Phosphorylation of Syntaxin 4 by the Insulin Receptor Drives Exocytic SNARE Complex Formation to Deliver GLUT4 to the Cell Surface.

A major consequence of insulin binding its receptor on fat and muscle cells is the stimulation of glucose transport into these tissues. This is achieved through an increase in the exocytic trafficking rate of the facilitative glucose transporter GLUT4 from intracellular stores to the cell surface. Delivery of GLUT4 to the cell surface requires the formation of functional SNARE complexes containing Syntaxin 4, SNAP23, and VAMP2. Insulin stimulates the formation of these complexes and concomitantly causes phosphorylation of Syntaxin 4. Here, we use a combination of biochemistry and cell biological approaches to provide a mechanistic link between these observations. We present data to support the hypothesis that Tyr-115 and Tyr-251 of Syntaxin 4 are direct substrates of activated insulin receptors, and that these residues modulate the protein's conformation and thus regulate the rate at which Syntaxin 4 forms SNARE complexes that deliver GLUT4 to the cell surface. This report provides molecular details on how the cell regulates SNARE-mediated membrane traffic in response to an external stimulus.

Reframing context specificity in team diagnosis using the theory of distributed cognition.

Context specificity refers to the vexing phenomenon whereby a physician can see two patients with the same presenting complaint, identical history and physical examination findings, but due to specific situational (contextual) factors arrives at two different diagnostic labels. Context specificity remains incompletely understood and undoubtedly leads to unwanted variance in diagnostic outcomes. Previous empirical work has demonstrated that a variety of contextual factors impacts clinical reasoning. These findings, however, have largely focused on the individual clinician; here we broaden this work to reframe context specificity in relation to clinical reasoning by an internal medicine rounding team through the lens of Distributed Cognition (DCog). In this model, we see how meaning is distributed amongst the different members of a rounding team in a dynamic fashion that evolves over time. We describe four different ways in which context specificity plays out differently in team-based clinical care than for a single clinician. While we use examples from internal medicine, we believe that the concepts we present apply equally to other specialties and fields in health care.

Twine virtual patient games as an online resource for undergraduate diabetes acute care education.

BACKGROUND: Virtual patients provide a safe way to simulate authentic clinical practice. Twine is an open-source software that can be used to create intricate virtual patient games, including elements like non-linear free text history taking and time-related changes to the game's narrative. We evaluated the incorporation of Twine virtual patient games into a diabetes acute care online learning package for undergraduate medical students at the University of Glassgow, Scotland. METHODS: Three games were developed using Twine, Wacom Intuous Pro, Autodesk SketchBook, Camtasia Studio, and simulated patients. Online material included three VP games, eight microlectures, and a single best answer multiple choice question quiz. The games were evaluated at Kirkpatrick Level 1 with an acceptability and usability questionnaire. The entire online package was evaluated at Kirkpatrick Level 2 with pre- and post-course multiple choice and confidence questions, with statistical analysis performed using paired t-tests. RESULTS: 122 of approximately 270 eligible students provided information on resource utilisation, with 96% of these students using at least one online resource. 68% of students who returned surveys used at least one VP game. 73 students provided feedback on the VP games they had played, with the majority of median responses being "agree" on positive usability and acceptability statements. The online resources were associated with a mean multiple choice score increase from 4.37 out of 10 to 7.96 out of 10 (p < 0.0001, 95% CI + 2.99 to + 4.20, n = 52) and a mean total confidence score increase from 4.86 out of 10 to 6.70 out of 10 (p < 0.0001, 95% CI + 1.37 to + 2.30, n = 48). CONCLUSIONS: Our VP games were well-received by students and promoted engagement with online material. The package of online material led to statistically significant increases in confidence and knowledge in diabetes acute care outcomes. A blueprint with supporting instructions has now been created to facilitate rapid creation of further games using Twine software.

Distributed cognition: Theoretical insights and practical applications to health professions education: AMEE Guide No. 159.

Distributed cognition (DCog) is a member of the family of situativity theories that widens the lens of cognition from occurring solely inside the head to being socially, materially and temporally distributed within a dynamic system. The concept of extending the view of cognition to outside the head of a single health professional is relatively new in the healthcare system. DCog has been increasingly used by researchers to describe many ways in which health professionals perform in teams within structured clinical environments to deliver healthcare for patients. In this Guide, we expound ten central tenets of the macro (grand) theory of DCog (1. Cognition is decentralized in a system; 2. The unit of analysis is the system; 3. Cognitive processes are distributed; 4. Cognitive processes emerge from interactions; 5. Cognitive processes are interdependent; 6. Social organization is a cognitive architecture; 7. Division of labour; 8. Social organization is a system of communication; 9. Buffering and filtering; 10. Cognitive processes are encultured) to provide theoretical insights as well as practical applications to the field of health professions education.

New approach to undergraduate quality improvement.

BACKGROUND: While quality improvement (QI) is an essential component to modern day clinical practice, some foundation doctors fail to engage. This is compounded by a lack of formalised undergraduate QI teaching. We trial an undergraduate active learning workshop and evaluate it using a concurrent triangulation mixed methods design. APPROACH: We constructed a 2-hour interactive QI workshop utilising near-peer educators for third year undergraduate medical students. Our workshop demonstrated an exemplary project and a template featuring evidenced-based QI tools to grasp key concepts. Informal support was provided for student QI projects, undertaken in small peer groups. Utility was assessed using linked pre-and-post event questionnaires with Likert scales, free text thematic analysis and project completion rates. EVALUATION: We recruited 74 students to attend our workshops delivered over 3 months. We achieved high event satisfaction and significant improvements on baseline confidence. Free text comments suggested students perceive QI as an important part of the undergraduate curriculum, described barriers to engagement and the value they place on project autonomy. The workshop eased student feelings of anxiety and intimidation regarding change ideas. Nine projects were completed with one winning a poster prize at a regional conference. IMPLICATIONS: We demonstrate a popular resource light model that can be scaled up to a variety of centres. Targeting QI teaching at the undergraduate level may be instrumental in developing QI culture in health care systems and address barriers to postgraduate involvement. Our study furthers the understanding of undergraduate students' perspectives of QI and demand for further sessions.

Increasing Collaborative Discussion in Case-Based Learning Improves Student Engagement and Knowledge Acquisition.

Background: In the transition from academic to clinical learning, the development of clinical reasoning skills and teamwork is essential, but not easily achieved by didactic teaching only. Case-based learning (CBL) was designed to stimulate discussions of genuine clinical cases and diagnoses but in our initial format (CBL'10) remained predominantly tutor-driven rather than student-directed. However, interactive teaching methods stimulate deep learning and consolidate taught material, and we therefore introduced a more collaborative CBL (cCBL), featuring a structured format with discussions in small breakout groups. This aimed to increase student participation and improve learning outcomes. Method: A survey with open and closed questions was distributed among 149 students and 36 tutors that had participated in sessions of both CBL formats. A statistical analysis compared exam scores of topics taught via CBL'10 and cCBL. Results: Students and tutors both evaluated the switch to cCBL positively, reporting that it increased student participation and enhanced consolidation and integration of the wider subject area. They also reported that the cCBL sessions increased constructive discussion and stimulated deep learning. Moreover, tutors found the more structured cCBL sessions easier to facilitate. Analysis of exam results showed that summative assessment scores of subjects switched to cCBL significantly increased compared to previous years, whereas scores of subjects that remained taught as CBL'10 did not change. Conclusions: Compared to our initial, tutor-led CBL format, cCBL resulted in improved educational outcomes, leading to increased participation, confidence, discussion and higher exam scores.

Something Stinks! Finding Ways to Manage Noxious Odours in the Operating Room and Other Clinical Settings A Randomized Controlled Trial.

Objectives: The hospital can be saturated with noxious smells. Anecdotally, medical staff apply products to surgical masks to lessen the impact of these smells. This study aimed to determine the odour-masking ability of 4 inexpensive and convenient products. Methods: A randomized, single-blinded crossover study was conducted in Vancouver, Canada. Participants, 19 to 30 years old, were invited to participate. Participants with active allergies, upper respiratory tract infection, alteration to sense of smell, or failure of olfactory screen were excluded from the study. An experimental odour was used in lieu of a noxious surgical odour. After smelling the experimental odour without barriers, participants were re-exposed to the odour using 5 surgical masks in randomized order. Each mask was lined with a test product (cherry lip balm, tincture of benzoin, Mastisol, mint toothpaste, and control [plain mask]). Participants rated the effectiveness of products at masking the experimental odour from 0 to 100 (0 = completely ineffective, 100 = completely effective). Participants also rated the pleasantness of the products, recorded if the products made them feel unwell, and identified their preferred product overall. Results: Eighty participants were included in the study (33 male, 47 female), averaging 24.2 years of age. Mean odour-masking effectiveness for cherry lip balm was 66.5 (±24.6), tincture of benzoin: 62.6 (±25.0), Mastisol: 61.3 (±23.9), mint toothpaste: 57.5 (±27.4), and control: 21.9 (±21.8). All products performed better than the control (P < .001), but there was no significant difference in performance between products. Cherry lip balm was the most preferred odour-masking product (29 participants), followed by mint toothpaste (22), Mastisol (14), tincture of benzoin (10), and control (5). Conclusions: All tested products demonstrated equivalent odour-masking abilities. If health care professionals choose to use an odour-masking product, they should consider their own olfactory preferences.

Objectifs: L'hôpital peut être saturé d'odeurs nauséabondes. On rapporte que le personnel médical applique des produits sur leurs masques chirurgicaux pour atténuer l'impact de ces odeurs. Cette étude visait à déterminer l'efficacité de quatre produits masqueurs d'odeurs, et ce de façon pratique et peu couteuse. Méthodes: Une étude croisée a simple insu et a répartition aléatoire a été menée à Vancouver, Canada. Des participants âgés de 19 à 30 ans, ont été invités à participer. Les participants souffrant d'allergies actives, d'une infection des voies respiratoires supérieures, d'une altération olfactive, ou aillant échoué la procédure de sélection ont été exclus de l'étude. Une odeur expérimentale a été utilisée au lieu d'une odeur nauséabonde chirurgicale. Après avoir senti l'odeur expérimentale, les participants ont été réexposés à la même odeur à cinq reprises. A chaque reprise, le participant était muni d'un de 5 masques tapissé d'un agent masquant d'odeur (baume à lèvres aux cerises, teinture de benzoïne, mastisol, dentifrice à la menthe, et contrôle [masque standard]). L'ordre des masques a été déterminé de façon aléatoire. Les participants ont noté sur une échelle de 0 à 100 l'efficacité des produits à masquer l'odeur (0: complètement inefficace, 100: complètement efficace). Les participants ont également évalué la qualité plaisante des agents, si ceux-ci les rendaient nauséeux, et ont ensuite identifié leur produit préféré parmi l'ensemble. Résultats: Quatre-vingts participants ont été inclus dans l'étude (33 hommes, 47 femmes), âgés en moyenne de 24,2 ans. L'efficacité des produits à masquer l'odeur expérimentale étaient d'une moyenne de 66,5 (+24,6) pour le baume à lèvres aux cerises ; 62,6 (+25,0) pour la teinture de benzoine ; 61,3 (+23,9) le mastisol ; 57,5 (+27,4) pour le dentifrice à la menthe, et 21,9 (+21,8) le contrôle. Tous les agents testés ont reçu une note supérieure au contrôle (P < .001). Par-contre, il n'y avait pas de différence significative entre les agents. Le baume à lèvres aux cerises était le produit préféré (29 participants), suivi du dentifrice à la menthe (22), du mastisol (14), de la teinture de benzoine (10), et finalement du contrôle (5). Conclusions: Tous les produits testés ont démontré une efficacité similaire, celle-ci supérieure comparée au contrôle. Si les professionnels de la santé souhaitent d'utiliser un produit qui masque les odeurs, ils devraient tenir compte de leurs propres préférences.

Diabetes is accompanied by changes in the levels of proteins involved in endosomal GLUT4 trafficking in obese human skeletal muscle.

INTRODUCTION: The regulated delivery of the glucose transporter GLUT4 from intracellular stores to the plasma membrane underpins insulin-stimulated glucose transport. Insulin-stimulated glucose transport is impaired in skeletal muscle of patients with type-2 diabetes, and this may arise because of impaired intracellular trafficking of GLUT4. However, molecular details of any such impairment have not been described. We hypothesized that GLUT4 and/or levels of proteins involved in intracellular GLUT4 trafficking may be impaired in skeletal muscle in type-2 diabetes and tested this in obese individuals without and without type-2 diabetes. METHODS: We recruited 12 participants with type-2 diabetes and 12 control participants. All were overweight or obese with BMI of 25-45 kg/m2 . Insulin sensitivity was measured using an insulin suppression test (IST), and vastus lateralis biopsies were taken in the fasted state. Cell extracts were immunoblotted to quantify levels of a range of proteins known to be involved in intracellular GLUT4 trafficking. RESULTS: Obese participants with type-2 diabetes exhibited elevated fasting blood glucose and increased steady state glucose infusion rates in the IST compared with controls. Consistent with this, skeletal muscle from those with type-2 diabetes expressed lower levels of GLUT4 (30%, p = .014). Levels of Syntaxin4, a key protein involved in GLUT4 vesicle fusion with the plasma membrane, were similar between groups. By contrast, we observed reductions in levels of Syntaxin16 (33.7%, p = 0.05), Sortilin (44%, p = .006) and Sorting Nexin-1 (21.5%, p = .039) and -27 (60%, p = .001), key proteins involved in the intracellular sorting of GLUT4, in participants with type-2 diabetes. CONCLUSIONS: We report significant reductions of proteins involved in the endosomal trafficking of GLUT4 in skeletal muscle in obese people with type 2 diabetes compared with age- and weight-matched controls. These abnormalities of intracellular GLUT4 trafficking may contribute to reduced whole body insulin sensitivity.

Evaluation of Low-Dose Versus High-Dose Opioid Pathway in Opioid-Naïve Patients After Total Knee Arthroplasty.

BACKGROUND: Pain control after total knee arthroplasty (TKA) remains challenging. Tramadol is a weak opioid with potentially lower side effects and risk for dependency than stronger opioids. The purpose of this study was to evaluate efficacy and safety of tramadol after TKA in opioid-naïve patients compared with stronger opioids. METHODS: A retrospective review of patients who underwent primary TKA was performed. In September 2018, opioid-naïve patients were prescribed tramadol instead of oxycodone. Patients receiving tramadol (low-opioid group) were matched to patients discharged with oxycodone before this transition (high-opioid group). We compared morphine milligram equivalent (MME) consumption and outcomes up to 3 months postoperatively. RESULTS: Two-hundred and five patients underwent TKA, with 126 receiving tramadol. Fourteen patients were converted to stronger opioid (11.2% conversion rate). Seventy patients from the low-opioid group were matched to 70 patients in the high-opioid group. Average daily inpatient MME consumption was higher in the high-opioid group (40.0 ± 27.4 vs 16.3 ± 10.9, P = .000). Outpatient prescribed MME was significantly higher in the high-opioid group (135.5 ± 71.5 vs 75.3 ± 51.3, P = .000) along with a higher number of refills (0.53 ± 1.1 vs 0.886 ± 0.94, P = .041). Knee range of motion was not statistically different at any timepoint postoperatively. There was higher adverse event rate in the low-opioid group (8.6% vs 5.7%) but not statically significant. CONCLUSIONS: Low opioid regimen following TKA showed lower MME consumption than high opioid regimen with no effect on outcomes up to 3 months. Use of low opioid regimen should be considered for TKA surgery.

Virtual Wards: A Rapid Adaptation to Clinical Attachments in MBChB During the COVID-19 Pandemic.

When the COVID-19 pandemic suddenly prevented medical students from attending their clinical attachments, the faculty involved in the third year of medical school (MBChB3) at the University of Glasgow created Virtual Wards. The focus of the Virtual Wards was to continue teaching of clinical reasoning remotely whilst COVID-19 restrictions were in place. Virtual Wards were mapped to the common and important presentations and conditions and provided opportunity for history-taking, clinical examination skills, requesting investigations, interpreting results, diagnosis and management. The Virtual Wards were successful, and further wards were developed the following academic year for MBChB4 students. This chapter describes the theoretical underpinnings of the Virtual Wards and the technological considerations, followed by a description of the Wards themselves. We then analyse an evaluation of the Virtual Wards and provide both a faculty and student perspective. Throughout the chapter, we provide tips for educators developing Virtual Ward environments.

Knockout of syntaxin-4 in 3T3-L1 adipocytes reveals new insight into GLUT4 trafficking and adiponectin secretion.

Adipocytes are key to metabolic regulation, exhibiting insulin-stimulated glucose transport that is underpinned by the insulin-stimulated delivery of glucose transporter type 4 (SLC2A4, also known and hereafter referred to as GLUT4)-containing vesicles to the plasma membrane where they dock and fuse, and increase cell surface GLUT4 levels. Adipocytokines, such as adiponectin, are secreted via a similar mechanism. We used genome editing to knock out syntaxin-4, a protein reported to mediate fusion between GLUT4-containing vesicles and the plasma membrane in 3T3-L1 adipocytes. Syntaxin-4 knockout reduced insulin-stimulated glucose transport and adiponectin secretion by ∼50% and reduced GLUT4 levels. Ectopic expression of haemagglutinin (HA)-tagged GLUT4 conjugated to GFP showed that syntaxin-4-knockout cells retain significant GLUT4 translocation capacity, demonstrating that syntaxin-4 is dispensable for insulin-stimulated GLUT4 translocation. Analysis of recycling kinetics revealed only a modest reduction in the exocytic rate of GLUT4 in knockout cells, and little effect on endocytosis. These analyses demonstrate that syntaxin-4 is not always rate limiting for GLUT4 delivery to the cell surface. In sum, we show that syntaxin-4 knockout results in reduced insulin-stimulated glucose transport, depletion of cellular GLUT4 levels and inhibition of adiponectin secretion but has only modest effects on the translocation capacity of the cells. This article has an associated First Person interview with Hannah L. Black and Rachel Livingstone, joint first authors of the paper.

Sharing the Bandwidth in Cognitively Overloaded Teams and Systems: Mechanistic Insights from a Walk on the Wild Side of Clinical Reasoning.

IssuePrevious work from the diagnostic error literature has provided indirect evidence that faulty clinical reasoning may be the most frequent cause of error when attaching a diagnostic label. The precise mechanisms underlying diagnostic error are unclear and continue to be subject to considerable theory informed debate in the clinical reasoning literature. Evidence: We take a theoretical approach to merging these two worlds of literature by first zooming out using distributed cognition as a social cognitive lens (macro theory) to develop a view of the process and outcome of clinical reasoning occurring in the wild - defined as the integrated clinical workplace - the natural habitat of clinicians working within teams. We then zoom in using the novel combination of cognitive load theory and distributed cognition to provide additional theoretical insights into the potential mechanisms of error. Implications: Through the lenses of distributed cognition and cognitive load theory, we can begin to prospectively investigate how cognitive overload is represented and shared within interprofessional teams over time and space and how this influences clinical reasoning performance and leads to error. We believe that this work will help teams manage cognitive load and prevent error.

Recent developments in adjunct therapies for type 1 diabetes.

INTRODUCTION: There have been many recent advances in the treatment of type 1 diabetes (T1D) including in insulin formulations, continuous glucose monitoring (CGM) technology and automated insulin delivery. However, long-term optimal glycemic control is still only achieved in a minority. AREAS COVERED: Adjunct therapy - the use of therapeutic agents other than insulin - is one strategy aimed at improving outcomes. An ideal adjunct agent would improve glycemic control, reduce weight (or weight gain), reduce insulin requirement and prevent complications (e.g. cardiorenal) without increasing hypoglycemia. The amylin analogue pramlintide has been licensed in the USA, while the sodium glucose co-transporter-2 inhibitor (SGLT2i) dapagliflozin, was briefly (2019 - 2021) licensed for type 1 diabetes in Europe and the UK. However, other agents from the type 2 diabetes (T2D) arena including metformin, other SGLT2is, glucagon-like peptide-1 receptor agonists (GLP-1RA) and dipeptidyl peptidase-IV (DPP-4) inhibitors have been investigated. EXPERT OPINION: As evidence emerges for cardiorenal protection by SGLT2is and GLP-1RAs in T2D, it has become increasingly important to know whether people with T1D can also benefit. Here, we review recent trials of adjunct agents in T1D and discuss the efficacy and safety of these agents (alone and in combination) in an era in which continuous glucose monitoring is becoming standard of care.

Improving the quality of weekend medical handover on non-receiving medical hospital wards.

INTRODUCTION: Handover is the system by which the responsibility for immediate and ongoing care is transferred between healthcare professionals and can be an area of risk. The Royal College of Physicians (RCP) has recommended improvement and standardisation of handover. Locally, national training surveys have reported poor feedback regarding handover at Glasgow Royal Infirmary. AIM: To improve and standardise handover from weekday to weekend teams. METHODS: The Plan-Do-Study-Act (PDSA) quality improvement framework was used. Interventions were derived from a driver diagram after consultation with relevant stakeholders. Four PDSA cycles were completed over a 4-month period:PDSA cycle 1-Introduction of standardised paper form on three wards.PDSA cycle 2-Introduction of electronic handover system on three wards.PDSA cycle 3-Expansion of electronic handover to seven wards.PDSA cycle 4-Expansion of electronic handover to all non-receiving medical wards.The outcome of interest was the percentage of patients with full information handed over based on a six-point scale derived from the RCP. Data were collected weekly throughout the study period. RESULTS: 18 data collection exercises were performed including 525 patients. During the initial phase there was an improvement in handover quality with 0/28 (0%) at baseline having all six points completed compared with 13/48 (27%) with standardised paper form and 21/42 (50%) with the electronic system (p<0.001). When the electronic handover form was expanded to all wards, the increased quality was maintained, however, to a lesser extent compared with the initial wards. CONCLUSION: A standardised electronic handover system was successfully introduced to downstream medical wards over a short time period. This led to an in improvement in the quality of handover in the initial wards involved. When expanded to a greater number of wards there was still an improvement in quality but to a lesser degree.

Ethnic differences in prevalence of actionable HbA1c levels in UK Biobank: implications for screening.

INTRODUCTION: Early detection and treatment of diabetes as well as its prevention help lessen longer-term complications. We determined the prevalence of pre-diabetes and undiagnosed diabetes in the UK Biobank and standardized the results to the UK general population. RESEARCH DESIGN AND METHODS: This cross-sectional study analyzed baseline UK Biobank data on plasma glycated hemoglobin (HbA1c) to compare the prevalence of pre-diabetes and undiagnosed diabetes mellitus in white, South Asian, black, and Chinese participants. The overall and ethnic-specific results were standardized to the UK general population aged 40-70 years of age. RESULTS: Within the UK Biobank, the overall crude prevalence was 3.6% for pre-diabetes, 0.8% for undiagnosed diabetes, and 4.4% for either. Following standardization to the UK general population, the results were similar at 3.8%, 0.8%, and 4.7%, respectively. Crude prevalence was much higher in South Asian (11.0% pre-diabetes; 3.6% undiagnosed diabetes; 14.6% either) or black (13.8% pre-diabetes; 3.0% undiagnosed diabetes; 16.8% either) participants. Only six middle-aged or old-aged South Asian individuals or seven black would need to be tested to identify an HbA1c result that merits action. CONCLUSIONS: Single-stage population screening for pre-diabetes or undiagnosed diabetes in middle-old or old-aged South Asian and black individuals using HbA1c could be efficient and should be considered.

Time in Range as a Research Outcome Measure.

Time in range (TIR) is gaining ground as an outcome measure in type 1 diabetes trials. However, inclusion of TIR raises several issues for trial design. In this article, the authors begin by defining TIR and describing the current international consensus around TIR targets. They then expand on evidence for the validity of TIR as a primary clinical trial outcome before concluding with some practical, ethical, and logistical implications.

Metformin and carotid intima-media thickness in never-smokers with type 1 diabetes: The REMOVAL trial.

AIM: To determine whether metformin's effects on carotid artery intima-media thickness (cIMT) in type 1 diabetes differ according to smoking status. METHODS: Regression model effect estimates for the effect of metformin versus placebo (double-blind) on carotid IMT were calculated as a subgroup analysis of the REMOVAL trial. RESULTS: In 428 randomized participants (227 never-smokers, 201 ever-smokers), averaged mean carotid IMT progression (per year) was reduced by metformin versus placebo in never-smokers (-0.012 mm, 95% CI -0.021 to -0.002; p = .0137) but not in ever-smokers (0.003 mm, 95% CI -0.008 to 0.014; p = .5767); and similarly in non-current smokers (-0.008 mm, 95% CI -0.015 to -0.00001; p = .0497) but not in current smokers (0.013 mm, 95% CI -0.007 to 0.032; p = .1887). Three-way interaction terms (treatment*time*smoking status) were significant for never versus ever smoking (p = .0373, prespecified) and non-current versus current smoking (p = .0496, exploratory). Averaged maximal carotid IMT progression (per year) was reduced by metformin versus placebo in never-smokers (-0.020 mm, 95% CI -0.034 to -0.006; p = .0067) but not in ever-smokers (-0.006 mm, 95% CI -0.020 to 0.008; p = .4067), although this analysis was not supported by a significant three-way interaction term. CONCLUSIONS: This subgroup analysis of the REMOVAL trial provides additional support for a potentially wider role of adjunct metformin therapy in cardiovascular risk management in type 1 diabetes, particularly for individuals who have never smoked cigarettes.