The SisterTalk study was designed to evaluate the effectiveness of a culturally tailored weight control program delivered by cable TV and also to test the separate and combined effects of two potential strategies (interactive programming and provision of telephone support) for enhancing effects. The overall results indicate modest effects of the TV interventions, taken together, in lowering BMI and dietary fat intake and increasing physical activity relative to no intervention when assessed soon after delivery of the 12 week program, but only the effect on dietary fat intake was sustained for the remainder of the year. The cultural tailoring and inclusion of theoretical constructs seems to be effective, but cannot be compared to a non-tailored intervention, or one tailored on theoretical constructs with this design. Kreuter, et al. found that cultural tailoring combined with tailoring on theory-based psychosocial constructs was more effective than tailoring on theory-based constructs alone , but further research should replicate these findings.
The finding of significant initial effects in the expected direction and for both behavior changes and BMI is encouraging as to the potential utility of this general approach with Black women. Effects on weight were modest (0.65 kg weight loss at 3 months. This is likely attributable mostly to the program content, which allowed flexible goal setting and emphasized weight control more strongly than weight loss, as described under Methods. Process evaluation data indicating that two-thirds of the women (70%) actually watched most of the shows or read most of the written materials (57%), which may also explain the modest changes observed, however group attendance in other programs is often at 70% or lower [21, 25, 64, 65]. An effect of intervention dose was evident for dietary fat intake, and dietary fat intake was also the outcome variable showing the most responsiveness to the intervention overall, over time, and to the intervention enhancements. While differences in fat intake persisted over time, this did not translate into a sustained difference in weight or BMI. Thus, future enhancements of the dietary content could include a stronger focus on reducing overall caloric intake and portion control, which may have a stronger effect on weight than reducing dietary fat intake.
The longer term findings clearly indicate a need to strengthen intervention delivery after the initial period in order to achieved sustained effects on BMI. Several studies that have included Black women report that initial weight loss is a major predictor of longer term weight loss [66, 67], which suggests that alterations to improve the potential for early weight loss may also improve results over the longer term. This could involve identifying culturally salient strategies to motivate overweight and obese women to lose weight rather than only control weight initially and strengthening advice regarding how to do this.
The attenuation of physical activity effects over time suggests that strengthening the physical activity intervention may be particularly important. Physical activity is difficult to measure [68, 69], and studies of the validity of PA assessments in Black populations are limited , but it does not seem likely that measurement issues would explain attenuation of effects over time and the attenuation over time is consistent with the similar finding for BMI. National survey data indicate that Black women report lower levels of leisure time PA as well as higher levels of sedentary behavior compared to white women [71, 72]. Physical activity has been identified as an important predictor of weight maintenance overall [73, 74], and in observational data for Black women , although physical activity levels did not predict weight maintenance among Black women in a controlled trial . In any case, it is likely that the intervention failed to achieve enough change during the initial program period to be sustained over time. The main activity message was to increase activity during daily life, but this message did not seem specific enough, and the exercise portion on the TV show was only 10 minutes. Future interventions should advocate more specific activity increases that can be planned and carried out consistently.
Other ways to improve behavior change and/or enhance maintenance of changes might include more gradual tapering of program frequency, e.g. providing additional shows bi-weekly before reducing the frequency to monthly, repackaging and re-airing core content, or combining these strategies. Another possibility would be developing a sufficient number and variety of shows to permit ongoing weekly access. Given the advent and popularity of digital communication channels in recent years, producing shows with web-based delivery might be another strategy. These approaches would be compatible with an entirely home-delivered program. The exercise component of the shows could also be given greater emphasis, e.g., by adding a set of exercise tapes rather than only including an exercise component in a multifocal TV show and encouraging more walking throughout the day.
Adding interactive programming or telephone support did not appear to improve effectiveness. The one exception was for fat intake, which improved among all of the experimental groups at most time points. The absence of any measurable effects of the interactive and social support components was disappointing given that each of these had a sound theoretical basis. For both components we can attribute the lack of effects at least, in part, to incomplete implementation, as indicated by the process evaluation. Omitting the interactive component would be favorable from feasibility, replicability, and cost perspectives in that future programming could be delivered via prerecorded rather than live programs. However, the aforementioned considerations related to how to strengthen the intervention would still need to be addressed. Perhaps interactivity could be added with a telephone hotline, an internet chat-room or email/texting. With respect to the telephone support, the feasibility issues encountered were not entirely unexpected. Non-completed calls have been an issue with telephone counseling in other research, [77–79] but it is difficult to disentangle issues of access to participants, from passive refusal to participate in that portion of the intervention. Updated research is needed in this respect given that the current context for telephone communications has changed dramatically: fewer people have telephone land lines, cell phones may be answered at times when people are not at home, and caller identification makes it easier to be selective about which calls are answered . Also, as noted above, the promise of digital strategies for remote support interventions such as internet chat rooms might be relevant [81, 82]. On the other hand, approaches that involve some direct personal contact or networking among participants would also be worth evaluating. An original plan for networking of participants for “out of class” get-togethers proved infeasible based on geographic dispersion of the sample and concerns about contamination of the waiting list comparison group and subsequent cohort waves. However, inability to join with a friend or network with others in the study were the most frequent criticisms of the program by participants, as women wanted to work with others as part of the program. Thus, it might be helpful to add a social component to future interventions.
The experience the COEs was mixed, raising questions that require careful consideration about whether or how to incorporate peer-counselors in this type of intervention. Involvement of the COEs was invaluable during the formative assessment phase of the project, e.g., provider “insider” perspectives of many aspects of the intervention and facilitating recruitment through personal connections in the targeted neighborhoods. However, their performance was less predictable when the role shifted to conducting peer counseling that involved specific skills, meeting deadlines, and recording and reporting data. Lessons learned include the need to have different criteria for hiring, training, and supervising outreach workers hired for formative activities and recruitment vs. counseling. Other researchers have had success with lay counselors when strict screening, training and certification procedures were used [77–79, 83–89]. Consistent with this interpretation, several COEs resigned as poor performance was brought to their attention and more was demanded from them. Thus, when forming partnerships with representatives of a targeted community, a great deal of care is necessary to match the skills and experience of the community members employed with the skills and abilities required for the specific tasks they are going to be responsible for.
Strengths of this study include the randomized design with multiple follow-ups; the extensive community-engaged formative research used to plan all aspects of the study; the comprehensive cultural-tailoring of the intervention; and the use of valid culturally adapted evaluation measures. Moreover, the targeting of Black women is a strength given the marked disparities in BMI and health outcomes.
Limitations include the aforementioned incomplete delivery of the intended intervention dose and our inability to completely judge the quality of telephone support. In addition, the lower than planned recruitment is a limitation given that it reduced the statistical power to see significant effects on outcomes, particularly for comparisons between the different intervention formats. However, as discussed earlier, because the intervention developed with target audience input ultimately focused on weight control rather than weight loss, a higher sample size would likely not have mattered with respect to the weight outcome; however, more power would likely have enhanced the significance of the other observed changes. Future weight control intervention studies need to find ways to improve recruitment of Black women.
Yancey, et al., found that Black women of higher weight status were more likely to be recruited by face-to-face engagement [90, 91]. While we conducted some face-to face recruitment, this was somewhat limited by the geographic distance of the research team and relied mostly upon the COEs for recruitment. A bias toward recruitment of higher SES women is implied by the levels of education, employment, and income, although higher SES African Americans do tend to live in lower income neighborhoods than might be expected . Having a higher SES sample would not affect internal validity in our randomized design, but would potentially affect salience of the intervention content, which was informed by formative research with a lower income sample. This lack of correspondence between the audience the intervention was developed for and the sample that actually joined the study, could also have limited effectiveness. Attrition was also moderate (>30% for in-person follow-up measurements). While the follow-up incentives were raised and shifted from gifts to cash over-time and this improved participation retention, overall retention was still not ideal. More research is needed on how to retain participants, especially Black women in weight control studies.