This study was a convergent mixed methods secondary analysis of a nonrandomized controlled trial [32,33]. It aimed to investigate the influence of HCWs’ well-being and cognitive variables on their adherence to MSM use. This study was reported in compliance with the Good Reporting of a Mixed Methods Study [34]. We acquired secondary data from HCWs who participated in the MSM program at a large general hospital in China from August 11, 2021, to January 31, 2022 [32,33]. Information on the research intervention can be found on the ClinicalTrials.gov website (NCT05239065).

The inclusion criteria were (1) currently employed, (2) self-reported having internet access, (3) did not have any serious medical problems (eg, cancer or stroke), and (4) were willing to participate in this study and sign the consent form. Participants were excluded if they (1) intended to leave the hospital or units for any reason or (2) were in insufficient medical condition to be followed up on.

Sociodemographic characteristics included sex (male or female), age (y), marriage status (married or unmarried), educational background (undergraduate degree or postgraduate degree), position (physician, nurse, or other), working hours (h), income (Chinese Yuan [CNY]), and BMI.

A convenience sample of HCWs was recruited from a large general hospital in China between August 11, 2021, and January 31, 2022. Participants were enrolled from an ongoing MSM program.

This study was a secondary analysis of the original MSM program data. In the primary trial, 179 participants were ultimately allocated to and received the intervention. For the purposes of the current analysis, which designated the Fatigue Assessment Scale (FAS) as a key outcome, we applied a strict criterion of complete data: any record with one or more missing items on the 10-item FAS was excluded to ensure the integrity of the primary outcome measure without imputation. According to the inclusion and exclusion criteria, after applying this data completeness rule, 157 participants were ultimately enrolled and completed the quantitative survey. The sample size for the qualitative component was guided by the principle of information saturation, whereby recruitment continued until no substantially new themes emerged from the data. Ultimately, 96 individuals voluntarily agreed to participate.

This study was approved by the Institutional Review Board (IRB) of the Medical Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, China (approval number: IRB number 2021S141). All participants provided informed consent electronically prior to participation. For the secondary analysis reported in this manuscript, the original IRB approval and consent procedures permitted the use of deidentified data. All data were collected and stored anonymously using coded identifiers. No personal identifying information was linked to the survey responses, ensuring participant confidentiality. Participants did not receive any financial compensation for their involvement in the study. This manuscript does not contain any personal images, audio or video recordings, or text excerpts that could lead to the identification of an individual participant.

There were 157 participants in total, with 135 (86%) female participants and 22 (14%) male participants. In terms of age, the mean age was 33.7 (SD 4.9) years, 35 (22.3%) participants were under the age of 30 years, 97 (61.8%) were between the ages of 30 and 40 years, and 25 (15.9%) were over the age of 40 years. In terms of marital status, 115 (73.2%) participants were married, while 42 (26.8%) were single. In terms of position, 30 participants (19.1%) were physicians, 104 (66.2%) were nurses, and 23 (14.7%) were other HCWs. When it came to family income, 18 (11.5%) of the participants had less than 10,000 CNY (1 CNY≈0.155 USD). 81 (51.6%) participants had a family income of 10,000 to 15,000 CNY, 35 (22.3%) had a family income of 15,001 to 20,000 CNY, and 23 (14.6%) had a family income of more than 20,000 CNY. When it came to working hours, 22 (14%) participants worked less than 35 hours per week, 74 (47.1%) worked 35 to 44 hours per week, and 61 (38.9%) worked more than 44 hours per week (Table 1).

Table 2 displays descriptive results for the main study variables. Participants rated perceived usefulness as 2.96 out of 5.0 (SD 0.85), perceived feasibility as 3.71 out of 5.0 (SD 0.72), and perceived enjoyment as 3.59 out of 5.0 (SD 0.69), in that order. The HCWs rated their experience with MSM as 32.03 out of 40.0 (SD 5.01), and the “quality of weekly feedback reports” received the highest mean score of 4.17 (SD 0.73). Participants rated their adherence to MSM for physical activity as 2.91 out of 5.0 (SD 1.15), deep breathing as 3.25 out of 5.0 (SD 1.34), and mindfulness as 2.92 out of 5.0 (SD 1.33), on average.

For the hierarchical regression analysis, 4 models were obtained. As shown in Table 3, the demographic variables (sex, age, marital status, position, income, and working hours) were entered into the model in the first step, and no variable was found to be significant in explaining the variance of treatment adherence. When perceived stress and fatigue symptoms were included in step 2, the percentage of treatment adherence explained by the predictors significantly increased from R²=0.043 in model 1 to R²=0.093 in model 2, and fatigue symptom was demonstrated to be a significant predictor (b=−0.050; t₁₄₉=−2.859; and 2-tailed P=.005). Then, after user experience was included as the predictor in step 3, the percentage of treatment adherence explained by the predictors significantly increased from R²=0.093 in model 2 to R²=0.205 in model 3, and user experience was demonstrated to be a significant predictor (b=0.074; t₁₄₈=4.569; and 2-tailed P<.001). In the last step, when attitude was included as the predictor, the percentage of treatment adherence explained by the predictors significantly increased from R²=0.205 in model 3 to R²=0.315 in model 4, while attitude was shown to be a significant predictor (b=0.238; t₁₄₇=4.863; and 2-tailed P<.001), the significance of user experience was shown to be disappearing in model 4 (b=0.031; t₁₄₇=1.770; and P=.08). All the predictors of model 4 together explained 31.5% of the variance of treatment adherence (R²=0.315; P<.001).

According to Hu and Bentler’s [38] suggested model fit indicators, the model in this paper had a good fitting degree. Table 4 shows that the χ²/df values were less than 3, the root mean square error of approximation and standardized root mean square residual values were less than 0.80, and the nonnormed fit index, comparative fit index, and adjusted goodness of fit index values were all greater than 0.90, all of which were within the range of acceptable fit indexes. To summarize, the results of confirmatory factor analysis confirm that the empirical model comprising attitude, experience, and adherence has good validity; the collected data are highly reliable; and the research findings are robust.

Figure 1 depicts the results of attitude mediation between experience and adherence to MSM. The path coefficient from experience to attitude was 0.079 (P<.001), and the path coefficient from attitude to adherence was 0.568 (P<.001). Meanwhile, the Bootstrapping estimates revealed that the direct influence of experience on adherence was not significant (direct effect=0.002; P=.86), implying that experience had a full indirect effect on adherence via the attitude toward MSM (indirect effect=0.047; P<.001).

The open-ended survey received 96 responses from 157 participants, which were identified and categorized into four themes: (1) individualized intervention, (2) effective feedback, (3) reward and constraint mechanisms, and (4) duration of intervention. Each theme could have 3 subthemes. Table 5 summarizes the themes, subthemes, and sample quotes.

This research focused on how individual treatment adherence differed depending on their well-being and cognitive factors. By using a convergent mixed methods approach, we specifically examined the direct effects of well-being factors on treatment adherence and the indirect effects of cognitive factors on treatment adherence, as well as learning about the specific requirements of HCWs for MSM interventions.

Overall, adherence to the internet-based MSM was generally modest, with mindfulness and physical activity techniques showing lower adherence compared to deep breathing exercises. The results are consistent with previous studies [20,22], despite the reduction in dropouts in the internet-based intervention compared to face-to-face, participants often fail to have complete adherence. While physical activity and mindfulness are evidence-based techniques, their lower adherence in our sample may relate to the demanding work schedules of HCWs [8,16,19,39]. A previous study showed that breathing exercises can reduce stress levels even in the short term (mean time of 1.9, SD 1.2 min) [40], which means that breathing exercises may be more available for HCWs who spend most of their time providing health care. HCWs are a special group of people with heavy workloads and limited job autonomy. The time spent setting requirements and the accessibility of management programs were priorities in stress management for this group of people. Therefore, in the future, the stress management of specific interventions needs to be based on the working conditions such as the working hours or environment of different HCWs, taking into account the feasibility of developing a personalized management plan to improve adherence.

On the issue of burnout symptoms, this study revealed that fatigue symptoms were a significant predictor rather than perceived stress. An explanation for this finding is that perceived stress did not reach a statistically significant level in relation to burnout. One possible reason for this could be attributed to insufficient training hours, while another factor may be the nurses’ anticipation of encountering stressful work in their future shifts. Additionally, according to Maslow’s hierarchy of needs theory, fatigue symptoms represent fundamental and immediate needs, whereas perceived stress can potentially be managed through various methods by HCWs. As professional practitioners, HCWs consciously adapt themselves to reduce stress levels; however, it is important to consider potential biases resulting from psychological effects during the design phase of this study.

Our findings regarding the prominence of fatigue over perceived stress align with broader postpandemic research on psychological well-being, which suggests a shift in the salient components of well-being. Babazadeh [41], using a multidimensional measure of psychological well-being, found that factors like health locus of control and social support were significant predictors, underscoring the importance of targeting specific well-being components and cognitive factors in interventions rather than focusing solely on global stress measures. The significant role of fatigue in our study may reflect its centrality within the somatic symptoms dimension of well-being, which appears highly relevant for HCWs. Furthermore, fatigue and burnout may have intertwined effects, as evidenced by a prospective cohort study reporting that individuals in the “burnout and fatigue” group had a higher proportion of chronic course type compared to those experiencing “pure burnout” or “pure fatigue.” Severe burnout and fatigue symptoms become interconnected with each other as well as with other psychological and/or somatic complaints [42]. However, the specific association between fatigue and specific factors remains unclear [43].

The most interesting finding is that the attitude mediates experience and adherence. Our mediation analysis revealed a significant indirect effect of user experience on adherence through attitudes, while the direct effect was not significant. Most existing studies have emphasized the importance of user experience on treatment adherence [8,20]. However, the direct relationship between the experience and adherence is limited to a definite explanation. On the contrary, the discussion of attitude and adherence is a classic issue, and the amount of evidence reports the relationship between attitude and adherence in stress management interventions [8,44,45]. Our research indicates that the direct impact of user experience on treatment adherence disappeared after inputting the variable of attitudes. An implication of the further relationship is the possibility that the improvement of experience with the object of adherence increasing is to make the attitude of the internet-based mobile program or management positive, despite the weak significant association between the experience and attitude. This may provide another perspective on retention in stress management, and a similar relationship in other psychological health of HCWs should be developed in the future.

We also identify the 4 aspects of individualized intervention, effective feedback, reward or constraint mechanisms, and duration of intervention as certain factors to adherence to internet-based MSM. Generally, all the factors highlight the personalized service throughout the whole intervention. The high quality and individual needs of intervention contents are more vital; inadequately supported contents are a hindrance. Furthermore, HCWs, as a special population with a health care education background, face the problem of their personal situations, may not have time to integrate exercises into their daily lives, and may need more personal conversations with the eCoach during the staging and feedback stages. This is in line with previous observations [46-48]. In the reward or constraint mechanisms, HCWs required the increasing demands outwards, which addressed the role of the organization in the interventions. Proving reminders, encouraging each other, and using rewards as the external triggers were taken as effective measures to increase adherence in other evidence [49,50]. To the duration of intervention, extensive studies have pointed out that stress was reduced with the increasing duration of intervention [9,51,52]. Considering that health care often requires demanding physical and psychological efforts, HCWs often have little energy to implement stress management strategies on a daily basis. Therefore, future studies may need to be more focused on less exhausting interventions for those working in health care contexts.

Some limitations in the paper also need to be taken into consideration. First, the exploratory study investigated a small sample size with a short-term intervention. Second, the data collection of qualitative research conducted online could result in lacking information. Furthermore, the implication of this paper is that participants in qualitative research may have been positively biased in their support of the intervention because all the interviewees finished the intervention. It would be valuable in future work to extend the sample, focus on those who chose to drop out of the intervention, and explore the reasons for dropping out.

The findings of this study highlight a central pathway for engagement in digital interventions and point to clear practical directions. By innovatively integrating quantitative modeling with qualitative assessment, this research moves beyond merely describing barriers to adherence. It thereby differentiates itself from prior work by quantitatively establishing the critical mediating role of participant attitudes between user experience and sustained use. This contributes a key insight to the field: successful engagement depends not just on the intervention itself, but on fostering positive attitudes through that experience. Consequently, the real-world implication is straightforward. To optimize adherence, MSM programs for HCWs must prioritize enhancing user experience via personalized content, actionable feedback, incentive mechanisms, and flexible duration, all essential for cultivating the positive attitudes needed for long-term use in high-demand settings.

In summary, by innovatively integrating quantitative modeling with qualitative assessment, this study demonstrates that the value of user experience in stress management adherence to internet-based mobile intervention is fully mediated by participant attitudes, highlighting a key insight that differentiates our work from previous barrier-focused research. For HCWs, especially in high-demand settings, adherence is enhanced by interventions that are perceived as personalized, with actionable feedback, incentive mechanisms, and flexible duration. Future research should incorporate additional mediating variables through larger-scale and longer-term studies to further elucidate these mechanisms.