The trial was registered with the ISRCTN registry (ISRCTN88405328). Ethical approval was granted by local institutional ethics committees at the Universities of Leeds (17-0082; 27 February 2017), Lisbon (17/2016; 20 February 2017) and the Capital Region of Denmark (H-16030495; 8 March 2017).

The NoHoW trial (ISRCTN88405328) was an 18-month, 3-center, 2-by-2 factorial, single-blinded, randomized controlled trial, evaluating a digital weight loss maintenance intervention. The participants were required to be aged ≥18 years, have a verified ≥5% weight loss in the last 12 months with current weight at least 5% below their highest weight, and have had a BMI of ≥25 kg/m² before weight loss. A total of 1627 participants were recruited and randomly assigned to one of the following four arms: (1) control or self-monitoring (n=400), (2) motivation and self-regulation (n=403), (3) emotion regulation (n=416), and (4) combined arm (n=408). All participants received activity trackers (Fitbit Charge 2) to be worn throughout the trial, weight scales (Fitbit Aria), and access to the web-based NoHoW Toolkit (TK) tailored to their respective arm. Participants in intervention arms were encouraged to complete 18 intervention modules in the TK during the first 6 months of the trial. The participants received weekly emails during the first 18 weeks as reminders recommending visiting a specific module. A detailed description of the trial is presented by Scott et al [7]. The TK design and content are presented in detail by Marques et al [8].The motivation and self-regulation and combined arms had 2 modules focusing on physical activity: physical activity goal (Goal) and physical activity barriers (Barriers). The Goal module (see Multimedia Appendix 1 for screenshots) addressed goal setting and action planning, contained information on how to set goals, and had a form to set a goal and detailed plan for either the number of steps per day or other type of physical activity. The estimated duration to complete the Goal module was 10 minutes. The Barriers module (see Multimedia Appendix 2 for screenshots) was introduced later in the intervention and contained a testimonial on potential barriers of physical activity and an interactive exercise for identifying personal barriers and creating a coping plan to deal with them. The estimated duration to complete the Barriers module was 8 minutes.

Visits to the Goal and Barriers modules were identified based on log files of the TK. The duration spent in the modules was calculated based on log events signifying entering and leaving the module. A module visit was considered complete if it lasted at least 33% of the estimated duration of the module (ie, 3.3/10 min for the Goal module and 2.6/8 min for the Barriers module) or if the duration of the visit could not be determined due to a missing end event. The threshold of 33% was determined by intervention designers as the minimum time required to become exposed to the behavior change mechanisms in these modules.

Daily summaries provided by Fitbit were used as the physical activity metrics and included daily steps; activity energy expenditure; active minutes categorized to fairly active, very active, and total active minutes; and distance. Activity metrics were averaged over the 7 days prior to visiting the modules and 7 days after visiting them. Days with less than 1000 steps were considered missing data and were not included. A similar threshold has been used in several previous studies (eg, [9,10]). It was also required that the 7-day periods contained at least 4 days of activity data. The immediate effects are presented both for all visits to the modules and for complete visits. If an immediate effect was found, maintenance of the effects was evaluated for 8 weeks following the module visits, considering only complete module visits. A comparison of overall changes in activity metrics between the first and sixth month of the study is presented in Multimedia Appendix 3.

As most of the physical activity metrics had skewed distributions, nonparametric methods were used. Median and interquartile ranges were calculated, and nonparametric tests (Wilcoxon signed-rank test) were used for comparisons. All analyses were conducted with Matlab R2017a (Mathworks) and SPSS Statistics software (version 26; IBM Corp). Statistical significance was set at P<.05.

The modules were available for 811 participants (ie, the participants randomized to the motivation and self-regulation and to the combined arms). Of the 811 participants, 498 (61.4%) visited the Goal module (252/403 motivation and self-regulation and 246/408 combined), and 406 (50.1%) participants visited the Barriers module (217/403 motivation and self-regulation and 189/408 combined). There were 628 visits to the Goal module, of which 309 were complete visits. The Barriers module had 514 visits, with 345 complete. The background characteristics of the visitors and nonvisitors of these modules are presented and compared in Multimedia Appendix 3.

The Goal module was first visited a median of 55 days (IQR 48-64) after the first login to TK, and the Barriers module was first visited a median of 98 days (IQR 91-110) after the first login. Tables 1 and 2 present the median and IQR for the activity metrics before and after visiting the Goal and Barriers modules, respectively. Visiting the Barriers module increased very active and total active minutes.

When only completed module visits were included, the results for the Goal module remained nonsignificant. For the Barriers module, the effects remained and became slightly stronger (for very active minutes: before median 22.4, IQR 10.1-41.0 vs after median 25.0, IQR 10.1-46.1; P=.007; and for total active minutes: before median 42.6, IQR 22.4-73.4 vs after median 46.6, IQR 21.8-79.4; P=.008).

Table 3 presents the 8-week maintenance of the Barriers module effect for the 3 active minute metrics based on completed module visits only. Total active minutes and fairly active minutes were higher than before module completion for most of the 8-week period following the module. Also, very active minutes remained higher for 3 weeks after module completion. Values that significantly differed from the before value were denoted.

This paper investigates the immediate changes in measured physical activity after visiting 2 web-based intervention modules targeting physical activity goals and action planning (Goal), and coping planning (Barriers). The Barriers module had a significant but small, immediate effect on very active and total active minutes during the week after visiting the module. When only completed module visits were considered, increases were larger. Module completion was also associated with sustained increases in all categories of active minutes, of which fairly active minutes and total active minutes were sustained for most of the 8-week period following module completion, and very active minutes for the first 3 weeks. Coping planning, addressed in the Barriers module, has previously been identified as a core self-regulation technique that can directly impact behavior (eg, [11,12]), and our findings support this effect for physical activity. The Goal module did not show similar effects on physical activity. One explanation for the differences in the results is that the coping planning activity in the Barriers module asked participants to identify barriers and strategies to overcome them in the immediate future (following week), while in the Goal module participants were asked to set a goal to increase their activity moderately in an unspecified timeframe.

The strengths of this study include the large number of participants and the ability to monitor the exposure to specific intervention techniques as well as the subsequent behavior.

The reliability of the variables may be affected by differences in individual wear times of the trackers. This was mitigated by setting a threshold of 1000 steps to consider the day valid. There is also a known tendency for Fitbit (and other trackers) to overestimate the amount of moderate to vigorous activity [13,14]. Furthermore, we used multiple observations for some volunteers to maximize the sample size. Although the rank-based nonparametric tests could not accommodate a random effect to account for the multiple observations, we checked an analysis, which averaged the observations per participant and found that the results and conclusions were the same.

The timing of the modules may have impacted the results. Although the order of visiting the modules was not technically restricted, the Goal session appeared earlier in the intervention flow and was visited by more participants than the Barriers session. It is thus possible that participants visiting the Barriers session were more committed to the intervention and, therefore, more likely to adhere to behavioral guidance as well. Further, engagement was only assessed based on log files and not by the quality of the action and coping plans done by the user, or the enactment of these.

As far as we know, no prior research to which we can directly compare these results exists. This shows the need for more research to examine the direct and immediate as well as longer-term effects of engaging with DBCI content on the enactment of the target behavior. Specifically, such work constitutes an important step toward identifying which behavior change techniques can have a differential impact on physical activity. This knowledge can, in turn, contribute to optimizing DBCIs in an adaptive and personalized way.

A self-regulation–based intervention module addressing physical activity barriers induced a significant increase in active minutes, and the effect was stronger when the module was completed. Module completion was further associated with sustained increases, especially in fairly active and total active minutes.