Background
Understanding the implementation of effective interventions is critical to promoting their sustained translation into practice-based settings and enhancing their potential impacts on population health. Successful implementation impacts both service and client outcomes [
1,
2]. In combination with understanding the generalizability of an intervention’s reach into the intended population and its adoption and maintenance at the organizational level, implementation is key factor in replicating effective interventions in typical community or clinical settings [
1]. Further, additional implementation outcomes including acceptability, appropriateness, costs, and feasibility can inform the external validity of evidence-based interventions when delivered outside of the research context [
2].
SIP
smartER is a six-month, multi-component, community-based, behavioral intervention designed to reduce the intake of sugar-sweetened beverages (SSBs) among rural, Appalachian adults [
3,
4]. SSBs are non-alcoholic beverages that contain sugar and few other nutrients, such as soda/pop, energy drinks, sweet tea, and fruit drinks. SIP
smartER is grounded in the Theory of Planned Behavior [
5] and health literacy principles [
6]. The intervention design is described in more detail elsewhere [
3]. It is one of only two known interventions targeting SSB intake among adults that have demonstrated significant improvement in adults’ SSB consumption [
3,
4,
7‐
9]. SIP
smartER also is the only SSB-focused intervention included in the National Cancer Institute’s repository of Research-Tested Intervention Programs (RTIPs) [
10]. Translating SIP
smartER into practice-based settings is important given the numerous preventable health conditions associated with excessive SSB intake (e.g., obesity, diabetes, heart disease, cancer, dental caries) [
11‐
15], particularly among populations with low socio-economic status and/or living in rural areas [
16‐
19]. Given SIP
smartER’s demonstrated effectiveness [
4], it is important to explore how it could be disseminated, implemented, and integrated within a system that reaches health disparate communities in Appalachia.
A pilot dissemination and implementation (D&I) trial was collaboratively developed with medical directors and leadership staff from the four local health districts within the Virginia Department of Health (VDH) that service the rural Appalachian counties [
20]. The trial design was grounded in the RE-AIM framework [
1] and the Interactive Systems Framework [
21,
22], and the evaluation was guided by RE-AIM. Specifically, this pilot trial was designed to measure the reach, effectiveness, adoption, and implementation of SIPsmartER when delivered through local health districts. To support the delivery of SIP
smartER, the research team and health department stakeholders developed and applied an implementation strategy [
23] that would build both the general and innovation specific capacity necessary to deliver SIP
smartER. This strategy utilized the key elements of consultation identified by Edmunds (e.g., on-going instruction, self-evaluation, and feedback) [
24].
The purpose of this paper is to describe SIP
smartER’s adoption and implementation when delivered through rural, local health districts [
1,
2]. In addition, determinants of adoption, implementation, and organizational maintenance that align with the Interactive Systems Framework were assessed from the perspective of the delivery agents from the local health districts: (i) acceptability (satisfaction with aspects of the innovation), (ii) appropriateness (perceived fit, relevance, and suitability), and (iii) feasibility (actual fit). This paper specifically focuses on outcomes related to both delivery expectations and the implementation strategy.
Methods
This study is a mixed-methods process evaluation of a pilot type 2 hybrid effectiveness-implementation trial of SIP
smartER [
20]. A type 2 hybrid effectiveness-implementation trial allows for the simultaneous testing of an intervention and an implementation strategy to supports its delivery [
20]. It specifically reports on SIP
smartER’s adoption and implementation at the organizational-level. A complete description of effectiveness outcomes are outside the scope of this manuscript. However, during this trial, significant improvements from baseline to 6-months in SSB intake (− 403(CI = − 528, − 278) kcals/day, (
p < 0.001)), which were comparable to findings from the effectiveness trial [
4]. Also, significant changes (all
p < 0.05) in SSB-related attitudes, perceived behavioral control, behavioral intentions, and media literacy were observed.
Study procedures were approved by the Institutional Review Boards of Virginia Tech, the University of Virginia, and Virginia Department of Health. Written informed consent was obtained from health department staff.
SIPsmartER type 2 hybrid effectiveness-implementation trial
Identification and logistics of health districts and delivery agents
Each of the four southwest Virginia health districts invited to participate in this trial agreed to participate. These four districts serve the same counties and cities as the effectiveness trial [
3]. These areas consistently score among the poorest across Virginia on the Health Opportunity Index [
25], are federally designated as medically-underserved [
26] and have an average rurality status of 6/9 [
27].
Medical directors were asked to identify the staff within their district who would be ideal delivery agents to implement SIPsmartER. The number of SIPsmartER cohorts each district agreed to deliver was determined by budget and power calculations, with each health district expected to deliver two to four cohorts. Budgets were planned collaboratively with medical directors during the grant writing process. Each health district was provided with a sub-contract reflecting the expected percent effort necessary for delivery agents to implement SIPsmartER.
Following awarding of the grant, planning meetings were held between research staff and VDH medical directors and delivery agents to help ensure the compatibility of the approach with the health department delivery system [
22,
28] and to devise a plan for the division of SIP
smartER implementation and research tasks between VDH delivery agents and research staff. This plan, detailed in Table
1, was adjusted to better reflect the needs of the delivery agents following the in-person training, the first implementation strategy activity. Notably, the extent of the role of the delivery agents in (i) the completion of teach-back and missed class calls and (ii) participant engagement activities was reduced. During one of these meetings, delivery agents created specific action plans for recruitment of participants within their districts.
Table 1
Distribution of SIPsmartER delivery tasks between Virginia Department of Health (VDH) staff and the research team
Cohort Recruitment |
Administer screening surveys | VDH | VDH-led; Research team to help as needed |
Call and schedule participants for baseline enrollment | Research team | Research team |
Intervention Delivery |
Teach 3 SIPsmartER lessons | VDH | VDH |
Conduct missed class calls | Researchers & VDH split | Researcher-led; VDH to assist after 1st cohort |
Conduct teach-back calls | Researchers & VDH split | Researcher-led; VDH to assist after 1st cohort |
Reminder calls | Researchers & VDH split | Research team |
Reengagement calls | Research team | Research team |
Track completion/attendance | Researchers & VDH split | Research team |
Manage IVR system | Research team | Research team |
Research Outcome Assessment |
Conduct health assessments | Research team | Research team |
Appointment postcards | Research team | Research team |
SIPsmartER intervention components
SIP
smartER consists of three lessons delivered through group classes, one teach-back call, and eleven interactive voice response (IVR) calls. Participants who did not attend classes had the opportunity to complete the lesson as a missed class phone call. In the classes, participants received instruction on core content necessary to increase motivation and skills to decrease SSBs. During the teach-back call, participant to review content from the first class and complete a personal action plan with a trained research assistant. Through the IVR calls, participants identified their ounces of SSB intake, completed an action plan, and received a motivational message. Intervention activities and materials have been described in detail elsewhere [
3,
10,
29‐
31]. In addition to intervention components, there were specific activities to support participant retention, including re-engagement calls to participants who had not completed two activities in a row [
32].
SIPsmartER delivery timeline
An initial staggered plan of implementation per district was planned with districts starting SIPsmartER cohorts at different times. A goal of 10 or more participants per cohort was set. However, during the trial, districts completed cohorts at similar times. First cohorts were completed between Fall 2016 and Spring 2017 (n = 6) with additional cohorts completed between Summer 2017 and Spring 2018 (n = 6).
Consultee-centered implementation strategy
An implementation strategy that utilized a consultee-centered approach was drafted by the research team based on the principles outlined by Edmunds and colleagues. This approach involves non-hierarchical interactions between a consultant (e.g., researcher) and consultee (e.g., delivery agent) through which the consultant provides guidance to the consultee related to a current work problem that is within the scope of the consultant’s expertise [
24,
33]. Through these interactions, consultees master general skills and build specific skills related to problem-solving implementation barriers and appropriately adapting intervention components. They are also held accountable for delivering the evidence-based program [
33].
Then, to ensure the training was compatible with the health department processes for adopting new interventions, the plan for the implementation strategy was presented to the medical directors and delivery agents for feedback and changes were made based on their feedback. Final implementation strategies addressed three of the four consultation techniques identified in Edmund’s review: on-going instruction, self-evaluation, and feedback [
24]. The expectations for engagement in each of these activities varied by timing of cohort, as delivery agents were expected to complete more implementation strategy activities during their first cohort(s) than subsequent ones.
On-going instruction
Five implementation strategies related to on-going instruction were utilized.
Self-evaluation and feedback
Delivery agents completed a fidelity checklist immediately following the delivery of each lesson and were provided with feedback through lesson observations and post-lesson meetings. A member of the research team observed each delivery agent deliver SIPsmartER the first two times they taught each lesson. During the observations, the researcher completed a fidelity sheet and took field notes. Following observed lessons, the delivery agent(s) and the researcher had a short (< 10 min) audio-taped discussion about the lesson delivery, including highlights of the lesson and areas for improvement. Additional instruction was provided if aspects of the lesson delivery (e.g., execution of activities, inclusion of improper content) needed improvement.
Measures
To assess engagement in and perceptions of implementation strategy activities and actual implementation, data from eight measures were used: (i) cohort recruitment logs, (ii) delivery agent engagement logs, (iii) post-training surveys, (iv) post-training interviews, (v) fidelity checklists, (vi) post-cohort interviews, (vii) post-cohort surveys, and (viii) capacity surveys. These measures allowed for a concurrent mixed methods assessment of SIPsmartER’s implementation and adoption.
Cohort recruitment logs
Logs of health district recruitment activities were maintained as a means of tracking the number of cohorts each district recruited.
Delivery agent engagement log
Logs were maintained to track delivery agents’ fidelity to delivery expectations and implementation strategy activities.
Post-training surveys and interviews
Post-training surveys and interviews were completed after the 2 day in-person training and before each delivery agents’ first cohort. Interviews were audio-recorded. These measures captured information from the delivery agents related to the appropriateness of SIP
smartER and its components within the health district and the delivery agents’ regular job functions. Survey items included question about delivery agents’ confidence to complete delivery expectations and implementation strategies and their perceived feasibility of doing so. Items were measured using 6-point Likert scales. Post-training interviews also assessed the adoption and feasibility of program recruitment. Please see Additional file
1 for these instruments.
Fidelity checklists
Unique fidelity checklists were developed for each of the three lessons. These checklists assessed the degree to which a specific lesson’s activities were completed (none = 0, partial = .5, all = 1) and if the activity was modified (no = 0, yes = 1). There were also sections to enter specific notes about the implementation. These checklists were completed by delivery agents after each delivered lesson and by a researcher after each observed lesson. Please see Additional file
2 for these instruments.
Post-cohort surveys and interviews
Post-cohort surveys and interviews were completed after each delivery agent completed a round of cohorts. Interviews were conducted by a researcher who had limited involvement with delivery agents during intervention delivery activities. These measures captured information from the delivery agents related to the feasibility and acceptability of SIP
smartER, its components, and the implementation strategy. Post-cohort surveys also assessed the fidelity to delivery expectations. Scaled items on the post-cohort survey were measured using 4-point Likert scales. Please see Additional file
3 for these instruments.
Capacity surveys
After completing all their cohorts, delivery agents were asked to complete a survey with open-ended questions. Questions were related to the acceptability and appropriateness of maintaining SIP
smartER in their health district, including the resources they would need to sustain the program. Please see Additional file
4 for this instrument.
Analysis
A concurrent mixed-methods approach was used to analyze data [
34]. Data from each measure were analyzed independently using the methods described below. Then, qualitative and quantitative data measuring the same adoption or implementation indicators were converged to allow findings to be compared across measures.
Quantitative analyses
Frequencies of completing delivery expectations and engaging in implementation strategy activities were calculated. Means and standard deviations were calculated for items on post-training surveys, post-cohort surveys, and fidelity checklists and are presented by district and overall. To make results from post-training and post-cohort surveys more comparable, post-cohort survey scores were transformed from 4-point Likert scale scores to 6-point Likert scores using linear stretch [
35]. An average fidelity score and average activity modification for each delivered lesson was calculated by averaging the fidelity ratings and modification ratings for each lesson activity.
Qualitative analyses
Transcripts of post training and cohort interviews and open-ended questions from capacity surveys were coded using a constant comparative approach by two researchers [
36]. Transcripts and open-ended responses on capacity surveys were first organized into categories that reflected the major delivery expectations (recruitment, lesson delivery, and teach-back and missed class calls), implementation strategy, and sustainability. Content within these categories were reviewed for emerging themes. These themes were reviewed and organized to create codes that were applied to the categories. One researcher applied the codes to the transcripts and surveys while another reviewed coding to ensure the appropriate text was captured. Researchers discussed and resolved any differences. This process was repeated within codes to identify more discrete units as needed [
37].
Discussion
This hybrid effectiveness-implementation trial examined adoption and implementation factors related to SIPsmartER, an evidence-based intervention to reduce sugary beverage consumption in a region experiencing health disparities. Through the use of a multi-component implementation strategy that included a consultee-centered approach, it was demonstrated that SIPsmartER could be adopted and implemented with high fidelity across four rural public health districts and by delivery agents with different roles within the health districts. This study contributes to the ISF literature in that findings demonstrate that the packaging an evidence-based intervention to reduce complexity and use of a facilitation process including consultee-centered training can result in the adoption and high-fidelity implementation of an intervention to address sugary beverage consumption in underserved communities. However, findings also found that the potential for sustainability and broader adoption could be jeopardized by intervention features that were less feasible and required research facilitation due to a lack of compatibility with the health department context and a lack of perceived relative advantage. Specifically, delivery agent-initiated telephone contacts were difficult to consistently implement while the in-person small group sessions were more compatible with health department practices. Similarly, qualitative feedback indicated that agent delivered telephone calls and the recruitment processes may not be as feasible.
While public health systems are beginning to play an increasing role in the implementation of evidenced-based interventions targeting chronic health conditions, little is known about how evidence-based programs can be best implemented and sustained in these systems [
39‐
42]. Related to recruitment, fidelity to the protocol was high in terms of delivery agents being able to recruit 89% of the cohorts. However, findings from post-training and post-cohort interviews indicate low perceptions of feasibility, with most delivery agents clearly expressing their frustration with the recruitment process. Delivery agents employed two broad strategies to recruit participants: (1) surveying of the community through canvassing health department customers or the general community through local events and (2) targeting established groups, including housing and work sites [
43]. The former was the strategy most districts started with and it was much less efficient. Recruiting for adult participants within the community for a program consisting of group classes and having to reach a threshold of approximately ten participants to start a class was not a common practice for any of the districts. The lack of experience and protocols within the districts may have weakened the potential reach of the program. This finding highlights the potential usefulness of systematically collecting and recording key patient health behaviors as a means to efficiently identify patients who would be good candidates for interventions [
44]. It also highlights the need to consider efficient recruitment strategies that could vary based upon community resources [
32].
Delivery agents’ engagement with and perceptions of the consultee-centered implementation strategies demonstrated high fidelity to these activities and reported high perceptions of acceptability, appropriateness, and feasibility. This finding is notable as it provides support for the use of consultation as an implementation strategy in community-based interventions when using professional health district staff as delivery agents. Although consultation is regularly used as an implementation strategy and has potential for use across contexts [
45], its use is most commonly reported within the context of supporting community and clinic-based mental health professionals to implement evidence-based programs [
33]. The strong implementation fidelity may be due to the design of the consultee-centered approach [
24]. Particularly, allowing for a non-hierarchical relationship between the researchers (consultant) and delivery agent (consultee) acknowledged the past training and professional experiences of the delivery agents. Also, being able to adjust the intensity of instruction and feedback activities to reflect the growing skill and content mastery of the delivery agents with the intervention allowed them to still feel well-supported without overburdening them. However, it is important to note, that clear explanations of the purpose of a consultation approach and the specific reasons activities were chosen is needed, as the relative intensity of the strategy made one delivery agent uncomfortable initially. She felt like she was being judged and that the level of support was not needed; yet, after engaging with the implementation strategy, she recognized its purpose was not to judge her but to support the program delivery.
Our findings about the implementation of SIP
smartER in these rural, local health districts reflect previously identified benefits, facilitators, and barriers of implementing evidence-based programs in public health agencies. Related to benefits, findings suggest that by implementing SIP
smartER, their districts added programming that better addressed common risk factors for disease [
46]. Delivery agents demonstrated contrasting views of some agents on the relative advantage of the approach from three districts. Three districts mentioned SIP
smartER was similar to another program they implement (a statewide intervention to reduce sugar-sweetened beverages that includes social marketing and single workshops for children, adolescents, and adults). However, delivery agents from two districts reflected that SIP
smartER was better designed to foster behavior change and could serve as a next step to that program. Future training approaches may want to address how SIP
smartER compares to existing programs to underscore the uniqueness and benefit in helping participants change behavior.
Our findings reflect barriers previously identified related to staff turnover, leadership, and agency structure [
46‐
48]. Staff turnover was particularly noticeable in this D&I pilot as 75% of the districts had changes in medical directors early in the study. The medical directors were key in the promotion and support of SIP
smartER within their districts. This occurrence is important as past research has demonstrated that leadership is a necessary factor to further implement evidence-based programs within health departments [
48‐
50]. Allen and colleagues’ findings suggest that effective leaders within health departments did not just talk about valuing and supporting evidence-based programs but also created an environment that fostered consistent conversation about the evidence-based programs and provided a supportive organizational environment [
48]. Hu and colleagues identified that public health agencies with “high agency leadership” and “supportive workplace” were 2.08 times and 1.74 times more likely to use research evidence in the workplace compared to unsupportive environments [
50]. In this trial, these changes in medical directors in the three districts may have impacted specific leadership actions related to this pilot study, which could have impacted implementation outcomes. However, researchers cannot control staff turnover, so these finding stresses the need to cultivate multiple formal and informal leaders within the health districts (e.g., opinion leaders, internal implementation leaders, and champions) from early on. In doing so, if one leader leaves the organization or the project, others remain to maintain the support and legitimacy of the program and to drive forward the implementation of the intervention [
51].
Agency structures and processes influence the ability and motivation of delivery agents to complete delivery expectations and implementation strategy activities. The impact of agency structure and processes on implementation outcomes was noticeable in this trial. Notably, from interviews and capacity surveys, it was evident that in three of the four districts, delivery agents adhered to work schedules that were within normal business hours. While there was flexibility to adjust schedules and there was specific funding for the delivery agent time during the work day, they did not feel that it was appropriate, acceptable, or feasible both in terms of time and resources for them to make calls to participants during normal work hours and/or to adjust their schedules to accommodate the calls. Therefore, those agents who attempted teach-back and missed class calls did so during normal business hours. Also, as this type and scale of participant recruitment was not common practice in any of the districts, the health districts were lacking both the structure and capacity to efficiently recruit for the program.
The approach to this trial was pragmatic. Planning was guided by the RE-AIM framework with a goal to design an implementation approach that would allow SIP
smartER be able to significantly impact SSB intake, have a broad reach, and be readily adopted, implemented, and sustained in a typical community delivery setting, i.e., local health departments. Additional pragmatic decisions were made to allow for health departments to test out the program (i.e., trialability) [
22,
28]. For example, it was decided from the outset that delivery agents would not play a role in the administration of the IVR call system and later decided they would not manage participant retention activities. These decisions were purposeful in order to allow delivery agents the ability to gain experience with recruitment, lesson delivery, and execution of teach-back and missed class calls.
Taken together, the findings and their implications identify next steps and implications for the translation of SIPsmartER into practice. Potential next steps include working with these four and other rural, local health districts to create systems to identify potential participants and streamline recruitment efforts. Additionally, it will be important to assess the feasibility of health department staff managing the automated call and participant retention portions of the program while also delivering the classes and teach-back and missed class calls. Finally, testing the implementation and effectiveness of the program with different combinations of components (lessons, automated calls, teach-back calls, and missed class calls) would aid in determining the most effective and feasible model.
Limitations
Findings are limited by the small number of health districts and delivery agents included in this study and that the geographic location of the health districts are within one state health department. While this may impact generalizability, it is important to note that our mixed-methods design allowed for a robust analysis and found differences in implementation experiences and perceptions across districts. Also, the districts represent the targeted region – rural Appalachia – for future dissemination.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.