The characteristics of the studies, including year, country, theoretical framework, sample, method, design, and analysis, are shown in Supplementary material 1. Almost all interventions examined in the studies were inherently digital, utilizing tools such as internet browsers, smartphone applications, and instant messaging platforms. However, two studies [47–48] conducted comparative analyses between online and face-to-face formats of distinct interventions. The nature of these digital interventions varied considerably. Twenty-two interventions were exclusively web-based, employing internet browsers as the primary medium for delivery, encompassing websites and online platforms. Nine interventions were solely app-based, delivered through smartphone, tablet, or computer applications. Additionally, four interventions incorporated a hybrid of web- and app-based components. Two interventions leveraged instant messaging platforms compatible with both smartphones and computers. Most interventions were initially designed for digital delivery. Nonetheless, there were digital adaptations of previously validated face-to-face interventions, particularly evident in mindfulness-based interventions [e.g., 49] and those translated to online formats due to Covid-19 restrictions [e.g., 50].
Context Factors of Digital Interventions
Table 1 shows the summary of Contexts, Mechanisms and Outcomes in our selected papers. Regarding the Context (C), at the individual level, the most frequent omnibus context factors related to employees’ roles or job positions, gender, health and well-being, previous knowledge, and personal resources/skills. Most studies included employees from various occupations, roles, and areas within organizations, regardless of gender, age, or other sociodemographic factors. However, some research focused on specific sectors or job positions, including healthcare professionals and school system employees [51–52] social workers [53], self-employed workers [54], IT employees [55], and manufacturing/sales company employees [56].
Also, the studies predominantly involved female participants and generally included individuals over 18 years of age (Supplementary material 1). Although gender and age were sometimes used as control variables, they often did not significantly relate to the outcomes, with few exceptions, such as Keller et al. [57] who found age positively influenced self-efficacy and Pandya [58], who noted the DI's greater effectiveness among males in reducing emotional exhaustion and increasing resilience.
Of the 44 selected studies, 24 considered employees’ prior health and well-being levels. These were divided into two categories: a) studies targeting employees with specific baseline health conditions, such as high levels of burnout or work-related stress [e.g., 59-62] and b) studies focusing on non-clinical populations without severe mental or physical disorders [e.g., 55-56, 63-64]. The remaining 29 studies targeted a universal population, with some examining how baseline health or well-being affected DI outcomes [e.g., 48, 52, 65].
Eight studies considered participants' previous knowledge and familiarity with the DI's subject matter. Six studies excluded those with prior meditation experience from mindfulness-based Dis [52, 63, 66-69], while Nadler et al. [70] found mindfulness DIs beneficial even for those with prior experience. Makowska-Tłomak et al. [61] assessed prior Information and Communication Technology (ICT) use, noting its impact on engagement during the pandemic.
Five studies explored the influence of employees' resources or skills on DI outcomes. Makowska-Tłomak et al. [61] examined readiness for change; Bormann et al. [71] focused on attitudes toward religion/spirituality; Li et al. [63] investigated trait self-compassion; Althammer et al. [49] looked at segmentation preferences as a moderator; and Bazarco et al. [66] used nurses’ performance standing as an inclusion criterion.
At the interpersonal level, two contextual factors were identified, one omnibus and another one discrete factor. Cantarero et al. [72] explored the impact of interaction and communication types on study outcomes as omnibus context factor. They examined the effect of the number of people participants had contact with during COVID-19, both in person and via internet/phone. The results indicated that employees who interacted via phone or the internet experienced greater satisfaction with basic psychological needs and well-being than those who had in-person interactions. On the opposite, as discrete context factor, three studies addressed the impact of support from leaders/supervisors and colleagues during the implementation of DIs. Shann et al. [73] found that support and commitment from other leaders in the workplace influenced training transfer during a digital leadership intervention. Ouweneel et al. [74] observed that the lack of support from supervisors and colleagues participating in similar interventions negatively influenced the content and effectiveness of the DI on work engagement. Tonkin et al. [75] demonstrated that organizations where senior managers encouraged participation and provided resources saw higher uptake and engagement in a well-being DI.
Several context factors at the organizational level were identified, which can be categorized into four groups, namely organizational culture, organizational change, economic incentives, and societal and cultural issues. As factors related to the organizational culture, participation in well-being-related interventions, considered as discrete context factor, was often voluntary. Neumeier et al. [54] found that self-selected employees in digital well-being interventions were more motivated and reported greater well-being gains. On the contrary, Shann et al. [73] identified the collective readiness and capability of the organization, as well as existing workplace activities and strategies related to mental health, as an omnibus context factor influencing training transfer.
Shann et al. [73] reported that training transfer was affected by organizational changes as variations in government and political priorities. IJntema et al. [76] found that resilience improved in response to changes in the work environment and conditions due to governmental policies and during a merger process.
Five studies used economic incentives as a motivational strategy for DI participation. Examples include gift vouchers [48, 54, 75], monetary rewards [66-67], continuing education credits, and wellness points [77]. However, Smith et al. [77] also required a $50 out-of-pocket fee for participation.
Finally, eight studies (47, 50, 52, 53, 67, 61, 72, 78) noted the influence of the societal and cultural issues, particularly during infectious disease outbreaks like the Covid-19 pandemic, on the implementation period of DIs.
Mechanisms of Digital Interventions
Out of 44 reviewed studies, 29 addressed the working mechanisms of interventions, with 20 testing their effects on outcomes. These mechanisms were primarily found at the individual level (n=9), and one at the interpersonal level. No mechanisms were discerned at the organizational level. Of the nine mechanisms identified at the individual level, six correspond to process, two at perception and one at the content. The following process mechanisms were detected in the studies: DI usage, frequency of practice, implementation adherence, training transfer, modality, duration of the DI, and external support.
Examples of DI usage triggering the study outcomes include the use of a digital transformation stress intervention for coping with stress [61], DI usage behavior tested in the form of participants using the App once or twice a day, perusing videos and learning sessions, and self-practicing regularly [58], reading the psychoeducational content related to stress and management practice [79], the use of an activity tracker to collect and self-monitor health information such as daily number of steps or energy consumption [80] and the usability of the DI itself in the form of information to read, video clips about mental health in the workplace, interactive exercises, and action plan implementation [73].
Frequency of practice was mostly addressed as mindfulness meditation practice duration (i.e., amount of time) in using web or mobile apps interventions [59, 66-68, 71]. Other examples of frequency of practice were the number of recovery activities per week [51] and the time spent on well-being-related activities in day-to-day life [75].
A third process mechanism was implementation adherence. This mechanism was reported in terms of the dose received by participants as compared to the dose delivered. Specifically, some of the studies assessed the number of sessions [50, 59; 64] or modules [51, 81] attended. In all studies, higher implementation adherence predicted post-intervention improvements in well-being and mental health outcomes.
Fourthly, training transfer was considered an important process mechanism in two studies. In one of these [73] the transfer of learning back to the workplace was activated by specific context factors (i.e., collective readiness, attitudes of others, organizational changes, low levels of stigma) and, as a result, mental health was promoted, and depression-related stigma reduced. In the other study [82] training transfer was addressed in the form of deploying knowledge and skills learned through resilience-building. Greater opportunities to put learned skills to use led to improvements in well-being-related outcomes (ibidem).
To a lesser extent, the mechanism of modality was also addressed. For example, [82] assessed the method of program delivery (computer-based/group-based classroom/one-on-one/ train-the-trainer) and found that, while interventions employing a one-on-one format were the most effective, the computer-based delivery formats were the least effective in triggering well-being outcomes. Carolan et al. [14] conducted a systematic review and revealed that studies that utilize secondary modalities for delivering the DIs and engaging users (i.e., e-mails and text messages, SMS) and use elements of persuasive technology (i.e., self-monitoring and tailoring) may achieve greater engagement and adherence, which lead to increases in psychological well-being and work effectiveness.
The mechanism of duration of the DIs was also considered in Carolan and colleagues’ [14] systematic review, in terms of DIs delivered over a shorter time frame (i.e., 6 to 7 weeks) leading to higher engagement and adherence than DIs of longer duration.
Finally, one process mechanism, namely external support, was identified at the interpersonal level and referred to the guidance or supervision provided by a facilitator during the intervention process. Two studies addressed this mechanism. Carolan and colleagues’ [14] systematic review suggested that interventions that achieve the greatest engagement and adherence offer some form of guidance, such as therapist, coach, a coordinator or member of staff, and clinical psychologist. More recently, Ijntema and colleagues [76] tested the role of the strength of the coach-client working relationship and found that it was related to most of the immediate program effects. The intervention seemed most effective for employees who experienced a stronger coach-client working relationship.
Shann and colleagues [73] identified two perception mechanisms, namely attitude change and relevance of (the DI) content, which impact the study outcomes. Utilizing a qualitative method through interviews with a selected group of leaders, the authors explored which mechanisms were activated by the participants that influenced these outcomes. The results revealed that the sustainability of attitude change and the relevance of the implementation content (DI) during the implementation period assisted participants in overcoming workplace stigma and improving mental health.
Finally, one content mechanism was tested in one study [69] in which facets of mindfulness (acting with awareness, describing, nonjudging, and non-reacting) were assessed as mechanisms of change. The intervention’s effect was primarily explained by increased levels of only one facet of mindfulness, that is, acting with awareness. Another content mechanism of change was addressed by Uglanova and Dettmers [83] where participants trained job crafting competencies on a single task to be able to apply this to other tasks. However, no changes in task crafting were observed over time. Thus, the authors were unable to explicitly demonstrate that task crafting skills were trained within the course of the intervention.
Outcomes of Digital Interventions
Thirty-three studies highlighted positive well-being outcomes, while thirty-six studies focused on reducing or preventing ill-health outcomes. All these outcomes were observed at the individual level.
Positive psychological outcomes were grouped in three categories. The first group was related to personal growth and well-being. DIs significantly enhanced various resources and aspects of psychological well-being. Prominent among these were mindfulness [i.e., 67], resilience [i.e., 78], self-compassion [i.e., 63], self-efficacy [i.e., 74], purpose in life [i.e., 52, 76], positive relationships [i.e., 84] and positive affect [i.e., 74]. Some studies reported additional benefits like spiritual well-being [i.e., 71], secondary posttraumatic growth [i.e., 60], hope [i.e., 76], empathy [i.e., 66], emotional intelligence [i.e., 70], flourishing [i.e., 81], and satisfaction of basic psychological needs [i.e., 72]. The second group of positive outcomes found among the included studies was related to health and recovery and included general health (i.e., [i.e., 80], sleep quality [i.e., 67, 69], and recovery from stress [i.e., 76]. A third group consisted of positive work skills and attitudes enhanced by DIs, and included job performance/effectiveness [i.e., 82], work engagement [i.e., 47], return to work [i.e., 85], resourcefulness and coping style [i.e., 47], psychological flexibility [i.e., 53], goal-striving reasons [i.e., 47], and work competences [i.e., 70].
Negative psychological outcomes were grouped into two categories. The first one was related to mental health challenges. The adverse outcomes mitigated through participation in the DIs that were most frequently documented by the studies referred to stress [i.e., 65], depressive symptoms [i.e., 55], burnout [i.e., 62], anxiety [i.e., 56], and negative affect [i.e., 77]. Less frequently documented ill-being outcomes included perseverative thinking [i.e., 52], secondary traumatic stress symptoms [i.e., 60, 50], social distance [i.e., 84], depression-related stigma [i.e., 73], and fear of COVID-19 [i.e., 67]. The second group consisted of work-related challenges that were improved after participating on a Dis, such as work–family conflict [i.e., 49], turnover intention [i.e., 48], work-related fatigue [i.e., 69], and digital transformation stress [i.e., 61].
[PLEASE INSERT TABLE 1 HERE]
Realist propositions
Based on the findings from the previous sections, we first selected the mechanisms that were most used or addressed within the studies we included. Next, we identified contextual factors (C) linked to such mechanisms (M), and the outcomes they triggered. As a result, we developed the five CMO configurations that are expressed in the form of the following five realist propositions. The goal of these propositions is to guide future research and provide new theoretical and practical insights once they are tested.
Realist proposition 1: DI usage
IF certain contextual factors exist (e.g., collective readiness, leadership support, employee willingness to discuss mental health), THEN DI usage (the mechanisms) improves mental health knowledge, decreases stress, and increases resilience.
Realist proposition 2: Frequency of practice
IF certain preconditions exist (e.g., prior work-related illness, receiving incentives), THEN frequent practice of DI activities (the mechanisms, i.e., mindfulness meditations, recovery activities, and well-being-related activities) improves outcomes like mindfulness and decreases ill-being.
Realist proposition 3: Implementation adherence
IF certain context factors exist (e.g., manager encouragement, persuasive technology), THEN strong implementation adherence (the mechanism, in terms of the dose received by participants) improves outcomes like personal resources and psychological well-being. However, online overload may hinder engagement.
Realist proposition 4: Training transfer
IF certain context factors exist (e.g., collective readiness, leadership support, organizational strategies and willingness to discuss mental health), THEN training transfer back to work (the mechanism) promotes positive affect and reduces psychological difficulties.
Realist proposition 5: Modality
For at-risk employees experiencing stress and lacking core protective factors (context factor), one-on-one and technologically enhanced DIs (the mechanism) promote engagement, well-being, and work effectiveness over computer-based interventions alone.