Study design

This study adopted a qualitative interpretive approach to deeply explore how referral practices are experienced and understood within Australian primary care. Guided by a constructionist ontology and an interpretivist epistemology, it acknowledges that reality is co-constructed through social interaction and that knowledge is inherently shaped by context and lived experience [23]. This philosophical foundation supported the use of qualitative methods to examine the complex social and organisational processes influencing the implementation of digital referral systems. The study adhered to the Consolidated Criteria for Reporting Qualitative Research (COREQ) guidelines (S1 File)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Thematic map illustrating interconnections among key themes.

https://doi.org/10.1371/journal.pone.0344663.g001

Setting and context

The study was conducted in general practice settings across both urban and rural areas in five of Australia’s eight states and territories, providing a broad representation of healthcare contexts and service delivery environments. These settings encompassed diverse practice sizes, patient populations, and resource availability, offering a varied backdrop for exploring experiences with COPD care and digital referral systems.

Participants

Participants were general practitioners (GPs) and practice managers (PMs) who were currently involved in COPD care and had experience using digital referral systems. Inclusion criteria required active engagement in clinical or administrative aspects of COPD management within general practice. Individuals who were not currently practicing or who lacked familiarity with digital referrals were excluded to maintain the relevance and quality of the data [24]. A total of (16) participants were recruited, comprising (10 GPs and 6 PMs), with representation from both metropolitan, regional and rural practices.

While traditional approaches often rely on data saturation to determine sample adequacy, this study followed Braun and Clarke [25] Reflexive Thematic Analysis (RTA) which rejects the notion of data saturation as a maker of completion and adopts meaning-making as an ongoing, interpretive process rather than a definite endpoint. In keeping with Braun and Clarke’s RTA, this study conceptualised sample adequacy through information power by Malterud and colleagues [26]. Recruitment was guided by both the diversity and depth of insights required to answer the research question. However, practical and contextual constraints, including limited response rates despite multiple follow-ups, also influenced the final sample size.

Recruitment

Recruitment was initiated through purposive snowball sampling through professional networks and extended via targeted outreach using email and social media platforms such as LinkedIn and Facebook. This mixed recruitment strategy was selected to ensure broad reach and engagement with diverse participants, particularly those with direct experience in digital referral processes [27]. Of the 35 invitations distributed, 16 participants consented, 6 declined, and 13 did not respond. To ensure a diverse participant pool while upholding ethical standards, the recruitment process included tailored invitations and measures to safeguard participants’ privacy and confidentiality. While snowballing enabled access to information-rich participants, it may have introduced homogeneity within professional circles. This limitation was mitigated through reflexive consideration of sample diversity, and adherence to qualitative trustworthiness principles of credibility, transferability, dependability, and confirmability. None of the participants were known to any of the authors.

Data collection

Semi-structured interviews were conducted one-on-one by the lead author (TJ), a female Registered Nurse with five years of experience in general practice, as part of her higher degree by research training. Interviews were conducted between 1^st August 2024–5^th September 2024. Each interview lasted between 30 and 60 minutes and were video/audio recorded via Microsoft Teams. This platform was selected for its transcription capabilities and its ability to connect geographically dispersed participants. Interview discussions centered on digital transformation, interoperability, and the integration of clinical workflows in referral processes [28]. The flexible format allowed participants to reflect on their real-world experiences, revealing perceived challenges, benefits, and contextual influences associated with digital referrals in COPD care [29]. Field notes were also taken to support transcription accuracy and enrich contextual interpretation. Participants were offered the opportunity to review their interview transcripts; however, none elected to do so.

Interview guide

Interview questions were developed based on a review of current literature and further refined through input from subject matter experts and feedback from the supervisory team to ensure clarity, relevance, and alignment with the objectives of the study. The questions were piloted with 10 GPs and 6 PMs and iteratively adjusted to enhance their relevance, clarity, and ethical appropriateness [30].

Researcher reflexivity

The research team comprised experienced nursing and public health academics with backgrounds in rural, reginal and metropolitan health, chronic disease management, and qualitative research. Awareness of team’s professional perspectives informed the design and interpretation of the study. To minimise bias, the study used a semi-structured interview guide, engaged in regular peer debriefing, and maintained an audit trail throughout the analysis process. Reflexivity was conceptualised not as bias minimisation but as an interpretive practice, acknowledging that the researchers’ professional perspectives inevitably shaped analytic engagement. Regular discussions fostered critical awareness and deepened interpretation, consistent with Braun and Clarke’s [31] reflexive thematic analysis framework.

Data management

All interview recordings and transcripts were stored on a secure, password-protected university server. Participant identifiers were removed during transcription, and each transcript was assigned a unique code (e.g.,-’GP4, NSW’ or ‘PM2, VIC’). Only authorised research team members had access to the data, which were managed in accordance with institutional ethics requirements.

Data analysis

Reflexive thematic analysis was used to analyse interview transcripts. The six-phase approach involved familiarisation with the data, initial coding, theme development, reviewing and refining themes, defining themes, and producing a final report [25]. All transcripts were verified for accuracy and assigned unique identifiers based on participant role, interview sequence, and state location. For example, identifiers such as ‘GP4, NSW’ or ‘PM2, VIC’ were used to provide contextual depth and support interpretation of participant quotes.

Transcripts were manually coded using Microsoft Excel (Version 2507), allowing for flexible and immersive engagement with the data. Codes were generated inductively and then grouped into broader themes that reflected patterns across participants’ experiences. Coding was initially performed by the lead researcher and subsequently reviewed in collaboration with other researchers involved in the study to ensure rigour, consistency, and credibility [32]. Themes were iteratively refined to ensure coherence, analytical depth, and alignment with research objectives. Quotes were selected to illustrate each theme, enhancing credibility and confirmability. Regular supervisory meetings and reflective discussions supported the rigour and dependability of the analysis.

Study rigour

Credibility was supported through peer debriefing within the research team and ongoing reflection on emerging findings. Dependability was maintained by following a consistent data collection protocol and documenting all methodological decisions in an audit trail. Confirmability was enhanced by retaining de-identified transcripts and analytic memos for review, ensuring interpretations were grounded in participant data. While transferability was supported through detailed descriptions of the study context, participant characteristics, and recruitment process, the inherently interpretive nature of RTA [25] limits the extent to which findings can be applied beyond this specific context. This has been acknowledged as a study limitation.

Ethical considerations

Ethical approval for the study was granted on 29^th September 2023 by the University of Southern Queensland Human Research Ethics Committee (HREC Approval Number: ETH2023−0488). All participants received a detailed participant information sheet outlining the study’s purpose, procedures, risks, and benefits. Written informed consent was obtained prior to participation, and verbal confirmation was reconfirmed immediately before each interview, in accordance with institutional guidelines and the Declaration of Helsinki. Participants were given adequate time to consider their involvement and were encouraged to ask questions. They explicitly agreed to audio and/or video recording and to the use of de-identified quotes in publications. Participants were reminded of their right to withdraw at any time without consequence. All data were stored securely on password-protected servers accessible only to the research team and will be retained for five years in accordance with institutional policy. No vulnerable populations were involved. The study adhered to ethical reporting standards for qualitative research, including transparency in consent documentation and data handling.

Systemic disruptions in digital workflows

Participants frequently described digital referral systems as unpredictable and difficult to integrate into daily workflows. While platforms such as Smart Referrals, HealthLink, Medical Objects, and Communicare offered structured formats and embedded logic, they often failed to perform reliably in real-world practice. System crashes, file upload restrictions, navigation issues, and misalignments between platform capabilities and clinical needs were commonly reported.

GPs and PMs described how digital referral tools, despite their potential, created operational inefficiencies. Participants felt these systems lacked interoperability, required significant manual workarounds, and offered limited flexibility during referral creation.

One GP noted, “Smart Referrals can be hit or miss... it doesn’t always work smoothly, particularly when the hospital system is not fully integrated” (GP1, QLD). Another participant reported that their platform required manual re-entry of patient details, stating, “It doesn’t self-populate... it’s still a blank page” (GP9, WA).

Training and onboarding also contributed to limited adoption. New users were often expected to self-navigate the systems or rely on informal support. This approach led to inconsistent use, especially in practices where there was high turnover of GPs, or those serving complex health populations. As one participant explained, “Training is minimal; we learn as we go” (GP9, WA), while another noted that “Proper training would save time and improve efficiency” (GP1, QLD). The lack of structured onboarding not only reduced confidence and accuracy but also amplified workload pressures, reinforcing perceptions that digital referral systems were inefficient and burdensome.

These recurring breakdowns extended beyond technical inconvenience; they disrupted professional routines, diminished clinicians’ sense of control, and eroded trust in the reliability of digital referral tools that were intended to streamline practice. Limited training and onboarding compounded these issues, as new users were often expected to self-navigate unfamiliar systems or rely on ad hoc peer support. The combination of technical fragility and inadequate preparation heightened frustration and reinforced perceptions that digital referrals increased workload rather than improved workflow efficiency.

Fragmentation of communication and feedback

Participants consistently highlighted gaps in information flow that undermined continuity of care. A persistent theme across interviews was the unidirectional nature of digital referrals reflecting broader issues of fragmented communication between primary and secondary care. While referrals could be submitted digitally, feedback regarding their receipt, triage, or outcome was infrequent or absent. GPs and PMs described a communication void with one participant noting:

“It feels like sending something into a black hole; you don’t know if it’s been seen, let alone what happens next.” (GP4, NSW/ACT).

That placed additional burden on administrative staff and jeopardised follow-up care. This disintegration led to concerns about incomplete patient journeys and missed appointments. Respective participants expressed that they could not confirm whether referrals had been acted upon unless patients returned or manual follow-up was initiated. As one GP stated,

“Once you refer the patient, it’s off your hands… you might not see the notes unless you follow up intentionally” (GP9, WA). A PM added, “Feedback isn’t consistent unless followed up” (PM2, QLD).

Fax remained in use, largely because it offered perceived reliability and confirmation of receipt, particularly when digital platforms failed due to technical issues, poor training, or systemic outage despite its inefficiency and lack of data security. One participant noted,

“Referrals were paper-based, printed and faxed - at least with fax you know they’ve received something” (GP2, NSW/ACT). Others echoed similar sentiments: “Technical issues or system outages affect eReferral reliability, so a backup method is always necessary” (GP4, NSW), and “Fax is still used out of habit and lack of training in digital platforms, despite being outdated and error-prone” (GP7, VIC).

Participants also criticised the rigid structure of referral templates. Many systems required fields to be completed even when they were not clinically necessary. Inflexibility limited the ability to personalise referrals, particularly in urgent or nuanced cases. One PM explained:

“It’s difficult to pin down one reason for a visit, patients often have multiple issues, but the system forces you to choose one, even if it’s not the main concern.” (PM6, QLD).

The lack of feedback and communication transparency extended beyond administrative inconvenience; it shaped clinicians’ sense of accountability and trust in the referral process. Without confirmation of receipt or outcome, many participants described feeling “disconnected” from patient progress and uncertain about whether referrals were acted upon. This disconnection reinforced the perception of digital referrals as one-way transactions rather than collaborative tools for continuity of care. The cumulative effect of unreliable feedback mechanisms and poor system interoperability compounded clinicians’ workload and undermined confidence in the broader promise of digital health integration.

Pragmatic enablers and emerging preferences

Despite the barriers, several features were identified as enablers of effective digital referral use. Automation, particularly the auto-population of patient information and test results, reduced administrative workload and enhanced accuracy.

“Once you open the patient’s file, you click on HealthLink, and then the data is transferred. You don’t have to manually fill anything. It automatically fills your investigations” (GP7, VIC). Similarly, another stated “Even if you’re new to it, it’s user-friendly. I think it’s idiot-proof”, and another GP added “If Communicare could populate the eReferral platform, like history, medications, allergies, it would be smooth. That’s the beauty of integration” (GP9, WA).

Participants praised the clinical logic embedded in some platforms, especially Smart Referrals. For example, the system could prompt users to include mandatory investigations such as spirometry, improving referral completeness. As one GP explained,

“The essential referral information, particularly the investigations that are necessary, automatically get sucked out of the practice software… you can attach spirometry results easily with Smart Referrals” (GP5, QLD).

Participants appreciated referral systems that offered suggestions based on patient proximity as well as visibility into triage categories and hospital wait times. These features supported clinical decision-making, streamlined referral choices, and enhanced communication with patients about expected timelines and service availability.

“The ability to see referral wait times by category and hospital has improved GP decision-making” (PM5, QLD). In addtion, another participant explained, “I like the idea that you can see the category of the hospital and how long the wait time is. That actually helps us explain things to the patient better” (GP6, NSW).

These features helped reduce referral rejection and improved transparency in service access. However, usability varied significantly across states, and lack of interoperability between systems remained a major concern.

While features such as automatic data entry and referral tracking improved efficiency, their benefits were undermined by ongoing interoperability issues and inconsistent usability across states. The combined effects of technical failures and poor feedback mechanisms continued to disintegrate clinicians’ confidence. Despite some progress, limited integration and uneven implementation highlighted the need for a nationally aligned, user-informed digital referral system.

Aspirations for an inclusive and integrated ecosystem

Across the interviews, there was a clear aspiration for a nationally consistent and integrated referral ecosystem. Participants envisioned platforms that connected public and private sectors, offered two-way communication, and embedded feedback mechanisms that closed the referral loop.

Several GPs expressed frustration that their systems only covered public hospital pathways, forcing them to maintain parallel processes for private referrals. This often-involved manual duplication of information and reliance on fax or email for private sector communication. One GP explained, “Smart Referrals is limited only to the public system, so you still have to write a separate referral letter for private” (GP8, QLD).

Participants also described the need for referral platforms to better support patients from underserved populations, including First Nations communities and individuals experiencing homelessness. The inability to track referral outcomes or receive discharge summaries created considerable gaps in care continuity for these patients. For example, a GP working with homeless populations remarked, “We need a two-way system that actually gives us updates, especially for patients experiencing homelessness who don’t have stable addresses or phones, which complicates follow-up care” (GP3, QLD).

In addition, a PM added: There needs to be better testing and implementation strategies, particularly for Aboriginal Medical Services, as Smart Referrals often fail to accommodate the complexity of our patients and do not always align with their needs. It is also difficult to capture accurate data about COPD during a single consult due to the multitude of issues typically addressed. (PM6, QLD)

This theme also captured frustrations with cross-jurisdictional barriers. For example, GPs practicing in border towns between states described being unable to seamlessly refer patients to services across state lines. One GP explained

The system works well for referrals within our state, but if we need to refer to a hospital in the neighbouring state, we’re back to printing and faxing. It feels like we have one foot in digital and the other in the 1990s. (GP4, NSW)

While recognising that referral systems in Australia are state-based and do not currently interlink, participants across jurisdictions reported similar frustrations with fragmentation, unidirectional communication, and inefficiencies.

Overall, participants described a healthcare environment where digital tools offered promise but often required manual workarounds and supplemental processes to ensure safe and effective care. In addition, participants consistently advocated for platforms that are reliable, intuitive, and interoperable across states, sectors, and systems.

These usability constraints heightened cognitive and administrative workload, prompting clinicians to question whether digital systems genuinely supported rather than complicated their clinical autonomy and patient-centred decision-making.

The lack of system interoperability compounded administrative burden and diminished clinicians’ trust in digital referral tools. Collectively, these interconnected challenges generated a sense of technological fatigue and highlighted the urgent need for an integrated, context-sensitive referral ecosystem. This relationship among themes is depicted in Fig 1, which illustrates how technological fragility, communication gaps, and workflow pressures converge to shape clinicians’ engagement with digital referral systems in COPD care.

Although participants described common challenges across Australia, variations in digital referral policies and system infrastructure between states influenced their experiences. For example, clinicians in Queensland and Victoria reported differing levels of integration with My Health Record and local Primary Health Networks. These jurisdictional inconsistencies affected interoperability, feedback mechanisms, and the perceived usability of digital platforms.

Across all themes, participants envisioned an inclusive digital referral ecosystem linking primary, secondary, and allied health services through a single interoperable platform. Their aspirations reflected frustrations with disrupted workflows, fragmented feedback, and inconsistent usability across systems. Beyond policy reform, participants called for human-centred design, national interoperability standards, and equitable access across regions. These aspirations reflect a deeper professional yearning to restore trust, autonomy, and efficiency through genuinely integrated digital referral practices.

Limitations

While the sample size was appropriate for qualitative inquiry, it may not fully capture the breadth of geographical and demographic diversity across Australia. Most participants were based in QLD and NSW, potentially limiting the representation of experiences from other states and territories. The use of purposive snowball sampling may have introduced selection bias, as individuals more engaged with digital referral systems or holding stronger views may have been more inclined to participate. This may result in underrepresentation of clinicians who are less digitally confident or disengaged from referral technologies. Although the study included participants from a range of general practice settings, rural, regional, and metropolitan, it did not incorporate perspectives from hospital-based clinicians, digital health platform developers, or respiratory specialists. While these groups were outside the scope of the focus of the study on general practice referral processes, their exclusion may limit the broader applicability of the findings across the wider referral ecosystem. As COPD was used as an exemplar condition, the findings should be interpreted as contextually situated insights into digital referral practices rather than disease-specific generalisations. This framing may limit direct transferability to other conditions but strengthens the conceptual understanding of system-level barriers and enablers in primary care digital referral implementation.

Implications

The findings highlight the urgent need for coordinated national efforts to strengthen digital referral infrastructure, particularly for chronic disease management such as COPD. Key priorities include system interoperability, real-time feedback loops, and inclusive platform design that supports both public and private sector workflows. Addressing digital literacy and infrastructure gaps, especially in rural and underserved areas, will be critical to ensuring equitable access. However, achieving these improvements will require navigating the complexities of Australia’s state-based health systems, where digital infrastructure and policy maturity vary across jurisdictions. Collaborative engagement between federal and state governments, digital health developers, and clinical stakeholders will be essential to align priorities and ensure sustainable implementation. These insights can inform co-design initiatives, policymaking, and implementation strategies that reflect the lived experiences of frontline users and contribute to broader goals of health equity and system resilience.