Examination of program exposure across intervention delivery modes: face-to-face versus internet

Current epidemiological evidence suggests that approximately 60% of the global population do not participate in sufficient physical activity for health benefit [1], defined as participating in 30-minutes of moderate-intensity physical activity (of least 10-minute bouts) on 5 or more days of the week [2]. Inactivity is associated with many chronic diseases including cardiovascular disease, type 2 diabetes, coronary heart disease and some cancers [3–7]. Therefore, developing effective physical activity programs that are widely accessible is a necessity, given the rising rates of sedentary living and associated health implications.

Group-based programs that offer face-to-face treatments are common and popular strategies in behaviour change research [8]. Unfortunately, face-to-face programs often have high running costs, limited access and availability, and high attrition rates [9–12]. Drop out rates have been reported as high as 50% within 6-months of beginning a program [13], which further limits the public health impact of face-to-face programs. Recently, researchers have focused upon innovative interactive health communications (IHC) such as the internet and computer expert systems, as a means of engaging populations and individuals in health-related behaviour change [14–17]. Studies in this area suggest that IHC mediums, particularly the internet, can offer an alternative method or mediated approach to intervention delivery. Reported advantages of the internet include; widespread access, tailoring of information, online social support, instantaneous interactivity, and confidentiality and anonymity [10].

In physical activity research only a small number of internet-based trials have been reported [16–19]. A study by Marshall and associates (2003) showed that a stage-based physical activity website significantly decreased sitting time, and increased motivational readiness for physical activity. They also reported that only 26% of participants logged on to the website more than once, and a later publication [20], reported that of the 327 participants randomised to the website condition, only 46% accessed the website a minimum of once. A meta-analysis also highlighted the difficulties associated with user engagement and retention in internet-based interventions [21]; which can result in high drop-out rates and reduced intervention exposure. Examining strategies to enhance program exposure is important, however we also need to explore participant characteristics and identify target groups that may be more responsive to internet-based program compared to traditional face-to-face delivery.

We recently reported the results of a randomised trial (RT) comparing intervention delivery modes for a 12-week physical activity intervention (Health-eSteps) [22]. The RT compared traditional face-to-face delivery (FACE) with a combined face-to-face plus internet delivery (internet-mediated [IM]), and an internet-only (IO) group. Results of the RT showed all groups significantly increased in self-reported mean minutes of activity; there were no differences between intervention groups; and all groups were statistically equivalent from baseline to the post-intervention follow-up [22]. The Health-eSteps intervention was based upon social cognitive theory (SCT) [23] and self-management [24], and aimed to provide users with the skills required to adopt and maintain an active lifestyle. We aimed to build upon previous research, and enhance website retention and engagement by (i) conducting a rigorous formative evaluation process, (ii) including additional face-to-face contact (Internet-Mediated group), (iii) ensuring a dynamic and changing website as opposed to a static information-only site, (iv) including up-to-date local community information and, (v) providing individualised and personally relevant feedback. Additionally, one of the unique aspects of the website was that the modules were delivered on a week-by-week basis; the entire website could not be accessed in one sitting, a disadvantage noted by others [18, 19].

The focus of this study was to; (a) describe program exposure [website usage and face-to-face attendance] across the three delivery modes of the Health-eSteps RT, (b) examine the association of program exposure and change in physical activity behaviour and, (c) identify predictors of program exposure. Specifically, we examined characteristics of participants who were exposed to at least 75% of the program (face-to-face attendance or website usage). Participant exposure has been found to be a strong influencing factor on the success of program outcomes [25] including internet-based interventions [16].

Previous studies have acknowledged low participation rates [21, 36] and increased drop-out in internet-based interventions [18]. This study found similar declines in weekly program exposure across each intervention group. A significant association between program exposure and participation in physical activity at follow-up (self-report mean minutes of physical activity) was found, suggesting that improved physical activity outcomes were related to being exposed to at least 75% of the Health-eSteps program material. In this study, the IO group had higher rates of program exposure and were more likely to have had exposure to at least 75% of the program material compared to the FACE group. Participants with higher PASE in the IM, and IO groups were also more likely to have had at least 75% program exposure. The FACE group participants were 79% less likely to reach at least 75% program exposure if they were classified as obese.

The FACE group showed similar face-to-face attendance rates (exposure) and drop-out, to other physical activity interventions delivered via the same modality [34, 37]. However, drop-out in the IO group was lower then the average drop-out rate of 21% reported in a recent internet-based behaviour change meta-analysis [21]. Glasgow et al., [38], reported an overall decrease in website access over time throughout a randomised controlled trial of the Diabetes Network self-management intervention. They reported between 11.4 – 16.7 logins per person during the first three months of the intervention, which decreased to 5.0 – 5.3 per person during the follow-up period (7–10 months). In a study of web-based nutrition counselling for patients at risk of cardiovascular disease, Verheijden [36], reported that only 33% (24/73) of participants used the internet-based program and the study failed to show any increase in outcome measures between the intervention group (internet) and the control group (usual care). They also reported that participants visited the website an average of once over an 8-month period. Another study also reported low exposure to a physical activity website, in which only 46% of participants visited the website at least once, a decline in website access was also observed [20].

One of the strong points associated with the Health-eSteps website that may have assisted in user engagement similar to that observed in the FACE group, was the delivery of modules/information sessions on a weekly basis. In this respect, participants could not log on to the website and review the entire website in the one sitting. Further, the website material was built-upon on a week-by-week basis, possibly enticing participants to return to the site the following week. Strategies incorporated into the website such as a "What's New at Health-eSteps" section, weekly reminder emails and the numerous interactive features (quizzes, activities, self-appraisal, feedback) may have also enhanced participant engagement in the website. However, this study did not examine specific strategies of engagement, therefore we can only speculate as to why participants returned back to the website.

The Health-eSteps program also used incentives to assist in enhancing program exposure across all three intervention groups, however without a comparison group (control with no incentives) this study can not conclude if the incentives had an effect upon program exposure or not. Although the limited observation of participant engagement with the incentive opportunities, suggests that new innovative incentives idea and engagement strategies are required. This may include the examination of different 'types' of incentives, combination of engagement strategies (e.g. email reminders and electronic pedometers that are synchronised with online personal diaries), addition of telephone counselling/support and/or the provision of more automated tailored feedback.

Previous research has also highlighted that poor program outcomes are associated with poor program attendance [13, 39]. Tate, Wing and Winett [16], reported greater changes in body weight with increased frequency of website log-on in an internet-based weight loss program. Similarly, this study showed higher program exposure was associated with higher physical activity levels immediately post-intervention. Identifying avenues to increase internet engagement will therefore not only increase program exposure but also lead to an improvement in behavioural outcomes. The internet, as suggested by others [40, 41], may be best used as a supplement to face-to-face interventions or standard care. However, we did not show any added advantage for the IM group over the IO, as a result of receiving additional face-to-face program exposure.

A potential avenue for future research is the examination of online chat rooms, support groups, and discussion boards for enhancing treatment effects and participant retention. For the Health-eSteps website it was decided not to include a discussion board or online chat room based upon formative evaluation results [26], and the lack of research supporting online chat rooms and discussion boards [17, 36]. Further, Harvey-Berino (2002) reported that 'attendance' at an online chat session for the maintenance of weight loss, did not reduce the decline in attendance observed during the intervention. The authors' suggested this is due to participants feeling that they were unable to communicate with other online participants, and the online therapist effectively [40, 42]. However, as technology advances and people become more familiar with using various internet components, the use of online support groups, and discussion boards are likely to increase [43]. Therefore, continual investigation of the role and mechanisms that facilitate effective online support (support groups, chat-rooms, discussion boards) is warranted.

To date, few studies have focused upon individual moderators of program effects [44, 45], such as the type of person a program is more likely to be effective for [46]. In this context a moderator such as sex or age may strengthen or weaken the relationship between the intervention and the outcome variable [45]. The results of this study, suggest that an internet-based intervention may offer an alternative delivery method to traditional face-to-face delivery for obese participants. This implies that internet-based programs may have a potentially greater outcome effect than traditional face-to-face interventions for this sub-population. This may be partly due to the fact that the internet removes perceived barriers related to weight image and embarrassment as suggested by others [40], and anecdotally expressed by participants in the FACE group. Further, the internet offers advantages over traditional face-to-face programs in terms of autonomy and confidentiality [47, 48].

Participants who had higher PASE at baseline were more likely to have been exposed to at least 75% of the program material in the IM and IO groups. Physical activity self-efficacy is commonly reported in the literature as an important mediator [45] across a diverse range of physical activity interventions [49–51]. We did not examine PASE as a mediating variable nor did we examine the relationship between PASE, intention to use the internet, and actual physical activity behaviour. However, the results found may indicate participants who have high PASE may also posses high outcome expectations, and therefore see the potential benefits of using an internet-based behaviour change program.

Interpretation of the findings presented in this study should be mindful of the following limitations. A methodological limitation to this study was that participants allocated to the IM and IO group still received face-to-face contact in Week 1 and Week 12 for data collection. Future studies should therefore, investigate the efficacy of changing behaviour with no face-to-face contact, and use internet-based data collection methods that may also increase the number and reach of the target population [16].

This study compared two types of 'exposure'; face-to-face program exposure (in-person sessions each week for an hour) with internet 'exposure' (logon frequency) in which the access time, day and location was not restricted. These two definitions are inherently different and plausibly reflect differences in the 'dose' of the intervention received. Examination of the amount of 'time' spent on the website (in comparison to a 1-hour session) was unable to be determined. However, the Health-eSteps RT was the first study of its kind to compare the efficacy of face-to-face with an internet-based behaviour change program, therefore examining program exposure across delivery modes, as presented here is of interest to researchers. The findings of this study were also based on self-report information and no objective measures of physical activity were used. The sample population were mainly female, living in a rural area of Queensland, Australia, therefore the results are not generaliseable to the population as a whole. Additionally, the study sample was self-selected and may be more motivated to change their behaviour than the general population.

The findings of this study support the growing body of evidence for the use of the internet in behaviour change research. Weekly exposure showed a similar decline over time, however the internet groups were more likely to have been exposed to at least 75% of the program material. Future research should investigate reasons for internet-based program drop-out and examine strategies for increasing participant exposure. It is also important to explore which target groups or populations that internet-based programs are best suited. Identifying factors that predict drop-out and program exposure is critical for the future development of physical activity internet-based interventions. In this context, we can increase the public health impact of internet-based programs, whilst providing effective alternate delivery modes for increasing physical activity in sedentary populations.