Academic posters

Here you will find detailed information about the academic posters presented at Usher Institute Annual Lecture & Showcase 2025.

Many of the posters are available to view/download as PDFs should you wish to do so.

  • Marlene Stewart1, Cathryn Broderick1, Wei Xu1, Candida Fenton1

1 NESSIE, Centre for Global Health, Usher Institute, The University of Edinburgh

NIHR Evidence Synthesis Scotland InitiativE (NESSIE) is one of nine evidence synthesis groups funded by the National Institute for Health and Care Research (NIHR) evidence synthesis programme. The aim of the evidence synthesis groups is to address knowledge gaps or to answer a specific need for healthcare, public health and social care stakeholders/audiences.

We undertake evidence synthesis projects in response to requests from the NIHR. The evidence synthesis projects will have a direct impact on decision-making, patient and client care, reducing inequalities and identifying future research needs.

We promote and advertise our projects using a wide range of dissemination activities including research reports, scientific posters, and meeting presentations and other methods such as social media posts, infographics, visual abstracts, blogs, podcasts, videos etc. Activities are tailored to the intended audience.

For members of the audience who have a topic for an evidence synthesis that addresses knowledge gaps or answers a specific need for healthcare, public health and social care stakeholders, we can advise on how to submit the topic to the NIHR.


  • Kelly Fleetwood1, Raied Alotaibi1, Stine Scheuer2,3, Daniel J Smith4, Sarah H Wild1, Caroline A Jackson1

1 Usher Institute, The University of Edinburgh 

2 Steno Diabetes Centre Copenhagen

3 Novo Nordisk

4 Centre for Clinical Brain Sciences, The University of Edinburgh

People with severe mental illness (SMI) have a higher risk of premature mortality than the general population. Our aim was to investigate whether the mortality gap for people with SMI is widening, by determining time-trends in excess life-years lost.

We performed a population-based study, including people with SMI (schizophrenia, bipolar disorder and major depression) alive on 1 January 2000. We ascertained SMI from psychiatric hospital admission records (1981-2019), and deaths via linkage to the national death register (2000-2019). We used the life-years lost (LYL) method to estimate LYL by SMI and sex, compared LYL to the Scottish population and assessed trends over 18 3-year rolling periods.

We included 28,797 people with schizophrenia, 16,657 with bipolar disorder and 72,504 with major depression.  Between 2000 and 2019, life expectancy increased in the Scottish population but the gap widened for people with schizophrenia. For 2000-2002, men and women with schizophrenia lost an excess 9.4 (95% CI 8.5-10.3) and 8.2 (7.4-9.0) life years, respectively, compared to the general population. In 2017-2019, this increased to 11.8 (10.9-12.7) and 11.1 (10.0-12.1). The life expectancy gap was lower for bipolar disorder and depression and unchanged over time.

The life expectancy gap in people with SMI persisted or widened from 2000 to 2019. Addressing this entrenched disparity requires equitable social, economic and health policies; healthcare re-structure and improved resourcing, and investment in interventions for primary and secondary prevention of SMI and associated comorbidities.


  • Aryelly Rodriguez1, Steff Lewis1, Tracy Jackson1, Christopher Weir1, Sandra Eldridge2

1 Usher Institute, The University of Edinburgh 

2 Queen Mary University of London

This research explores whether it is possible to identify individuals from data that has been "anonymised." First, recommendations for anonymising clinical trial data were reviewed. Common suggestions include removing personal details such as names and dates of birth, as well as carefully handling information such as sex, race, height, and weight. Sharing anonymised data under certain restrictions and conditions is also frequently recommended. Next, anonymised datasets available in the public domain were analysed using three equations to estimate the risk of re-identification. The analysis showed that it is possible to calculate this risk, that some datasets carry a higher risk than others, and that such calculations can provide useful insights for researchers.


  • Sasha Lewis-Jackson1

1 Advanced Care Research Centre, The University of Edinburgh

Restrictive practices, which include seclusion, restraint, and rapid tranquillisation, are used in mental health settings to address behaviour exhibited by patients due to stress and distress. These practices are intended as a last resort due to their negative impacts, including worsened mental and physical health and extended hospital stays. Recognising these adverse effects, the Scottish government and various organisations aim to reduce their usage. However, there is limited understanding of their application and impact on those living and working in old-age psychiatric settings in Scotland.

Our research aims to understand the lived experiences of patients and staff involved in restrictive practices, with specific objectives to (1) explore how restrictive practices affect individuals with dementia or delirium, (2) identify the contextual and environmental factors that contribute to their use, and (3) examine the effects on the care environment. Using a focused ethnography approach that combines semi-structured interviews and participant observations over six months on two psychiatric wards in Scotland, we observe types of restriction, triggers of distress in patients, and resulting impacts.

The paper will discuss preliminary findings from ongoing fieldwork, integrating data from a qualitative literature review that includes empirical, legal, and medical documents. By triangulating these findings, we aim to create a decision-making framework. This framework will guide clinicians on when restrictive interventions are necessary and when alternative behavioral techniques can be used. We suggest that these findings be used to implement evidence-based changes that can improve care and reduce reliance on restrictive practices.


  • Jorge Crespo1

1 Centre for Biomedicine, Self and Society, Usher Institute, The University of Edinburgh

Global efforts to close the ‘treatment gap’ have driven the expansion of community-based mental healthcare services. In Chile, this has taken shape through three major national mental health programmes since the 1990s, with psychologists playing a key role in primary care institutions. Yet we still know little about how they navigate these institutional settings.

Drawing on medical sociology – particularly the sociology of mental health and illness, and the sociology of professions – I examined how psychologists negotiate their professional role in Chile’s public primary care system, and what this reveals about wider processes of medicalisation. Adopting an interpretivist approach, this research combined policy analysis with qualitative interviews. I conducted semi-structured interviews with 16 primary care psychologists, 10 policy officials, and three psychologists based in specialised community mental health centres.

Thematic analysis showed that medicalisation is not a uniform or fixed process but one that is fluid and shaped by contrasting policy discourses, institutional arrangements, contexts and practices. Psychologists’ roles are ambiguously constructed – oscillating between disorder-focused clinicians, mental health promoters, and productivity-driven practitioners. Rather than occupying a stable position, they move fluidly, negotiating professional and institutional expectations while responding to both material and symbolic constraints. They described a range of formal and informal strategies to adapt their work to different demands and professional expectations, often reinforcing a narrow, clinical approach to mental health despite official commitments to more comprehensive care. In this context, the medicalisation, and sometimes demedicalisation, of distress and everyday life appears to be advancing through a dominant biomedical–psychiatric framework, underpinned by the instrumentalisation of care within a biomedical and managerial hierarchy.

These findings contribute to medical sociology by showing how professional roles are shaped within specific policy and institutional frameworks, how they might respond to them, and how such dynamics can drive medicalisation through unexpected, diverse, context-dependent, and sometimes contradictory pathways.


  • Helmi Hietanen1, Mark Woolhouse2

1 School of Biological Sciences, The University of Edinburgh

2 Usher Institute, The University of Edinburgh

RNA viruses cause infectious diseases which are responsible for a significant global health burden, and have potential to cause epidemics and pandemics. Currently, 239 human-infective RNA virus species are known. Based on extensive literature searches carried out over more than two decades, we have catalogued these 239 species in a database alongside extensive metadata on virus characteristics, all updated from an earlier 2018 version. Here, we present the metadata on species distribution across families, accumulated counts of species, genera and families since 1901, geographical locations of first reported human infections, levels of transmissibility, and host categories. These data can be used in comparative studies to identify virus characteristics associated with public health burden and potential pandemic risk.


  • Martine Miller1, Angela Niven1, Fiona Dobbie1

1 Usher Institute, The University of Edinburgh

This display will showcase findings from recent research on preventing gambling and nicotine-related harm among children and young people. It combines evidence from two separate projects in which we conducted extensive consultation with school staff, students, parents/carers, and stakeholders from policy and practice. Our aim is to deepen our understanding of the risk factors and develop school-based interventions focusing on peer support and social networks.


  • Bruce Mason1, Tricia Tooman1, Stewart Mercer1, Maria Wolters2, Sue Lewis1, Katie Hawkins3, Helen Frost1

1 Advanced Care Research Centre, Usher Institute, The University of Edinburgh 

2 School of Informatics, The University of Edinburgh/OFFIS Institute for Information Technology, Oldenburg, Germany

3 NHS Lothian, Scotland, UK

Green Social Prescribing (GSP) is a policy response aimed at mitigating the effects of health inequalities in the UK; however, its provision is inconsistent. Targeting patients from deprived areas for GSP referrals could help reduce inequalities. This project aimed to develop and test a digital toolkit to support referrals to GSP in primary care.

We used a mixed-methods approach following MRC guidelines for complex interventions. Toolkit development was guided by a systematic review, a co-design workshop, and stakeholder interviews to identify contextual factors. Thematic analysis, using a socio-ecological framework, highlighted issues at personal, institutional, and policy levels.

We screened 4,303 papers and included 34. Twenty stakeholders participated in a workshop, and twenty were interviewed. Thematic analysis highlighted the need for multiple GSP pathways, improved education, effective monitoring, and secure funding. Main barriers included low awareness of GSP, limited time to discuss it with patients, and inconsistent funding. Key facilitators were link workers, a ‘green ethos’ in practices, local GSP champions, and access to reliable information for practitioners and patients.

Findings indicate enthusiasm for GSP as a holistic approach, but concerns exist regarding its implementation in primary care. To help address these issues, we developed deepgreen.scot. It provides evidence of GSP’s effectiveness, bespoke lists of local providers tailored for each practice, clear guidelines for prescribing and documenting GSP, comprehensive patient information, and options for self-directed activities in both online and printed formats. We are currently testing the feasibility of the toolkit in primary care.


  • Denise Cranley1, Sarah Dunn2, John-Paul Taylor2, Michael Desborough3, Jennifer Craig4, Nikola Sprigg5, Rachel McComish6, Thomas Foltynie7, Joanna M Wardlaw8,9, Katherine Oatey1, Anna Heye1, Philip M Bath10, Karen Innes1, Lynn Dinsmore1, Jessica Griffiths11, Lisa Fox11, Paula R Williamson12, Rustam Al-Shahi Salman (corresponding author)1,8

1 Edinburgh Clinical Trials Unit

2 Newcastle Clinical Trials Unit 

3 Department of Clinical Haematology, Oxford 

4 Stroke Trials Unit, Nottingham

5 Stroke, Medicine Division, Nottingham

6 UCL Comprehensive Clinical Trials Unit, London

7 UCL Institute of Neurology, London 

8 Centre for Clinical Brain Sciences

9 UK Dementia Institute Centre, Edinburgh

10 Stroke Trials Unit, Nottingham

11 The Institute of Cancer Research Clinical Trials & Statistics Unit (ICR-CTSU), London

12 MRC-NIHR Trials Methodology Research Partnership, Liverpool

This study quantified the carbon footprint of six neurology clinical trials, using the guidance and data collection tools developed by the Low Carbon Clinical Trials (CiCT) group.

Climate change is a significant global health threat and reducing carbon consumption needs to be prioritised, including within clinical research.

We searched two clinical trial registries for phase 2-4 randomised controlled trials who enrolled people with any of the 15 neurological disorders with the highest global burden of disease (GBD). We invited eligible trials to share data to allow us to quantify estimated carbon emissions of each trial.

The primary outcome measure was kilograms (kg) of Carbon Dioxide Equivalent (CO2e). 

Results showed the highest contributors to the carbon footprint of these six trials was the Clinical Trial Unit staff emissions, with commuting accounting for >50% of estimated emissions. The second and third highest contribution for these six trials were trial-specific participant assessments, and trial supplies and equipment. Other notable contributing factors for these trials include archiving, imaging and IMP distribution. 

These findings are in line with existing research in this field and highlight the modifiable areas where greener practices can be considered in clinical trials. Areas where we can potentially reduce the carbon footprint of clinical trials include awareness and development of sustainable travel practices, increased use of technology and a change in practice from funders and governing bodies to encourage greener and more sustainable research practices.


  • Lili Yu1, Mr Kening Gao2, Danijela Gasevic3, Anna L Hansell4, Xue Li2, Mark R Miller5, Evropi Theodoratou1, 6

1 Usher Institute, The University of Edinburgh

2 Department of Big Data in Health Science, Zhejiang University

3 School of Public Health and Preventive Medicine, Monash University

4 Centre for Environmental Health and Sustainability, University of Leicester

5 Queen's Medical Research Institute, The University of Edinburgh

6 MRC Institute of Genetics and Cancer, The University of Edinburgh

Ambient air pollution is closely linked to respiratory and cardiovascular disease; however, uncertainty remains regarding the causal pathway in a wide range of health-related conditions. Observational phenome-wide association study (PheWAS) was initially conducted to explore disease outcomes associated with air pollution in 460,028 individuals from the UK Biobank. Then, the causality of air pollution-related DNA methylation (DNAm) and identified clinical outcomes was examined by applying weighted allele score-based Mendelian randomization (MR) (discovery) in UK Biobank (N = 385,917), and summary level-based MR (replication) and subsequent Bayesian colocalization analyses in FinnGen study (N = 500,348). PheWAS revealed that each interquartile range (IQR) increase in exposure to particulate matter with aerodynamic diameters of ≤ 2.5 μm (PM2.5) and ≤ 10 μm (PM10), nitrogen oxides (NOx) and nitrogen dioxide (NO2) was associated with 55, 13, 42, and 64 phenotypes, respectively. These included circulatory, metabolic, genitourinary, neurological, respiratory, musculoskeletal, hematologic and neoplastic conditions. Allele score-based and summary level-based MR provided causal relationships between genetically determined DNAm levels of 16 NO2-related cytosine-phosphate-guanine (CpG) sites and metabolic, genitourinary and neoplasm conditions. Colocalization analysis supported strong evidence of DNAm alteration at cg06240200 [CCDC12], cg09367967 [CALR], and cg18603577 [RILPL2] with type 2 diabetes, cg20959189 [ZBTB38] with hyperplasia of prostate, and cg01543603 [ANKRD11] and cg25670076 [BACH2] with non-epithelial skin cancers. Our study suggest that ambient air pollution is associated with the risk of multiple health conditions via DNAm of specific genetic loci, providing insights into epigenetic targets for disease prevention and health benefits of environmental policy interventions.


  • COALESCE consortium

The display will present the findings of the first ever study using routinely collected electronic health care records for the entire UK population. The study reports on COVID-19 vaccination coverage and the risks associated with being undervaccinated—defined as receiving fewer than the recommended number of vaccine doses. Drawing on anonymised, harmonised health data from across England, Northern Ireland, Scotland, and Wales, the study analysed the records of nearly 65 million people.

Visitors will see an accessible visual summary of key findings: who was most likely to be undervaccinated (including patterns by age, ethnicity, deprivation, and health status), and how this affected the risk of hospitalisation or death from COVID-19 during the summer of 2022. The display will highlight that undervaccination was common—affecting up to half the population in some nations—and that even missing one or two doses significantly increased the risk of severe outcomes, especially in people aged 75 and older.

The display will also present estimates from a counterfactual scenario, showing how many severe outcomes might have been avoided across the UK if everyone had been fully vaccinated by June 2022. This includes nearly 5,500 severe cases prevented in the 75+ age group alone.


  • T. Ashcroft1, E.McSwiggan2, E. Agyei-Manu1, M. Nundy1, N. Atkins1, J.R. Kirkwood1,3, M. Ben Salem Machiri1, V. Vardhan1, B. Lee1, E. Kubat1, S. Ravishankar1, P. Krishan1, U. De Silva1, E.O. Iyahen1, J. Rostron1, A. Zawiejska1, K. Ogarrio1,4, M. Harikar1, S. Chishty1, D. Mureyi1, B. Evans5, D. Duval5, S. Carville6, S. Brini6, J. Hill6, M. Qureshi6, Z. Simmons5, I. Lyell7, T. Kavoi6, M. Dozier8, G. Curry2, J.M. Ordóñez-Mena9, S. de Lusignan9,10, A. Sheikh3,9, E. Theodoratou1 and R. McQuillan1

1 Usher Institute, Centre for Global Health, The University of Edinburgh

2 Usher Institute, Centre for Population Health Sciences, The University of Edinburgh

3 Usher Institute, Centre for Medical Informatics, The University of Edinburgh

4 School of Public Health and Tropical Medicine, Department of Social, Behavioral, and Population Sciences, Tulane University, New Orleans

5 Science Evidence Review Team, Research, Evidence and Knowledge Division, UKHSA

6 Clinical and Public Health Response Evidence Review Team, Clinical and Public Health, UKHSA

7 Health Protection Operation, UKHSA

8 Information Services, The University of Edinburgh

9 Nuffield Department of Primary Care Health Sciences, University of Oxford

10 Royal College of General Practitioners (RCGP), Research and Surveillance Centre

Although non-pharmaceutical inventions (NPIs) were used globally to control the spread of COVID-19, their effectiveness remains uncertain. We aimed to assess the evidence on NPIs as implemented in the UK, to allow public health bodies to prepare for future pandemics.

We used rapid systematic methods (search date: January 2024) to identify, critically appraise and synthesise interventional, observational, and modelling studies reporting on NPI effectiveness in the UK.  

Eighty-five modelling, nine observational, and three interventional studies were included. Modelling studies had multiple quality issues; six of the 12 non-modelling studies were high quality. The best available evidence was for test and release strategies for case contacts (moderate certainty), which was suggestive of a protective effect. Although evidence for school-related NPIs and universal lockdown was also suggestive of a protective effect, this evidence was considered low certainty.  Evidence certainty for the remaining NPIs was very low or inconclusive.

The validity and reliability of evidence on the effectiveness of NPIs as implemented in the UK during the COVID-19 pandemic is weak. To improve evidence generation and support decision-making during future pandemic or other public health emergencies, it is essential to build evaluation into the design of public health interventions.


  • Pinnock H1, Habib M2, Agarwal D3, Engkasan J4, Paul B, Hammersley V1, Rabinovich R6, and the teams from BPCRS, CMC, KEM, UM and UoE.

1 Usher Institute, The University of Edinburgh, UK

2 Community Respiratory Clinic, Bangladesh Primary Care Respiratory Society, Khulna, Bangladesh

3 Vadu Rural Health Program, King Edward Memorial Hospital Research Centre, Pune, India

4 Department of Rehabilitation Medicine, Universitii Malaya, Kuala Lumpur, Malaysia

5 Rural Unit for Health and Social Affairs, Christian Medical College, Vellore, India

6 NHS Borders

Co-designing research with local stakeholders improves relevance of findings and is especially pertinent to developing and evaluating complex interventions in low- and middle-income countries when technical expertise may lie with colleagues in high-income countries, but the experience and local knowledge is embedded in a low-resource setting.  Our experience in designing the PuRe trial during the RESPIRE programme illustrates some key steps and considerations.

PuRe, a ‘Hybrid-1’ implementation trial testing low-resource pulmonary rehabilitation for people with chronic respiratory disease, will be conducted in four diverse settings in Bangladesh, India and Malaysia.

  • Community engagement work conducted in the four centres, ensured the research question reflected the context and healthcare needs of the population, optimising the relevance of research to end-users.
  • The research gap was identified by systematic reviews undertaken by the team.
  • Feasibility studies in the four centres explored a range of approaches to delivering pulmonary rehabilitation and confirmed practicability, acceptability and potential utility in the proposed settings.
  • The core components of pulmonary rehabilitation, systematically adapted from global guidelines by local colleagues, are being defined for the trial in a detailed manual.
  • On-going discussions allow/encourage adaptations that will enable delivery in the diverse settings.
  • A robust process evaluation will be needed to monitor fidelity and record local adaptations 

All four centres have started recruitment and to date we have 89 participants randomised to an internal pilot. The embedded process evaluation has also started with interviews from participants receiving pulmonary rehabilitation.


  • Catherine MacLeod1, Brigitte Delaney2, Viv Marsh1, Vicky Hammersley1, Kirstie McLatchey3, Liz Steed4, Jess Sheringham5, Steph Taylor4, Hilary Pinnock1

1 The University of Edinburgh

2 University of Sheffield

3 University of Dundee

4 Queen Mary University of London

5 University College London

Supported self-management for people with asthma reduces the burden on patients, health services and wider society.

Implementation, however, remains poor in routine clinical practice. IMP2ART is a multidisciplinary, theoretically-informed strategy which includes patient resources, professional education, and organisational strategies to improve the implementation of supported self-management in UK primary care. The impact of the IMP2ART strategy is being tested in a UK-wide cluster randomised implementation trial (ISRCTN15448074).

To undertake a mixed-methods process evaluation within the IMP2ART trial to understand how supported self-management was implemented (or not) by primary care practices.

Data and analysis strategies were informed by mid-range and programme level theory. Data collected from all implementation practices include practice context and IMP2ART delivery/response. Longitudinal case studies (n=5) and interviews with practice staff from implementation and control sites explore the interaction of practice context, delivery and response. Synthesis, informed by theory, combines both qualitative and quantitative data. 

Preliminary findings highlight the importance of facilitation as a catalyst for successful implementation, providing a focus for practices to consider their asthma care, and the influence of practice culture and external influences on implementation. Regular audit promoted review of organisational processes and monitored progress. 

A process evaluation provides insights into the delivery and response to a whole systems implementation of supported asthma self-management in primary care.


  • Durga Kulkarni1, Richard Osei-Yeboah1, Kate Templeton2, Harish Nair1

1 Centre for Global Health, Usher Institute, University of Edinburgh, UK

2 Viral Genotyping Reference Laboratory Edinburgh, NHS Lothian, Royal Infirmary of Edinburgh, UK

We estimated the annual laboratory and extrapolated hospital incidence of hMPV RTIs in older adults in Scotland over six seasons (2017-2023) using national hospital and laboratory data. Hospital incidence in Scottish health boards other than Lothian, where testing practices were uncertain, was extrapolated using Lothian’s comprehensive laboratory data. Additionally, we developed similar estimates for RSV and Influenza A for comparisons.

This analysis included 626 laboratory-confirmed hospital episodes of hMPV. Only 26% of laboratory-confirmed hMPV episodes were clinically coded. The extrapolated hMPV hospital incidence ranged from 3.57/100,000 to 49.53/100,00 in adults aged ≥60y in Scotland and the extrapolated incidence was 1.27 to 3.78 times higher than the laboratory incidence. The hospital incidence was higher in the 60-74y than in ≥75y. hMPV incidence dropped substantially during the COVID-19 pandemic.  The hospital incidence of RSV and Influenza A were higher than those of hMPV for both laboratory-confirmed and extrapolated estimates. 

hMPV RTIs pose a considerable hospital burden in older adults in Scotland. Adults aged ≥75y reported higher hMPV incidence than those aged 60-64y, indicating increased associated morbidity in the very old. There is a considerable year-on-year variation in hMPV hospital incidence and the COVID-19 pandemic was associated with a substantial drop in hMPV incidence in older adults in Scotland. The differences between clinically coded, laboratory-confirmed and extrapolated estimates highlight the need for improved surveillance, diagnosis and coding practices to develop robust burden estimates.


  • Ruonan Pei1, Dr Catalina A Vallejos1,2, Prof Sir Aziz Sheikh1,3, Dr Ting Shi1

1 Centre for Medical Informatics, Usher Institute, The University of Edinburgh

2 MRC Human Genetics Unit, Institute of Genetics and Cancer, The University of Edinburgh

3 Nuffield Department of Primary Care Health Sciences, University of Oxford

Acute respiratory infections place significant strain on healthcare systems, particularly affecting high-risk populations such as the elderly and individuals with underlying comorbidities.

Our aim is to develop and validate a robust risk prediction model for winter respiratory disease (WRD)-related hospitalisations, Accident and Emergency (A&E) attendances, and mortality, supporting evidence-based winter preparedness strategies across the National Health System (NHS).

Adults (≥18 years) from 16 UK-based longitudinal cohort studies will be investigated, linked to the NHS digital health records via the UK Longitudinal Linkage Collaboration research environment (01/03/2017-28/02/2025). Outcomes included WRD-related hospitalisation, A&E usage, and 28-day post-hospitalisation mortality. Exposures comprised acute respiratory diseases and respiratory viral-triggered chronic disease exacerbation. Covariates included location, socio-demographics, clinical characteristics, comorbidities and meteorological exposures (e.g., air pollution).

Statistical and machine learning algorithms, such as logistic regression and XGBoost-derived survival models, will be applied. Nested cross-validation will be utilised for hyperparameter tuning and model selection; class imbalance will be addressed via down-sampling. Model performance will be evaluated using the concordance statistic and calibration-in-the-large. An ensemble of top-performing models will be compared to single models, with the final tool chosen based on superior performance.

We anticipate to identify key risk factors and quantify variations across population subgroups to generate individualised risk estimates for WRD-related outcomes. It will support targeted prevention and early intervention for high-risk groups, reducing the winter healthcare burden and improve outcomes across the UK.


  • Eleonore Sophia Heri1, Felicity Vidya Mehendale2

1 Glasgow Royal Infirmary

2 Centre for Global Health Research, Usher Institute, The University of Edinburgh

Missed diagnoses of cleft palate (CP) can turn the first weeks of life into a struggle for survival - feeding is difficult, growth falters and aspiration pneumonia threatens. Yet many cases are still missed at birth globally. The cause is often not clinical oversight, but a lack of clear newborn screening (NBS) guidance. Without practical instructions, healthcare professionals - especially in resource-limited settings - are left to rely on experience alone, and opportunities for early diagnosis are lost.

To assess the extent, clarity and consistency of CP screening recommendations in national and international NBS guidelines, and to identify gaps that may contribute to missed diagnoses.   We searched PubMed, Google Scholar, national/international organisations, grey literature for NBS and CP-specific guidelines in English. Data extracted included country, timing of newborn examination, designated examiner, mention of CP, clarity of CP vs cleft lip (CL) distinction, examination method, and reference to uvula visualisation.  Fifteen NBS guidelines from eleven countries were found. Although most mentioned CP, over half made no distinction between CP (that can be diagnosed only by intra-oral examination) and CL (that is easily visible externally on the face). Guidelines that did mention intra-oral examination, failed to describe what structures to visualise/examine, except one guideline which mentioned uvula inspection - a key step for detecting submucosal clefts.  Inconsistent terminology and vague instructions risk delayed or missed diagnoses, with lifelong consequences. Standardised protocols - defining examination methods, tools, and CP–CL distinction - are essential to enable early detection and prevent avoidable neonatal morbidity.


  • Anna Chiara Corriero1,2, Felicity Vidya Mehendale2  

1 NHS Lothian, Edinburgh

2 Usher Institute, The University of Edinburgh

Disparities in research, caused by financial, geographical and gender differences, are widespread, and cleft palate studies are no exception. These challenges highlight the need for a diverse academic landscape in cleft palate research. This paper identifies inequalities in cleft research by examining publications across 182 years (1842-2024) and advocates for a more inclusive scholarly environment in the field of cleft.

This analysis examined articles published in Elsevier journals, using a Python script developed by the ICSR Elsevier Lab. The selection process involved a keyword search, followed by an evaluation of factors including the journal of publication, geographical trends and first author gender data.

A total of 39176 articles met the inclusion criteria, yet only 22% were published open access.

In 2023, the top publishing countries were the United States, China, and the United Kingdom. Since 1985, Europe and North America have consistently led in publication output, while East Asia and the Pacific have shown significant growth. Although high-income countries have historically dominated research, their lead is gradually decreasing, allowing for more diverse contributions. However, Latin America, Southeast Asia, and Africa still remain underrepresented.

First author gender analysis showed approximately 75% male first authorship early on, but this narrowed to 38% by 2024, reflecting an improved gender balance.

While inequalities in cleft palate research persist, progress has been made in gender representation and the global distribution of research output. A stronger international effort is needed to ensure continuous improvement in fostering a more inclusive cleft research culture.


  • Anna Chiara Corriero1,2, Felicity Vidya Mehendale2

1 NHS Lothian, Edinburgh

2 Usher Institute, The University of Edinburgh

Orofacial clefts (OFC) are among the most common congenital conditions, affecting over 8 million newborns annually. They have lifelong impacts - nutritional, educational, and psychosocial - requiring multidisciplinary approaches, which include surgical intervention in paediatric ages. In 2010, the World Health Organization passed a resolution to strengthen global responses to congenital conditions, aligning with the 2016 United Nations Sustainable Development Goals (SDGs) to be met by 2030. Our research evaluated how well represented SDGs are in OFC publications.

A Python script developed by Elsevier was used to conduct a comprehensive keyword search and content analysis of OFC-related articles (1842-2024). The script inferred SDG themes from the content. The study focused on SDG mentions, thematic relevance, and World Bank country income classification.

Of 39176 publications, 5598 (14%) referenced at least one SDG, with 6069 unique mentions. 87% were linked to author country income group. Unsurprisingly, SDG3-Good Health was the most cited (4255 references). SDG17-Partnerships being mentioned more frequently by low- and middle-income countries, alongside SDG10-Reduced Inequalities and SDG2-Nutrition. Instead, higher income countries tend to focus more on SDG3-Good Health and SDG4-Education. This shows diverse research OFC-related priorities globally, which are directly linked with OFC patients’ needs.

These findings highlight how partnership, nutrition and education represent different OFC-related challenges and priorities based on country income. This can guide sustainability and global health discussion worldwide, with the objective of achieving the SDGs by 2030 – aiming to address diverse patients’ needs globally.


  • Martina Schatz1, Felicity Vidya Mehendale2

1 Medical student, The University of Edinburgh

2 Centre for Global Health Research, Usher Institute, The University of Edinburgh

The importance of National Congenital Anomaly Registries (CARs) in improving healthcare quality, access and planning is well recognised. However, few countries report the use of a national CAR. Every year, about 295,000 babies die within the first 4 weeks of birth due to congenital anomalies (CA) (WHO).

This study aims to identify whether there are differences in key health and economic indicators, in countries with and without CARs.

Countries were grouped based on CAR status into those that had:

  • a national CAR with or without regional CARs (national_CAR)
  • only a regional CAR (regional_CAR)
  • no CAR data available (no_CAR).  

CAR status was correlated with Birth registration rate(BRR), Infant mortality rate(IMR) from UNICEF; Income Group, GDP and percentage health expenditure, from the World Bank(WB).Of the 196 countries with data available, CAR status was national_CAR = 30, regional_CAR = 19 and no_CAR = 147 national_CARs exist in 35% of HICs, 13% of UMICs and 0% of LMICs/LICs. Countries lacking a national_CAR have lower GDPs, increased IMR and lower BRR. Despite facing the greatest burden of disease, of severe CAs, none of the LMICs/LICs had a CAR.

Our study suggests that countries without CARs face broader and more severe economic and health systems challenges Collaborations to develop and maintain national CARs, need to take into account the impact of these wider economic and health systems challenges.


  • Mahwish Fatima1, Shazia Junejo1, Ammar Ali Muhammad1, Maryam Sherwani1, Felicity Mehendale2, Lubna Samad3

1 Interactive Research and Development (IRD) Pakistan, Pakistan

2 Usher Institute, The University of Edinburgh  

3 Interactive Research and Development (IRD) Global, Singapore

Orofacial clefts (OFCs) affect 1 in 700 newborns globally, but in the absence of active newborn screening (NBS) programmes in low-middle-income countries, prevalence estimates are often derived from surgical data. Presurgical outcomes are often undocumented, including mortality from unrepaired OFCs. We aimed to establish the prevalence of OFCs in a facility-based birth cohort and to determine the neonatal mortality in affected infants.

Health-worker-led NBS for external congenital anomalies was conducted from July 2023 to June 2024 at three facilities in Karachi, Pakistan. At the end of the neonatal period, a follow-up telephone interview was conducted to determine anomalies that manifest late and to record neonatal mortalities. Telephonic verbal autopsies were conducted to determine the cause of death. 

Of 18,728 births screened, OFCs were identified in 45 births, establishing a prevalence of 1 in 417 births. Cleft lip (n=13), cleft palate (n=12), and combined cleft lip/palate (n=20) were identified. One child was stillborn, and another died before the mother’s hospital discharge. Of the 43 remaining babies, nine died during the neonatal period, bringing the overall mortality rate in babies born with OFCs to 24.4%. Pneumonia secondary to OFCs was determined to be the cause of death in five neonates, three succumbed to multiple anomalies, while one was uncontactable.  

We document a higher prevalence of OFCs than previously reported, underlining the importance of NBS. Children with OFCs are at a high risk for malnutrition and infections, leading to mortality. Newborn pre-operative support is essential to ensure better outcomes.


  • Mahwish Fatima1, Shazia Junejo1, Ammar Ali Muhammad1, Maryam Sherwani1, Felicity Mehendale2, Lubna Samad3

1 Interactive Research and Development (IRD) Pakistan, Pakistan

2 Usher Institute, The University of Edinburgh  

3 Interactive Research and Development (IRD) Global, Singapore

Globally, 8 million children are born with congenital anomalies (CAs) annually; 94% are from developing countries. Neonatal screening is imperative for early diagnosis to prevent mortality and disability. Orofacial clefts (OFCs) contribute significantly to this burden. We aimed to establish a newborn screening program for external CAs, including OFCs, by training health workers (HWs) to identify, counsel and refer affected newborns.

HW training: An educational intervention was developed to enable screening of newborns for external CAs, which included subject knowledge, head-to-toe examination, counselling and referral guidelines. Pre and post-tests were conducted to assess HWs' knowledge, followed by an evaluation to assess concordance in anomaly identification between them and the specialists. Implementation: The program was conducted at three high-volume facilities. Newborns were enrolled after obtaining caregiver consent and all identified anomalies were documented. Parents of affected babies were counselled and referred using standardized expert-developed guidelines.

Fifteen HWs underwent classroom training; thirteen were selected based on performance. A statistically significant increase in median knowledge score was observed in the pre and post-test assessments (5.0 vs 18.0). From the 18,728 births screened, 19.9% (n=3,730) were identified with anomalies, 334 (1.8%) major and 3,384 (18.1%) minor anomalies. OFCs (n=45) represented 13.5% of all major anomalies. The concordance of OFC identification between HWs and photographic review by specialists was 98%.

Capacity building of HWs is an effective and feasible strategy for reliable newborn screening for external CAs, including OFCs, linking them to care pathways.


  • Lizzie Wastnedge1, Haris Haseeb2, Sharon Cameron1,3, Chelsea Morroni1

1 Centre for Reproductive Health, The University of Edinburgh

2 Usher Institute, The University of Edinburgh

3 Institute for Regeneration and Repair, The University of Edinburgh

Understanding reproductive goals for People Living with HIV (PLHIV) in Sub-Saharan Africa (SSA) is key to reducing HIV transmission and helping people to realise fertility aspirations. Pregnancy intention with HIV is complex and dynamic, and factors underlying this decision-making are not well understood. This review uses a meta-narrative approach to summarise qualitative evidence exploring these factors.

A literature search was performed identifying 26 studies performed between 2010-2024 in SSA exploring factors influencing fertility intention for PLHIV. A codebook was developed and themes were extracted using NVivo. Papers were assessed for quality and bias using GRADE-CERQual criteria.

Confidence in antiretroviral-therapy was a key factor influencing fertility desire, however there were ongoing concerns about horizontal and vertical transmission. The data described pronatalist societies where childbearing was the norm and significant societal pressure was placed on women to reproduce. PLHIV described a difficult choice between the stigma of childbearing with HIV and both having a stigmatising illness such as HIV, and remaining childless. Childbearing was seen as fundamental to identity as person and this did not change following HIV diagnosis.

Fertility intention is complex, conflicting and fluctuating, and made more so by a diagnosis of HIV. Developing a nuanced understanding of the influences on fertility desires and intentions for PLHIV is crucial for developing person-centred, applicable and acceptable sexual and reproductive health services. This is centric to a rights-based approach to sexual healthcare and to enabling people to achieve childbearing goals whilst minimising risks of HIV transmission.


  • Christine Campbell1, Heather Cubie1, Beatrice Kabota2, Anna Balley3, Hariet Chirwa2, Angella Chiwaya4, Miriam Deeny5, Wendy McMullen6, Isabel Bruce7, Yowati Nthenga2, Agness Nyanda2, Themba Mzilahowa8

1 Usher Institute, The University of Edinburgh, 

2 Nkhoma CCAP Hospital, Central Region, Malawi

3 Mlambe Mission Hospital, Southern Region, Malawi

4 Mzuzu Health Centre, Northern Region, Malawi

5 NHS Greater Glasgow and Clyde (retd)

6 NHS Tayside 

7 Project Solutions Consultancy

8 Centre for Health, Agriculture and Development Consultancy, Blantyre, Malawi

Malawi has very high rates of cervical cancer, and the highest global mortality. This poster describes thirteen years of partnership working with colleagues in Malawi and Scotland and alongside the Malawi Ministry of Health to strengthen sustainable health care of women in rural areas of the country, specifically through provision of cervical cancer screening using Visual Inspection with Acetic Acid (VIA) and treatment with thermal ablation or biopsy and referral for subsequent care.

Cervical cancer screening was implemented in eight hospitals and forty-two health centres, located in under-served rural areas; health facilities were refurbished, and screening clinics equipped with screening supplies and gynecological beds.

There was a strong emphasis on empowerment of women through development of accessible education and awareness materials, including through TV and radio promotion. 

The initiative increased clinical skills of nurse providers through iterative and experiential training in screening skills, mentorship and Continuous Professional Development, leading to over 115 nurse providers receiving independent accreditation, promoting an ongoing and sustainable service. Safeguarding and Mentoring policies and practices developed by the project have now been adopted nationally. 

Over 100,000 women received an initial screen, with follow-up and subsequent screen pathways implemented. Services were integrated with family planning and clinics for women living with HIV. All sites adoption of an inclusion health perspective, ensuring vulnerable and excluded women including disabled women, women living with albinism and transactional sex workers have the opportunity to access cervical cancer screening.


  • Jaime Garcia-Iglesias1, Martyn Pickersgill1, Tom May2, Jeremy Williams3, Maurice Nagington4, Sophie Buijsen5, Ciara McHugh6, Jeremy Horwood2

1 Centre for Biomedicine, Self and Society, The University of Edinburgh

2 University of Bristol

3 University College London

4 University of Manchester

5 University of Edinburgh

6 Queens University Belfast

The 2022-23 mpox outbreak was a critical moment for sexual and public health in the UK. It required developing responses that were rapid and engaged with a multitude of stakeholders, including community organisations, clinicians, and policy actors.

Our interdisciplinary team conducted rapid research to explore how the response had been articulated, its strengths and weaknesses, and develop guidance for future pandemic preparedness and outbreak response. 

In this poster we focus on two areas:

  1. The involvement of community organisations was key to tackling health inequalities but faced important barriers related to funding allocation and 'ways of working'. Furthermore, the was no clarity regarding the role of novel partners, such as social media influencers or dating apps. We suggest that communities should be strengthened as part of pandemic preparedness and the ethical implications of new partnerships with digital health actors.
  2. Collaborative networks involving a diversity of stakeholders were essential for effective communication and message development. 

However, these networks were often informal and based on personal relationships. We provide recommendations to develop 'infrastructures of collaboration' that can support future outbreak response. 

This project was funded by the ESRC [ES/ X010805/1] and led at the University of Edinburgh by the Centre for Biomedicine, Self and Society (Jaime Garcia-Iglesias and Martyn Pickersgill) with partners at UCL (IPPO), UKHSA, University of Bristol (NIHR HPRU) and University of Manchester.


  • Jeni Harden1; Nicola Boydell1; Marie Larsson1

1 Usher Institute, The University of Edinburgh

Access to safe, high-quality abortion and contraception care is an important aspect of the provision of sexual and reproductive health care. Access to contraception, including post-abortion contraception, was disrupted during the COVID-19 pandemic and challenges in access to post-abortion contraception persist. We know from our past research with people with abortion experience and abortion care providers, that there are differences in the ways people access post-abortion contraception in different areas of Scotland. Our research shows that these things really matter to people’s experiences with this care. We believe there is a need to work together with both people who have used (or in the future might use) abortion and contraceptive services and those that help provide.

The aim of the SCOPE project is to co-design adaptable models for post-abortion contraceptive (PAC) services and improve patients’ choice of, and access to, contraception across Scotland. We have worked with people with experience of abortion and abortion care providers to 1) identify the core principles of a feasible PAC service that meets the needs of people from diverse communities, including those that may struggle to access healthcare (such as those living in remote and rural areas) and 2) establish a PAC framework to support abortion services across Scotland to implement PAC services based on these principles. In this poster, we outline the stages of the project and the co-design process.


  • Cathryn Broderick1, Marlene Stewart1, Katie Thomson2, Ceri Sellers2, Candida Fenton1, Julie Cowie2, Wei Xu1, Christa St Jean3, Keira Charteris3, Madhurima Nundy3, Vaishali Vardhan3, Prerna Krishan3, Jolie Pistol3, Leonor Rodríguez4, Alex Todhunter-Brown2, Frederike van Wijck5, Sheila Cameron2, Catriona Keerie6, Rod S Taylor7, Gerry Stansby8, Gillian Mead9

1 NESSIE, Usher Institute, The University of Edinburgh

2 NESSIE, Glasgow Caledonian University

3 UNCOVER, Usher Institute, The University of Edinburgh

4 School of Health in Social Science, The University of Edinburgh

5 Glasgow Caledonian University

6 ECTU, Usher Institute, The University of Edinburgh

7 MRC/CSO Social and Public Health Sciences Unit & Robertson Centre for Biostatistics, University of Glasgow 

8 University of Newcastle

9 Ageing and Health, Usher Institute, The University of Edinburgh

Polyvascular disease, i.e. having more than one cardiovascular condition, is becoming increasingly common, yet systematic reviews of interventions such as exercise are traditionally targeted at people with a single condition. We aimed to determine the effect of exercise in the secondary prevention of major cardiovascular events and health-related quality of life (HRQoL) in people with vascular disease and to assess the impact of polyvascular disease.

We searched Cochrane Register of Studies Online, MEDLINE, Embase, CINAHL, and trials registries in February 2025 for randomised controlled trials (RCTs) of exercise in people with stroke (including TIA), coronary artery disease, heart failure (HF), and peripheral arterial disease (PAD). We excluded studies where exercise was delivered for less than six weeks. Critical outcomes were mortality (all-cause and cardiovascular-specific), vascular events (stroke, myocardial infarction, amputation, acute limb ischaemia), vascular hospitalisations, and HRQoL. We assessed risk of bias (RoB1 tool) and certainty of evidence (GRADE). PROSPERO: CRD42024517019

This systematic review (280 RCTs, 23,419 participants) showed that in people with stroke, coronary artery disease, HF, and/or PAD exercise improved HRQoL and reduced hospitalisations with no difference in all-cause or cardiovascular-specific mortality. Most trials were designed to target people with a single vascular condition and did not report the presence of additional vascular diseases thus restricting the applicability of the evidence to people with polyvascular disease.

This systematic review highlights the safety and benefits of exercise for people with vascular disease and the evidence gap regarding guidance on exercise for people with polyvascular conditions.


  • Cat Tottie1, Stephen Malden2, Debra Hertzberg3

1 School of Engineering, The University of Edinburgh

2 School of Health in Social Science, The University of Edinburgh

3 Homeless Link

People experiencing homelessness have the most severe health outcomes of any population. Homelessness is associated with high prevalence of multi-morbidity, preventable disease and early onset of frailty. These inequalities mean that people experiencing homelessness have an average age of death of just 43 – over 30 years younger than the general population.

These inequalities are widely recognised as being related to challenges in healthcare access, with severe chronic illness compounded by barriers into community care which drive up use of urgent and emergency care. This is so much the case that homelessness is recognised as an independent risk factor for intensive unplanned care use.

But while this connection is recognised, less is known about what factors presuppose a person experiencing homelessness to higher rates of unplanned care use. Evidence suggests disproportionate unplanned care use among a small proportion of people experiencing homelessness; there is therefore value in identifying what other factors indicate higher rates of unplanned care use overall.

This poster examines how factors such as housing status, historic trauma, multiple disadvantage and demographic status influence rates of unplanned care use among people experiencing homelessness. This is achieved by outlining findings from a secondary analysis of a large-scale English survey of people experiencing homelessness. The findings are of interest to those working in research, policy or practice with focus on secondary care, health inequalities and homelessness.


  • Susanne Maxwell1, Caroline Jackson1, Nazir Lone1, Debbie Cavers1, Bruce Guthrie1

1 Usher Institute, University of Edinburgh

Cancer remains a major public health challenge in the UK, with more than 392, 000 new cases diagnosed across the UK each year. Understanding the co-existing health conditions of people with cancer, both at diagnosis and throughout survivorship, is essential for effective public health planning and optimising treatment pathways.

This cross-sectional study included patients registered with a CPRD Aurum–participating primary care practice in the UK on 1 January 2018. Primary care records were linked to Hospital Episode Statistics Admitted Patient Care (HES-APC), Office for National Statistics (ONS), death registration data, and the Index of Multiple Deprivation (IMD). Eligible patients were aged ≥18 years, alive, and actively registered on the index date. Four cancer types were analysed: breast, prostate, lung, and colorectal. Multimorbidity was assessed using 45 chronic physical and mental health conditions, defined as ≥2 of these conditions in addition to cancer. Incident and survivor cohorts were described by cancer type, and negative binomial regression estimated associations with condition count.

There were 33,258 incident cancers diagnosed during the study period and 16,807 cancer survivors. Multimorbidity was common at diagnosis in 59.4, 70.4, 73.5 and 85.3% of breast, prostate, colorectal and lung cancer patients respectively. Hypertension was the most common comorbidity across all cancers. Deprivation and age significantly increased the prevalence of multimorbidity across all cancer types.

Multimorbidity is prevalent among cancer patients and is associated with age and socioeconomic deprivation. Cancer care should be designed to address the substantial pre-existing multimorbidity in this population.


  • Holger Schulze1, Mohana Bhattacharjee4, Ananta Chakraborty5, Angela Towers3, Jay Patel6, Abbie Gao1, Bhaskar Jyoti Das4, Chandrayee Deshmukhya4, Subhankar Bhattacharjee5, Rakib Ahmed5, Ananya Deb5, Barnamoy Bhattacharjee5, Delina Evans3, Ada Zwetlana6, Kathakali Nath4, Prabhu Balasubramanian6, Neha Khaware6, Amitabha Bhattacharjee4, Debadatta Dhar5, Alison Prendiville3, Ravikrishnan Elangovan6, Till T. Bachmann1,2

1 Centre for Inflammation Research, Institute for Regeneration and Repair, The University of Edinburgh 

2 Usher Institute, The University of Edinburgh

3 University of the Arts London (Professor Dr Alison Prendiville)

4 Assam University, India (Professor Dr Amitabha Bhattacharjee) 

5 Silchar Medical College and Hospital, India (Professor Dr Debadatta Dhar)

6 Indian Institute of Technology Delhi, India (Professor Dr Ravikrishnan Elangovan)

Urinary tract infections (UTIs) are one of the most frequent infectious disease globally and the second leading cause of empiric antibiotic consumption, contributing to the threat of antimicrobial resistance (AMR). Better diagnostics could help, but UTI patients in the community of Low- and Middle-Income Countries (LMIC) often cannot seek medical care and suffer in silence, with UTI diagnostics currently unavailable in primary care. The DOSA project addresses the challenge of UTIs in rural India as a representative setting for a LMIC. We developed a UTI point of care strip test as part of a diagnostic bundle including healthcare information for the patient and data capture which interfaces with the public health system. The UTI strip test will be deployed through community healthcare workers for home testing directing patients to access primary care. The DOSA UTI strip test was co-designed with community healthcare workers and community members and its field performance evaluated in the field in 2024/25. Once fully deployed, our diagnostic bundle will improve access to care, therapeutic outcome, and reduce the inappropriate use of antibiotics. The bundle will also deliver the surveillance data on UTI and AMR urgently needed for policy interventions to tackle UTI in the community in LMICs. This will increase the understanding and change behaviour towards UTI and enable policy makers to implement better solutions for patients. We currently operate in rural India but plan to expand to similar settings in Africa as well as the UK. DOSA1 (2019-23) was funded by UKRI/DBT, DOSA2 (2024-25) was funded by LifeArc.

Poster not published online as per author request.


  • Nandhni Chirganam Gandhi1, Karolina Anna Scahill2,3, Vidar Nimer4, Christine Campbell1

1 Usher Institute, The University of Edinburgh

2 College of Medicine and Veterinary Medicine, The University of Edinburgh

3 Evidensia Södra Djursjukhuset Kungens Kurva, Sweden

4 Evidensia Djurklininen Roslagstull, Sweden

Dentists prescribe around 10% of all human antibiotics worldwide, with evidence showing that a large proportion are inappropriate, up to 80% in the USA and UK. Such misuse contributes to adverse effects and drives antimicrobial resistance (AMR). Despite initiatives like the WHO’s Global Action Plan (GAP), AWaRe classification, stewardship programmes, and prescribing guidelines, inappropriate use persists. Antibiotic prescribing is complex and influenced by multiple modifiable factors. This thesis aimed to (i) estimate the magnitude of reported inappropriate antibiotic prescribing (IAP) in dentistry, and (ii) explore the barriers and facilitators to appropriate prescribing among dental professionals.

Two systematic reviews were conducted. The first, registered in PROSPERO, screened 12,352 records, identifying 21 studies. Study quality was assessed using the MMAT, with most ratings being medium to high quality. High heterogeneity precluded meta-analysis, so a narrative synthesis was undertaken. The second review screened 14,498 records, identifying seven qualitative studies that were assessed using CASP. Thematic synthesis identified key themes influencing prescribing behaviour.

The prevalence of IAP was high, with a mean of 64.03% and a median of 77.66%. Middle-income countries reported the most significant burden (mean 85.67%), followed by high-income countries (mean 53.25%); no studies were identified from low-income settings. Five themes and 17 sub-themes were identified as barriers and facilitators. Key barriers included time pressures, limited access to care, medico-legal concerns, and patient expectations. Facilitators included access to information, communication, education, awareness of AMR, and guideline use.

Tackling AMR in dentistry requires integrated policies, tailored guidelines, ongoing education, and interprofessional collaboration, with greater research needed in LMICs. 


  • Hannah C. Lepper1, Emma Pujol-Hodge2, Jada Hackman1,3, Kevin van Zandvoort4, Kate Mellor-Wright3, Oliver Lorenz3, Stephanie Lo3,5, Duc Anh Dang6, Hung Thai Do7, Gerry Tonkin-Hill8,9, Stephen D. Bentley3, Lay Myint Yoshida10, Stefan Flasche4,11, Nicholas G. Davies4, Katherine E. Atkins1,4

1 Usher Institute, The University of Edinburgh, UK

2 Public Health Scotland, Edinburgh, UK

3 Parasites and Microbes, Wellcome Sanger Institute, UK

4 Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, UK

5 Department of Life Sciences, University of Bath, UK

6 National Institute of Hygiene and Epidemiology, Vietnam

7 Pasteur Institute of Nha Trang, Vietnam

8 University of Melbourne, Australia

9 Peter MacCallum Cancer Centre, Australia

10 Nagasaki University, Japan

11 Charité Centre for Global Health, Germany

Antibiotic resistant pneumococcal infection causes significant morbidity. The impact of interventions to reduce antibiotic resistance depends on complex ecology governing serotype and resistant strain dynamics, for which we currently lack mechanistic explanations. We aimed to evaluate whether multi-serotype transmission models of pneumococcal strains can recapitulate empirical resistance data, thus providing plausible explanations for how antibiotic-sensitive and resistant strains coexist at intermediate frequencies, both between and within serotypes.

We extracted a global dataset of pneumococcal carriage prevalence in healthy, unvaccinated children by serotype and penicillin susceptibility from a previously-published systematic review. Next, we developed a suite of multi-serotype individual-based models based on mechanisms previously shown to plausibly explain coexistence in single serotype models, incorporating transmission and clearance rate differences between serotypes and serotype-specific immunity. Finally, we calibrated the suite of models to the relative serotype prevalence and overall resistance frequency drawn from the extracted dataset.

Overall resistance frequency varied considerably between studies, but we found a positive association between frequency of a serotype and resistance frequency within that serotype, and coexistence of sensitive and resistant strains within serotypes. Each individual-based model predicted different serotype-specific resistance frequencies, and combinations of models performed best at capturing features of the empirical data.

Serotype-specific resistance patterns may give important clues to understand fundamental antibiotic resistance epidemiology. We show that combining multiple independently plausible mechanisms can capture resistant pneumococcal carriage, but additional data are needed to determine the strength of individual mechanisms within these combinations.


  • Victor Velasco-Pardo1, Nathan Constantine-Cooke1, Evangelin Shaloom Vitus1,2, Begoña Bolós1,3, Charlie W Lees1,4, Catalina A Vallejos1,4,5

1 Institute of Genetics and Cancer, The University of Edinburgh, UK

2 Aalborg University, Copenhagen Campus, Denmark

3 Alan Turing Institute, UK

4 Usher Institute, The University of Edinburgh, UK

5 Health Data Research UK

Time-to-event, or survival analysis, is used to analyse the time until an event of interest occurs. For example, in clinical applications, death, relapse or disease progression are often events of interest. Whilst classic survival methods assume model covariates are static, longitudinal measurements related to the outcome of interest, such as biomarker, are often collected. Such longitudinal information can be used to dynamically predict risk, updating those predictions as new data is available.

Whilst several methods have been proposed for this purpose, landmarking stands out as a computationally efficient approach suitable for large cohorts with an extensive longitudinal follow-up. Landmarking is a two-step approach which combines an arbitrary longitudinal model with an arbitrary time-to-event model. Despite this flexibility, existing landmarking software offers restricted options.

Here, we present the landmaRk package, a flexible tool for dynamic risk prediction. Our aim is to provide an end-to-end pipeline: from data pre-processing steps to model fitting and the evaluation of predictive performance. Using a modular approach, landmaRk can support a wide array of longitudinal and survival models whilst allowing others to incorporate their own models. landmaRk also allows users to compare the performance of different model specifications using concordance index (C-index) and Brier score.

Finally, we introduce a novel approach which incorporates latent population heterogeneity into the landmarking framework. This improves risk prediction by borrowing information across individuals with similar longitudinal trajectories. This is particularly critical when monitoring patients with chronic conditions (e.g. inflammatory bowel disease) which can be highly heterogeneous in nature.


Bavo Kempen (1), Bart Depreiter (1,2), Ian Piper (3,4), Maria Poca (5), Stefan Mircea Iencean (6), Mireia Garcia (7,8), James Weitz (9), Gayathri Subramanian (10), Roddy O’Kane (11), Julian Zipfel (12), Arta Barzdina (13), Stefano Pazzato (14), Hari Krishnan (15), Patricia Jones (16), Laura Smith (3), Tsz-Yan Milly Lo (3,4) on the behalf of the KidsBrainIT Consortium

1 Department of Neurosciences, KU Leuven, Leuven, Belgium

2 Department of Neurosurgery, University Hospitals Leuven, Leuven, Belgium

3 Usher Institute, The University of Edinburgh, Edinburgh, UK

4 Paediatric Critical Care Unit, Royal Hospital for Children & Young People, Edinburgh, UK

5 Department of Neurosurgery, Vall d’Hebron University Hospital, Barcelona, Spain

6 Department of Neurosurgery, GrT Popa University of Medicine and Pharmacy, Iasi, Romania

7 Paediatric Intensive Care Unit, Bristol Royal Hospital for Children, University Hospitals Bristol NHS Foundation Trust, Bristol, UK

8 Paediatric Intensive Care, Parc Taulí University Hospital, Sabadell, Spain

9 Paediatric Critical Care, John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust, Oxford, UK

10 Paediatric Intensive Care Unit, Royal Manchester’s Children’s Hospital, Manchester University Hospitals NHS Foundation Trust, Manchester, UK

11 Glasgow Paediatric Neurosurgery, Royal Hospital for Children & Institute of Neurological Sciences, Goven Road, Glasgow, UK

12 Section of Paediatric Neurosurgery, Department of Neurosurgery, University Hospital Tübingen, Germany

13 Children’s Clinical University Hospital, Clinic for Anesthesiology and Intensive Care, Riga, Latvia

14 Neonatal and Paediatric Intensive Care Unit, IRCCS Istituto Giannina Gaslini, Genova, Italy

15 Birmingham Children’s Hospital, Birmingham, UK

16 Child Life and Health, The University of Edinburgh, Edinburgh, UK

Cerebral perfusion pressure (CPP) dose-response on outcome following paediatric traumatic brain injury (TBI) remains unknown. We aim to produce the first paediatric TBI CPP dose-response visualisation plot to define a ‘safe’ CPP zone. 

A prospective international multi-centre observational study was performed to recruit 104 paediatric TBI patients, aged 2 to <16 years-old, within the KidsBrainIT consortium. Their fully-anonymised prospectively collected routine minute-by-minute intracranial pressure (ICP), mean arterial blood pressure (mABP), and CPP time-series data were categorised into CPP intensity-duration episodes, albeit CPP above or below a range of thresholds. These episodes were then correlated with the 6-months modified Glasgow Outcome Score (GOS) and depicted in ‘3-D’ colour-coded CPP dose-response plots. The effects of cerebrovascular reactivity (CVR) patterns and ICP were also examined.

Our paediatric CPP dose-response plots resembled the previously published adult CPP dose-response plots. An exponential ‘black’ transition curve on the CPP pressure-time plots separated CPP episodes associated with poor (‘red’, GOS <4) and good (‘blue’) outcome. Lower and higher ends of CPP intensity were only tolerated for shorter durations. A ‘safe’ CPP zone (56-89 mmHg) was identified for paediatric TBI with active CVR pattern. ICP levels >20 mmHg were associated with worse outcome, irrespective of CPP dose.Conclusions: The paediatric CPP dose-response on poor outcome was visualised successfully for the first time defining a ‘safe’ CPP zone. The identified ‘critical’ lower CPP limit exceeded the current recommended minimum CPP target and warrants urgent validation in an independent cohort to provide evidence-based future CPP clinical targets.


  • Iris Soare-Nguyen1, Javier Escudero Rodiguez2, Shima Abdullateef1, Brian Jordon3, Laura Smith1, Samson Gwer4, Ailsa McLellan3, Vera Nenadovic5, Tsz-Yan Milly Lo1,3

1 Usher Institute, The University of Edinburgh

2 School of Engineering, The University of Edinburgh

3 Royal Hospital for Children and Young People, Edinburgh

4 Gertude’s Children’s Hospital, Kenya

5 BrainsView Inc., Mississauga, Canada

Seizures are common in acutely ill children with the majority (74%) being subclinical seizures making diagnosis without electroencephalograms (EEG) difficult. Gold-standard multi-channels EEG require specialist neurophysiologists to acquire which is undeliverable at the bedside globally. We previously demonstrated that brain connectivity computational assessment of 4-channels EEG allowed seizure detection with over 80% accuracy without specialist neurophysiologists input. We aim to determine the feasibility of machine learning seizure detection using only 2-channels of EEG to facilitate future point-of-care uptake including in resources limited countries. 

A prospective data informatics study was performed through the Global Window-in-the-Brain (G-WiB) consortium. Twenty-nine fully anonymised multi-centre routinely collected ‘clinical grade’ EEG were used to adapt and test our brain connectivity assessment algorithm’s seizure detection performance.  Six different 2-channels EEG montages were tested. Seizure detection accuracy was determined by comparing the algorithm’s output with expert neurologist identified seizures. 

Seizure detection using only 2-channels of EEG with our G-WiB algorithm was possible and achieved over 75% accuracy in 4 of the 6 selected 2-channels montages. 

Machine learning seizure detection with very low-density EEG montage is feasible and warrants further investigations and validations.


  • Junyi Gao1,4, Zifeng Wang2, Cao Xiao3, Jimeng Sun2, Ewen Harrison1

1 Centre for Medical Informatics, The University of Edinburgh

2 School of Computer Science, University of Illinois at Urbana-Champaign

3 GE Healthcare

4 Health Data Research UK

The clinical trial pipeline suffers from major inefficiencies in planning, site selection, patient recruitment, and documentation, leading to significant delays and costs. We developed a suite of targeted artificial intelligence solutions to address these critical bottlenecks by leveraging diverse datasets, including EHRs, claims, and trial documents.

For patient recruitment, our cross-modal deep learning model, COMPOSE, achieved 83.7% patient-trial matching accuracy and 98.0% AUC, significantly outperforming baselines. To optimize site selection, our DocTr model utilized graph networks and large language models (LLMs) to integrate heterogeneous data, yielding a 58% higher clinician-trial match similarity than traditional methods while improving site fairness scores by up to 25%.

For regulatory documentation, we created InformGen, an LLM-driven copilot that achieved nearly 100% compliance with core rules for informed consent forms and over 90% factual accuracy under human supervision.

These tools are supported by TrialPanorama, a curated database of approximately two million trials linked to biomedical ontologies, which also serves as a benchmark for future AI development. Collectively, these AI-driven innovations demonstrate a powerful capacity to enhance the speed, efficiency, and fairness of the clinical research ecosystem, promising to reduce operational bottlenecks and accelerate the delivery of novel therapies to patients.


  • James C Porter1; Miguel O. Bernabeu1; Baljean Dhillon2,3

1 Centre for Medical Informatics, Usher Institute, The University of Edinburgh

2 Centre for Clinical Brain Sciences, The University of Edinburgh

3 Ophthalmology, Princess Alexandra Eye Pavilion

Typically imaged via optical coherence tomography (OCT), retinal layer thicknesses have been proposed as biomarkers for neurodegenerative disease. We demonstrate that these thicknesses can be estimated from more abundant and simple colour fundus photography (CFP) via a ridge-regression model. We examine the performance influence of data stratification by sex, ethnicity and health.

We sourced UK Biobank CFP images and corresponding OCT thicknesses: retinal nerve fibre layer (RNFL), ganglion cell-inner plexiform layer (GCIPL), and inner nuclear layer (INL). CFP features were extracted using the pretrained RETfound foundation model. Ridge regression (10-fold CV) was used to predict layer thickness from CFPs with performance evaluated via mean absolute error (MAE) and coefficient of determination (R2).

To investigate the effects of sex, ethnicity and health we subset our data via these variables. A healthcare interaction score (HS) was calculated as a proxy for worse health and used to filter participants at different thresholds. 

Thickness prediction MAEs across all layers were comparable to the natural variability and had positive R2 (≈0.09 to 0.3) indicating the models captured the measurement variance. Across subsets, GCIPL was not significant, RNFL MAE was significantly lower for right eye or female, INL MAE was left/right eye and white/nonwhite significant but with very small MAE changes (≈0.09μm). HS filtering had negligible effects (<0.1μm MAE change) event at strict thresholds.

Our model successfully estimates retinal thickness from CFP images. GCIPL predictions are robust, but RNFL and INL require confounder-adjusted models. Performance was robust to HS filtering, suggesting invariance to general health.