The Delphi technique was selected as the most appropriate methodology for developing these educational standards due to its established effectiveness in achieving expert consensus through structured, iterative discussions (37, 38). The Delphi method is particularly valuable when developing professional standards or guidelines that require both theoretical rigour and practical applicability (39). This approach provided a structured “ground-up” approach to gaining reliable expert consensus, ensuring that the resulting standards would reflect both academic expertise and practical implementation experience. The methodology also aligned well with our aim to develop standards that would be responsive to Australia’s unique health system characteristics and geographical diversity. By engaging participants (defined in section 3.3) from various regions and professional backgrounds, the Delphi process facilitated the development of standards that could be implemented effectively across different healthcare settings while maintaining consistency in quality and outcomes. As well as using a large group of Australian participants, a small panel of international participants were also sampled to benchmark the standards globally. This study was conducted and reported with consideration of both the Conducting and Reporting Delphi Studies guidelines (40) and the reporting guidelines for Delphi techniques in health sciences (41). Alternative methodologies were considered. Consensus development conferences or nominal group techniques would have required synchronous participation, which was impractical given the geographic dispersion of Australian experts and the inclusion of international participants for benchmarking purposes. Mixed methods approaches combining surveys with focus groups were considered, but the Delphi’s structured iterative process was deemed more appropriate for achieving systematic consensus across a large, geographically dispersed expert panel while maintaining anonymity to reduce dominance effects and social pressure that can occur in face-to-face consensus methods.

The draft educational standards that formed the basis for this Delphi study were developed through an iterative synthesis process appropriate for an emerging field with limited published frameworks on link worker education. The development combined three evidence sources: (1) the lead author and colleague’s ongoing engagement with peer-reviewed and grey literature on social prescribing and link worker roles from Australia and internationally, including publications examining various aspects of social prescribing in Australia (17, 21, 22, 42–46) and international practices from statutory entities such as the World Health Organization and the NHS; (2) consultations with implementation leaders across six countries (Canada, Netherlands, UK, Singapore, Austria, and Poland) exploring their local link worker models, training elements, and contextual adaptations; and (3) over a decade of practical experience implementing and supervising social prescribing programs across diverse Australian contexts.

To mitigate potential bias from practice-based insights, the draft framework prioritised competencies that appeared across multiple evidence sources and aligned with documented implementation experiences. This preliminary framework was then subjected to systematic expert consensus through the Delphi process described below. The development of the draft training standards considered the Australian context to ensure relevance. In Australia, social prescribing operates within a complex healthcare landscape characterised by multiple funding streams, diverse geographical contexts, and rich cultural diversity, which includes the largest living first nations culture (32, 47–50). This complexity creates distinct challenges for implementing social prescribing effectively. The draft standards comprise two main components: graduate attributes and core competencies (Supplementary Table 1). The graduate attributes outline key characteristics expected of social prescribing Link Workers upon completion of their training while the core competencies detail the specific abilities required for effective practice in the Australian context.

Participants were recruited using purposive sampling, with expertise defined as: ASPIRE membership; authorship of academic or grey literature on social prescribing; research involvement in social prescribing; health or social care provision involving social prescribing; Link Worker experience; or healthcare administration/management of social prescribing services. Given the emerging nature of the field, participants were included even when their work was not explicitly labelled as ‘social prescribing’ or ‘link work’. Diversity in professions was prioritised to mitigate professional biases in cross-disciplinary Delphi studies with heterogeneous samples. Participants were included if they understood the Australian social prescribing landscape and able to read and understand English.

Sample size considerations for expert panels in a Delphi study varies, with the average sample size in the health sciences being 40 participants (39) and most ranging from 20 to 60 (6), while specialised topics may only require15-20 (51). Accounting for the expected dropout rate of 20–30% between subsequent rounds (51), we aimed to recruit 40 Australian experts, with an additional subgroup of 10 international experts for benchmarking.

Potential participants were identified through networks developed during previous studies on social prescribing, contacts known to the research team, relevant published research and news articles, attendees at social prescribing conferences, and member directories of professional associations relevant to social prescribing. Invitations to participate were emailed to potential participants, along with a detailed information sheet, contact details of the researchers, and the link to Round 1 of the study. Additional invitations were sent to a purposive selection of ‘snowballed’ contacts whose professional expertise was not yet adequately represented in the sample to ensure diversity.

To enhance representativeness within the constraints of an emerging field, several strategies were employed. First, we systematically sought participants across multiple expertise types (providers, researchers, policy makers, consumers) and geographical regions within Australia to capture diverse implementation contexts. Second, the snowball sampling approach specifically targeted gaps in representation identified during initial recruitment, such as ensuring adequate representation from rural/regional areas, diverse organisational types (government, NGO, private), and practitioners with direct link worker experience. Third, international expert inclusion provided external benchmarking perspectives. However, we acknowledge that true population-level representativeness was not feasible given the nascent state of social prescribing in Australia, where the total population of individuals with relevant expertise is unknown and formal professional registration for link workers does not yet exist. Therefore, our sampling aimed for diversity of expertise and perspective rather than statistical representativeness.

The Delphi studies were conducted over 4 months using the Qualtrics™ survey management platform. Data were collected using unique survey links to enable follow-up while maintaining participant confidentiality. Participants provided informed consent electronically before accessing the survey content. Unique survey links were emailed to maintain participant confidentiality while enabling follow-up communications. Two follow-up reminder emails were sent at one-week intervals to non-responders.

Although there is not much agreement on what constitutes consensus in the literature (39, 51), there is a suggestion that for the development of medical guidelines, consensus is 75% agreement, and a high level of consensus is 95% agreement (41). For this study, consensus was defined a priori if ≥80% of participants rated the item as Completely Agree or Somewhat Agree combined. Neutral responses were included in the denominator but not counted toward the consensus threshold, treating neutrality as absence of agreement rather than disagreement. This conservative approach ensures that consensus reflects not only lack of disagreement but also sufficient clarity for experts to take definitive supportive positions. Stopping criteria were not predetermined by number of rounds; rather, the decision to conduct additional rounds was guided by whether consensus thresholds were achieved and whether qualitative feedback patterns indicated stable consensus or continued conceptual concerns requiring revision. Previous social prescribing Delphi studies have demonstrated that items achieving 78–79.99% agreement (within 2% of threshold) may indicate practical consensus (6). Items not reaching the 80% threshold were evaluated for whether qualitative feedback indicated fundamental conceptual concerns requiring revision or focused primarily on implementation considerations suggesting practical consensus had been achieved. Data was analysed in Microsoft Excel. Quantitative data were expressed as percentages and qualitative data were analysed using Braun and Clarke’s (52) thematic analysis methodology. This approach was selected at it provided a rigorous and systematic approach to identifying patterns in qualitative data and is well suited to examining different perspectives and highlighting differences and similarities.

The thematic analysis followed Braun and Clarke’s six-phase approach. The second author (AT) led the qualitative analysis with oversight from the first author (JB). Phase 1 (familiarisation) involved multiple readings of all qualitative responses from both Australian (n = 36) and International (n = 11) participants, with documentation of initial impressions. Phase 2 (initial coding) involved line-by-line coding focusing on explicit and implicit statements about requirements, challenges, role boundaries, scope, and implementation considerations. Phase 3 (theme development) grouped codes into potential themes, creating initial thematic maps and identifying relationships between codes. Phase 4 (theme review) checked themes against original data to ensure they captured the breadth of responses and were coherent and distinct, with cross-checking against quantitative agreement levels to understand patterns of consensus and dissent. Phase 5 (theme definition) developed clear definitions for each theme and selected representative quotes. Phase 6 (reporting) organised themes to address feedback on the standards and broader implementation requirements. The analysis used a combination of inductive (data-driven) and deductive (theory-driven) coding approaches, maintaining distinction between Australian and International responses throughout. Credibility was enhanced through regular discussion between analysts (AT and JB) to review emerging themes, resolve any interpretive differences, and ensure themes accurately reflected participant perspectives. The integration of qualitative and quantitative data ensured that revisions between rounds were informed by both consensus levels and the substantive concerns expressed in open-text responses.

This study was reviewed and approved by the human research ethics committee from Monash University (approval no. 44727).

The initial email was sent to 96 experts. In total, 47 completed round 1 (49% response rate). Participants represented diverse expertise types, including healthcare providers/practitioners (n = 15), researchers/academics (n = 11), government and policy representatives (n = 7), and consumers/service users (n = 3), with some participants holding multiple roles. Of the 47 participants in Round 1, 43 provided complete responses for round 2 (91% completion rate among those who started the survey) (Table 1). Australian experts had a mean age of 30.3 years (SD = 11.8), were 78% female (N = 28), and represented diverse professional backgrounds including health care providers, researchers, educators, and social prescribing practitioners. Most participants were from Queensland (44%, N = 16), New South Wales (25%, N = 9), and Victoria (22%, N = 8), with primary affiliations to non-government organisations, and averaged 17.2 years of relevant expertise (SD = 12.5). International experts had a mean age of 26.7 years (SD = 14.2), were 70% female (N = 7), primarily from Canada and the United Kingdom, with backgrounds in research and education/training, particularly in public and population health. They were typically affiliated with not-for-profit organisations and averaged 18.4 years of expertise (SD = 10.5). The sample had a high proportion of people identifying as female, which is consistent with the distribution of the health, social care, and community-sector workforces in Australia, although the underlying reasons for this distribution in our sample cannot be verified.

Figure 1 illustrates the flow of the study. Round 1 achieved high-level consensus on all seven graduate attribute domains and all nine core competency domains. Among the detailed indicators and skills, 34 of 37 graduate attribute indicators (92%) and 50 of 63 competency-associated skills (79%) reached the 80% consensus threshold. The three graduate attribute indicators and 13 associated skills that did not reach consensus primarily reflected concerns about role scope boundaries, the emphasis on evidence versus lived experience, and resource navigation responsibilities (detailed in Table 2).

Perspectives were analysed collectively for the primary consensus assessment, consistent with standard Delphi methodology. Australian and international responses were examined descriptively as a secondary step to validate alignment between national and global expertise and to ensure the resulting standards were coherent with international models of social prescribing. At the completion of round one, consensus (≥80% selecting “Completely Agree” or “Somewhat Agree”) was reached on all 7 graduate attributes (Ethical Practitioner, Inclusive and Respectful, Effective Communicator, Critical Thinker and Problem Solver, Systemic Change Catalyst, Community Empowerment and Collaboration, Reflective and Lifelong Learner) and all 9 core competencies domains (Working across Australian Health and Social Systems, Working with people on what matters, Resource Navigation and Management, Community Development and Social Capital Building, Culturally Safe and Inclusive Practice, Safe and Effective Practice, Data Management and Evaluation, Supportive Partnerships and System Change, Integration of Environment and Contextual Factors) (see Supplementary material for detailed results). Among the 37 key indicators of graduate attributes, 34 reached consensus, and 50 of the 63 associated skills of the core competencies reached consensus. International experts (recruited for benchmarking) showed similar patterns, with 35 key indicators of graduate attributes reaching consensus. The two remaining attributes also did not reach consensus among the Australian group. Of the 12 associated skills that did not reach consensus among international experts, 9 also did not reach consensus with the Australian group. Table 2 lists all the items that did not meet consensus and the outcomes for each one.

74% participants left at least one comment. In total, there were 293 comments across 19 feedback options. Analysis of Round 1 qualitative feedback revealed several significant gaps in the original competency framework that led to the development of four new skills. First, participants emphasised that relationship building was foundational to effective social prescribing but was not explicitly captured, with feedback noting “a relational way of working and building trust is imperative… the relational aspects and trust/empathy elements are missing.” This led to the development of the “Trust and Relationship Building” skill. Second, concerns about role definition emerged strongly, with participants questioning “what distinguishes a Link worker from a Social worker” with the suggestion that “link workers should focus on connecting people with community resources to support their well-being, rather than acting as therapists or clinical practitioners.” This informed the development of the “Scope Recognition and Referral Navigation” skill. Third, participants frequently discussed the need for creativity within resource constraints, noting how service delivery “really depends upon the funding and case load of the link worker,” leading to the “Enterprising and Adaptive Practice” skill. Finally, analysis revealed significant concerns about digital access and inclusion, with participants noting that while “digital literacy is pretty crucial,” there was also a need to consider those who cannot engage with technology, leading to the development of the “Digital Literacy and Inclusive Engagement” skill. These new skills were developed to address these identified gaps while maintaining alignment with the broader objectives of social prescribing, strengthening the framework by explicitly addressing core elements of practice that were previously implicit or absent.

The results from Round 2 are presented in Table 3. 63.0% participants left at least one comment. In total, there were 554 comments across 13 feedback options. In Round 2 all the items reached consensus. Two items—‘Funding and Resource Navigation’ (78.79%) and ‘Environmental Health Awareness’ (78.79%)—achieved consensus slightly below the predetermined 80% threshold, they were within the 2% margin that previous social prescribing Delphi studies have considered acceptable (6). Crucially, qualitative feedback analysis revealed a distinct pattern from Round 1 items that failed to reach consensus. Unlike Round 1, where qualitative feedback indicated fundamental concerns about scope and role boundaries requiring substantial revision, Round 2 feedback focused primarily on implementation considerations rather than conceptual disagreement. For example, feedback on ‘Funding and Resource Navigation’ centred on clarifying practical application rather than questioning its inclusion in the framework. Similarly, comments on ‘Environmental Health Awareness’ focused on level of emphasis and practical implementation rather than fundamental changes to the concept. This pattern of feedback, combined with the high level of quantitative agreement, supported the retention of these items without need for further rounds. For educators and policymakers, these items with marginally lower consensus (78.79%) should be understood as competencies requiring particular attention to organisational context and resource availability during implementation, with expected variation in how they are enacted across different practice settings.

Round 2 results showed strong alignment between Australian and international experts across all revised items. The revised graduate attributes - Evidence-Informed Practice, Supporting Community-Focused Initiatives, and Resilience and Adaptability - achieved consensus above 90% from both groups. Among core competencies, Evidence Collection for System Impact, Awareness and Communication of Policy Impacts, and Navigating Complex Situations similarly achieved strong consensus. Environmental Health Awareness received slightly higher support from international experts (90.00%) than Australian experts (78.79%), reflecting different contextual emphases. The newly proposed skills were well-received by both groups (see Figure 1).

Analysis of Round 2 qualitative feedback focused on refinement and implementation considerations. Participants emphasised the importance of organisational support and clear role boundaries, particularly for Funding and Resource Navigation, noting that “some aspects should be organization-led rather than link worker responsibility.” The newly added skills received strong endorsement, with particular emphasis on Trust and Relationship Building as “foundational to effective social prescribing” while requiring “clear boundaries and scope documents” for implementation. For Environmental Health Awareness, feedback highlighted the need to “separate into different levels (awareness vs. implementation)” while maintaining the core competency. The newly added skills received strong endorsement with particular emphasis on their practical application. For example, feedback on Trust and Relationship Building noted that while “relationship building was foundational to effective social prescribing,” there needed to be “clear boundaries and scope documents” to support implementation. Similarly, commentary on Digital Literacy and Inclusive Engagement acknowledged its importance while emphasising the need for “IT support” and consideration of contexts where “technological barriers” could not always be addressed. This feedback pattern reinforced the shift from Round 1’s fundamental concerns to Round 2’s focus on practical implementation considerations.

Following the two-round Delphi process, consensus was reached on the educational standards for Australian social prescribing Link Workers. The final framework comprises seven graduate attributes that define the professional characteristics expected of qualified link workers, and nine core competencies that detail the specific skills and knowledge required for effective practice (Figures 2, 3) These standards provide a foundation for curriculum development, professional training programmes, and quality assurance in social prescribing link worker education across Australia. The complete final standards with detailed indicators and skills are provided in Supplementary material.

These consensus-based educational standards provide a foundation for policy development that addresses social prescribing workforce gaps in Australia’s evolving healthcare landscape. The framework proposes link workers as a distinct professional category within Australia’s existing health workforce structure, complementing rather than duplicating established roles such as social work, nursing, and allied health. The standards align directly with national policy priorities outlined in the National Preventive Health Strategy 2021–2030 (33) and the Primary Health Care 10-Year Plan 2022–2032 (34), both of which advocate for community-based interventions to address social determinants of health.

Similarly to other high-income countries, Australia faces mounting healthcare challenges that demand innovative workforce solutions. Australia’s rising burden of chronic conditions among young people (2), increasing mental health challenges in young adults (3), and an ageing population (4) are placing unprecedented pressure on clinical services (2–4). Simultaneously, workforce shortages across general practice, nursing, and mental health represent critical undersupply areas underscoring the urgency of deploying a non-clinical workforce to support early intervention and community-based care (5). By providing consensus on the educational preparation required for link workers, the framework offers a foundation that could support efforts to redirect non-medical needs away from overstretched clinical services towards appropriate community-based supports. This systematic approach to workforce development provides commissioners with a credible tool to guide investment decisions and quality assurance processes, enabling evidence-based resource allocation and performance monitoring. These standards enable rapid deployment of a non-clinical workforce capable of addressing psychosocial factors while reducing pressure on clinical services.

The framework addresses Australia’s unique implementation challenges through competencies specifically designed for the nation’s multi-funder healthcare environment, geographic diversity, and cultural complexity. Unlike single-payer systems, Australia’s blend of federal and state governance, Medicare, NDIS, My Aged Care, and community funding streams requires link workers skilled in complex system navigation. The standards’ emphasis on cultural safety for First Nations peoples and rural-remote adaptability reflects Australia’s distinctive demographic and geographic characteristics.

Implementation science perspectives highlight how these standards address fundamental barriers to social prescribing adoption. Qualitative feedback revealed concerns about role duplication with social work; however, comparison with the Australian Social Work Education and Accreditation Standards demonstrates distinct yet complementary scopes of practice (53). Rather than duplicating existing professional roles, link workers address a distinct gap in the healthcare workforce. While social workers typically manage complex cases requiring statutory intervention, therapeutic support, or intensive case management, link workers operate with a broader population-level approach, connecting people to community resources and non-clinical supports. For example, while a social worker might provide ongoing case management for someone experiencing domestic violence, a link worker would connect someone to community gardening groups, walking clubs, or volunteering opportunities to address social isolation. This differentiation enables social workers to focus on their specialised scope while link workers address the broader population’s social determinants of health through community engagement. In this way, link workers complement rather than duplicate social work roles, addressing a gap at a lower scope of practice but with a strong orientation toward community building and wellbeing. This differentiation also points toward the value of conceptualising link worker practice within a broader framework of intervention levels and intensity—a potential direction for future work on social prescribing typologies.

The standards provide policymakers, funders, and commissioning bodies with a consensus-based framework for workforce planning and quality assurance. By proposing clear competency expectations and training pathways, the framework offers a foundation for systematic scaling of social prescribing across diverse healthcare settings, with the aim of promoting consistent quality and outcomes pending validation through implementation. If validated through implementation research, such standards could support efforts to leverage social prescribing as a cost-effective and environmentally conscious strategy for addressing healthcare capacity constraints and improving population health outcomes through community-based rather than clinic-based interventions.

The strengths of this study include robust mixed-methods Delphi methodology combining quantitative consensus measurement with qualitative feedback analysis. The diverse stakeholder representation across healthcare disciplines, research backgrounds, and practice settings strengthened the validity of consensus. High retention rates between rounds (91%) and substantial qualitative engagement (74% of Round 1 participants providing comments) indicate strong participant investment in the standards development process. The integration of international benchmarking provided valuable external validation while maintaining focus on Australian contextual requirements.

Limitations include potential sampling bias, with 78% female participation among Australian experts, which may not fully represent the diversity of perspectives in the emerging field. While we did not analyse consensus patterns by participant gender, it is possible that the gender distribution may have influenced consensus levels on competencies emphasising relational and care-oriented aspects of practice. Ethnicity and cultural background were not captured because expertise was the primary recruitment criterion. As social prescribing grows nationally and globally, future validation should ensure that a more diverse demographic representation is included to strengthen applicability across varied populations and workforce contexts. As an emerging field, the definition of “expertise” was necessarily broad, which may have included varied levels of social prescribing-specific knowledge. Additionally, the sample comprised predominantly social prescribing advocates and practitioners, which may have introduced positive bias towards comprehensive standards. Participants’ professional investment in advancing social prescribing could have influenced consensus levels, particularly for competencies expanding the link worker role. However, this limitation is inherent to developing standards for emerging fields where expertise necessarily derives from advocacy and practice experience. The inclusion of critical feedback and items that failed to reach consensus demonstrates that participants engaged critically rather than endorsing all proposed standards uncritically.

The sampling approach, while achieving diversity across expertise types and geographical regions, cannot claim statistical representativeness of all potential stakeholders in Australian social prescribing. The reliance on professional networks and purposive sampling may have systematically excluded perspectives from practitioners working in very remote areas, those from culturally and linguistically diverse backgrounds who may face language barriers, and community organisations without established connections to social prescribing networks. These limitations may have influenced consensus results by over-representing perspectives from well-connected, professionally established individuals and organisations with existing social prescribing experience. However, the high consensus levels achieved (>90% for most items) and the integration of international benchmarking suggest that the standards reflect broadly applicable principles rather than narrow, context-specific perspectives. Future validation research should deliberately seek input from underrepresented stakeholder groups, particularly from remote and very remote communities, culturally and linguistically diverse populations, and emerging practitioners new to the field.

The consensus thresholds, while established in literature, remain somewhat arbitrary, and the standards require validation through implementation and evaluation before being considered definitively established. A limitation is that the study focused on professional and expert perspectives, and future research should incorporate direct input from community members and service users who are the intended recipients of social prescribing services Future research should examine how these standards perform in diverse educational and practice settings, measure their impact on workforce preparation and practice quality, and incorporate perspectives from community members and service users who are direct recipients of social prescribing services.