Delta-Like Homologue 1 (DLK1), also known as Preadipocyte Factor 1 (PREF-1), is a non-canonical EGF-like NOTCH ligand. It is maternally imprinted at the Dlk1-Dio3 imprinted locus, and has been shown to regulate embryonic growth, lipid metabolism and skeletal development. However, the precise role of Dlk1 in haematopoiesis, leukaemia and angiogenesis, processes in which it has been previously implicated, is unknown. We generated Dlk1 knockout and conditional knockout mice and used a constitutive overexpression system via retroviral transduction to specifically study Dlk1 in these contexts. Dlk1 knockout mice showed distinctive phenotype of increased perinatal mortality and growth retardation. In foetal livers, significant expression of Dlk1 was detected in the haematopoietic cells, with higher level of expression in the haematopoietic stem cells compared to lineage positive mature cells, with overall expression decreasing with embryonic age. Dlk1 knockout mice were not significantly different from wild type mice in the mature haematopoietic lineages, but serial competitive transplant assays demonstrated Dlk1 knockout bone marrow cells were inferior to controls in reconstituting lethally irradiated recipient mice in short term haematopoietic reconstitution assays, suggesting that Dlk1 knockout led to a defect in adult short term haematopoietic stem cells. Despite frequent overexpression of DLK1 found in many human acute myeloid leukaemias, constitutive overexpression of Dlk1 did not lead to increase in acute leukaemia or death in reconstituted mice. However, Dlk1-overexpressing haematopoietic cells demonstrated competitive repopulation advantage compared to MIG-transduced controls. Using a retinal model of angiogenesis, Dlk1 was found to be expressed by the pericytes rather than endothelium of newly developing blood vessels in postnatal murine pups, in contrast to the published data. Conditional knockout of Dlk1 in endothelial cells using the endothelial specific Tie2 Cre transgene did not lead to significant abnormality in postnatal retina, confirming that Dlk1 did not have a functional role in the retinal endothelium. These new findings add to our current knowledge of stem cell biology and leukaemia, and the role of Dlk1 in angiogenesis.

Aggressive B-cell lymphomas include diffuse large B-cell lymphoma, Burkitt lymphoma and intermediate forms. Despite high response rates to conventional immuno-chemotherapeutic approaches, an unmet need for novel therapeutic strategies is required in the setting of relapsed and refractory disease, typified by resistance to chemotherapy and radiotherapy. The proto-oncogene MYC is frequently dysregulated in the aggressive B-cell lymphomas, however, it has proven an elusive direct therapeutic target. A significant body of evidence is accumulating to suggest that MYC-dysregulated disease maintains a ‘transcriptionally-addicted’ state, whereby perturbation of RNA polymerase II activity may indirectly antagonise MYC activity. Furthermore, very recent studies implicate anti-apoptotic myeloid cell leukaemia 1 (MCL-1) as a critical survival determinant of MYC-driven lymphoma. This thesis utilises pharmacologic and genetic techniques in MYC-driven models of aggressive B-cell lymphoma to demonstrate that cyclin-dependent kinase 9 (CDK9) and MCL-1 are oncogenic dependencies of this subset of disease. The cyclin-dependent kinase inhibitor, dinaciclib, and more selective CDK9 inhibitors are used to demonstrate efficient apoptosis induction conferred at least in part by downregulation of MCL1 transcription. Furthermore, a genetic screen identifies other transcriptional cyclin-dependent kinases that are required for viability of MYC-driven lymphoid disease. Finally, having established MCL-1 as a critical oncogenic dependency of MYC-driven lymphoma, this thesis demonstrates the significant activity that is conferred by direct pharmacologic antagonism of MCL-1 using a small molecule BH3-mimetic inhibitor of MCL-1. These findings confirm a druggable pathway of oncogenic cMYC dependency involving CDK9 regulated RNA polymerase II-mediated transcription of MCL-1, and proposes pharmacologic inhibition of CDK9 and MCL-1 as novel anti-lymphoma strategies.

Written communication within the health professions has been rapidly changing over the last decade. Implementation of Electronic Health Records (EHR) in health services is now widespread. Medical student teaching and learning of the skills specifically required for EHRs has lagged behind the implementation. Very few original studies have focused on EHR skills and there are no validated measures by which to assess any of the EHR skills students are expected to develop. Our study explored the attributes of quality EHR documentation recorded by medical students, with the purpose of the EHR documentation being the communication between health care professionals to share or transfer the clinical care of a patient. Recently there have been published validated instruments for measuring quality in physician EHR documentation, one being Physician Documentation Quality Instrument (PDQI-9). The purpose of this study was to explore the attributes of quality of EHR documentation written by first-year clinical medical students by building upon existing literature. The PDQI-9 was used as a basis for defining the attributes of quality in EHR documentation as a foundation for assessing and providing feedback on the performance of documentation to medical students. With the focus on assessment, and providing a content validated test domain for assessment in quality EHR documentation, we utilised Kane’s framework for validity to structure the study and a mixed method study design to achieve the depth of exploration required to examine the performance of quality documentation fully. The study was conducted in two stages. In the first stage of the study, an expert panel of assessors applied the PDQI-9 to existing EHR data recorded by first clinical year medical students in a graduate entry program. The assessors both scored the records and justified their grading. Descriptive statistics and thematic analysis were undertaken on the data collected, and the findings triangulated with the literature review. The second stage employed explanatory semi-structured interviews with the expert assessors to better understand the findings of the first stage and reach consensus on a test domain for assessing quality documentation recorded by medical students. Outcomes from our study indicated that the PDQI-9 in its current format was not valid in a medical student setting, however most of the attributes assessed by the PDQI-9 were deemed relevant and meaningful to assess if their interpretations were clarified. In addition, Professionalism of documentation was regarded as a quality attribute. Consensus was reached on modifications that have the potential to improve the validity of the assessment of quality documentation recorded by medical students. Further studies need to complete Kane’s framework of validity for an assessment instrument and collect evidence to broaden the validity of the scoring, the generalization of the assessment items, the extrapolation to the real world and the implications of this assessment for students and health services.

Online support groups (OSGs) are popular sources of both health information and social support. Though early research into OSGs highlighted a concern that non-expert members may give harmful advice, more recent studies have typically shown that engagement with OSGs can increase consumer satisfaction with the treatment process, enhance wellbeing, and ultimately improve health outcomes. OSGs are well-researched within applied linguistics. Qualitative studies have focussed on member roles within OSGs, as well as the ways in which group members discursively construct their identities, especially with respect to their illness. These approaches have generated rich insights into consumer healthcare discourse that can inform strategies for fostering consumer-centred care. Qualitative approaches, however, are resource-intensive, di cult to reproduce, and limited in terms of generalisability and representativeness. Quantitative and computational approaches are able to overcome these shortcomings, adding transparency, reproducibility, scalability, and reducing the potential for researcher bias. Current computational approaches to consumer healthcare discourse, however, tend to rely on simplified conceptualisations of language, prioritising lexis over grammar, and thus ignoring the central role played by grammar in the meaning-making process. To address current methodological shortcomings in OSG discourse research, this thesis presents an interdisciplinary, corpus-based investigation of lexicogrammatical and discourse-semantic choices made by members over the course of membership in an online bipolar disorder support community. 8.2 million words in over 66,000 posts from approximately 3,500 members were transformed into a metadata-rich, grammatically annotated corpus and investigated from a systemic-functional linguistic (SFL) perspective using purpose-built corpus/computational linguistic tools. An analysis of MOOD and MODALITY choices made over ten stages of membership highlights differences in the ways members negotiate role-relationships, with changes in Mood Type, Modality and Speech Function reflecting a longitudinal increase in the provision of advice and social support. An analysis of the TRANSITIVITY system shows longitudinal changes in the kinds of participants and processes construed by Forum members, as well as changes in how these participants and processes behave lexicogrammatically. The diagnosis Event, for example, is represented by newcomers as a process and modi ed temporally; at later stages of membership, it is more often reconstrued as a participant in discourse, framed in terms of veracity. Longitudinal shifts were also observed in the preferred ways of ascribing/attributing bipolar disorder to Forum members: new members use bing forms (I’m bipolar), while veteran members prefer having constructions (I have bipolar). The thesis has implications for corpus linguistics, systemic-functional linguistic theory, and healthcare communication research. For corpus linguistics and corpus-assisted discourse studies, the main contribution is corpkit, an open-source software tool designed to build and analyse parsed and metadata-rich corpora. it is suggested that the developed tools and methods can circumvent theoretically problematic current practices, and increase the accuracy and automatability of the analytical process. For healthcare communication research, the case study demonstrates the importance of expanding the conceptualisation and analysis of the consumer healthcare journey to include intra–consumer communication that occurs outside of hospitals and clinics. The thesis also advances an argument that the emerging eld of clinical natural language processing stands to benefit from increased engagement with functional linguistic theory and insights generated within the qualitative paradigm. I argue that combining the insights from functional linguistics and discourse analysis with automated computational workflows is a step toward an important future goal of improvement of consumer health outcomes through analysis of large, digital collections of spoken and written healthcare discourse.

Understanding naturally acquired immunity to malaria and how human antibodies protect against clinical malaria is essential for vaccine development. Antibodies against Plasmodium falciparum apical membrane antigen 1 (AMA1), a leading vaccine candidate, can inhibit merozoite invasion of erythrocytes and protect from P. falciparum malaria. However, polymorphism in antigens like AMA1 is a common mechanism for immune evasion and presents major challenges in vaccine development. This study aims to understand antigenic diversity of AMA1, the correlation between sequence polymorphism and antigenic differences, the impact of polymorphism on potential vaccine escape, and the structural differences of the protein amongst various alleles, with the goal to ultimately ascertain which AMA1 alleles should be included in an effective multi-allele vaccine.

The work presented here is the product of several years of clinical trial and laboratory endeavour focused on the theme of understanding and improving currently available epigenetic therapies for blood cancers, and understanding fundamental aspects of clinical trial design. Several classes of epigenetically active drugs with proven clinical usefulness have received regulatory approval for the therapy of haematological malignancies in the last 5 years. At the same time, advances in high throughput sequencing technologies have provided landmark discoveries that reinforce the hypothesis that epigenetic mechanisms play a central role in the neoplastic process, as well as giving us new opportunities to develop targeted therapies. Some of the most exciting new targeted biological therapies arising from these discoveries such as inhibitors of histone methyl-transferases, the BET bromodomain inhibitors and the isocitrate dehydrogenase inhibitors are only now in the earliest phase of clinical development with initial signals that many of these agents will be both tolerable and active therapies for blood and solid tumours. In the context of the field at the start of 2010, only two classes of agent were available in clinical trials and for early investigator-initiated development. Even today, only these classes of epigenetic agents have marketing approval. As a consequence, histone deacetylase inhibitors and DNA methyl-transferase inhibitors are the focus of this thesis. The thesis presents investigator-initiated trials of agents in these two drug classes as well as the theory associated with the design of the trials. The literature review in Chapter 1 establishes the foundation for the clinical trial work. Here, the mechanism of action of the histone deacetylase inhibitors and the DNA-demethylating agents is reviewed. Activity of these agents in industry-sponsored clinical trials is presented as well as key toxicity data that informs the hypotheses and design of the clinical research presented in the thesis. The theories underlying the combination therapies that form the backbone of the thesis are discussed. Eltrombopag, a thrombopoietin agonist, is given particular prominence for its potential as a supportive care agent. The clinical trials that represent the majority of the thesis were designed to explore the critical questions of improving the tolerability of the DNA-demethylating agent azacitidine, and developing novel combination therapies for HDAC inhibitors. As experimental design is a critical question, a discussion on clinical trial design is presented in Chapter 2, which closes with a relevant case study. Several clinical trials were developed during the conduct of this higher degree from concept through to funding and clinical recruitment. Three, presented in Chapters 3, 4, and 7 have completed accrual and a further trial (Chapter 5) has been interrupted due to administrative challenges and is currently being reworked. The protocol for each study was in itself an extensive body of novel work contributing to the thesis. These are included in the appendices but summarised within the body of the thesis. Chapter 3 presents finalised and novel data analysed in depth of a trial of the combination of eltrombopag, a thrombopoietin-receptor agonist and azacitidine. The purpose of the study was to develop a supportive care therapy for patients with myelodysplastic syndromes, a disease that causes low platelets (thrombocytopenia), and has a risk of transforming to acute myeloid leukaemia, an almost invariably fatal event. Thrombopoietin mimetic agents have a somewhat complicated relationship with myeloid malignancies in the minds of researchers and while the receptor (mpl) and the role of the mpl receptor in haematopoietic stem cell survival and fate has been the subject of the research for at least 20 years, many questions about their theoretical safety remain open. There is considerable debate over the theoretical safety of mpl in agonists in patients with myeloproliferative diseases. Hence, attention is given to the rationale for the choice of eltrombopag for myelodysplastic syndrome in Chapter 1; also, the established body of preclinical data is presented to provide a context. This particular trial, “AzaE”, has been presented in international fora and the data has provided a foundation for a 170-centre, 43-country international randomised study that, if positive, may change clinical practice for patients with myelodysplastic syndrome and thrombocytopenia. Hence Chapter 3 is a central part of the novel data in this thesis. Based on the hypothesis that HDAC inhibition would up-regulate CD20 on the surface of malignant B-cells, a collaborating colleague, Mark Bishton, developed a phase 1 clinical trial of the combination of the radio labelled rituximab and the HDACi panobinostat. Dr Bishton and Prof Seymour wrote the trial protocol. As Principal Investigator of the study, I set up the study, moving it from protocol to execution. I oversaw the conduct of the study as the principal investigator and reviewed and approved the clinical data. The combination was toxic and the lessons learned are cogent and argue against further developing this sort of combination therapy. This study is presented in Chapter 4. In order to address the moderate activity of the HDACi inhibitor romidepsin and to broaden its clinical indications, a phase I/II study was designed in collaboration with academic partners from Yale University. The strategy behind the design of that study, the protocol itself, and initial data from six patients that have been recruited is presented in Chapter 5. That study has been halted due to regulatory challenges in the USA and Yale University that interrupted the research without warning. The trial strategy is discussed and preliminary results are presented. HDACi are highly active in cancers of the immune system, where the symptomatic and physical manifestations are to a considerable extent driven by cytokine changes which serve a both disease epiphenomena and drivers of the malignancy itself. In Chapter 6, I investigate the effect of the HDACi on human monocyte derived dendritic cells, identifying a mechanism by which these agents are likely shift the T-cell populations, altering patient symptoms and potentially inducing clinical responses. The findings support the contention that HDACi are immunosuppressive and do not make suitable partner drugs for treatments that benefit from an intact antigen presentation apparatus. Histone deacetylase inhibitors and DNA demethylating agents are often conceptually explained as driving a single cellular mechanism. However, the effects within the cell are pleiotropic and there are very few studies that provide data on predictors for response, especially so for the HDAC inhibitors. Therefore, a part of this work was to design a clinical trial that enables future scientific studies on predictors of response to HDAC inhibitors. Several academic questions were faced in developing the approach to this study. Chapter 7 presents the approach to the design of the clinical study, and presents the clinical features of the patients who have been recruited and key response data. An overview of planned experiments is presented. Having concerned itself with the design of investigator-initiated clinical trials of novel combinations and presented successes and failures of trial design and execution as well as original laboratory work that may inform combinations, the thesis concludes with a discussion on future directions for the projects arising from this work and epigenetic agents in general.

Menopausal hot flushes affect up to 90% of women and pose a significant health and financial burden. Many women are reluctant to use hormone replacement therapy, an effective treatment, because of fear of serious adverse events. Acupuncture is a safe treatment with conflicting evidence for efficacy as a hot flush treatment. The aim of this thesis was to assess the efficacy of acupuncture for menopausal hot flushes. The principal study was a fully powered randomised sham-controlled trial which enrolled women aged over 40 in the late menopausal transition or postmenopause with a moderate hot flush burden, and who met criteria for the Chinese medicine diagnosis of Kidney Yin deficiency. Women were randomised to receive eight weeks of either Chinese medicine needle acupuncture or non-insertive sham acupuncture. The primary outcome was hot flush score at the end of treatment. Secondary outcomes included quality-of-life, anxiety, depression, and adverse events. Women were followed up at 4 and 8 weeks (end of treatment) and at 3 and 6 months post end of treatment. Analysis was by intention-to-treat using mixed effects modelling. 327 women were randomised to real (n=163) or sham acupuncture (n=164). There was no evidence for a between-group difference in hot flush scores at end-of-treatment (mean difference 0·33; 95% CI -1·87 to 2·52; p=0·77) or for secondary outcomes, and no serious adverse events. Both groups improved by an average of 40% from baseline for hot flush score. There was a small, clinically irrelevant difference for hot flush severity at end of treatment. Limitations of the study are that participants were predominantly Caucasian and none had breast cancer or surgical menopause, conditions which predispose to more severe hot flushes. In conclusion, an eight week course of Chinese medicine acupuncture was not superior to non-insertive sham needling for menopausal hot flushes. Women may have improved because of natural resolution of symptoms, frequent therapist interaction, or the placebo effect. A specific effect from acupuncture needling was not demonstrated. Future research may focus on the role of acupuncture in breast cancer survivors.

The career medical interview is rare in that it is a central discriminator for aspirant doctors and trainees. Not only does the outcome determine the future job prospects for the candidate, but it also decides access to continued education and training. Despite the importance of the interview and these tensions, little training is provided to aspirants and there are often conflicting messages. Furthermore, the interview is not a reliable test measure; it is susceptible to a range of bias and rating difficulties. Additionally, the number of places is outweighed by the number of potential applicants, hence adding further tension to the process. One possible outcome of the lack of training programs specific to the career medical interview is the susceptibility to impression management. Impression management is where a candidate intentionally influences the outcome of an interview by engaging in the active manipulation of their image portrayal. At worst, candidates might engage in faking (where facts are made up or deceitful responses are given). Presently, little is known about the impact or strategies of medical candidates and impression management in the career medical interview. It is known that a key factor in successful career interview performance is the management of anxiety. Too often, anxiety in regards to the interview or communication apprehension can cause distorted behaviour. This may result in excellent candidates being unsuccessful not because they may not be the best candidates, but because they were unable to manage their anxiety sufficiently well. This can result in clinical and training impacts if the wrong candidates are selected. This thesis successfully proposes a new model for communication training for the career medical interview to address these tensions. Drawing on the body of research known as applied theatre, a new training model is developed and tested using psychometric scales designed to measure anxiety. Kirkpatrick’s model of measuring training effect is utilised. This thesis also draws new insight into existing training strategies used by candidates undertaking interviews in Australian medical contexts as well as capturing data in regards to impression management being utilised by currently successful surgical trainees.

Medical training in Australia and comparable countries internationally is based on an apprenticeship model of training. However this has been challenged by contemporary healthcare practices as well as the limited training capacity of the acute care settings where medical training is largely undertaken. While this impacts all phases of medical training, it is the prevocational phase in Australia that is most vulnerable as its training outcomes are poorly defined and not routinely measured. Similarly, while a twelve‐month rotation‐based internship has existed in Australia for decades, the effectiveness of this model in providing core training and clinical competencies for prevocational doctors has not been rigorously evaluated, and is currently being questioned. This research seeks to better understand the clinical placement experience of prevocational doctors in Victorian Health Services, and to consider how this may be affected by contemporary healthcare delivery. The research has been designed as a mixed –methods study, where data obtained from a broad‐based survey of exposure to a range of clinical curriculum‐based activities of junior doctors from seven Victorian health services was explored by groups of junior doctor supervisors and managers. These research findings have concurred with limited previous literature, noting that prevocational trainees have limited exposure to a number of curriculum areas, particularly within the curriculum domains of procedures and emergency management, as well as teaching and learning activities, and other more complex patient management and interaction activities. It has identified that particular intern core terms and PGY2 clinical streams may provide better access to curriculum experiences than others, and that there may be particular curriculum strengths and weaknesses of prevocational training in metropolitan versus regional areas. It has also suggested that the current prevocational curriculum framework is variably understood by health service supervisors and managers, and that these findings have potentially significant implications for the trainees themselves, for length of training, and for the confidence and competence of the end practitioner. Essentially a key finding from this research is that acute health services are increasingly performing high‐risk activities within constrained environments, resulting in changes to healthcare teams and individual roles – and the legitimate peripheral participation of junior doctors in clinical care has decreased in proportion to the acuity and specialisation of the activity involved. While local educational initiatives to overcome this challenge are currently being variably utilised by health services, they do not ensure ongoing repeated curriculum exposure and therefore curriculum mastery. Instead, system‐wide issues require to be addressed by a system wide approach. Lave and Wanger’s Situated Learning theory (1991) provided a framework through which the research results were viewed, and provides a lens through which recommendations can be conceptualised. Essentially, this thesis recommends reform of prevocational training ‐ that redefines its role; that addresses curriculum ambiguity; that enhances junior doctor access to legitimate peripheral participation; and that re‐establishes communities of practice. These medical training reforms need to be undertaken within a clear, integrated governance model, and supported by ongoing evaluation and innovation.

The increasing use of online technology in medical education and junior doctor training raises questions about the efficacy of this approach to improve the preparedness of medical students transitioning to internship. This study examined the use of eOrientation resources for hospital and ward orientation by medical students during transition from medical school to internship, with the aim of determining what attributes make an online resource useful and what effect an eOrientation program can have on preparedness for hospital practice. More specifically, the research questions focussed on how intern preparedness changed following implementation of eOrientation, which resources were most and least used, which were found to be most and least useful and what common themes could be found with respect to the type of resources interns found most useful. A mixed method case study approach was used to explore the use of eLearning resources and preparedness of medical students transitioning to internship. The Preparation for Hospital Practice questionnaire was administered to a cohort of 74 interns at a metropolitan teaching hospital before they had undertaken any form of hospital orientation. The questionnaire was then readministered at the conclusion of their second hospital unit rotation. Learning management system data was used to examine resource usage, supplemented by personal interviews to extract detail rich information regarding which resources were found most useful. The study found that interns initially felt well prepared for internship with a minor increase in preparedness over time. Resources which were interactive, role relevant and assisted in application of identity formation tasks were deemed most useful. eOrientation was found to be a useful supplement to the suite of teaching and learning methods used for transition to internship. Findings from the study suggest that the value of eOrientation can be further enhanced when resources are directed to interns, interactive and role relevant while also considering the temporality of a resource. Resources made by or with input from interns were also found to be particularly useful, leading to a recommendation of utilising instructional designers in medical education teams in order to assist in managing stakeholder input, assess content suitability across the transition curriculum and design interactive resources. eOrientation can be an effective teaching and learning method if a considered and holistic approach to transitional curriculum development is adopted.

As participatory health develops, healthcare as a whole continues to cautiously embrace social media and the vast possibilities they can bring. They may be useful to transform the management of people living with chronic disease by enabling them to be more active participants and shared-decision makers in their own care. Unfortunately, limited definitive evidence exists indicating that social media use can improve health outcomes, nor is research mature enough to have established stringent methodologies for examining social media in this context. Unlike a variety of evidence-based guidelines suggesting best practice for an array of chronic diseases, evidence that could underpin guidelines for use of social media remains inadequate. There are many unknowns, for example: How social media are used as part of self-management, what mechanisms underpin social media’s therapeutic potential, the health outcomes and other effects most likely to be impacted by social media, and whether social media can form an effective part of management in the clinical setting. The key challenge addressed by this thesis is whether criteria and methods for generating and evaluating evidence about health outcomes from social media use can be defined. A framework for research and practice has been developed, tested and refined to address this challenge. With further research using the framework across a range of conditions and contexts, evidence about patient-reported outcomes from social media use will grow. This will improve evidence-based decision-making about social media and ultimately benefit a variety of stakeholders: patients, clinicians, health researchers and health informaticians.

Background: The majority of care for people with type 2 diabetes occurs in general practice in Australia, yet insulin initiation is often delayed in this setting and many people are referred to specialists (endocrinologists and registered nurse credentialed diabetes educators (RN-CDEs)). Stepping Up (SU) is a new model of care supporting insulin initiation in general practice which includes an enhanced role for practice nurses (PNs) with support and mentoring from a SU RN-CDE. Successful implementation of this model of care is likely to be shaped by interprofessional roles and relationships. Relational Coordination (RC) theory provides a conceptual framework to study key factors underpinning effective interprofessional work. Aim: To explore the roles and relationships between health professionals involved in insulin initiation, how these are affected by models of care and whether, within the Stepping Up model of care, they are associated with improved clinical outcomes. Methods: I conducted five related sub-studies utilising RC theory. The first two sub-studies were exploratory and utilised a survey and interviews to explore current levels of RC between endocrinologists, RN-CDE, General Practitioners (GPs) and PNs for the task of insulin initiation, perceived appropriateness of this task in general practice and contextual factors influencing RC. The last three sub-studies were nested within a randomised controlled trial testing the effectiveness of SU. I analysed a baseline cross sectional survey exploring characteristics of participating GPs, PNs and general practices associated with RC for the task of insulin initiation. I explored how GPs and PNs perceived that SU had impacted on professional roles and RC using qualitative methods. Finally, I performed a quantitative analysis to determine whether RC was a partial mediator of the effect of SU on glycaemic outcomes. Results: There was agreement from all four health professional groups that general practice-based initiation of insulin was appropriate. Potential barriers to achieving this in practice included a lack of communication and relationships between health professionals across secondary/specialist and primary care, with weak RC reported between PNs and RN-CDE. Specialists had concerns about the competence and capacity for PNs to have a role in insulin initiation. Factors associated with stronger RC included PN qualifications in diabetes education, PN prior experience with insulin initiation, younger PN age and increasing rurality of general practices. Dedicated space and protected time for PNs to participate in insulin initiation and RN-CDE mentoring were identified as important facilitators for an expanded PN role. While SU was associated with improved glycaemic outcomes, this was not mediated by RC with RN-CDE reported by PNs. Conclusion: SU was explicitly designed to change professional roles to support insulin initiation in general practice. SU was associated with stronger inter-professional relationships, particularly between PNs and the SU RN-CDE, and improved clinical outcomes for patients. Other important factors in normalising the SU model of care into Australian general practice include achieving interprofessional consensus on the role of PNs in insulin initiation, funding structures to support the PN and SU RN-CDE roles and organising general practice to support expanded PN roles.