Engaging diverse midlife and older adults in a multilevel participatory physical activity intervention: evaluating impacts using Ripple Effects Mapping

Abstract

Multilevel interventions are increasingly recommended to increase physical activity (PA) but can present evaluation challenges. Participatory qualitative evaluation methods can complement standard quantitative methods by identifying participant-centered outcomes and potential mechanisms of individual and community-level change. We assessed the feasibility and utility of Ripple Effects Mapping (REM), a novel qualitative method, within the context of a multi-level cluster randomized trial, Steps for Change. Housing sites with ethnically diverse, low-income aging adults were randomized to a PA behavioral intervention alone or in combination with a citizen science-based intervention (Our Voice) for promoting PA-supportive neighborhoods. Four REM sessions were conducted after 12 months of intervention and involved six housing sites (n = 35 participants) stratified by intervention arm. Interviews (n = 5) were also conducted with housing site staff. Sessions leaders engaged participants in visually mapping intended and unintended outcomes of intervention participation and participant-driven solutions to reported challenges. Maps were analyzed using Excel and Xmind 8 Pro and data were classified according to the socio-ecological model. Eight themes were identified for outcomes, challenges, and solutions. Most themes (6/8) were similar across intervention arms, including increasing PA and PA tracking, improving health outcomes, and increasing social connectedness. Groups (n = 2) engaged in Our Voice additionally identified increased community knowledge and activities directly impacting local environmental change (e.g., pedestrian infrastructure changes). Housing staff interviews revealed additional information to enhance future intervention recruitment, sustainability, and implementation. Such qualitative methodologies can aid in evaluating multi-level, multi-component interventions and inform future intervention optimization, implementation, and dissemination.

This first-generation study explored the feasibility and value of applying a novel participatory evaluation method, called Ripple Effects Mapping, within a behavioral lifestyle intervention trial to assess participant-centered outcomes, inform future intervention optimization and implementation, and complement more traditional researcher-driven evaluation measures.

Graphical Abstract

Graphical Abstract.

Implications.

Practice: Systematically assessing participant-centered outcomes, challenges and participant-driven solutions should be incorporated to help capture areas of importance and meaning for participants, helping to strengthen interventions and inform future implementation and dissemination.

Policy: Policymakers and other decision makers should more closely partner with citizen scientists to leverage their data and locally relevant solutions to support broader level impacts, including build environment changes that promote healthy and thriving communities.

Research: Future research should continue to apply Ripple Effects Mapping to the evaluation of multi-level or complex behavioral interventions, to assess mechanisms of change, and further determine its potential applications within the health promotion field.

Introduction

Although behavioral lifestyle interventions have been shown to effectively promote physical activity (PA) [1, 2], most focus almost solely on individual-level factors (e.g., increasing aerobic or strength exercises). Few interventions address the built or social environmental contexts that play a key role in PA participation [3, 4] and few studies have examined the impacts of multi-level, multi-component interventions among low-income, racially and ethnically diverse populations.

The effectiveness of lifestyle interventions are often evaluated via randomized controlled trials (RCTs) and other investigator-driven quantitative methods [5, 6]. These methods, however, often fail to assess unexpected outcomes of importance to participants as well as complex mechanisms of change, which can hamper future intervention implementation and dissemination. Complementary, participant-centered evaluation approaches can contribute additional critical insights, especially for multi-level or multi-component behavioral interventions [7, 8].

The Steps for Change trial [9] assessed the effectiveness of an individually-focused PA intervention when combined with an evidence-based community engagement approach for improving local environments to promote PA. In this cluster randomized trial, public housing sites for low-income and ethnically diverse midlife and older adults were randomized to receive the evidence-supported Active Living Every Day PA intervention (ALED) [10, 11] alone or in combination with Our Voice, a novel citizen science approach in which community members collect data on barriers to and facilitators of PA in their neighborhoods and use these data to advocate for feasible and locally-relevant changes to promote PA [12, 13].

Ripple Effects Mapping (REM) is a participatory evaluation approach [14] that is ideally suited to complement the quantitative approaches of multi-level and multi-component interventions, such as the Steps for Change trial. Grounded in qualitative inquiry, REM relies on structured participatory discussions with key stakeholders (e.g., participants, interventionists, community partners) to reflect on intervention challenges, assess intervention outcomes, and inform intervention translation, sustainability, and implementation. This process can also identify outcomes of interest to diverse stakeholders, which may facilitate the adoption and scaling of policies and programs. Originally developed within Cooperative Extension programs, REM has not been traditionally applied in the health promotion and behavior fields [14].

This first-generation investigation applied the innovative REM evaluation methodology with diverse midlife and older adults within the Steps for Change interventions’ to better understand its utility within the context of multi-level, multi-component behavioral lifestyle interventions. We anticipated that the REM would uncover novel targets for future intervention optimization, including expected and unexpected outcomes from the perspective of the ­participants, and elucidate potential mechanisms of individual- and ­community-level change. Drawing on the socio-ecological model, we also evaluated the interplay among various intervention components and relevant multi-level outcomes. A secondary goal was to introduce this methodology and highlight its potential to behavioral medicine audiences.

Methods

We conducted four REM sessions, two per study arm, engaging participants across six housing sites who participated in the NIH-funded Steps for Change cluster randomized trial, approved by the [institution] IRB.

Steps for change trial

In Steps for Change, 10 affordable senior housing sites for low-income racially/ethnically diverse aging adults in two Bay Area counties were randomized to receive Active Living Every Day (ALED) in combination with either supplemental health education or Our Voice. Housing sites represented geographically distinct neighborhoods. Full methodological details for Steps for Change were published previously [9]. Inclusion criteria were: 1) individuals ages 40 or older, 2) willing and physically able to increase walking time in their neighborhoods, 2) fluent in English and/or Spanish, 3) living in or near a designated affordable housing site, and 4) planning to reside in the area for the next 24 months. The major outcome was mean change in weekly minutes of walking.

Physical activity ntervention: active living every day (ALED)

The evidence-supported ALED intervention aimed to improve PA by increasing personal self-efficacy, social support for PA, and knowledge of relevant local PA resources such as recreational facilities, neighborhood walking routes, and local parks [10, 15–17]. ALED was delivered by trained research staff to both Steps for Change intervention arms in 12 semi-monthly group sessions over a 6-month period. In this two-stage intervention, both arms received the ALED intervention during the initial 6-month period. During the second stage, the control arm received the an evidence-supported “healthy aging” education module [10, 11] for an additional 6 months, for a total of 12-month intervention period. The ALED arm + health education (control) arm will hereafter be referred to as PA + HE. In this second, 6-month period or stage, the experimental study arm received the Our Voice intervention described below.

Our Voice

The Our Voice method engages residents in documenting and improving features of their local communities that promote or hinder healthy living. Using an accessible mobile app (“The Discovery Tool” [18]), residents become “citizen scientists,” gathering geo-tagged photographs, audio/text narratives that reflect their lived experience. The citizen scientists then review and analyze their collective data, identify and prioritize opportunities for change, and share their data-informed recommendations with local stakeholders and decision makers to activate realistic, potentially sustainable neighborhood-level improvements [12, 13]. For sites randomized to the PA plus Our Voice arm, hereafter be referred to as PA+Our Voice, the Our Voice activities were introduced around month 5 of the PA intervention, when environmental factors and their contribution to PA activity start being discussed. Our Voice activities also lasted 6-months.

Ripple effects mapping

REM combines qualitative methods, such as individual and group interviews, with mind mapping, a graphical way of displaying information that facilitates idea generation, reflection, and synthesis around a particular topic being discussed [14, 19]. REM session leaders “map”—i.e., visually displays of the information gathered—participants’ responses in real time and group similar responses together in emerging categories. Participants can see the developing “map” and further elaborate on key sections that are most relevant to them. Session leaders can probe further on specific “ripples,” or impacts, for a better understanding of underlying change mechanisms. Prompts such as “How did that [outcome] happen?” “What led to this happening?” “What happened as a result?” can help in understanding the sequence of events and exploring antecedents and factors facilitating a range of outcomes.

Procedures and setting

Recruitment for REM in-person sessions (which occurred before COVID-19 lockdowns) consisted of in-person, email, and text communications to 83 Steps for Change study participants across six housing sites that had completed at least 12 months of the intervention. While 10 sites were originally randomized, due to phased recruitment and intervention start, not all sites had completed 12 months of intervention at this time. Invitees from eligible sites were encouraged to also invite someone (e.g., spouse) who supported them during the PA intervention for an additional stakeholder perspective. Participants were assigned to one of four REM sessions, facilitated similar to a focus group, some of which combined neighboring housing sites assigned to the same intervention arm to facilitate logistics and ensure enough attendees. Research staff involved in interventions provided additional stakeholder perspectives. Sessions, held at the housing sites, lasted approximately 2 hours.

Sessions consisted of two segments. First, participants paired up for what is called in REM “appreciative inquiry” interviews [14], a strength-based approach focused on: (1) highlights or successes related to study involvement; (2) unexpected things that happened as a result of participation; (3) social connections made (new or strengthened); and (4) challenges and how they overcame them, as appropriate. In pairs, study participants asked each other about each area and recorded the responses. Nonstudy participants (i.e., stakeholders) discussed observations across the same four areas.

During the second segment, participants shared their appreciative inquiry interview responses. Real-time mapping of the responses began at this point, using post-it notes and Xmind—a “mind mapping” software [20]—on a projector screen. Using Xmind, REM session leaders would type participant responses for all to see on the screen. Similar responses would be grouped together and collectively given a descriptive label during the session, for example “challenges.” Participants could visually see the developing map (similar responses grouped and labeled, see Fig. 1, panel A for simplified example) and modify the content, if needed. Prompts were used to get additional details on specific responses, known as “rippling.” REM session leaders grouped responses into emerging themes and participants offered feedback on the themes to ensure proper documentation of their experience. After the session, session leaders refined the map and returned it to session participants for verification and data corroboration. Participants received the map via email. In addition, study interventionists utilized a group session to also present the map to participants for any additional feedback.

Fig. 1.

Uncovered similarities and differences across intervention arms. Panel A. Simplified ripple effect map showing similarities and differences across intervention arms. Note. Adapted from original ripple effect maps. White indicates themes found regardless of intervention arm. Gray indicates themes found only in groups assigned to the Our Voice intervention. Panel B. Frequencies of outcomes for each level of the Socio-Ecological Model.

Individual interviews. Due to scheduling conflicts, housing site managers and other relevant housing site staff were not able to attend the REM sessions. These stakeholders were later invited for individual interviews using the appreciative inquiry questions and inviting discussion of challenges and facilitators from an implementation perspective. After the interview, they were presented with the map developed from their housing site participants and had an opportunity to offer additional insights. These interviews lasted approximately 45 minutes.

Data analysis

Using participant responses as the essential data elements, REM session leaders employed Xmind 8 Pro [20] for live (real-time) mapping and to finalize the mind maps. Informed by thematic analysis [21], participants’ responses were organized into inductive themes during group discussion. After review during and after the session, themes were finalized and responses from each map were downloaded from Xmind into Excel and each response was further categorized according to the levels of the socio-ecological model [22, 23]. Two team members (PRE, LGR) independently coded all responses, compared results, and discussed any discrepancies. Data within each theme were classified according to the different levels of the model: (a) individual; (b) group or interpersonal (i.e., group activities during the intervention); (c) neighborhood or community; and (d) policy and built environment (including any participant activities aimed at promoting community-level changes to support PA). Results were summarized for each intervention arm by adding the number of “ripples” or outcomes for each level of the socio-ecological model.

Results

Participants

Of the 83 invited Steps for Change participants, 35 participated in the REM sessions and 48 declined (response rate = 42%). They were joined by three individuals who were spouses/close friends. As was the case with the overall Steps for Change population, participants were older adults on average (average age 72.14 [SD = 7.33] years), 82% women, and college educated (77%) [See Table 1]. An additional 48 invitees did not participate due to being unreachable during the session scheduling period (46%), having stopped active involvement in the intervention (42%), scheduling conflicts (10%), or current health issues (2%). Demographic characteristics did not differ among attendees and nonattendees.

Table 1.

Participants and site demographics

	PA + OV		PA + HE
	Session 1	Session 2	Session 3	Session 4
Participant demographics
# study participants	11	7	7	10
Spouses or friends of participants	0	0	1	2
Age in years, mean (SD)	73.2 (6.0)	74.0 (8.7)	68.0 (7.7)	72.6 (7.5)
Female, %	100	85.7	71.4	70
Marital status, %
Married	27.3	14.3	14.3	40
Divorced	45.5	57.1	57.1	40
Widowed	9.1	14.3	0	20
Other a	18.2	14.3	28.6	0
Education, %
High School or less	9.1	14.3	0	20
College	81.8	74.4	71.4	80
Post grad	9.1	14.3	28.6	0
Race/ethnicity, %
Non-Hispanic White	72.7	85.7	57.1	80
Asian	18.2	0	14.3	10
Black	0	14.3	14.3	10
Hispanic/Latino	9.1	0	14.3	0
Foreign born, %	18.2	28.6	28.6	40
Living alone, %	63.6	57.1	71.4	30

Note. ^aParticipant demographics reported for study participants only. ^bOther includes never married, refused and missing data.

Themes

Because no major differences (i.e., overall themes did not differ) were found for the two maps generated within each study arm, we combined the two maps within each arm to obtain one set of data representing themes for the PA + HE arm and one for the PA + Our Voice arm. Eight unique themes emerged across the categories of reported outcomes, challenges, and solutions (Fig. 1, panel A). Most of the themes (6 out of 8) were the same across intervention arms and included the following: PA tracking, PA levels, health outcomes, social connectedness, challenges faced, and solutions enacted. The remaining two themes were conditional upon the intervention arm (gray in Fig. 1, panel A) and were impacting community and build environment change for the groups engaged in PA + Our Voice, and doing activities in the community for the groups assigned to the PA + HE arm. We further describe the themes below as follows: (1) outcomes including PA tracking, PA levels, health outcomes, and social connectedness; and (2) challenges and solutions.

Outcomes reported

Social connectedness emerged as a prominent theme in both arms (26% of all outcomes discussed). Participants reported both new and strengthened connections with others as a result of their participation in Steps for Change. For example, participants reported “I expanded my world” (PA + Our Voice arm), “found like-minded people” (PA + Our Voice arm), “explored new areas and got outside more” (PA + HE arm), and “[we] had a lot of fun together” (PA + HE arm). In addition, social connections extended beyond the scope of the intervention. One participant reported receiving “support from [new] friends after surgery,” and some participants became group organizers, planning a range of social outings in the community (e.g., to local museums, coffee shops). These social connections and support were described by some as contributing to personal growth (“model self-disclosure & vulnerability for others”), peer accountability (“resilience and accomplishments being witness by all”), and appeared to have facilitated continuous engagement in the intervention (“Group serves as motivator for exercising,” and “They keep [me] motivated to keep going”).

The following interrelated themes also emerged across all groups: (a) increasing physical activity tracking (8% of the total outcomes reported); (b) increasing physical activity (18%); and (c) improving health outcomes (8%). As a result of the intervention, many of the participants discussed using tracking devices (e.g., pedometors, Fitbits) as a motivator for increasing their PA and for accountability purposes (see Table 2 for representative quotes). Some participants (14%) also reported that their significant others or family members also started to track PA after observing the participant doing so. Some participants also reported generalizing the process of tracking and goal setting to other health areas (e.g., weight).

Table 2.

Themes by socio-ecological model and exemplary quotes

Themes and sub-themes	Exemplary quotes	Level of the socio-ecological model
1. Outcomes
1.1 Increasing physical activity tracking	◦ “Made a new habit, made a goal, and do it every day” ◦ “Motivated family to make changes as well”	Individual level Interpersonal/group level
1.2 Increasing physical activity (primarily walking)
• Increasing quantity	◦ “Walking over 1 hr not a problem now” ◦ “Able to walk with grandsons when home in Ethiopia”	Individual level Interpersonal/group level
• Increasing quality	◦ “Walking in more difficult terrain (e.g., hills)” ◦ “Thinking of walking as an activity to socialize with others”	Individual level Interpersonal/group level
•Other	◦ “Increased awareness of what is happening in the community (e.g., neighborhood walks)”	Community level
1.3 Improving health outcomes	◦ “Improvements in mood”	Individual level
1.4 Increasing social connectedness	◦ “Formed coffee club and met outside of the intervention” ◦ “Gives you strength to stay involved and not give up on the things you are doing in the community”	Interpersonal/group level Community level
1.5 Impacting community and built environment change (only present in PA ± OV sessions)
• Increased knowledge about decision making	◦ “Learned about who is in charge of resources in the community. Learned how to make change in the community”	Policy/built environment
• Action oriented efforts	◦ “Got side walks repaired. Now we can walk to shopping.”	Policy/built environment
• Increased awareness around issues	◦ “Increased understanding of importance of getting involved in the community”	Community level
1.6 Doing Activities in the Community (only present in PA±HE sessions)
	◦ “Voluntering in the park”	Individual level
2. Challenges	◦ “Pain in joints, making hard to walk”	Individual level
	◦ “Violence in the community is a challenge for walking”	Community level
3. Solutions	◦ “Made a scrapbook about overcoming challenges to being active” ◦ “Walking as a group to make a statement about safety.”	Individual level Community level

Note. Themes were observed in all maps unless otherwise indicated.

Participants in all sessions described increases in both PA quantity (i.e., number of steps) and quality/location (i.e., faster pace, in different terrains). See Table 2 for examples. They engaged primarily in walking-related activities, although some discussed other forms of PA including aerobics and other types of PA classes. In addition, participants reported that they were now seeing PA as a social and rewarding activity. For example, some participants stated that they had begun thinking further about walking as an option for socializing with others and as a mode of transportation. In addition, participants discussed their new perceived connections between more PA and other health-related outcomes, including “improved sleep,” “weight loss,” “quicker recovery after being sick” or “after surgery,” mood improvements, increased “flexibility” and “better joints,” and increased motivation to leave their home and go outside.

Engaging in ongoing community-offered activities vs. making discernable changes in community environments or elements. While all groups spoke about increased community awareness such as increased knowledge of local opportunities for PA (e.g., parks, walking trails, community gardens), the content was qualitatively different between sites assigned to different intervention arms. Housing sites engaged in the PA + Our Voice intervention spoke about impacting policy and built environment change (34 total comments in the PA + Our Voice arm vs two comments in the PA + HE arm. See Fig. 1, Panel B). Referring to the data collected with the Discovery Tool, participants in one PA + Our Voice group described how they took specific action steps to promote local improvements, for example to repaint a crosswalk, repair a fallen stop sign, and install a speed bump near a park to improve traffic safety. Others engaged in a variety of systematic advocacy efforts which had been discussed as part of the Our Voice intervention, including calling and writing letters to policy makers, calling the authorities to raise awareness about speeding traffic, and speaking with local corporations about improving neighborhood-level conditions for pedestrians. The second PA + Our Voice group discussed how members had documented faded bus stop signs and worked to get them replaced, as well as successfully advocating for improved walkway accessibility for people with disabilities along with sidewalk repairs. Participants’ comments during the REM process showed that such solution-oriented efforts helped participants gain better insights into the decision-making process at the municipal level and encouraged participation in community processes (e.g., municipal infrastructure planning meetings). Our Voice participation also helped to build collective and personal efficacy for effecting change; as one participant stated: “we have a say.” Participants also expressed increased awareness of the need to systematically “document” features of the community to successfully advocate for change. In terms of challenges, for these groups engaged in the Our Voice process, dealing with setbacks or rejections from decision makers when advocating for changes was listed as part of the learning process. Normalizing this experience and working on “easy” or early wins was discussed as a way the group kept themselves motivated despite potential setbacks or perceived lack of progress.

In contrast, participants in REM groups that had been assigned to the PA+HE arm spoke about increased levels of awareness and involvement in ongoing community-offered activities (23 comments), such as volunteering, visiting local senior centers, and taking nature education walks. Although one participant joined an advisory board for a group working on HIV in the county and another discussed wanting to add lights to a pedestrian crossing, overwhelmingly, activities involved individual participation in events organized by others.

Assessing Outcomes Across the Socio-ecological Spectrum

Figure 1 panel B shows the frequencies of outcomes reported across the levels of the socio-ecological model [22]. Notably, Groups 1 and 2 that were assigned to the PA + Our Voice arm, had more frequent (total of 63, compared to 23 in the PA + HE arm) comments about community level and built environment level factors (e.g., advocacy activities aimed at improving the built environment around their housing site). In contrast, Group 3 (PA + HE arm) comments (n = 2) coded under “community level” focused on individual actions (e.g., volunteering at a local garden) and increased awareness of local PA resources and opportunities (e.g., senior walking groups, new trails).

Challenges and solutions

Challenges in all groups often (40% of comments involving challenges) involved individual level factors such as physical limitations and health struggles due to aging and/or chronic conditions, including physical setbacks such as a surgery or medical procedure, joint pain, and back issues. Solutions listed by participants in both intervention arms included being “persistent,” learning coping strategies from each other and the intervention instructors, and incorporating a variety of PA options (walking, yoga, light seated weights). At the group or interpersonal level, scheduling conflicts and having to attend sessions in person were also listed as challenges in both arms (40% of comments). For these challenges, solutions included having group members take ownership of group discussions, organization of walks, and other logistics as a way of increasing group accountability and contribution and reducing reliance on interventionists over time. At the neighborhood level (16% of comments), traffic, safety concerns, and living in less walkable areas were listed as challenges for increasing (or maintaining) PA. At the policy level, and for PA + Our Voice arm, challenges included gaining momentum in advocacy efforts and “dealing with no or setbacks” from policy makers regarding proposed changes. In addition to advocacy efforts in the PA + Our Voice arm, walking as a group was seen as a way of calling attention to needs of pedestrians: “We are making a statement by walking. They see us.”

Housing site management: supporting housing sites in making long-lasting changes

Key stakeholder interviews with housing site managers and staff across the study arms (n = 5) revealed additional information relevant to the study implementation. Noting widespread concern about social isolation as a risk factor for older adults, all housing staff discussed witnessing changes in the level of engagement of residents as a result of having the study on site, including the social connections formed among residents and between residents and neighbors of the housing site as a key benefit. Citing limited resources to deploy programs, managers and housing staff noted that partnering with outside institutions (in this case a local university) allowed them to expand their programming and enrich their residents’ experience. As the affordable housing sites provided meeting space and welcomed study participants living near their housing site, which allowed the study to expand its recruitment, they began to be seen as assets within the local community.

Housing site staff also underscored the importance of focusing on programs with outcomes of importance to participants (e.g., social connections in addition to PA benefits). To bolster sustainability, housing staff also recommended that they work with intervention delivery staff to learn about and potentially co-deliver sessions to further develop internal capacity. Since the culmination of the intervention, at least one of these six sites had added similar group-based programming for promoting PA and wellness in the context of aging.

Discussion

As multilevel, multicomponent PA interventions continue to emerge, additional participatory evaluation approaches can increase understanding of both intended and unintended outcomes and the constellation of factors associated with success, thereby informing their optimization, translation, and implementation. Such approaches currently are scant in the literature [7, 8]. In this first-generation exploratory investigation, we assessed the utility of REM—one such approach—as a complement to more standard quantitative approaches to evaluating a multilevel evidence-based PA intervention. This process elucidated a more diverse understanding of participant- and stakeholder-centered outcomes and solutions to challenges encountered during the interventions that likely would not have been generated from the investigator-chosen questionnaires and surveys often used in this type of behavioral health research.

Implications for behavioral lifestyle intervention research

Uncovering mechanisms by which interventions may work and factors that facilitate or hinder outcomes, particularly in multilevel or complex interventions, is a strength and application of REM. This low-cost and less time intensive method offers a systematic way of understanding the nuanced ways in which different aspects of complex interventions promote outcomes. Some of these pathways may be nuanced or unexpected, and potentially not currently captured by the quantitative measures initially proposed. For example, participants often shared what kept them returning, even after the more intensive 6-month phase of the PA intervention were completed. Strong social connections within the group appeared to be critical in maintaining and enhancing participant engagement over time, which in turn can contribute to outcomes related to PA and health (e.g., sustaining PA gains despite a health crisis, improving mood and well-being outcomes). This is consistent with literature suggesting social support and social connections can be crucial for promoting health and healthy behaviors [24, 25], and the role of social support in community settings (e.g., walking groups) for enhancing PA interventions [24, 26]. Our results suggest the potential positive impacts of including opportunities to foster social connections across the phases of behavioral lifestyle interventions involving midlife and older adults, irrespective of the outcomes being targeted. Thus, REM resulted in data that can further our understanding of drivers of long-term engagement and maintenance of behavior changes, a challenge in behavioral medicine research [27].

Moreover, REM helped elucidate which outcomes where most important to the various stakeholders, as well as challenges, and participant-driven solutions. This information can be used to engage and enroll future participants, to partner with additional stakeholders or relevant funders on outcomes they care most about, and that may be different than primary outcomes of the intervention. Thus, REM can complement more traditional and researcher-driven measures for capturing interventions impacts, challenges, and solutions.

In exploring the differences between the intervention arms, REM also highlighted the positive impacts of participating in a neighborhood-level citizen science component, the Our Voice intervention, as a catalyst for participation and potential facilitator of individual-level outcomes. Although participants in the PA + HE arm increased awareness of and participation in ongoing community activities, those in the PA + Our Voice citizen science arm discussed participating in activities to promote broader impacts, including changes to their local built environments. Findings suggest that purposefully connecting over a common goal (i.e., collecting and using data to promote actionable solutions) can enhance the quality of both the connections made and engagement in various aspects of the intervention. This broader engagement also appears to have promoted the development of natural “champions”—participant(s) who encourages others, plans activities, and/or who may serve as an accountability person or leader for the group—who took on leadership activities for the group and served as catalysts for further engagement. These factors should be explored as potential facilitators of positive outcomes in other interventions. Our results offer some insights into how to set up interventions in a way that facilitate this type of traversing of impacts across individual and local environmental levels. This has important implications for the development of behavioral interventions that can help to address health inequities, particularly as many of the participating affordable senior housing sites are in underserved neighborhoods with less access to health enhancing resources, including PA opportunities.

Implications for translation, dissemination, and implementation

By systematically incorporating open-ended questions about challenges and participant-driven solutions as a part of project evaluation, REM can help to capture areas of importance and meaning for participants themselves, which in turn may help to strengthen interventions and inform future implementation and dissemination. For example, pragmatic participant-identified solutions to challenges can then be implemented in the future iterations of the intervention. REM offers the additional advantage of involving diverse stakeholders in the information gathering process to shed additional light on intervention outcomes, impacts, and challenges. Discussions with housing staff further elucidated factors influencing implementation and impacts of these kinds of interventions. For example, advertising outcomes such as decreased isolation was discussed as crucial when working with some groups, including older adults. Behavioral interventions, regardless of their primary outcome, could consider additional steps to enhance social connection and discuss these outcomes when approaching community-based organizations for partnership. Moreover, findings also elucidated important insights for future scaling and implementation. For example, developing capacity for intervention delivery among housing staff was suggested, and currently being implemented at one site, to ensure sustainability of the intervention beyond the study period.

Limitations and future directions

This first-generation investigation of the potential utility of Ripple Effects Mapping in a community-based clinical trial has some limitations that should be noted. These include the inability of some individuals to participate in the group-based data gathering method, which prevented us from understanding what that subgroup of individuals might have contributed to the mapping of impacts that occurred. Moreover, while we purposefully selected groups that had completed 12 months of intervention, it is possible that this could have contributed to selection bias. To advance understanding of the potential utility of REM in the health promotion field future studies could benefit by applying the REM method with other populations and health areas and outcomes. While REM has been applied in studies focused on childhood obesity [28], community-based volunteer programs [29] and youth development [30], additional behavioral medicine applications will help to clarify the types of research for which this REM approach brings the most added value. Future studies may also implement REM at multiple time points (mid-intervention in addition to at the end) to identify modifiable challenges and optimize intervention adherence and motivation. This type of multi-phase evaluation can also help to further elucidate mechanisms of change and set the stage for larger studies examining causal pathways.

Conclusions

Participatory methods, such as REM, offer an opportunity to further understand outcomes of multilevel multicomponent and other complex behavioral interventions. These approaches can help to bring additional knowledge to the fore that can inform the optimization of behavioral lifestyle interventions and assist in future translation, implementation, and dissemination. In addition to aiding in uncovering unexpected outcomes and factors that promote behavior change and maintenance that may not have been measured in the larger study, REM can further advance the literature by highlighting the contributions of different personal and socio-environmental factors connected with an intervention and eliciting valuable participant and stakeholder-driven solutions to challenges experienced during the intervention.

Compliance with Ethical Standards

Conflict of Interest: None declared.

Funding sources

This work was funded by the National Cancer Institute (R01CA211048, PI: King) and by the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health under Award Number UL1TR003142. The content is solely the responsibility of the authors and do not necessarily represent the views of NCATS or the National Institutes of Health. Funding sources did not have a role in the study design; data collection, management, analysis, or interpretation; or writing and approval of this manuscript.

Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This article does not contain any studies with animals performed by any of the authors.

Informed Consent: Informed consent was obtained from all individual participants included in the study.

Transparency Statement: The study was registered at ClinicalTrials.gov (NCT03041415). The analytic plan was not registered. Deidentified data, statistical analysis code, and materials used to conduct the study are not publicly available. Deidentified data can be requested by contacting the corresponding author; however, data sharing will be contingent on institutional approval.