Development of a scalable weight loss intervention for low-income workers through adaptation of interactive obesity treatment approach (iOTA)

The current epidemic of obesity in the US and other countries is projected to greatly increase the prevalence of diabetes and other health consequences [1–6]. Obesity is most prevalent among socioeconomically disadvantaged populations, including racial and ethnic minorities [7, 8]. National data show that obesity is strongly associated with low socioeconomic status (SES), which includes those in hourly and working-class jobs and employees with low education [9–14]. SES factors are related to environmental characteristics and risk behaviors that promote obesity, as well as limited access to weight management resources [15–19]. Despite the compelling evidence for the importance of healthy weight, few weight loss interventions have been rigorously tested in low-SES populations [20–22]. Adaptation of existing intervention models is needed to improve reach and sustainability while maintaining the effectiveness shown in interventions such as the Diabetes Prevention Program (DPP) [23].

Worksites are targeted as a priority location for health intervention efforts by employers, the Centers for Disease Control (CDC), and the National Institutes of Health (NIH) [24–29], because they offer an efficient and effective means of delivering and evaluating such programs, and offer opportunities to reach socially disadvantaged groups. Worksites offer ready access to populations, natural structures for social support, and the opportunity to build health promotion activities on existing communication networks. Importantly, there has been minimal evaluation of the effectiveness of worksite-based health promotion among low-SES employees [30, 31], a population with more limited access to and participation in worksite wellness programs [30, 32–41]. This may be attributable to a number of barriers such as stress and limited communication to these groups [24, 35, 38, 42–47], particularly for low-wage hospital employees [48, 49].

One program that successful promoted weight loss in a low-income, racially diverse population, Be Fit Be Well (BFBW), was implemented in individuals with obesity and hypertension who received healthcare through community health centers [50, 51]. This lifestyle modification program targeted multiple levels, included dietary messaging around the Dietary Approaches to Stop Hypertension (DASH) diet, and incorporated electronic as well as interpersonal supports; participants received both types of supports. Depending on the participant’s available computer access and comfort with computer use, the electronic supports were delivered either through a Web-based system or through a telephone-based system using Interactive Voice Recognition; the electronic support component did not include personal contact. The interpersonal supports included community health workers and linkages to local resources. Behavioral targets for BFBW included diet, physical activity, and hypertension medication adherence, and outcomes assessed were weight loss, blood pressure control, and quality of life [52]. Overall, BFBW showed significant benefit for weight loss at 24 months [51].

While evidence-based interventions promoting weight loss and chronic disease prevention, such as BFBW, are readily available, they require adaptation for implementation in more diverse settings such as worksite settings with low-SES employee populations. Describing the process of adaptation is critical to advancing implementation science [53–56]. The application of a formal framework to an intervention and its use in another setting can advance implementation efforts by shedding light on the steps taken and the modifications required. This can help important stakeholders (e.g., other investigators, practitioners) understand why and how an intervention being implemented is different from the original intervention, and how they might further adapt the intervention for other settings and populations. For the current study, the BFBW program was adapted to fit a worksite intervention targeted at low-SES employees. This manuscript aims to use the model developed by Stirman et al. [55–57] to describe in detail the adaptation and process for adapting BFBW to enhance the external validity and transparency of the adapted intervention, Working for You (WFY) [55].

To put the adaptations made to BFBW in context, and inform others seeking to modify evidence-based weight management interventions for new contexts and populations, we used the adaptation framework developed by Stirman et al. [55–57] This framework was selected as it is recommended by Chambers et al. [54] as a method to capture adaptations to a specific evidence-based intervention. This framework was developed based on a review of articles describing modifications to evidence-based practices across a variety of interventions and settings [55, 57]. The most relevant components of the framework are summarized in Fig. 1, and these include four Contextual modifications (e.g., change in setting or format), three Content modifications (e.g., adding or omitting components), six levels at which modifications could occur (e.g., consistently throughout a system or organization or only for particular clients), and a code for modifications to Training or evaluation processes. Content modifications can be made at multiple levels.

First, we identified targets for modifying BFBW using surveys with 219 hospital workers (e.g., housekeepers, food service workers). The full survey methods and results have been published elsewhere [49]. Briefly, the survey found that workers with irregular schedules were less likely to participate in a worksite health promotion program [49]. These findings were incorporated into the content and context modifications described below, such as the inclusion of SMS text messaging and a workgroup-level intervention. Next, key informant interviews with five staff, including managers and wellness committee members, knowledge and expertise from the interdisciplinary team (described below), and a first round of 4 focus groups with 20 individuals (Focus Groups 1) were used to dive deeper into the modification targets, such as time and/or priority barriers to participation in the workplace programs offered [48]. We recruited a similar sample to that described for the survey; the methods and results for this qualitative work have been described elsewhere [48]. Briefly, experienced research team members used a semi-structured script to guide focus group discussions, which were audio recorded and transcribed. The full list of interview topics are published elsewhere [48], but include: work schedule, healthy eating priority, eating at work, current wellness programs. NVivo 10 software (QSR International Pty Ltd., Melbourne, Australia), was used to analyze transcriptions. Using a phenomenological approach, to find the “essence” or common themes across individual experiences [58], the thematic analysis aimed to answer: 1) “what impacts healthy eating and physical activity” and 2) “what can be modified at the workplace?”, and codes were merged and grouped under main themes [48]. We conducted an additional set of four focus groups to gauge interest and willingness to participate in a text-messaging weight loss program and test preliminary versions of the messages (Focus Groups 2), which included 13 individuals. The focus group topics included current usage of text messaging, workplace eating and physical activity behaviors, perceptions of the proposed intervention, and feedback on message content and wording (focus group discussion guide is available in the Additional file 1). Interviews were not recorded in order to protect potentially sensitive information. In addition, recording and transcription were not necessary since the purpose of the focus groups was to get feedback about the intervention components (e.g. message timing, message content, logistics) as well as to gather additional information to be used in creating goal domains specific to the workplace. Participant responses and notes captured during the discussions were reviewed by the research team and used to make decisions about intervention modifications. This supported the decisions for incorporation of SMS messaging as a communication channel to reach the target population.

During goal and message development, we conducted a third round (Focus Groups 3) of three focus groups (n = 24) with hospital employees from workgroups similar to those to be targeted for recruitment in the intervention (focus group discussion guide available in the Additional file 1). We used an iterative process for designing goals and messages based on information gathered during focus groups to inform subsequent focus groups. Consequently, recording and transcription were not necessary and summaries of focus group discussions, rather than systematic analysis of transcripts, were presented to the research team. Meetings of the research team were interspersed between the focus groups so that the responses and notes captured during the groups could be discussed and used to modify the intervention. This was aimed to ensure the goals and messages were relevant and understandable to the new target population. Topics for Focus Groups 3 were: perceptions of proposed intervention methods, additional workplace eating and physical activity questions including use of physical activity trackers for self-monitoring, and discussion around the logistics and feasibility of delivering the intervention in a workplace setting.

Table 1 includes a summary of the key findings from the formative work. Multiple modifiable barriers to improving workplace health promotion were identified from key informant interviews and Focus Groups 1. The barriers were primarily related to lack of engagement included 1) communication barriers (many workers were unaware of the wellness initiatives offered by their employer); 2) barriers to healthy eating and program participation (i.e., night shift workers had less access to health programs or to healthy food choices, work schedules caused meals to be hurried, and food brought for potlucks and employee appreciation did not provide healthy choices); and 3) the opinion of many workers that their work activities provided sufficient exercise, obviating the need for or making them too tired for leisure time physical activity. Content modifications, such as the tailoring of goals and messages and substituting goals and components, were made based on these findings.

The workgroup-level component of WFY incorporates the important social and physical environmental factors considered in BFBW, but includes a natural network (co-workers) and environment (workplace) where support and environmental changes can occur (Table 2). This is targeted through a participatory approach using the Intervention Design and Analysis Scorecard (IDEAS) tool [70] and a Human Centered Design approach [71–75] to engage members of the work group in designing and implementing interventions to promote healthy eating and physical activity through changes to the work environment. The remainder of the modifications focus on the iOTA, and are organized according to the Stirman model (Fig. 1 and Table 2).

We modified the evidence-based weight and hypertension BFBW intervention to provide a scalable weight loss intervention in the context of a worksite setting with low-wage employees who did not necessarily have hypertension. This process led to an intervention that should be able to be sustainably implemented in workplace settings and also promotes healthy weight behaviors in low-SES worker populations. To allow comparison of study results and to inform future intervention efforts, it is important to formally map the modifications made to this intervention for the new setting and population [54]. While considerable attention has been paid to cultural adaptation, adaptation for implementation has not been well documented [55–57]. Although intervention adaptation is common, it is uncommon to describe the process and theoretical underpinning. Mapping such changes onto an adaptation framework [56] contributes to the growing implementation science literature [53, 54]. This adaptation process fits with other concepts guiding intervention design, including methods such as Intervention Mapping [83], which has been used within worksite health promotion [84–86] to improve intervention fit [87, 88].

Relations of worksite factors with obesity and chronic diseases such as diabetes have been seen elsewhere in the literature [13, 89]. .The employee preferences identified in the current study and incorporated into the intervention adaptation have been identified in other employee populations [90, 91]. Factors identified in the formative work for the current adaptation process such as long work hours and hostile work environments have been found to be associated with obesity [13] and diabetes [89], highlighting the importance of incorporating these as intervention targets in an adapted intervention [91].

This study has limitations worth noting. For this study, we followed Chambers and Norton’s guidance and applied the model by Stirman et al. [54, 57] However, there are several models for adaptation available [88, 92, 93], and the results obtained might have been different were another model selected. Regardless of the model used, we believe formally documenting the steps required for adaptation leads to more reproducible science and improved interventions. While we employed a framework to systematically document how BFBW was adapted to WFY with the aim of enhancing the generalizability of this experience, the study was conducted within only one hospital system in one city in the United States. Therefore, while the approach is likely to be generalizable, the findings from the resulting WFY intervention may not be [54].

The WFY intervention is currently being tested in a large group-randomized trial. As described above, the individual-level intervention is nested within a workgroup-level intervention in which employees participate for 2 years. Employees with obesity in the workgroup are eligible for the individual-level intervention. Survey measures and height and weight are measured at baseline and (all but height) at 6-month, 12-month, and 24-month follow-up. The primary outcome is weight; secondary outcome measures include dietary and physical activity behaviors as well as work satisfaction. In total, 22 workgroups (11 intervention and 11 control) will be enrolled, and approximately 308 participants with obesity (of 990 total) are expected to participate in the individual-level intervention.

Attention to how and why an evidence-based intervention is adapted for a new population and setting can inform others seeking to modify evidence-based weight management or other health interventions. This has particular relevance if the adapted intervention is found to be effective. WFY is designed to be scalable and relatively inexpensive, such that transparency in the adaptation process may make this intervention easier to translate to other work settings to reduce obesity and diabetes risk among low-SES workers, a group with a higher prevalence of these conditions.