Essential Design Principles for a Family Digital Mental Health Intervention: A Delphi Study

ABSTRACT

Systemic family therapy aims to improve mental health and well‐being in the face of challenge by optimizing relational health between family members or people in close relationships. The potential of digital interventions for family mental health is underexplored. This study aimed to understand core principles that facilitate end‐user acceptability and engagement with such a resource. A modified Delphi study was conducted to identify key factors, barriers, and risks influencing family engagement with a digital mental health intervention (DMHI). The 18 participants comprised consumers (n = 5), family therapists (n = 8), and digital mental health professionals (n = 5). Participants completed online questionnaires over two rounds. An agreement was reached on 23 essential factors, within four design principles: informed choice, anonymity and privacy, accessibility, and support and connection. Implications for the design of family DMHIs are considered alongside existing knowledge about individually focused DMHIs.

Systemic family therapy aims to improve mental health outcomes and support healthy family development by addressing the relationships between individuals and members of their family and social network (Carr 2012). Globally, the accessibility of mental health support is recognized as a significant public health concern (Wainberg et al. 2017; Whiteford et al. 2013). However, barriers to accessing family therapy are considerably more pronounced than those affecting general mental health services. Family therapy is a specialist modality, and most services are concentrated in metropolitan areas, making it difficult for families in rural or remote locations to access care. Telehealth family therapy services assist (McLean et al. 2021), but equally are associated with wait times that hinder timely intervention.

Digital mental health interventions (DMHIs) encompass websites and mobile‐based applications that offer self‐help resources or deliver self‐guided interventions intended to support users' mental health and well‐being. They may include some contact with a therapist; however, in this paper, telehealth services alone are not classified as DMHIs. DMHIs for individuals address many accessibility challenges for solo access, by providing scalable, flexible, and geographically independent solutions with fewer required resources than in‐person delivery (Gan et al. 2022; Lipschitz et al. 2023; Opie et al. 2024). It can therefore be assumed that DMHIs for families would also reduce accessibility challenges and have the potential to reach underserved populations.

This said, DMHIs for individuals continue to demonstrate poor engagement in real‐world settings, characterized by limited adoption and high dropout rates. Engagement with a DMHI is a dynamic process occurring across many time points, including pre‐use and initial sign‐up, first use, returning and sustained use, and ultimate disengagement (Borghouts et al. 2021; Nadal et al. 2020; Yardley et al. 2016). Effective engagement with a DMHI is clearly necessary for improved clinical outcomes (Lo et al. 2022; Yardley et al. 2016). As such, understanding how to design a DMHI to promote engagement is a crucial first step in the development of a DMHI for families.

The design of DMHIs must accommodate a variety of factors influencing engagement at all stages. The term “factors” includes individual, technical, and environmental considerations that could act as a barrier, risk, or facilitator to engagement with a DMHI. A recent systematic review identified a complex set of factors influencing user engagement with DMHIs. These factors included: age, gender, and education; individual motivation levels; individual mental health status; individual technology experience and skills; perceived fit; and privacy and confidentiality (Borghouts et al. 2021). In addition, an individual's preference for content modality and the perceived personalized nature of the intervention were shown to influence user engagement. Another review attributed low engagement to poor usability, lack of user‐centric design, and lack of trust, echoing privacy concerns (Torous et al. 2018).

While design factors influencing engagement with DMHIs for individuals are increasingly evident, a recent systematic review highlighted a lack of evidence regarding the effective design of DMHIs for families (Welsh et al. 2024). This gap is significant because many existing engagement factors identified are inherently individualistic and may become more complex when engaging multiple members of a family. Families consist of members of diverse ages, mental health needs, preferences, and technology skills, all of which must be accommodated to ensure meaningful engagement. Privacy and confidentiality concerns are likely to be particularly nuanced in family‐based interventions, as they extend beyond protecting user data to managing the privacy dynamics between family members participating together. Further, the inherent clinical risks of engaging multiple individuals within a shared online space—including, but not limited to, potential interpersonal conflicts and unintended disclosures of information—require careful consideration and mitigation in the design of such platforms. These considerations underscore the importance of developing family‐focused design principles to ensure DMHIs are effective and safe for all users in the family therapy context.

This study employed a novel, multidisciplinary approach to understanding what factors could influence family engagement with a DMHI, incorporating perspectives from three groups of experts: consumers with lived experience of attending at least one family therapy session (hereafter referred to as “consumers”), family therapists, and digital mental health professionals (defined as having researched, designed, or evaluated an online DMHI). Particular attention was paid to the areas of consensus across these three groups. Expert consensus methods, as employed in this study, aim to build the foundation upon which further research can be conducted (Jorm 2015; Minas and Jorm 2010), thereby providing crucial information for informing the future development of family‐based DMHIs. As reported by consumers, digital mental health professionals, and family therapists, the aims of this study were to:

• 1.Identify factors, barriers, and risks likely to influence family member engagement with a family DMHI.
• 2.Describe uniformity and indicate consensus on the strength of each factor, barrier, and risk in influencing family engagement with a family DMHI.

1. Methods

1.1. Study Design

A modified Delphi approach was used to address the aims of this study. The Delphi method has demonstrated acceptable construct validity and reliability (Cross 2005; Hutchings et al. 2006; Minas and Jorm 2010) and is ideal for reaching a consensus about a topic where there is uncertainty or little prior evidence (Jones and Hunter 1995; Shang 2023). The Delphi method involves assembling a group of experts (hereafter referred to as the “expert panel”) whose expertise is the most significant guarantee of quality outcomes (Stone Fish and Busby 2005). Panel diversity is more important than panel numbers, with the quality of expert opinions and group dynamics the drivers of consensus (Domlyn and Wandersman 2019; Okoli and Pawlowski 2004).

Experts respond to an online questionnaire over a number of iterative rounds (Avella 2016; Turoff and Linstone 2002). Between each round, a group‐level summary of the panel's responses to the previous round is provided, and panel members are offered the opportunity to modify their previous response. Expert panel members complete Delphi questionnaires online individually, allowing for a methodological approach that maintains anonymity and provides panel members with the opportunity to express their opinion free of social or hierarchical pressures or judgments (Kodish et al. 2023; Shang 2023). The Delphi method has been used in prior studies addressing issues of engagement in digital mental health (Kodish et al. 2023) and where the expert panel comprised both professional and consumer expertise (Bailey et al. 2024; Lo et al. 2022; Murphy et al. 2020; Zelmer et al. 2018).

1.2. Participants

The expert panel for this study was comprised of family therapists, digital mental health professionals, and consumers with lived experience of attending at least one family therapy session. Family therapists and consumers were not expected to have experience engaging with DMHIs. Using a purposive sampling technique, participants were self‐selected based on the criteria outlined in Table 1. An expert panel size of 10–18 respondents is generally considered sufficient to enable consensus to be achieved in a Delphi study (Domlyn and Wandersman 2019; Trevelyan and Robinson 2015).

Table 1.

Selection criteria for each user group in the study.

User group	Inclusion criteria
Family therapists	Self‐report having at least 3 years of experience delivering family therapy services in the public mental health system.
Digital mental health professionals	Self‐reported experience of having researched and/or designed and/or evaluated an online self‐guided mental health platform.
Consumers	An individual member of a family over the age of 18 who has participated in at least one family therapy session.

Recruitment materials were distributed by email to established networks known to the research team. Family therapists and digital mental health professionals were contacted via colleagues and professional networks to facilitate the recruitment of qualified participants who were more likely to meet selection criteria, engage meaningfully with the study, and contribute high‐quality data. Family members who had completed family therapy with The Bouverie Centre were contacted via an established family advisory network. Potential participants were asked whether they met one of the selection criteria as outlined in Table 1 and informed consent was obtained. Upon consenting, participants responded to demographic information and provided further details about their expertise as it related to the study. Participants in the consumer group were remunerated for their participation in the study. Participant characteristics and completion rates are detailed in Table 2.

Table 2.

Participant completion rate and characteristics.

	Family therapists	Digital mental health professionals	Consumers	Total
Recruited, N	8	5	5	18
Delphi Round 1, n (%)	8 (100%)	4 (80%)	5 (100%)	17 (94%)
Delphi Round 2, n (%)	8 (100%)	3 (60%)	5 (100%)	16 (89%)
Females, n (%)	7 (88%)	4 (80%)	4 (80%)	15 (83%)
Age, n (%)
36–45 years	1 (12%)	2 (40%)	0 (0%)	3 (17%)
46–55 years	3 (38%)	1 (20%)	2 (40%)	6 (33%)
56–65 years	4 (50%)	2 (40%)	1 (20%)	7 (39%)
65+ years	0 (0%)	0 (0%)	2 (40%)	2 (11%)
Expertise	Minimum 9 years postqualification as a family therapist	Studied digital health; developed/evaluated digital mental health sites	Experience attending family therapy

Eighteen participants were recruited to the expert panel. Almost all panel members completed Survey 1 and were eligible to complete Survey 2. Of those who received Survey 2, only one participant was lost to follow‐up, resulting in an overall completion rate of 89%. All family therapists and consumers completed all components of the study.

As seen in Table 2, self‐reported expertise across expert panel members was diverse. The expert panel was mostly female, and the majority were aged between 56 and 65 years, followed by those aged between 46 and 55 years. Family therapists comprised most of the sample (8/18; 44%) and, on average, members of the consumer group were older than family therapists and digital mental health professionals.

1.3. Procedure

The Delphi study was conducted by researchers at The Bouverie Centre, La Trobe University in Melbourne, Australia between August and December of 2024. Ethics approval for this study was obtained by La Trobe University in August 2024 (HEC24299).

The Delphi survey was informed by a prior scoping review (Welsh et al. 2024) and was conducted over stages, as depicted in Figure 1.

Figure 1.

Delphi procedure overview. IQR, interquartile range. [Color figure can be viewed at wileyonlinelibrary.com]

1.3.1. Stage 1: Scoping Review and Reference Groups to Create the Delphi Survey

Recognizing the critical role of questionnaire design in ensuring the quality of Delphi study outcomes, a literature review was conducted to inform the development of questionnaire statements, consistent with established practice (Jorm 2015) and its application in other digital health Delphi studies (Bailey et al. 2024). The scoping review was used to identify DMHI build and design characteristics that promote accessibility and engagement and enable co‐completion by families and was conducted by the research team before this study (Welsh et al. 2024). The research team generated a set of 79 statements from the themes identified in this review.

Given the limited literature on family engagement with DMHIs, reference groups were conducted to validate the proposed Delphi questionnaire. The goal of the reference groups was to refine the questionnaire through the addition of any missing items, removal of redundant items, and clarification of existing items to ensure consistent comprehension. All panel members were invited to participate in a reference group. Twelve panel members attended the family therapist (n = 4), digital mental health (n = 3), and consumer reference groups (n = 5). Following the reference groups, the research team assessed all suggestions to ensure they were within the scope of the study and incorporated relevant feedback.

This process resulted in a Delphi questionnaire containing 62 statements across the following categories: initial engagement (n = 17 statements), subsequent engagement (n = 16 statements), accessibility of content (n = 15 statements), safety (n = 5 statements), co‐completion by multiple family members (n = 4 statements), and integration with family therapy service delivery (n = 5 statements).

1.3.2. Stage 2: Delphi Questionnaire Rounds and Analysis

Round 1. A link to the final online Delphi questionnaire was emailed to the expert panel. Panel members were asked to rate the perceived importance of each of the 62 statements as relevant to influencing family engagement with a DMHI using a seven‐point Likert scale (where 1 = not at all important and 7 = essential). For example, participants were instructed to consider the statement “… feel confident using the internet” and indicate its importance on a scale of 1 (not at all important) to 7 (essential).

Data collection for Round 1 remained open for 2 weeks, after which the responses were deidentified and analyzed to determine which had reached consensus. Consensus definition in Delphi studies varies depending on the study context (Hsu and Sandford 2019). Consistent with prior digital health Delphi studies (Blease et al. 2021; Kodish et al. 2023; Lo et al. 2022; Shen 2019), consensus was defined as an interquartile range (IQR) less than or equal to 1. Additionally, statements were classified as essential if the median response rating was 7 and achieved consensus. Statements that did not reach consensus in Round 1 were redistributed to the expert panel in a second‐round survey.

Round 2. A second online questionnaire was developed, containing 24 statements from the Round 1 Delphi questionnaire that had not reached consensus. Panel members who participated in Round 1 were emailed an invitation with a link to the Round 2 Delphi questionnaire and a personalized PDF “Participant Response Summary.” The Participant Response Summary included a table listing all statements that had not reached consensus (i.e., IQR > 1), along with that specific panel member's previous response (rated from 1 to 7) and the mean and range of responses from all panel members for each statement. Panel members were instructed to use the Participant Response Summary to review their prior rating of these statements in light of the group's feedback to consider re‐rating each statement on the same seven‐point Likert scale. If participants chose not to revise their previous response and their score differed from the group average, they were encouraged to provide an explanation in a free‐text box included in the questionnaire. This process allowed participants to reflect on their responses and offer insights into any disagreements with the group consensus. The review of participant free‐text responses did not influence the final results, but the research team viewed the results to explore possible response variations and gain insights into differing perspectives. Additionally, median response ratings were analyzed and compared across subsamples to identify any notable differences or patterns in uniformity. Within user groups, subsample sizes were smaller than the typical sample size used in Delphi studies to determine consensus (Domlyn and Wandersman 2019; Trevelyan and Robinson 2015), and therefore the term “uniformity” was used where a user group subsample achieved an IQR less than or equal to 1.

Data and analysis code for this study are available by emailing the corresponding author.

2. Results

Out of 62 statements assessed in Round 1, 38 statements (38/62; 61%) achieved consensus by the expert panel, while 24 statements (24/62; 39%) were redistributed to participants for further evaluation in Round 2. After completing both rounds of data collection, 53 statements (53/62; 85%) reached consensus. Among these, 50 statements (50/62; 94%) had a median of 5 or higher, and 43% (23/50) of these were classified as essential, having achieved a median rating of 7. Figure 2 contains the 23 essential statements identified by the expert panel through this process.

Figure 2.

Essential statements arising from two‐rounds of Delphi questionnaires. DMHI, digital mental health intervention.

Appendix 1 presents the results of the Delphi method. The statements are categorized and written as they were presented to participants. The 24 statements that did not achieve consensus in Round 1 and were redistributed in Round 2 are indicated. The IQR and median values are presented overall and within user groups.

2.1. Uniformity Across User Groups

Of the 53 statements that drew consensus, 21 (40%) showed uniformity within all three user groups. Uniformity was observed within only two user groups on 25 (25/53; 47%) statements, and six (6/53; 11%) showed uniformity within one user group. One statement (looks modern and contemporary) gained consensus overall, but there was variation within all groups.

Consumers were most likely to demonstrate uniformity, with 54 (54/62; 87%) statements achieving uniformity within that group. The family therapist and digital health groups showed uniformity on 37 (37/62; 60%) statements each. The consumer and digital health groups had the greatest overlap, both achieving uniformity on the same 34 (55%) statements, followed by consumers and family therapists (33/62; 53%) and finally, digital health and family therapist user groups (24/62; 39%).

Within the consumer user group, 32 (32/62; 52%) statements were deemed essential, followed by 10 (10/62; 16%) each for family therapists and digital health user groups. Consumers and digital health participants were most likely to agree on which statements were essential (9/62; 15%). This was followed closely by consumer and family therapist user groups (8/62; 13%) and digital health and family therapist user groups (4/62; 6%).

Of the 62 statements presented to participants, four (4/62; 6%) were deemed essential across all three user groups. These statements were: families can complete content and activities at their own pace; the hub is accessible for people with disabilities; the hub has crisis support links easily accessible; and, for families later engaging in family therapy, they would have continued access after they have finished therapy.

2.2. Variation Across User Groups

Figure 3 presents the median values where consensus was not achieved across groups. Of the nine statements that did not achieve consensus across groups, six (6/9; 67%) achieved uniformity within at least one user group. The remaining three (3/9; 33%) statements demonstrated variation within all user groups. These were: has the option to message or connect with other families via chat rooms or discussion boards; has a discrete name and web address; and feel hopeful things in the family can get better.

Figure 3.

Median response ratings to each statement where consensus was not achieved by the user group. [Color figure can be viewed at wileyonlinelibrary.com]

2.3. Findings by Category

Overall, the categories Accessing Content and Safety and Privacy had the greatest proportion of statements that were deemed essential, with 67% (10/15) and 60% (3/5) of statements, respectively. No statements in the category examining factors influencing family member co‐completion were deemed essential though all statements achieved consensus. Only two (2/16; 13%) statements in the category regarding sustained engagement were deemed essential.

3. Discussion

This study investigated factors influencing family engagement with DMHIs from the perspectives of consumers with lived experience attending at least one family therapy session, digital mental health professionals, and family therapists. Twenty‐three statements were deemed essential by the expert panel. For ease of interpretation and to support further research and development, these 23 statements have been summarized and discussed below under four categories, hereafter termed “essential design principles.” These design principles complement what is already known about barriers and facilitators to individual engagement and cover crucial considerations for designing a family DMHI. Most importantly, they encompass the key factors identified by the expert panel that influence a family's engagement with a DMHI. As such, these principles could usefully and readily inform the design of family DMHIs. These essential design principles will be discussed in turn, including implications for implementation and possible avenues for future research.

3.1. Informed Choice

In summarizing the relevant statements, the principle of informed choice emerged. This describes the provision of adequate information to family members to inform their decision to engage at all stages and embedding choice in how (by which mode of delivery and on any device), when (at their own pace), and with whom (alone or with family members) they engage with a DMHI.

Choice is not widely discussed in literature examining factors related to engagement with DMHIs. Personalization is more commonly referenced as a way to promote engagement, where an intervention seeks to provide users with content that is personally relevant and customizable according to their unique needs and preferences (Borghouts et al. 2021; Gan et al. 2022; Oinas‐Kukkonen and Harjumaa 2009). There are subtle differences between personalization and choice as expressed by the expert panel in this study. Personalization may be one way a DMHI offers choice; however, choice is inherently consumer‐led rather than intervention‐driven. Choice is also a cornerstone of effective systemic psychotherapies (Cook et al. 2017; Dourdouma et al. 2020).

Additionally, DMHIs are predominantly offered as course‐like programs, where content is completed sequentially at a pace dictated by the intervention (Lattie et al. 2022; Welsh et al. 2024). Though this structure provides advantages for measuring clinical effectiveness and efficacy, and in many cases is a direct translation of an in‐person model of care, the principle of informed choice does not align with this approach. Single Session Thinking (SST) acknowledges that a significant proportion of clients will only attend one session and therefore strives to provide adequate support in a single session, assuming it may be the last, while leaving room for further sessions if desired (Hoyt et al. 2020). SST has demonstrated clinical efficacy in family‐therapy contexts (Hartley et al. 2023) and could support embedding informed choice into the structure and design of a family DMHI. For example, the principles of SST could be applied to a family DMHI by developing content in standalone activities such that families have the choice to return, and if they choose not to, the content they completed offered something useful in itself. Conversely, another essential statement within this category related to families having the option to build upon work they have already done, which, as a standalone statement, may seem to contradict an SST approach. Rather, this demonstrates how fundamental the principle of informed choice is, whereby families should be offered options at all stages of their engagement with a DMHI (e.g., to complete discrete once‐off activities or not).

A further finding regarding informed choice related to an agreement among the expert panel that individuals using the DMHI should always have a choice about whether they participate with other family members. A family DMHI must, therefore, accommodate both individuals and multiple users, leading to further complexity in design. Although no statement regarding co‐completion was rated as essential, all achieved consensus and were rated as very important (6 out of 7) and offered some suggestions for ways in which both individuals and multiple users can be accommodated. Such design features included: activities designed to initiate conversations; the capacity to share activities with other family members and/or to “tag” such activities for future joint discussions; the option to complete activities collaboratively offline; and shared activities designed for individual completion at the user's convenience. Together, findings supported design features that embed the principle of informed choice in both content design and architecture, and as such, are foundational to family DMHI creation. Establishing this principle in the very early stages of DMHI development would be essential for success. Codesign with consumers is an important first step to ensure informed choice is embedded considerately.

3.2. Anonymity and Privacy

The expert panel agreed on the importance of being able to access a family DMHI anonymously, and that information about individuals and their activity completion is kept private from other participating family members. This aligns with previous research regarding concerns about data privacy as a barrier to engagement for individuals (Borghouts et al. 2021). Findings from this study suggest that anonymity would facilitate family engagement if families had the choice to access content without providing personal information. While anonymity protects privacy, it also limits personalization and human interaction. Similar to previous studies, this study highlights the juxtaposition between offering anonymity and being able to provide human or tailored support (Garrido et al. 2019). The principle of informed choice, as discussed earlier, may offer a solution to this tension.

In the context of a family DMHI, there are significant privacy considerations when data is collected from multiple individuals in relation to one another. Primarily, data provided by an individual in a relational context is likely to contain information about other family members. Further, if a family DMHI offers personalization, there are significant risks associated with doing so based on information provided by individuals. Consideration about what information could be used to offer suggested content or construct a tailored pathway would be essential for ensuring individual privacy is not breached. For example, if a family member discloses information about their sexual identity that has not been shared with other family members, it is vital that the DMHI does not inadvertently reveal this information by recommending related content.

This is a uniquely complicated yet fundamental issue for any designer of a family DMHI to tackle, and this study alone cannot address all considerations. It is clear that the maintenance of privacy, both between the individual and the intervention and between the individual and other family members, is essential for fostering engagement. The application of this principle in practice warrants codesign with families, family therapists, and digital mental health professionals to understand how to best maintain the safety of family members engaging in a family DMHI.

3.3. Accessibility

The third principle, accessibility, emerged in many of the essential statements identified by the expert panel. Within this principle, there are two subcategories: it is essential the hub can be readily accessed by everyone; and it is essential the content is easy to use. The expert panel agreed that a family DMHI must be readily accessible to everyone by ensuring it is free to use, is culturally inclusive, accessible for people with disabilities, can be used on any device with internet access, and offers content in different lengths, forms, and modalities to support different styles and preferences.

This principle is not unique to a family DMHI, with reviews highlighting similar findings (Lattie et al. 2022). What is unique in this context is the need to ensure families with a variety of presenting needs can all access the same DMHI. All families will present differently, and all members of a family will have different needs. It is essential, therefore, that a family DMHI is widely accessible. Web content guidelines exist to support website developers to design content for equal access (W3C World Wide Web Consortium, 2024). Applying these alongside representative consumer codesign workshops will ensure a family DMHI is designed with accessibility in mind. Further, user‐interface codesign will ensure the DMHI is intuitive and easy to navigate. This is particularly important when choice is offered to families, to ensure that an abundance of options does not restrict accessibility.

It is interesting to note that the statement “segments content by age” was not deemed essential. DMHIs reported in the literature are overwhelmingly targeted to particular age groups (Lattie et al. 2022), yet findings in this study suggest targeting content to specific ages may not be appropriate for a family DMHI. Comments by panel members suggest that information should be developmentally and life‐cycle appropriate instead. This aligns with fundamental principles of family therapy, whereby the family lifecycle is seen as a series of stages, each characterized by a set of tasks or activities that are completed before progressing to the next stage (Carr 2012). Given this statement did not gain consensus, however, codesign with families and family therapists is essential to progress this.

Though there are existing guidelines and a growing field of research into developing accessible DMHIs, the unique context and needs of families would likely result in unique design implications. Codesign activities with a wide representation of families are crucial to ensure that the user interface of a family DMHI is intuitive and easy to navigate for family members across the life‐course.

3.4. Support and Connection

The final principle identified through this study was “support and connection.” The expert panel agreed that a family DMHI must be supported and connected with access to technical help, links to other sources and crisis support, deliver alerts to therapists, and offer the ability to contact a therapist. This aligns with previous research suggesting that DMHIs with human guidance are more likely to have higher engagement rates than those that are entirely self‐guided (Borghouts et al. 2021; Carlbring et al. 2018; Garrido et al. 2019). However, enacting and embedding this principle requires additional therapeutic resources, which may not always be feasible. Blended approaches are becoming more common, where a DMHI is integrated into face‐to‐face service delivery. Practitioners have demonstrated greater interest and enthusiasm for blended approaches over self‐guided DMHIs (Lattie et al. 2022), and studies have identified service integration strategies as a means of increasing DMHI engagement (Knapp et al. 2021; Kodish et al. 2023). In the context of existing services, the implementation of a DMHI for families as a waitlist or complementary intervention aligns with a preventative approach to mental health care as it is typically less resource‐intensive, easier to implement on a broader scale, maximizes efficiency, reduces costs, and prevents unnecessary over‐servicing (Arango et al. 2018).

Peer connection is also often reported as a key driver of engagement (Borghouts et al. 2021; Saleem et al. 2021). In the context of a family DMHI, this could be an online forum where family members can message and connect with others outside of their own family. In the present study, however, perceptions of the importance of such tools were mixed, and expert consensus was not gained. Some experts commented that it would be a helpful tool, but that professional moderation would be necessary. One goal of family therapy is to support families to better manage challenges and difficulties in the future. Family therapists are often trying to develop connections that are unmoderated, and as such, the expert panel did not see peer support as essential for promoting engagement with a family DMHI.

The findings in this study align with research indicating human guidance promotes engagement (Borghouts et al. 2021; Carlbring et al. 2018; Garrido et al. 2019; Lattie et al. 2022). How this policy can be applied while maintaining enough anonymity and privacy to ensure family comfort and safety should be explored further through codesign with families. In addition, a balance will need to be struck between the needs of consumers and the resources available from service providers.

3.5. Implications and Future Directions

The Delphi methodology aims to build a foundation for further research (Jorm 2015; Minas and Jorm 2010). As such, these principles are intended to inform future research and development and could usefully inform clinicians, researchers, policymakers, and DMHI designers. Specifically, designers can use these principles in creating a family DMHI. It is recommended that these principles inform early codesign activities with families and professionals to test ways in which they can best be applied in practice. Adopting them early in the design will ensure they form the foundation for development, likely contributing to improved engagement when the DMHI is used in practice.

Researchers can use these principles to define benchmarks and means of evaluating engagement with family DMHIs. For example, the principle of informed choice could fundamentally shift traditional means of measuring engagement with a DMHI, which have typically been measured by completion and attrition rates. While not possible in the current study due to sample size restrictions, future research could conduct a factor analysis to assess the dimensionality of the 62 Delphi questionnaire statements. This could aid in the development of a psychometrically valid and comprehensive evaluation tool for DMHIs and other mental health services, which do not currently assess for the influence of design principles on intervention engagement.

Further, these design principles could usefully inform DMHI policy for services and peak governance bodies. For example, there are unique considerations for individual privacy in the development of a family DMHI. As is outlined, an individual's privacy must be maintained within the intervention, but also between family members. This is a perceived driver of engagement, but is also essential for ensuring the relational safety of individual family members while engaging with family DMHIs.

Finally, findings could also inform when, and for whom, a clinician might recommend a DMHI to a family, by understanding factors that could influence whether the family is likely to engage with a DMHI.

3.6. Strengths and Limitations

This study is novel in its focus on the complex dynamics of family DMHI engagement. A key strength of this study design was the inclusion of multiple user perspectives. The Delphi method effectively incorporates diverse expert perspectives while minimizing individual influence (Kodish et al. 2023). This multidisciplinary approach ensured a comprehensive exploration of factors influencing engagement with a family DMHI, encompassing both professional expertise and lived experience. Further, the statements were both evidence‐informed and codesigned, limiting the potential for researcher bias. This did result in a high proportion of statements being rated as essential; however, this is not unexpected given that panel members were involved in distilling the most important factors into statements.

While the study aimed to recruit user groups of equal size, this objective was not met. Further, while there are no fixed guidelines, a minimum of 10 participants is generally expected to achieve consensus in a Delphi study (Domlyn and Wandersman 2019; Trevelyan and Robinson 2015). As such, the user group subsamples in this study were not sufficient in size to describe them as having gained consensus. Combined, this constrained the ability to perform robust cross‐group comparisons. In addition, though the expert panel varied significantly in their expertise, demographically there was minimal variation with most participants identifying as female and aged between 46 and 65. Therefore, findings cannot be generalized to families with children and adolescents. Finally, the racial and ethnic composition of the expert panel was not collected, which limits the ability to assess the generalizability of the findings across diverse populations. It is recommended that any future codesign activities occur across age ranges and with diverse families of different constellations.

There are both strengths and weaknesses to using online questionnaires that are important to note. Most notable in this context was the inability to converse with participants, which restricted both participants' and researchers' ability to seek clarification or further information. This limitation was overcome through the use of a reference group, which informed the statements included in the questionnaire. For those who attended, the reference groups provided an opportunity to discuss the questionnaire and study context in detail. Additionally, a free‐text box was included in the second round of the Delphi questionnaire, allowing participants to elaborate on their responses. Finally, a key strength of the Delphi method is its ability to maintain participant response anonymity, enabling panel members to express their opinions freely, without social or hierarchical pressure or judgment (Kodish et al. 2023; Shang 2023). In this context, online questionnaires were an essential tool in supporting the methodological rigor of the present study.

4. Conclusion

In the rapidly developing field of digital mental health, interventions for families are almost nonexistent. Knowledge of how to design such an intervention is therefore limited. This modified Delphi study took the first step in addressing this knowledge gap by identifying 23 essential factors and four design principles for developing a DMHI that is safe and engaging for families. These principles include informed choice, anonymity and privacy, accessibility, and support and connection. Codesign with a representative sample of families is a crucial next step to bringing these design principles to life.

Appendix 1. Results of the Delphi Survey by user group

#	Statement	Overall		Consumers		Family therapists		Digital health
		IQR	Med	IQR	Med	IQR	Med	IQR	Med
Engagement: Chooses to start using the hub. For individuals and families to want to start using a family hub like this, it is important that they …
1.	can find information about who the hub could be helpful for	1.0+	7.0*	0.0	7.0	1.3	7.0	0.8	6.0
2.	can find information about the “steps” involved in using the hub, and what might come before or after the hub	1.0+	6.0	0.0	6.0	0.3	7.0	1.5	5.5
3.	know they can choose when and how often they use it^	2.0	6.0	1.0	6.0	1.3	6.5	1.5	5.5
4.	can find information about how other families used it, including how long they used it to find it helpful^	1.0++	5.0	2.0	4.0	1.0	5.0	0.8	5.0
5.	know they won't ever have to pay to use the hub	1.0+	7.0*	0.0	7.0	1.0	6.5	1.0	6.5
6.	know they can choose to use it alone or with other family members	1.0+	7.0*	0.0	7.0	1.3	6.5	1.3	6.5
7.	know they do not have to give any personal information to start using the hub	1.0+	7.0*	0.0	7.0	1.3	7.0	1.0	6.5
8.	feel confident using the internet^	2.0	5.0	1.0	5.0	2.0	5.0	1.5	5.5
9.	feel hopeful that things in the family can get better^	3.0	5.0	2.0	6.0	4.0	4.0	1.8	4.5
Engagement: Chooses to start using the hub. For individuals and families to want to start using a family hub like this, it is important that the hub …
10.	is one of the first options that appear on an internet search when looking for family and relationship help^	1.0++	5.0	0.0	6.0	0.3	5.0	2.0	5.5
11.	has been recommended by other credible services^	1.0++	6.0	1.0	7.0	1.0	5.5	0.5	6.0
12.	has a familiar and recognizable name	1.0+	6.0	2.0	6.0	0.3	6.0	1.5	5.5
13.	is based on what research has shown is helpful for families	1.0+	7.0*	0.0	7.0	1.0	6.5	1.0	6.5
14.	is based on what other families have said was helpful for them	1.0+	6.00	1.0	7.0	1.0	6.0	2.3	5.0
15.	is culturally inclusive and can be completed in different languages	0.0+	7.0*	0.0	7.0	0.0	7.0	1.8	6.5
16.	can be used on any device with internet access, including a phone, tablet, or laptop	1.0+	7.0*	0.0	7.0	0.3	7.0	1.8	6.5
17.	looks modern and contemporary^	1.0++	6.0	2.0	5.0	1.3	6.0	1.3	5.5
Engagement: Chooses to keep using the hub and come back another time. For individuals and families to want to keep using or come back to a hub like this, it is important that they…
18.	had a good experience the last time they used it	1.0+	7.0*	1.0	7.0	1.0	6.0	0.3	7.0
19.	feel there is more they could get out of it	1.0+	6.0	1.0	6.0	0.5	6.0	0.3	6.0
20.	can return to the spot they left off^	1.0++	6.0	0.0	7.0	0.3	6.0	1.0	5.5
21.	can review content and redo activities as many times as they like	1.0+	6.0	0.0	7.0	1.3	6.0	0.8	6.0
22.	know they can start fresh each time they come back	1.0+	6.0	0.0	7.0	0.3	6.0	0.5	6.0
23.	know they can choose to build upon work they have already done on the hub	1.0+	7.0*	0.0	7.0	1.0	6.0	1.0	6.5
24.	can set goals in the hub and track how they progress toward those goals over time	1.0+	6.0	1.0	6.0	1.0	6.0	0.3	6.0
25.	know the hub will help them to reflect on change (individual and family) over time through activities, such as reflective exercises and self‐assessments^	1.0++	6.0	1.0	6.0	1.3	6.0	0.5	6.0
26.	know the hub will check in on how they went practicing skills between hub sessions, and provide suggestions and reinforcement	1.0+	6.0	1.0	5.0	1.0	6.0	1.8	5.5
27.	experience personal achievement from the hub, for example, through earning points^	2.0	4.0	1.0	2.0	0.8	4.0	2.0	4.5
28.	can customize reminders or notifications from the hub, including when and how (e.g., text and email), and can change these preferences any time^	1.0++	6.0	0.0	6.0	1.3	5.0	2.0	5.0
29.	can customize how the hub looks, including colors, or adding personal photos^	2.0	3.0	1.0	2.0	1.3	4.0	0.8	3.0
Engagement: Chooses to keep using the hub and come back another time. For individuals and families to want to keep using or come back to a hub like this, it is important that the hub …
30.	gives feedback that feels personal	1.0+	6.0	1.0	6.0	0.3	6.0	0.3	7.0
31.	provides the option to “tag“ content to come back to at another time	1.0+	6.0	1.0	7.0	1.3	6.0	0.5	6.0
32.	offers a variety of content that is updated regularly	1.0+	6.0	0.0	7.0	0.3	6.0	1.0	6.5
33.	has the option to message or connect with other families via chat rooms or discussion boards^	2.0	4.0	2.0	4.0	1.3	4.5	1.3	4.5
Accessing and using content. For individuals and families using the hub, it is important that they …
34.	can choose to access content anonymously	1.0+	7.0*	0.0	7.0	1.0	6.5	1.0	6.5
35.	can choose at any time to tell the hub a bit about themselves or create an account so that the hub can offer personalized suggestions	0.0+	6.0	0.0	6.0	0.0	6.0	1.8	5.5
36.	can choose how they use content, for example, by watching videos, reading text, or listening to audio	1.0+	7.0*	1.0	7.0	0.3	7.0	1.3	6.5
37.	can complete content and activities at their own pace, speeding up or slowing down activities and content	1.0+	7.0*	0.0	7.0	1.0	7.0	0.3	7.0
38.	can find information about which content or activities other families found helpful	1.0+	6.0	0.0	6.0	2.3	6.0	1.8	5.5
39.	can access technical help while using the hub	1.0+	7.0*	1.0	7.0	1.3	7.0	0.8	7.0
40.	can schedule a time with a therapist when they need	1.0+	7.0*	0.0	7.0	1.3	6.0	1.3	6.5
Accessing and using content. For individuals and families using the hub, it is important that the hub …
41.	provides the option to download materials or activities to do off the hub^	1.0++	7.0*	0.0	7.0	1.0	7.0	0.5	6.0
42.	offers links to other credible sources that may be helpful for individuals and families	1.0+	7.0*	0.0	7.0	1.3	6.0	1.0	6.5
43.	includes interactive activities, such as quizzes, exercises, self‐assessment tools, and games^	2.0	5.0	0.0	5.0	2.3	5.0	1.3	5.5
44.	is accessible for people with disabilities	0.0+	7.0*	0.0	7.0	0.0	7.0	0.5	7.0
45.	offers content in short form (e.g., “what do I need to know now?”), with the option to view more details, or return to it later	1.0+	7.0*	0.0	7.0	0.3	6.0	1.0	6.5
46.	is easy to navigate and find relevant information without making too many clicks	1.0+	7.0*	0.0	7.0	1.0	6.5	0.8	7.0
47.	segments content by age^	1.0++	5.0	1.0	5.0	1.3	6.0	2.3	5.0
48.	includes video examples of other families practicing skills taught in the hub^	1.0++	5.0	2.0	4.0	0.5	6.0	0.3	5.0
Safety and privacy. When thinking about individual and family safety and privacy, it is important that the hub …
49.	keeps personal information and activity completion private between family members	0.0+	7.0*	0.0	7.0	1.3	7.0	0.0	7.0
50.	has a discrete name and web address^	2.0	6.0	3.0	6.0	1.3	5.0	1.8	6.5
51.	has crisis support links easily accessible	0.0+	7.0*	0.0	7.0	0.0	7.0	0.0	7.0
52.	helps the family to identify and make the most of their existing supports	1.0+	6.0	1.0	6.0	1.0	7.0	1.3	5.5
53.	alerts a therapist if safety concerns are raised	1.0+	7.0*	0.0	7.0	1.5	6.5	0.3	7.0
Co‐complete activities. When thinking about families working together, it is important that …
54.	the hub includes “conversation starter” activities to help with safe family discussions about family therapy, using the hub, or about other content on the hub^	0.0++	6.0	1.0	6.0	0.0	6.0	0.8	6.0
55.	the hub lets them share activities and content with other family members or tag for future discussion together^	1.0++	6.0	1.0	6.0	1.0	6.0	1.3	5.5
56.	the hub offers activities that can be completed together away from the hub	1.0+	6.0	1.0	6.0	1.3	6.0	1.0	5.0
57.	the hub lets individuals contribute to shared activities in their own time^	1.0++	6.0	1.0	7.0	1.0	5.0	2.0	5.0
When engaged with a therapist. For families about to, or already seeing a therapist, it is important that the hub provides the option for …
58.	their therapist to send recommended activities^	2.0	6.0	2.0	7.0	1.0	5.5	0.5	6.0
59.	family members to flag things to discuss at their next session	1.0+	6.0	0.0	7.0	1.3	5.5	0.3	6.0
60.	their therapist to view what they have completed and how they are progressing^	1.0++	6.0	1.0	6.0	1.3	5.0	0.3	6.0
61.	the family to keep track of their past and upcoming sessions^	0.0++	5.0	1.0	5.0	0.0	5.0	0.3	5.0
62.	continued access after they have finished therapy	0.0++	7.0*	0.0	7.0	0.3	7.0	0.3	7.0

Abbreviations: IQR, interquartile range; Med, median.

^Did not gain consensus at Round 1 and was redistributed to participants in Round 2.

+Gained overall consensus at Round 1, that is, the interquartile range was ≤ 1.

++Gained overall consensus at Round 2, that is, after Round 2 results, the interquartile range was ≤ 1.

**Deemed essential overall, that is, the median = 7.