to design and document modifications [31‐33], but more attention to the nature and extent of modifications is needed [30, 34]. We sought to understand the local context to proactively tailor our original implementation plan to achieve guideline-concordant LCS. This manuscript offers an emic, “inside look” into the process of collecting data to make planned modifications to an implementation plan.

We used an ethnographic study design to conduct a formative evaluation to identify barriers to SDM and inform the tailoring of our implementation plan. Data consisted of semi-structured qualitative interviews, fieldnotes of team meetings, and document review. The study setting was the VA New England Healthcare System. The local Institutional Review Board (IRB) designated this work as IRB-exempt, thus, this study was approved and overseen by the VA Bedford Research and Development Committee. All methods were carried out in accordance with relevant guidelines and regulations. Informed verbal consent was obtained from all subjects.

We interviewed 15 key people across seven VA New England medical centers, including regional and facility leadership, primary care providers, pulmonologists, a tobacco cessation psychologist, and LCS coordinators. Additionally, we observed one regional team meeting and collected six documents about LCS. These documents included one LCS script, one risk summary document, meeting minutes from one meeting, clinical reminder text from two sites, and an LCS decision aid. The observation and fieldnote data were used to describe the context of each barrier below. An eighth facility did not have an LCS coordinator or LCS program in place; no other participants were identified at this facility.

Our formative evaluation identified barriers to SDM for LCS. We used this information to tailor our WISDOM LCS implementation plan for VA New England. Before commenting on our use of tailoring, however, several of the barriers themselves merit discussion.

We found barriers to SDM for LCS at the organization, clinician, clinical encounter, and patient levels across all facilities. Many of these have been previously identified, especially barriers at the clinician and clinical encounter levels, such as not having time for SDM [22, 42, 43].

Across the VA New England Healthcare System, LCS was not consistently a priority relative to other clinical issues. This made implementing SDM for LCS challenging. This lack of prioritization of LCS could explain why the eighth VA facility, where we did not conduct interviews, lacked an LCS program, despite an offer from regional leadership to fund an LCS coordinator. We noted a tension between SDM and LCS at the organizational level. The current focus on increasing LCS uptake may undermine SDM implementation, though these efforts are not inherently working at cross purposes. Further, uncertainty about the future of the funding for the LCS coordinator role could tip the balance towards wanting to demonstrate strong LCS uptake, which is less complex to measure than SDM. LCS coordinators interviewed described their concern that, if they did not screen enough people for lung cancer, they would not be able to justify the funding for their role. These concerns highlight a larger issue; it is more difficult to routinely measure quality of SDM in comparison to counting the number of LCS orders placed. Services that are more easily measured may be perceived as more valuable.

At the clinician-level, we found a “get the cancer” mentality. There is a strong pro-screening bias in the United States, which clinicians typically share [44]. SDM is predicated on clinicians sharing their medical expertise to inform patients that LCS offers an opportunity to detect lung cancer at an earlier, more treatable stage. However, it is also important that clinicians share the potential harms of screening, such as radiation exposure or overdiagnosis. Thus, it is important that clinicians elicit and consider patients’ values and goals when deciding whether LCS is the right decision for that individual patient. It may be difficult for some clinicians to balance their mental model of their role to keep patients healthy while also respecting patient values and goals, which may sometimes conflict with what the clinician would choose for the patient [45]. Participants alluded to an assumption that SDM would prevent patients from agreeing to LCS, and because clinicians believed so strongly in the benefits of LCS, they did not want to engage in SDM and risk having patients decline screening. It’s difficult to ascertain the impact of SDM on LCS uptake. However, a study by Volk and colleagues found that patients who received a LCS decision aid, which presents both benefits and harms of LCS, were no less likely to be screened for lung cancer than patients who received basic patient education materials [45, 46]. While decision aids strive to present balanced information in a format patients can understand, decision aids are not routinely used to support SDM for LCS in practice, despite multiple studies finding that they increase patient knowledge [47‐50].

Clinicians’ pro-screening bias can be seen in the mission to “educate patients” until they agree to LCS. Framing SDM as purely patient education positions the patient as ignorant and the provider as possessing all necessary information, an orientation rooted in paternalism, as the clinician is sharing information, not power [51]. This framing does not allow patients to bring knowledge of their own lives and bodies into the conversation. Because the clinician is the expert in this scenario, following their recommendation is the “right” decision. This may lead to clinicians continuing to educate patients until they make the “right” decision. A paternalistic approach to decision-making disproportionately impacts people of color and other marginalized groups, given the history of racism and oppression of these groups [52, 53]. Work specifically focusing on empowering Veterans of color to engage in SDM for LCS is the focus of a parallel project by our team, funded by VA’s Office of Health Equity [54, 55].

Clinicians also exhibited bias against patients with a smoking history. Clinicians believed that needing to be screened for cancer is the natural consequence of smoking, and precluded patients from making future medical decisions. Clinicians’ stigmatization of people who smoke or smoked is well documented; fortunately, good patient-provider communication can lessen smoking-related stigma [56‐58]. Smoking-related stigma, along with pro-screening bias, can inhibit high-quality SDM. Clinicians may not spend time discussing LCS if they do not believe there is truly a decision to be made.

We mapped the barriers identified in our data to levels of the socio-ecological model. We then tailored our multi-level plan to implement SDM for LCS in VA New England. Following the Miller et al. framework, FRAME-IS, for recording and reporting modifications to implementation strategies, we described what is being modified and the nature of the modifications. The goal of the modification was twofold: (1) we made changes to increase the acceptability, appropriateness, and feasibility of the implementation effort and (2) we made changes to improve the sustainability of SDM for LCS, such that the practice continues after our project ends.

A strength of our work is that we proactively tailored our implementation plan by conducting a formative evaluation. To our knowledge, this is the first application of the FRAME-IS to an ongoing implementation project. The modifications made were proactive, rather than reactive, and the rapid, pragmatic approach to tailoring the implementation plan allowed us to improve our current work, rather than waiting for a future implementation project to put the changes into effect. For this to be possible, researchers must reserve resources for formative evaluation work and expect that changes to their implementation plan will be necessary.

Our study has limitations. We did not interview patients as part of this formative evaluation; therefore, we only present clinicians’ perceptions of patients. Understanding patients’ experiences of LCS conversations is the focus of our future work as we implement SDM for LCS. Our findings may be unique to VA or VA New England. One distinction between VA and other healthcare systems is the CMS requirement of SDM for LCS for reimbursement, which only affects non-VA settings but not VA. Although SDM is not required for LCS in the VA, it is notable that studies show that SDM is not occurring routinely outside of VA either [23]. Thus projects like ours that seek to implement tailored strategies to promote SDM for LCS are essential both in and outside VA [17].

Many implementation scientists agree that tailoring implementation strategies is a strength and a necessity to carry out successful implementation work [30, 59]. Tailored implementation plans improve the fit of an implementation effort to a specific context and audience, increasing their success. As such, we anticipate our tailored implementation plan will be more successful than it otherwise would have been without our formative evaluation and subsequent tailoring. Further, planned modification, in comparison to reactive modification, can be a stronger and more thoughtful approach to implementation efforts, highlighting the importance of formative evaluation work in implementation science.

Our formative evaluation using an ethnographic approach, which informed modifications to the implementation, can be used across implementation science studies. This work demonstrates how FRAME-IS can be used to guide and document modifications to reshape an implementation plan based on contextual barriers identified during a formative evaluation. Our future work will focus on measuring the outcomes of the implementation effort, namely, the frequency and patient-perceived quality of SDM for LCS, to evaluate this implementation plan. Tailoring the implementation plan to local contextual barriers identified through ethnographic data collection methods, guided by application of the FRAME-IS to our formative evaluation, represents an advance both for implementation science methods in general as well as implementation of SDM specifically.