This 20-week, 4-armed, pragmatic RCT’s primary outcomes were exercise adherence, aerobic capacity, and muscular strength. Hence, this study presents a secondary analysis of the intervention’s effect on perceived motivation, barriers to exercise, self-efficacy, social support, and health-related quality of life.

Participants were invited through Facebook and Instagram advertisements posted by various fitness clubs across rural and urban locations in Norway (7 different counties and 12 different municipalities). The recruitment was ongoing in February 2023, with eligible participants enrolled continuously. Unfortunately, data on the total number of interested participants were not collected. The inclusion criteria were BMI ≥30 kg/m², aged 18 to 65 years, no membership in a fitness club 6 months before recruitment, being classified as low activity (<150 minutes of moderate-intensity or 75 minutes of vigorous-intensity physical activity per week), Norwegian speaking, and having a smartphone. Exclusion criteria were chronic disease or injuries (eg, severe hypertension 180/110 mm Hg, heart disease, lung disease, or functional impairment) that hindered participation in exercise and planned leave during the intervention period. To ensure that none of the participants had any contraindications for exercise, all participants underwent a general health screening by their general practitioner.

A total of 188 participants were included, and informed consent was signed to participate. A blinded statistician performed random allocation (1:1:1:1) following a simple computer-based randomization program (R for Windows, version 4.2.2; R Foundation for Statistical Computing), and the participants were allocated to (1) once a week (in-person exercise coaching 100% [IP100], n=47), (2) twice a month (in-person exercise coaching 50% [IP50], n=47), or (3) once a month (in-person exercise coaching 25% [IP25], n=47) in-person coaching, or as controls (n=47). The intervention groups were provided the same weekly frequency of follow-up, but in IP50 and IP25, a total of 50% (2/4) and 75% (3/4) of the follow-up sessions were replaced with 15-minute digital sessions, respectively. An overview of intervention arms is shown in Textbox 1.

A total of 23 exercise professionals, all employed as personal trainers, were recruited to provide follow-up for study participants. Trainers needed at least 3 years of part-time experience and to have led at least 80 exercise sessions/month recently. Of the 23 trainers, 10 had acquired certification as personal trainers, 8 held relevant bachelor’s or master’s degrees in fields such as exercise science or physiotherapy, and 5 had bachelor’s or master’s degrees in health and social care education. The trainers were instructed to follow up with each participant and create customized weekly exercise programs based on their needs. They also offered lifestyle advice, such as coaching in maintaining a healthy diet. Our study design did not allow for further masking of study participants or the personal trainers.

The analyses were performed with SPSS statistics (version 28; IBM Corporation), following a predefined analysis plan and before unmasking the intervention arms. The normal distribution of the data was assessed with the Shapiro-Wilk test. We used appropriate statistical tests to compare baseline characteristics between the current sample and those who dropped out, as well as to identify differences between groups. These included a chi-square test, an independent 2-tailed t test, or a 1-way ANOVA, as dictated by the data. Postintervention, group comparisons were done using a 1-way ANOVA with Bonferroni post hoc comparisons. We also calculated Hedges g effect size, with values around 0.2, 0.5, and 0.8 interpreted as weak, medium, and strong effect sizes, respectively. Data are presented as mean (SD), numbers (%), mean changes, mean differences, 95% CI, effect sizes, and P value. The level of statistical significance was set at P≤.05 and adjusted as appropriate for the 1-way ANOVA and Bonferroni post hoc comparison (P≤.01).

In this study, targeting a specific population of women with obesity who have low activity presented challenges in accurately estimating attrition rates due to limited comparable data. Given the lack of reliable data on expected dropout rates within this demographic, the research group assumed recruitment of 200 participants was appropriate.

The study was conducted following the ethical standards of the Helsinki Declaration (as revised in 2000) and relevant national and institutional guidelines. All participants provided written informed consent after receiving detailed study information, including their rights to withdraw at any time without consequence. The study was reviewed by the Regional Committee for Medical and Health Research Ethics (REK 2022/552476), which determined that, under the Act on Medical and Health Research (the Health Research Act 2008), full review by the Regional Committee for Medical and Health Research Ethics was not required. Ethical approval was granted by the Norwegian School of Sports Sciences Ethical Committee (Applications number 262). The study was also registered with the Norwegian Social Science Data Service (Sikt 104437) and ClinicalTrials.gov (NCT05792657).

To protect participant privacy, no identifying information was collected or published. Participants in the intervention arms received regular follow-ups by a personal trainer, while all study groups, including the control group, had access to a fitness club and an exercise app (ABEL Technologies). No financial compensation was provided to participants.

This pragmatic RCT investigated the effectiveness of weekly in-person coaching compared with 2 different combinations of in-person and web-based coaching on motivation, barriers, self-efficacy, social support, and health-related quality of life among women with obesity. While no intervention effects were detected for most of the measured variables, a slight improvement was reported in self-perceived health in 1 of the 8 subscales of the SF-36. This improvement was consistent across all intervention groups when compared with the control group from before to after the intervention. However, it is important to note that the observed difference was limited to a single subscale of the SF-36, indicating that the interventions had only a minimal impact on overall health-related quality of life.

The improvement in the “change in health” subscale of the SF-36 across all intervention groups suggests that combining in-person and web-based exercise coaching may contribute to enhanced self-perceived health among women with obesity. This is consistent with a previous RCT involving 143 inactive women with excess body weight, which found that those who received in-person exercise coaching by a personal trainer reported improved self-perceived health after 12 weeks of intervention [42]. Interestingly, this RCT also found that those who exercised alone reported better self-perceived health compared with the control group [42]. This may be due to the fact that regular physical activity, even at low levels (≥1 day/week), has been shown to improve self-perceived health [57]. In a previous longitudinal prospective study among novice exercisers in a fitness club setting, regular exercise in the fitness club (≥2 days/week) was associated with high self-perceived health (odds ratio 3.53, 95% CI 1.60-7.82; P=.002) [58]. Therefore, regular exercise itself may improve self-perceived health, emphasizing exercise as a key factor in overall well-being. Although the use of a personal trainer may not directly influence self-perceived health, it seems to bolster exercise adherence, thus supporting long-term health benefits. This appears to be consistent regardless of in-person or web-based coaching, as we did not find differences in exercise frequency between the 3 intervention groups. However, since this improvement in our study was limited to just 1 of the 8 subscales of the SF-36, further research is necessary to determine whether this represents a true and clinically significant effect.

The intervention did not demonstrate significant effects on exercise motivation, which may be explained by several factors. One key factor is the baseline differences in intrinsic motivational regulation between the control group and the intervention arms, with the control group having higher intrinsic motivation at baseline. This disparity may have created a ceiling effect, limiting further improvements in motivation both within and between groups. However, all groups scored between 1.7 and 2.6 on the 5-point scale, indicating room for improvement, irrespective of the initial differences. In addition, variations in the coaching approaches of the personal trainers, which were not controlled for in this study, could influence exercise motivation. According to self-determination theory (SDT), interactions between the personal trainer and the client may enhance motivation for exercise by satisfying the basic psychological needs (autonomy, competence, and relatedness) [59]. Personal trainers who adopt a need-supportive coaching approach, such as providing meaningful choices, acknowledging client feelings, and offering constructive feedback, may foster these needs and thereby improve exercise motivation [30]. For instance, one study investigating the effects of an SDT-based physical activity program among adults with obesity (BMI ≥40 kg/m²) found that perceived autonomy support and the satisfaction of basic psychological were positively associated with increases in autonomous motivation and negatively associated with controlled motivation and amotivation [60]. On the other hand, certain coaching approaches, which include elements like guilt-inducing language or not fully considering the clients’ preferences, may inadvertently impact these psychological needs and reduce motivation [30].

By offering scheduled appointments, free follow-up sessions, and access to exercise equipment, the intervention effectively minimized many logistical and financial barriers to exercise. However, we found no changes in exercise barriers, with all intervention groups reporting high scores on barriers such as “priority” (lack of time and energy) after the intervention. These findings are consistent with previous studies [52,61-63], suggesting reducing logistical obstacles alone may not be sufficient to overcome exercise barriers. This highlights the importance of addressing more individual factors, such as personal priorities and attitudes toward exercise. Overcoming these barriers likely requires coaching strategies that go beyond logistical support to include more personalized approaches tailored to each individual’s needs and circumstances [64,65].

The lack of effect on self-efficacy in our study is in line with 2 previous RCTs, which reported no differences between in-person and phone-based exercise coaching [66,67]. While one cross-sectional study found an association between in-person coaching and improvement in self-efficacy compared with web-based coaching [68], this was not reflected in our results. One potential explanation for the varying results could be the high baseline self-efficacy scores across all groups, indicating that participants who already had high self-efficacy may have had limited room for further improvement [31,37]. Nevertheless, it is important to note that all intervention arms in our study exercised significantly more than the control group after the intervention, even without a corresponding increase in self-efficacy. This suggests that improvements in exercise adherence may occur independently of self-efficacy changes. Future research should explore additional support mechanisms, such as coaching approaches grounded in behavioral theories, to further enhance exercise participation. For instance, one study that applied social cognitive theory–based coaching, which focuses on psychological factors such as self-efficacy, found that women in the coaching group had higher exercise adherence (49% versus 31%) compared with the control group, alongside improvements in self-efficacy [69].

The measurement used in our study included perceived social support from friends and family, unfortunately not accounting for potential support the participants might have experienced from their personal trainer. Including data on social support from personal trainers would have offered a more comprehensive understanding of participants’ support networks, while also highlighting the trainers’ role in promoting exercise adherence. In addition, it could have provided valuable insights into the dynamics of the trainer-participant relationship. However, low social support scores, both before and after the intervention, indicate a lack of sufficient encouragement from the significant others. This finding is important, as social support from family and friends is a key factor that consistently influences exercise adherence [25,26].

Our findings align with the broader literature, highlighting the complex relationship between exercise interventions and psychosocial outcomes among individuals with obesity [70,71]. For instance, Pilates, yoga, tai chi, and qigong have demonstrated potential for improving mental well-being in populations living with obesity [70,71]. However, the evidence regarding their direct effects on broader psychosocial factors remains limited. Further, a recent review showed clear evidence of combined aerobic and resistance training in improving key health indicators, including glycemic control and mental health, among individuals with obesity and type 2 diabetes [72], while our study did not demonstrate significant changes in health-related quality of life across all dimensions. The small improvement in the “change in health” subscale may reflect the cumulative benefit of structured exercise on self-perceived well-being. These results also support that regular exercise may enhance specific quality-of-life dimensions, even without observable changes in other psychosocial variables.

Finally, while this study investigated the effect of hybrid coaching models, targeted exercise regimens, such as high-intensity interval training, are shown to have the potential to improve psychosocial factors, including enhanced self-perceived health [73]. Further, exercise modalities like Pilates and yoga could also complement exercise coaching to improve psychosocial factors, since these approaches offer a less demanding, more inclusive alternative, particularly for individuals facing physical or psychological limitations [70,71,74]. Future research may integrate such modalities within hybrid coaching models to evaluate their combined impact on psychosocial outcomes.

Increasing exercise participation among the population living with obesity could significantly impact public health. Since motivation, barriers, self-efficacy, and social support are key to exercise adherence [25,75], research should explore how various coaching approaches affect these factors. For instance, SDT-based coaching may help personal trainers create a more supportive and motivating environment, improving adherence [30]. Despite the lack of effect on motivation, the intervention’s potential for positive impact lies in highlighting the importance of SDT-aligned coaching strategies, which could be key to improving motivation in future interventions targeting populations living with obesity. By integrating behavioral theories into coaching interventions, researchers may better identify strategies to improve motivation, reduce barriers, and strengthen self-efficacy, ultimately promoting sustainable exercise habits and improved health outcomes [76]. While the intervention did not reduce self-reported barriers, it minimized external challenges, highlighting the value of interventions that address personal barriers. Last, although in-person and web-based coaching showed similar effects, the potential cost-effectiveness and accessibility of web-based coaching make hybrid or fully web-based approaches appealing alternatives for supporting exercise adherence among populations living with obesity. This is especially relevant given that logistical and financial barriers often limit consistent participation in in-person coaching [43,44,46]. While our RCT did not demonstrate differences in psychosocial outcomes, it shed light on the feasibility and value of hybrid coaching models to support exercise adherence in obese populations.

Our study is one of the first global pragmatic RCTs of its kind, with a long intervention period and a unique catchment area, including a nationwide intervention spanning both urban and rural regions of Norway. The inclusion of 23 experienced personal trainers, each with at least 3 years of experience, further enhances the robustness and generalizability of our findings in a real-world setting. Another strength is the hybrid approach, combining web-based coaching with in-person sessions. This approach reflects the contemporary trends in exercise supervision and meets the evolving needs of participants in modern society. Our study also contributes to the United Nations’ Sustainable Development Goal 5 by focusing on obese women and addressing psychosocial factors through in-person and web-based exercise coaching, thereby supporting efforts to improve health equity and empower women. Finally, we used a range of validated questionnaires to measure the psychosocial outcomes, including motivation (BREQ-2), self-efficacy (Self-Efficacy Survey), social support, and health-related quality of life (SF-36), providing a comprehensive assessment of the psychosocial effects of the intervention.

Some study limitations should be considered when interpreting our results. We unfortunately had a notable dropout rate, with only 64% (120/188) of participants completing both the baseline and postintervention questionnaires. However, no significant differences were found between those who completed the study and those who dropped out in terms of baseline data. Further, the 20-week intervention period may be too short to fully assess the intervention’s effects, particularly on psychosocial factors such as self-efficacy, motivation, and health-related quality of life, which often require longer durations to reveal changes or differences. Finally, while this study was conducted in Norway, limiting direct global generalizability, culture and societal similarities between Norway and other Western countries suggests that our findings could apply to similar populations and contexts. Further, the lack of a priori power calculation is another limitation of this study. The sample size was determined based on practical considerations rather than statistical power affecting the robustness and generalizability of our findings, as the study may be underpowered to identify small effects. Last, the Shapiro-Wilk test was used to evaluate whether our data met the assumption of normality. While this test is widely recognized for assessing normality, it is sensitive to sample size and outliers, which may limit its ability to fully characterize data distributions. In our study, as many as 120 participants completed both baseline and postintervention questionnaires. Thus, we believe the results of the Shapiro-Wilk test align with the assumption of normality.

This study indicates that both in-person coaching and 2 combinations of in-person and web-based coaching may lead to a slight improvement in self-perceived health, by higher scores in 1 of the 8 SF-36 subscales. However, these observed improvements are preliminary and may not generalize to broader health or psychosocial domains without further evidence. No significant effects were found on motivation, barriers, self-efficacy, social support, or other health domains. Further research is needed to fully understand how we can increase the ability to maintain regular exercise among women with obesity.