This study adhered to the CONSORT (Consolidated Standards of Reporting Trials) 2010 guidelines for randomized controlled trials (RCT) [19,20]. The original study was designed as a cluster-RCT [21]. However, disruptions caused by the COVID-19 pandemic and recruitment challenges necessitated its adaptation to a standard RCT with convenience sampling. Key changes included adjusting the estimated sample size from 495 (cluster-RCT design) to 135 (RCT design), replacing face-to-face workshops with e-workshops conducted via Zoom (Zoom Communications), and omitting workplace-level environmental changes.

The intervention was developed using the IM framework, a systematic and iterative approach for designing evidence-based health promotion programs. The IM framework comprises 6 steps, with the first 4 steps focusing on the development of the intervention. These steps provided a structured, theory-based, and evidence-based process to create a tailored intervention addressing the needs of the target population. Figure 1 summarizes the IM framework and the tasks involved in each step.

Step 1 identified the health problem, its determinants, and the needs of the target population. Step 2 defined the intervention’s expected outcomes, specified performance objectives, and developed a matrix of change objectives. In step 3, evidence-based behavior change methods were selected and translated into practical applications to address these objectives. Finally, step 4 focused on preparing and pilot-testing intervention materials to ensure feasibility, acceptability, and usability. This process enabled the development of a comprehensive and adaptable intervention, even under the constraints posed by the COVID-19 pandemic.

This study was a 24-week, 3-arm RCT. Participants were randomly allocated to one of the 3 groups—a web-based intervention group, a blended intervention group (combining web-based with e-workshops), or a control group. Assessments were conducted at baseline (T1), postintervention (T2, 12 wk), and follow-up (T3, 24 wk). All participants accessed the same website, which consisted of 2 main modules, a “Library” module with generic MVPA information and an “Intervention” module with personalized, theory-driven content (accessible only to the web-based and blended intervention groups). Additionally, participants in the blended intervention group attended 3 e-workshops at weeks 2, 4, and 8. All participants received fortnightly reminders to engage with the website.

This study was approved by the Research Ethics Committee of Hong Kong Baptist University (approval HASC/17-18/0529). All participants provided electronic informed consent via the QuestionPro platform after being fully informed of the study’s objectives, procedures, potential risks, and benefits. The informed consent process also granted permission for secondary analysis of the collected data without requiring additional consent.

Identifiable personal data will be destroyed after the study is completed, while deidentified datasets will be retained for research purposes and can be requested from the corresponding author (YG) upon reasonable request. Participants were compensated with HK $50 (approximately US $6.40) coupons for each completed evaluation session. Additionally, those who completed all study requirements were entered into a lottery to win a HK $1000 (approximately US $128; quota: 5) coupon. As part of the study benefits, participants also received a personalized physical activity and health tracking report summarizing their progress and providing actionable insights. No identifiable images or information about participants are included in the manuscript or supplementary materials.

Eligible participants were office employees in Hong Kong aged 18 years or older, engaging in less than 150 minutes of MVPA per week according to WHO guidelines, assessed using the short-form International Physical Activity Questionnaire [22]. To ensure safety, participants completed the Physical Activity Readiness Questionnaire, and those who answered “yes” to any question were required to obtain medical clearance before participation. Individuals enrolled in other MVPA promotion programs or with conditions preventing physical activity were excluded.

Recruitment was conducted remotely during the pandemic of COVID-19 (September-December 2022) when many employees were working from home. A prerandomized list of 440 companies, representing 20% of eligible Hong Kong companies with more than 100 employees, was obtained from the Hong Kong Census and Statistics Department. Due to pandemic-related concerns, most companies declined participation or did not respond, with only 4 agreeing to disseminate study materials to their office employees. Social media platforms, such as Facebook (Meta), Instagram (Meta), and WeChat (Tencent Holdings Limited), were also used, resulting in the recruitment of 88 participants. Referrals were encouraged through unique tracking links shared by interested individuals with colleagues, friends, or family members who met the eligibility criteria, which led to the recruitment of an additional 27 participants. In total, 172 individuals were assessed for eligibility, and 141 participants (82%) completed baseline enrollment.

Participants were randomly assigned to the 3 study groups using block randomization with a fixed block size of 6 to ensure balanced allocation throughout enrollment. The randomization sequence was generated using a computer-based random number generator and administered by an independent research team member who was not involved in the intervention or outcome assessment. Allocation concealment was ensured, and outcome assessors remained blinded to group assignments.

The sample size was calculated using repeated measures ANOVA within-between interaction approach (G*Power, version 3.1.9.6), where a power set at 0.80 and an α at .05, with 3 groups and 3 measurement waves. An effect size of 0.30 (Cohen d) was chosen based on a similar study in terms of design and intervention components [23], and on a median effect size of 0.30 (IQR 0.16-0.38), calculated from 19 studies reporting Cohen d in a recent systematic review of blended interventions [17]. To account for an expected attrition rate of 30%, the required sample size was estimated at 135 participants, with 45 participants allocated to each group [24].

MVPA was objectively measured by a hip-worn accelerometer (ActiGraph, Models wGT3X-BT) at T1-T3, and data were analyzed by using ActiLife software v6.13.4 (ActiGraph LLC). The participants were instructed to wear an accelerometer on the right hip during waking hours for 7 consecutive days, except when bathing, performing water-based activities, and sleeping [30]. Freedson cutoff points were used to classify VPA, MPA, and light-intensity physical activity [31]. Data of at least 10 hours per day for a minimum of 3 days (including at least 1 weekend day) were regarded as valid and included in data analysis [32,33]. Then, the total weekly MVPA was calculated by multiplying the average daily MVPA levels by 7.

The retention rate (%) was calculated as participants at the end of the intervention divided by the participants at baseline.

The engagement rate (%) was calculated by the number of intervention sessions attended by participants divided by the total number of sessions provided to them. In the blended group, engagement rates were calculated for web-based sessions and e-workshop sessions separately and in combination.

The acceptability of the intervention was evaluated using a 5-point Likert scale (from “1=strongly dissatisfied” to “5=strongly satisfied”) across several dimensions, such as ease of use, understandability, usefulness, enjoyability, time acceptability, and overall satisfaction [34,35]. Statements included, for example, “The content of the intervention is understandable” and “The time taken to complete the content of the intervention was acceptable.” For the evaluation of the e-workshops, similar statements were adapted. Responses were grouped, combining “agree” and “strongly agree” into one percentage, and “disagree” and “strongly disagree” into another, to reflect participant agreement or disagreement.

MVPA (min/wk) was summarized as mean and SD at each assessment time point based on valid ActiGraph data. Primary intervention effects at T2 and T3 were evaluated under an intention-to-treat (ITT) framework using generalized linear mixed models with a γ distribution and a log link. Participants were analyzed as randomized. The generalized linear mixed models included fixed effects for group, time, and the group×time interaction, with participant treated as a random effect. Models were adjusted for sex, ActiGraph wear time, and baseline MVPA. Between-group intervention effects were estimated as differences in change from baseline using the group×time interaction (difference-in-differences); pairwise contrasts were obtained by changing reference groups where needed. For completeness, estimated within-group changes from baseline are also presented.

At baseline, MVPA values were unavailable for 15 participants because ActiGraph recordings did not meet the prespecified wear-time validity criteria. These baseline values were replaced using within-group mean imputation for the primary analyses [36]. Sensitivity analyses used multiple imputation (MI) for MVPA across all 3 time points, generating 5 imputations [37,38]. The MI model incorporated baseline values and other variables associated with MVPA to improve estimation accuracy.

Retention and engagement rates were compared across groups using chi-square tests. Statistical significance was set at P<.05, with 95% CIs reported. All analyses were performed in IBM SPSS Statistics (version 29.0.1).

This study evaluated the effectiveness of a blended intervention, combining web-based content with interactive e-workshops, in promoting MVPA among physically inactive office employees. The findings demonstrated that the blended intervention significantly increased MVPA levels at both T2 and T3, outperforming the web-based intervention and the control group. Retention rates were high across all groups, with the blended group achieving the highest rate, further supporting the feasibility and acceptability of the intervention. These results underscore the potential of blended approaches in addressing physical inactivity among office employees, a population particularly vulnerable to sedentary behaviors.

The significant improvements in MVPA levels observed in the blended group at both T2 and T3 underscore the effectiveness of integrating web-based tools with interactive e-workshops. Compared with the web-based and control groups, the blended intervention achieved the most substantial increases in MVPA, highlighting its potential to surpass web-only or traditional approaches. In contrast, the web-based group showed modest improvements, while the control group exhibited minimal changes, with both groups showing significant changes at T3.

The success of the blended intervention can be attributed to its comprehensive, theory-based design, which incorporated a total of 34 BCTs targeting motivation, self-regulation, and habit formation. Previous meta-analyses have demonstrated that interventions using multiple BCTs grounded in behavioral theories tend to produce stronger effects on physical activity [39,40]. While the web-based intervention also used 31 BCTs, the blended intervention expanded upon these by incorporating 3 additional BCTs unique to the e-workshops. These additional BCTs further enhanced the intervention’s overall effectiveness by promoting interpersonal interaction, fostering peer support, and providing real-time feedback and accountability.

The group-based nature, combined with the digital person-to-person component of the e-workshops, played a pivotal role in the success of the blended intervention. The group-based format fostered interpersonal interaction and collaboration, encouraging participants to share experiences, brainstorm solutions to common barriers, and build self-efficacy [41]. Peer support within the group likely strengthened participants’ motivation and commitment, as they could learn from others’ strategies and successes, while also feeling a sense of collective accountability. At the same time, the digital person-to-person component added a human support that is often missing in purely digital interventions [42]. Unlike the impersonal nature of human-computer interactions, face-to-face communication offered warmth, empathy, and dynamic support. Facilitators were able to provide real-time feedback and adaptive guidance, promptly addressing participants’ individual challenges. Together, these features fostered a supportive and interactive environment that not only promoted encouragement, belonging, and accountability but also strengthened the application of behavior change theories targeting motivation, volition, and habit formation. By reinforcing these theoretical foundations, the blended approach amplified and consolidated the impact of the web-based content, ensuring a more robust and sustainable behavior change process.

In contrast, the web-based and control groups lacked these collaborative and interactive elements, which may explain the superior effectiveness of the blended intervention. While the web-based intervention provided valuable tools for self-regulation and goal-setting, it did not offer the same level of social reinforcement, dynamic problem-solving, or accountability as the blended approach. These findings align with previous research emphasizing the superiority of blended interventions, which effectively combine the flexibility of digital tools with the social and interactive benefits of face-to-face components [17].

The long-term improvements in MVPA levels observed at T3 suggest that the blended intervention may have facilitated long-term behavior change. This aligns with habit formation theory, which posits that consistent behaviors are more likely to be maintained when reinforced in stable contexts and supported by environmental cues [43]. Periodic e-workshops likely served as motivational “boosters,” helping participants overcome lapses and maintain engagement over time. This is particularly critical for workplace interventions, where competing demands and time constraints often challenge long-term adherence.

However, the web-based intervention alone demonstrated limited impact on MVPA levels compared with the control group, despite offering tailored feedback based on behavioral determinants. This is consistent with previous studies reporting mixed results for digital-only interventions, especially those lacking interactive or social components [36,44]. The absence of interpersonal interaction and real-time feedback in the web-based group may have reduced participants’ sense of accountability and support—both key drivers of long-term behavior change [16]. Without the reinforcement provided by workshops, participants may have struggled to translate digital content into consistent physical activity behaviors. These findings highlight the critical role of interactive components, such as group-based e-workshops, in enhancing the effectiveness of digital interventions.

The findings highlight the feasibility and acceptability of the intervention, as evidenced by high retention rates and positive participant feedback. Notably, the retention rates in this study were relatively high compared with similar interventions, which typically report rates ranging from 59% to 82% [36,45-47]. This can be attributed to the systematic and evidence-based design of the intervention using IM, which ensured the intervention was theory-driven and tailored to meet the needs of the target population [48]. Furthermore, participant retention was supported by strategies such as providing MVPA and health progress reports, as well as incentives like supermarket coupons and a grand prize for completing all outcome assessments. These measures created a structured and supportive environment, encouraging long-term participation.

Directly comparing engagement across studies is challenging due to differences in how engagement is defined and measured. While our study calculated engagement as a rate, other studies have used metrics like mean session completions or participant distributions, complicating direct comparisons. Despite this, previous studies have shown that blended interventions often achieve better engagement than platform-only or control groups. For example, 1 study found that adding coaching significantly improved session completion rates [49], while another highlighted the value of interactive components in enhancing adherence [50]. In our study, the lack of significant differences in engagement may be explained by limited participation in e-workshops, likely due to scheduling challenges and competing responsibilities during personal time, which diminished the potential advantages of the blended format. Future research could explore strategies such as offering more flexible time slots to overcome these barriers and improve engagement in blended interventions.

The strengths of this study lie in its rigorous design, guided by the IM framework, which ensured a systematic and theory-driven approach to intervention development and evaluation. The use of objective measures, such as actigraphy, provided robust and reliable data on MVPA outcomes. Additionally, the high retention and engagement rates observed in the blended group demonstrate the intervention’s feasibility and acceptability in a workplace setting.

However, several limitations should be acknowledged. First, the ActiGraph device used to measure MVPA did not capture water-based activities, potentially underestimating MVPA levels among participants who engaged in swimming or similar activities. Second, the sample was skewed toward younger, highly educated, and digitally literate individuals, limiting the generalizability of the findings to other populations. Third, baseline MVPA levels in this study were relatively high, possibly due to discrepancies between self-reported activity during recruitment and objective measurements at baseline. This might have resulted in a ceiling effect, underestimating the intervention’s true potential. Finally, the study was conducted exclusively among office employees in Hong Kong, a dense urban setting with unique cultural and environmental factors, which may limit the applicability of the findings to other contexts.

This study provides strong evidence for the effectiveness of a blended intervention in promoting MVPA among physically inactive office employees. By integrating web-based platforms with interactive e-workshops, the intervention effectively addressed key barriers to MVPA and achieved significant improvements in engagement, retention, and behavioral outcomes. These findings highlight the potential of blended approaches to address physical inactivity in workplace settings and beyond.

Future research should build on these findings by exploring the long-term sustainability of behavioral changes achieved through blended interventions and their scalability across diverse populations and settings. Additionally, leveraging emerging technologies, such as artificial intelligence, to deliver personalized feedback and adaptive interventions could further enhance engagement and effectiveness. Addressing these questions will be critical for advancing workplace health promotion and tackling the global physical inactivity crisis.