The Fogg Behavior Model (FBM) provides a structured framework for understanding human behavior in the context of technology use and digital engagement (18). The model aims to support behavior change through persuasive design, and asserts that behavior is a product of three factors: motivation, ability, and triggers, and that for a person to perform a target behavior, they must be sufficiently motivated, have the ability to perform the behavior, and be triggered to perform the behavior (18). This model is particularly useful in the design of digital health technologies, where engaging users effectively requires understanding and influencing their motivations, enhancing their ability to use the technology often by making the technology easy to use, and providing timely and appropriate prompts or reminders. Persuasive system design (PSD) builds upon FBM to describe technology design techniques that can motivate behavior change and user engagement, comprising of 28 persuasive strategies grouped into four categories—primary task support, dialogue support, system credibility support, and social support (19). The evidence on how effective these approached are in improving engagement is limited. In one recent review, a negative association between PSD features and engagement (as measured by completion rate) was observed (20). Another found mixed evidence that a specific strategy or group of strategies is relatively more efficacious in promoting engagement than others (21).

Self Determination Theory (SDT) is one of the most prominent and empirically supported theories of human motivation that has demonstrated efficacy in predicting motivated behavior in multiple contexts and populations and for a variety of health behaviors such as physical activity, healthy eating, and smoking cessation (22). The quality of motivation is influenced by the extent to which individuals experience support for three basic psychological needs: autonomy, competence, and relatedness. Therefore, behaviors or messages from agents that support the satisfaction of these needs are likely to promote autonomous motivation and sustained behavior change within the individual, while those that do not may undermine autonomous motivation and lead to negative outcomes, such as disengagement or poor adherence.

In acknowledging the value of support in improving engagement and adherence with psychological interventions, Mohr and colleagues constructed the Supportive Accountability (SA) model (23), which is particularly relevant in the context of remote interventions where there is limited direct human interaction. Support emphasizes the encouragement, guidance and feedback from a trained coach or therapist. Accountability refers to the user justifying their actions or inactions to another person such as the coach or therapist. Both SDT and SA emphasize the importance of interpersonal interactions and relationships on motivation and engagement.

The above theories and models are widely used and referenced when developing digital interventions, though each when used on its own has limitations when applied to the use of digital interventions to treat mental ill health. For example, in their extensive review of engagement with DMHI, Borghouts and colleagues (17) coded 3 categories of barriers and facilitators: 1. user characteristics, including demographics such as age and education, personality traits, mental health status and severity, beliefs, technology experience and skills and life integration, 2. program characteristics including type of content, perceived fit, perceived usefulness, level of guidance or support, social connectedness, perceived impact or effectiveness of the intervention, and 3. technology and environment characteristics which include technology related factors, technical issues, usability and delivery platform, privacy and confidentiality, social influence by external others, and implementation factors. Based on this evidence, it’s clear that a more comprehensive and holistic models of digital psychological treatment engagement are required. Models that consider the complex interaction between personal capacity and the burdensomeness of the technologies and interventions themselves. For example, each of the above models do not adequately explain the complex interacting relationship between the personal resources of the patient/user, the burden of their illness, the burden of their current life circumstance and the burden placed upon them by the DMHI (including both the technology and the intervention itself). In the following section we outline an approach to conceptualizing treatment engagement and adherence used more widely in chronic disease research that has direct relevance for digital psychological interventions.

Being a patient can be hard work (30–32). In addition to the burden of illness (BOI) experienced by the individual (expanded on below), individuals may experience concurrent burdens in the ordinary demands of daily living (life demands). When seeking treatment for a mental health problem, there is additional burden or work placed on the patient (or a user of a digital intervention). This is known as the Burden of Treatment (BOT). Both impact complexity through their toll on the persons time, effort, and attention (24).

Capacity can be thought of as the sum-total of resources and abilities that a patient can draw on to access care, use care, and enact self-care. Every structure or system (including patients) has a maximum performable workload that is determined by its unique capacity (32). Capacity is affected by the availability or scarcity of personal resources, be they intrinsic (such as physical, mental, personal attitudes and beliefs, personality, intelligence, literacy) or extrinsic (such as socio-economic resources, social supports, education level, income, amount of free time). Furthermore, capacity is negatively affected by the burdens associated with their disease or illness itself (BOI). The interaction between personal resources and BOI affect the person’s ability or readiness to do the ‘work’ of treatment (24). Importantly, capacity varies between individuals; some persevere despite tremendous workloads; others falter even when relatively unencumbered (24). According to Shippe et al. (24), measurable attributes of capacity include the following: 1. amount/magnitude (e.g., greater/lesser symptoms, finances, or social support); 2. controllability (some factors, such as literacy, are more responsive to personal efforts than others, such as pain); 3. extensiveness or scope of impact (symptoms may have limited or widespread effects on functioning).

The core assertion of the model is that where there is an imbalance between the person’s workload and their capacity to meet those demands, poor engagement and subsequently poor outcomes ensue (24). Collectively, a person’s capacity may or may not reach that required to meet the prerequisites of sustained engagement in any psychological treatment. When workload exceeds capacity, people feel the effects of treatment burden and this can lead to de-prioritization of various aspects of care and ultimately result in poor fidelity to treatment programs and to treatment failures (28, 30, 62–64). An example is reading content in a DMHI. The act of reading is considered work and people vary in their capacity to complete this work. Those with high reading ability will not experience this work as burdensome. Those with reading difficulties will find that the reading workload exceeds their reading capacity and will find this burdensome, leading to disengagement. Furthermore, people experiencing this imbalance may disengage or clinically deteriorate, leading service providers to provide more care, often unintentionally fueling burden and creating a perpetuating cycle of increasing complexity. Approaches like stepped care models or adaptive interventions increase the intensity and nature of treatment in response poor engagement or poor treatment outcome (65) but in doing so, may increase the burden of treatment in a potentially already ‘overloaded’ person. This is why most stepped care models recommend greater levels of professional support as a means to increase capacity in the face of increase illness and treatment burden (65).

Given these complexities, multiple interactions may shape patient demands, capacity, and the interplay between them ( Figure 1 ). Even though there is significant evidence of both burden and capacity related factors in isolation influencing engagement, limited research has been conducted on these complex interactions over time. Yet it is their interaction at the individual level which is most likely to be predictive of engagement and subsequent outcome, which may explain the variability in findings when these factors are studied in isolation. The goal is to balance the recommended treatment workload with the unique capacity of the individual to avoid overload and to adapt this balance over the course of care. Engagement research, however, rarely examines how these complex combinations interact, which inhibits our ability to design, test and deliver more appropriate interventions.

The workload-capacity imbalance is most apparent in standardized DMHI (and other psychological treatments for that matter), where all users receive the same intervention regardless of their capacity. While useful for research standardization and fidelity, these ‘one-size-fits-all’ approaches disregard individual variation in illness burden and personal resources. Treatment demands are fixed and usually specify a set amount of time to spend, a set amount to read, what tasks to complete, and the frequency and length of support appointments to participate in. The result as outlined above is poor engagement for large numbers of participants in these standardized programs.

Shared decision making and informed consent are critical before commencing any health intervention (68) following an adequately comprehensive assessment of the user at entry ( Figure 2 ). Users should have a complete understanding of the potential benefits, burdens, demands, harms and risks of starting a proposed intervention. These kinds of discussions with MHPs are valued by users of DMHI (69). Indeed, a recent meta-analysis showed that synchronous contact before the digital intervention commenced significantly improved intervention adherence and outcome (70). For example, if the intervention requires an 8-week enrolment, weekly reading, regularly questionnaire completion and regular telephone contact with the clinician, these requirements should be stated clearly up front, and the user themselves will decide as to their capacity to engage in the suggested work to obtain their desired goal. Just like other mental health interventions, DMHIs are not suitable for all. It is important that the intended user’s assessed capacity is balanced against the fixed burden of the intended intervention, and that this is openly and transparently discussed and consented to prior to commencement. If this were to occur, less users might start these interventions, but conceivably more would engage. At present, we have many users who fail to complete an intervention, which either maintains or exacerbates their burden of illness, and can lead to ‘failure demand’, where failing to adequately treat the condition the first time creates later downstream demand on services (64, 71).

Providing additional professional support has been associated with improved engagement (39). A meta-analysis of 40 studies with a total of 7313 participants found an overall completion rate of 43%, with completion rates increasing with greater levels of support: 26% without support, 62% administrative support only, and 72% for therapeutic support (40). Studies have also explored the role of social support systems in psychological treatments, including peer support groups, community-based programs, and social networking (72, 73). These interventions aim to provide emotional and practical support, reducing feelings of isolation, shape more positive attitudes toward intervention and provide additional encouragement or motivation to do the work of therapy (74). Providing additional social supports may also be warranted, such as access to career consultants or services (75), information or direction to services that can assist with financial counselling, domestic violence, housing and accommodation. Digital Navigators are becoming increasingly popular as a way to support MHP and users with digital literacy, workflow integration (for clinicians) and engagement, although further work is required to standardize the scope of these roles (76).