Background Maternity care is a complex system involving treatments and interactions between patients, healthcare providers, and the care environment. To enhance patient safety and outcomes, it is crucial to understand the human factors (e.g. individuals' decisions, local facilities) influencing healthcare. However, most current tools for analysing healthcare data focus only on biomedical concepts (e.g. health conditions, procedures and tests), overlooking the importance of human factors. Methods We developed a new approach called I-SIRch, using artificial intelligence to automatically identify and label human factors concepts in maternity investigation reports describing adverse maternity incidents produced by England's Healthcare Safety Investigation Branch (HSIB). These incident investigation reports aim to identify opportunities for learning and improving maternal safety across the entire healthcare system. Unlike existing clinical annotation tools that extract solely biomedical insights, I-SIRch is uniquely designed to capture the socio-technical dimensions of patient safety incidents. This innovation enables a more comprehensive analysis of the complex systemic issues underlying adverse events in maternity care, providing insights that were previously difficult to obtain at scale. Importantly, I-SIRch employs a hybrid approach, incorporating human expertise to validate and refine the AI-generated annotations, ensuring the highest quality of analysis. Findings I-SIRch was trained using real data and tested on both real and synthetic data to evaluate its performance in identifying human factors concepts. When applied to real reports, the model achieved a high level of accuracy, correctly identifying relevant concepts in 90% of the sentences from 97 reports (Balanced Accuracy of 90% ± 18% (Recall 93% ± 18%, Precision 87% ± 34%, F-score 96% ± 10%). Applying I-SIRch to analyse these reports revealed that certain human factors disproportionately affected mothers from different ethnic groups. In particular, gaps in risk assessment were more prevalent for minority mothers, whilst communication issues were common across all groups but potentially more for minorities. Interpretation Our work demonstrates the potential of using automated tools to identify human factors concepts in maternity incident investigation reports, rather than focusing solely on biomedical concepts. This approach opens up new possibilities for understanding the complex interplay between social, technical and organisational factors influencing maternal safety and population health outcomes. By taking a more comprehensive view of maternal healthcare delivery, we can develop targeted interventions to address disparities and improve maternal outcomes. Targeted interventions to address these disparities could include culturally sensitive risk assessment protocols, enhanced language support, and specialised training for healthcare providers on recognising and mitigating biases. These findings highlight the need for tailored approaches to improve equitable care delivery and outcomes in maternity services. The I-SIRch framework thus represents a significant advancement in our ability to extract actionable intelligence from healthcare incident reports, moving beyond traditional clinical factors to encompass the broader systemic issues that impact patient safety.

This study applies Natural Language Processing techniques, including Latent Dirichlet Allocation, to analyse anonymised maternity incident investigation reports from the Healthcare Safety Investigation Branch. The reports underwent preprocessing, annotation using the Safety Intelligence Research taxonomy, and topic modelling to uncover prevalent topics and detect differences in maternity care across ethnic groups. A combination of offline and online methods was utilised to ensure data protection whilst enabling advanced analysis, with offline processing for sensitive data and online processing for non-sensitive data using the `Claude 3 Opus' language model. Interactive topic analysis and semantic network visualisation were employed to extract and display thematic topics and visualise semantic relationships among keywords. The analysis revealed disparities in care among different ethnic groups, with distinct focus areas for the Black, Asian, and White British ethnic groups. The study demonstrates the effectiveness of topic modelling and NLP techniques in analysing maternity incident investigation reports and highlighting disparities in care. The findings emphasise the crucial role of advanced data analysis in improving maternity care quality and equity.

In healthcare, thousands of safety incidents occur every year, but learning from these incidents is not effectively aggregated. Analysing incident reports using AI could uncover critical insights to prevent harm by identifying recurring patterns and contributing factors. To aggregate and extract valuable information, natural language processing (NLP) and machine learning techniques can be employed to summarise and mine unstructured data, potentially surfacing systemic issues and priority areas for improvement. This paper presents I-SIRch:CS, a framework designed to facilitate the aggregation and analysis of safety incident reports while ensuring traceability throughout the process. The framework integrates concept annotation using the Safety Intelligence Research (SIRch) taxonomy with clustering, summarisation, and analysis capabilities. Utilising a dataset of 188 anonymised maternity investigation reports annotated with 27 SIRch human factors concepts, I-SIRch:CS groups the annotated sentences into clusters using sentence embeddings and k-means clustering, maintaining traceability via file and sentence IDs. Summaries are generated for each cluster using offline state-of-the-art abstractive summarisation models (BART, DistilBART, T5), which are evaluated and compared using metrics assessing summary quality attributes. The generated summaries are linked back to the original file and sentence IDs, ensuring traceability and allowing for verification of the summarised information. Results demonstrate BART's strengths in creating informative and concise summaries.

Humans lack the enzyme that produces the sialic acid N-glycolyl neuraminic acid (Neu5Gc), but several lines of evidence have shown that Neu5Gc can be taken up by mammalian food sources and replace the common human sialic acid N-acetyl neuraminic acid (Neu5Ac) in glycans. Cancer tissue has been shown to have increased the presence of Neu5Gc and Neu5Gc-containing glycolipids such as the ganglioside GM3, which have been proposed as tumor-specific antigens for antibody treatment. Here, we show that a previously described antibody against Neu5Gc-GM3 is binding to Neu5GC-containing gangliosides and is strongly staining different cancer tissues. However, we also found a strong intracellular staining of keratinocytes of healthy skin. We confirmed this staining on freshly isolated keratinocytes by flow cytometry and detected Neu5Gc by mass spectrometry. This finding implicates that non-human Neu5Gc can be incorporated into gangliosides in human skin, and this should be taken into consideration when targeting Neu5Gc-containing gangliosides for cancer immunotherapy.

Virtual Reality (VR) environments have been used for training, education, and entertainment due to the interactive and embodied experiences the technology provides. Studies have shown that VR can be used to help support individuals with intellectual disabilities in various aspects of their lives. Likewise, conversational agents such as chatbots can be used to bolster competence training and well-being management for this user population. This paper addresses the need for inclusive job interview practice for individuals with intellectual disabilities by discussing the development of a VR application, AllyChat. A two-part development phase is presented with pilot testing included for each phase. First, a conversational AI chatbot is tested and with positive feedback iterated upon to develop an immersive mock job interview experience in VR. A second pilot study is conducted with university students to test the functionality of a high-fidelity prototype. Future work will include improvement upon the developed VR application and further testing.

Learning objectivesBy reading this article you should be able to:•Discuss the properties of complex systems and a systems approach to incident investigation.•Describe the differences between reactive and proactive safety approaches.•Describe some of the system-focused models applied to patient safety incident investigations.Key points•Healthcare has properties of a complex system.•A systems approach to investigation considers how a system's structure influences behaviour.•A system's resilience is contributed to by people's ability to adjust to varying conditions. By reading this article you should be able to:•Discuss the properties of complex systems and a systems approach to incident investigation.•Describe the differences between reactive and proactive safety approaches.•Describe some of the system-focused models applied to patient safety incident investigations. •Healthcare has properties of a complex system.•A systems approach to investigation considers how a system's structure influences behaviour.•A system's resilience is contributed to by people's ability to adjust to varying conditions. Patient safety incidents are events where a patient experienced or could have experienced harm during an encounter with healthcare. The aim of investigating an incident is to learn from it by identifying contributory factors. An investigation also seeks to understand the interactions between these factors and how they give rise to either safe or unsafe care. Patient safety investigations have historically focused on solutions at the end of a perceived linear ‘chain of events’ with little attention applied to the background of the incident. Healthcare has, in its recent history, been encouraged to learn from safety-critical industries such as transport. This has led to modified tools and procedures from these sectors being transposed into different areas of healthcare with variable effects on improving safety.1Macrae C. Stewart K. Can we import improvements from industry to healthcare?.BMJ. 2019; 364: l1039Crossref PubMed Scopus (16) Google Scholar One approach, imported from the aviation industry, stemmed from a report released by the US Institute of Medicine, called To Err is Human: Building a Safer Health System.2Institute of MedicineCommittee on quality of health care in America. To Err is human: Building a safer health system. IOM, Washington, DC2000Google Scholar This report recommended a nationwide mandatory public reporting system for incidents. A range of such incident reporting tools now exist internationally. However, recently it has been argued that there has been too much focus on reporting rather than high-quality analysis of the incident itself.1Macrae C. Stewart K. Can we import improvements from industry to healthcare?.BMJ. 2019; 364: l1039Crossref PubMed Scopus (16) Google Scholar Studies of accidents in safety-critical industries have led to the development of incident investigation tools focusing on a broader understanding of accident causation with less focus on individual error and more on organisational and wider system factors. One such tool, now widely used in healthcare, is root cause analysis (RCA), which originated in the engineering industry. However, there have been criticisms of the technique and how it is implemented at a local level despite its intentions to look at the wider system and not just individual actions.3Wiig S. Braithwaite J. Clay-Williams J. It’s time to step it up. Why safety investigations in healthcare should look more to safety science.Int J Qual Health Care. 2020; 32: 281-284Crossref PubMed Scopus (9) Google Scholar Critics suggest that system components are studied in isolation and the importance of their interactions and dependencies with other components is not appreciated.4Peerally M.F. Carr S. Waring J. Dixon-Woods M. The problem with root cause analysis.BMJ Qual Saf. 2017; 26: 417-422PubMed Google Scholar Kellogg and colleagues5Kellogg K.M. Hettinger Z. Shah M. et al.Our current approach to root cause analysis: is it contributing to our failure to improve patient safety?.BMJ Qual Saf. 2017; 26: 381-387PubMed Google Scholar reviewed 302 RCAs and found that despite repeated RCAs into several event types the same events recurred multiple times. Another concern is that most RCAs propose behaviour-focused solutions such as training and policy reinforcement. System-focused solutions involving interactions between people and technology and underlying design-related factors are rarely considered.5Kellogg K.M. Hettinger Z. Shah M. et al.Our current approach to root cause analysis: is it contributing to our failure to improve patient safety?.BMJ Qual Saf. 2017; 26: 381-387PubMed Google Scholar,6Trbovich P. Shojania K.G. Root-cause analysis: swatting at mosquitoes versus draining the swamp.BMJ Qual Saf. 2017; 26: 350-353PubMed Google Scholar On 1 April 2017, the Healthcare Safety Investigation Branch (HSIB) became operational. Its remit is to conduct independent investigations of patient safety concerns in NHS-funded care across England. It is the first organisation of its type in the world. The investigations are learning- and system-focused and do not look to apportion blame or liability. In its National Investigation Programme, investigations are selected based on their potential to interrogate different levels of the healthcare system and their impact on patients and services. In March 2020, NHS England/Improvement in the UK, as part of its new patient safety strategy, published a new patient safety incident response framework (PSIRF).7NHS England and NHS ImprovementThe NHS patient safety strategy: safer culture, safer systems, safer patients. NHS, London, UK2019Google Scholar It describes a broader systems approach to local incident management moving away from ‘root cause analysis’ focused on person-based factors.7NHS England and NHS ImprovementThe NHS patient safety strategy: safer culture, safer systems, safer patients. NHS, London, UK2019Google Scholar In January 2020, the UK Academy of Medical Royal Colleges published the first iteration of a national patient safety syllabus to guide the training of all NHS staff.8Academy of Medical Royal Colleges in collaboration with Health Education England, NHS England and NHS ImprovementNational patient safety syllabus 1.0. Training for All NHS Staff, London2020Google Scholar Its learning outcomes include ‘understanding the systems-based approach to investigating patient safety incidents’. Safety management encompasses the practices that direct, monitor and intervene in operations for the purpose of minimising the risks to patient safety. Incident analysis is just one aspect of a wider safety management system. Safety-critical sectors such as aviation have very low accident rates and advocate a systems approach to safety management.9EUROCONTROLSystems thinking for safety: ten principles. A white paper.Moving Towards Safety-II. 2014; (Available from: https://www.skybrary.aero/bookshelf/books/2882.pdf (Accessed 31 May 2020))Google Scholar A systems approach argues that it is unexpected system interactions rather than individuals that are responsible for accidents and safety improvement cannot be achieved simply through more individual training.10Leveson N. Samost A. Dekker S. Finkelstein S. Raman J. A systems approach to analyzing and preventing hospital adverse events.J Patient Saf. 2020; 16: 162-167Crossref PubMed Scopus (33) Google Scholar Following on from that, a systems-approach advocates that interventions for improvement should be focused on better design and system interactions. These should occur alongside person-based interventions such as re-training and local policies which are believed to be less effective for longstanding change and improvement11Braithwaite J. Changing how we think about healthcare improvement.BMJ. 2018; 361: k2014Crossref PubMed Scopus (193) Google Scholar (Fig. 1). Healthcare has been described as having the characteristics of a complex system (Table 1). A complex system is a collection of individual agents (patients, staff, departments) that undertake work that cannot be fully specified or prescribed in advance. These agents' actions are interconnected so that one agent's actions change the context for other agents in the system. Within these complex systems are interactions between people, technologies, environments and external factors. Different systems interact with other systems with different properties, characteristics and healthcare delivery goals, for example an operating theatre and ICU working together during a patient's perioperative care.Table 1Properties of complex systems relevant to healthcarePropertyDescriptionPoorly defined boundaries with other systemsOther hospitals, ambulance trusts, community services influence the functioning of an individual hospital. Healthcare is an open system in which external agents (regulators, media, social care policy, weather, pandemics) can influence its functioning.Membership can change and members can have membership of more than one systemWithin anaesthesia and intensive care this is apparent with rotational and temporary staff, unfamiliar anaesthetists working across different external teams and potentially at very short notice (interventional radiology, endoscopy).Dispersed and decentralised controlDecisions are made constantly by individual agents. An ICU may have different protocols and policies to the rest of a hospital for certain circumstances or conditions (e.g. cardiac arrest, medication practices). Clinicians have autonomy to deviate from policy to accommodate patient complexity and uncertainty. Open table in a new tab These interactions can unpredictably affect the system's capacity, resources and demand, requiring it to adapt to maintain performance. This may be achieved through unplanned workarounds or trade-offs. A workaround or trade-off is a deviation from an intended work process or policy, which is used to overcome an obstacle in order to meet demands, for example not double-checking a medication before it is given. This particular example trades efficiency over thoroughness when throughput and output are the dominant concerns. A workaround may also be necessary to overcome poor design or availability of equipment. A systems approach during an investigation allows an opportunity to understand the role of the system in influencing frontline healthcare workers behaviours. Dekker and Leveson12Dekker S. Leveson N. The systems approach to medicine: controversy and misconceptions.BMJ Qual Saf. 2015; 24: 7-9Crossref PubMed Scopus (29) Google Scholar highlight however that a systems approach should not be misunderstood as blaming the system rather than holding people accountable; neither is it removing their autonomy and creating more policies, rules and checklists within the system. Despite the assertions of the benefits of system-based interventions, it has been difficult to evaluate their impact by the traditional measures that assume a linear relationship between processes and outcomes.11Braithwaite J. Changing how we think about healthcare improvement.BMJ. 2018; 361: k2014Crossref PubMed Scopus (193) Google Scholar Models have emerged to incorporate the science of human factors and ergonomics (HFE) to understand how humans interact with the complex systems in which they work, referred to as work systems. A work system is a system in which humans or machines perform processes or activities using information and technology. Ergonomics and human factors is the scientific discipline concerned with the understanding of interactions among humans and other elements of a system, and the profession that applies theory, principles, data and methods to design in order to understand human well-being and overall system performance.13International Association of Ergonomics. Available from: https://iea.cc/what-is-ergonomics/(accessed 29 December 2020).Google Scholar The systems engineering initiative for patient safety (SEIPS) model illustrates that patient safety risks may develop from several work system factors and the interactions between them.14Holden R.J. Carayon P. Gurses A.P. et al.SEIPS 2.0: a human factors framework for studying and improving the work of healthcare professionals and patients.Ergonomics. 2013; 56: 1669-1686Crossref PubMed Scopus (476) Google Scholar These interactions affect the processes (admission to the ICU, flow in and out of the postoperative care unit) required to deliver safe care. This influence on the processes can produce different outcomes, which can be patient-, staff-, organisation-related (medication safety, staff well-being, organisational reputation). Fig. 2 shows an example of work system elements, processes and outcomes applied to analysing an incident involving insertion of a central venous catheter. Insertion of a central venous catheter is a task that is affected by numerous work system factors and necessary during the process of admitting a patient to the ICU for the management of septic shock. Application of the SEIPS model to the work system in which the incident occurred allows investigators to visualise the numerous interactions that occur which impact the process and outcomes. This allows for an analysis of the work system factors that may have contributed to the incident occurring, thus allowing for a structured approach to developing the terms of reference and scope of an investigation. In the graphical representation of SEIPS, the person is deliberately placed at the centre of the work system, emphasising that design should support people (healthcare worker or patient) and that their performance is affected by interactions between the other work system factors. In terms of incident investigation, SEIPS provides a framework for analysing a particular work system and understanding the emergence of patient safety issues within it. The model has been applied to a number of relevant areas of healthcare including the implementation of electronic health records in ICU.15Hoonakker P.L.T. Cartmill R. Carayon P. Walker J.M. Development and psychometric qualities of the SEIPS survey to evaluate CPOE/HER Implementation in ICUs.Int J Healthc Inf Syst Inform. 2011; 6: 51-69Crossref PubMed Scopus (19) Google Scholar Patient safety approaches have in the past assumed that linear cause-effect chains can explain incidents.3Wiig S. Braithwaite J. Clay-Williams J. It’s time to step it up. Why safety investigations in healthcare should look more to safety science.Int J Qual Health Care. 2020; 32: 281-284Crossref PubMed Scopus (9) Google Scholar They have also assumed that incidents will always have a cause that can be found and that these causes are noticeably different from those actions that allow the achievement of successful incident-free care. This is the basis of the Safety-I approach to safety management which tends to be reactive to incidents.16Hollnagel E. Wears R. Braithwaite J. From Safety-I to Safety-II: a white paper.https://www.england.nhs.uk/signuptosafety/wp-content/uploads/sites/16/2015/10/safety-1-safety-2-whte-papr.pdfGoogle Scholar Accidents and near misses are believed to be the result of deviations from prescribed work, and therefore remedies traditionally focus on increasing compliance and training. However, as already described healthcare is complex and unpredictable and the problems that occur within it cannot be deconstructed into a linear causal chain. There will, however, be instances where the incident complexity and context may allow for linear ‘cause and effect’ thinking. Non-linear approaches to safety question several assumptions made by the Safety-I approach. These assumptions include:•Systems can be easily divided into meaningful elements or activities.•Events occur in a predictable and predetermined sequence.•Incidents occur in a logical and understandable way. The performance and the work within complex systems both vary. Although variability can be thought of as a weakness, contemporary safety models see it as a strength and as the primary reason why complex systems function well despite external pressure. Resilience engineering (RE) is a perspective on safety in complex systems that emphasises how outcomes emerge from the complexity of the clinical environment. Complexity creates the need for flexible adaptation by humans to achieve outcomes in the face of expected and unexpected conditions.17Provan D. Woods D. Dekker Rae A. Safety II professionals: how resilience engineering can transform safety practice.Reliabil Eng Syst Saf. 2020; 195: 106740Crossref Scopus (65) Google Scholar RE focuses on understanding the nature of adaptations, learning from success and increasing adaptive capacity.18Anderson J. Ross A. Back J. et al.Beyond ‘find and fix’: improving quality and safety through resilient healthcare systems.Int J Qual Health Care. 2020; 32: 204-211Crossref PubMed Scopus (10) Google Scholar Indeed, a study using RE principles to examine emergency department escalation policies intended to deal with increased demand, found the policies to be inadequate.19Back J. Ross A. Duncan M. Jaye P. Henderson K. Anderson J.E. Emergency department escalation in theory and practice: a mixed-methods study using a model of organizational resilience.Ann Emerg Med. 2017; 70: 659-671Abstract Full Text Full Text PDF PubMed Scopus (35) Google Scholar It was discovered that escalation processes were adapted to manage pressures informally. This adaptive capacity (work as done) was found to be incompletely specified in policies (work as prescribed/imagined). RE has, particularly in healthcare, come to be known as Safety-II. The Safety-II approach to safety management is proactive and bases its definition of safety on what normally goes right to prevent incidents from occurring. The approach argues that safety exists in the absence of incidents, so the system should be examined at these times to see how it prevents these from occurring. A Safety-II approach sees human performance variability as the reason things go right most of the time and focuses on work-as-done in its study of safety. Humans are seen as an essential resource required for flexibility and resilience. Approaches to analysing the work of humans attempt to discriminate between the assumptions that people have about how work should be done and the observations and descriptions of how work is actually done.20Dekker S. The field guide to understanding human error.3rd Edn. CRC Press, Boca Raton, FL2014Google Scholar Work is imagined by a variety of stakeholders in healthcare including policy makers, regulators, patients and frontline clinicians. It affects the way we expect things to be done and how particularly as anaesthetists and intensivists we integrate into teams who regularly work together without us. Work-as-imagined (WAI) describes what should happen under normal or ideal working conditions but it does not consider how performance must be adjusted to compensate for the changes in working conditions. This inconsistency is not always considered when developing guidelines, policies or work processes (work-as-prescribed) and can lead to inefficiency, misunderstanding and conflict.21Moppett I.K. Shorrock S.T. Working out wrong-side blocks.Anaesthesia. 2018; 73: 407-420Crossref PubMed Scopus (24) Google Scholar Work-as-done (WAD) describes what actually happens and how it is influenced by the complex system within which it occurs. It describes how clinicians make continuous small and large changes to their daily work to satisfy the needs of patient care and efficiency. To maintain efficiency, workarounds may be used. These may include avoiding basic tasks perceived to have low value, skipping middle steps in guidelines or escalating care through non-standard processes. Moppett and Shorrock discuss how thinking about the different types of work can be used when approaching the problem of wrong-sided regional anaesthetic blocks.21Moppett I.K. Shorrock S.T. Working out wrong-side blocks.Anaesthesia. 2018; 73: 407-420Crossref PubMed Scopus (24) Google Scholar They highlight that WAI and work-as-prescribed (the formalisation of work-as-imagined in procedures) does not account for variations in human behaviour when designing the stop-before-you-block initiative. These stop moments rely on an independent second check. The WAI includes a fully engaged assistant who has the confidence, authority and engagement to stop the process and does not consider distractions. The discrepancy between WAI and WAD highlights the necessity to observe how work is performed on ‘the shop floor’ to ensure that recommendations are relevant and appropriately designed. The functional resonance analysis method (FRAM) is a method to explore complex systems and has been used to analyse incidents in many domains.22Hollnagel E. Functional resonance analysis method: modelling complex sociotechnical systems. Ashgate Publishing Limited, Surrey, UK2012Google Scholar FRAM has been applied to healthcare including understanding the variations in the process of sepsis management in different settings.23McNab D. Freestone J. Black C. Carson-Stevens A. Bowie P. Participatory design of an improvement intervention for the primary care management of possible sepsis using the Functional Resonance Analysis Method.BMC Med. 2018; 16: 174Crossref PubMed Scopus (20) Google Scholar,24Raben D. Viskum B. Mikkelsen K. Hounsgaard J. Bogh S. Hollnagel E. Application of a non-linear model to understand healthcare processes: using the functional resonance analysis method on a case study of the early detection of sepsis.Reliabil Eng Syst Saf. 2018; 177: 1-11Crossref Scopus (27) Google Scholar FRAM depicts a system as a number of functions – a function being something the system does in everyday work to achieve an outcome, for example refer to neurosurgical centre leads to decision to operate. The method looks at each function and examines its potential variability in everyday work and unexpected circumstances and helps to describe work-as-done. Variability here might include neurosurgeon occupied in the operating theatre or decision requires more senior input. FRAM provides an opportunity to highlight where control or dampening of variability can be implemented so the system remains predictable and within the boundaries of safe performance. FRAM allows the examination of how one function may affect the other without the need to describe a strict direct-cause relationship. The relationships and interactions of the functions are studied in terms of six aspects (see Supplementary Table 1). The characterisation of these aspects helps define the potential variability of functions and how the functions relate to each other. A FRAM analysis does not attempt to find a cause for an incident but attempts to describe what should happen for the work to succeed.25Hollnagel E. Hounsgaard J. Colligan L. FRAM – the functional resonance analysis method – a handbook for the practical use of the method. Centre for Quality, University of Southern Denmark, 2014Google Scholar Once this is described, the investigator can start to look at how the variability of different functions can combine to explain the incident. By doing this, the system's potential interdependencies is described as opposed to depicting a sequence of individual steps as would be done in a linear model. FRAM can also be used for a proactive analysis of a system to see how potential variability creates risks and judge if a system is fit for purpose. The FRAM method is used to compile a model of the interactions and dependencies between system functions being studied. Supplementary Fig. 1 shows an example of a FRAM model. A key part of FRAM is describing the type of variability relevant to a function. Functions can be divided into three broad categories: human, technological or organisational (Table 2).Table 2Describing variability of functions in FRAM. Each type of function can be affected by Endogenous (internal) and Exogenous (external) variability26The Norwegian healthcare investigation board. https://www.ukom.no/(accessed 29 December 2020).Google ScholarType of functionDescriptionEndogenous variabilityExogenous variabilityHumanCarried out by individuals or small groups. Variability is often of high frequency and amplitude. High amplitude can lead to positive and negative outcomes.Physiological and psychological factors such as fatigue and stress. These may be induced by workload or different working practices.Social factors such as peer pressure, social norms and expectations. Other factors such as public expectation, and standards. Commercial and political considerations can affect human function.TechnologicalEquipment and devices. Technological functions are assumed to be stable and reliable most of the time but it is appreciated they can be variable.Inner workings are complicated and there is inevitable component degradation.Inappropriate maintenance and inappropriate working conditions.OrganisationalOrganisational functions are performed by groups with defined work activities. Organisational variability is thought to be of low frequency but high amplitude when it does happen. This is thought to be that that organisational functions are mostly systemic.Organisational functions are affected by reasons such as communication, authority gradient and culture.Environmental factors: regulation, public requirements, financial pressure, weather, politics. Open table in a new tab Both SEIPS and FRAM are models that have been used by HSIB to investigate patient safety incidents. An initial scoping investigation into the local circumstances of the incident is carried out before broadening the investigation nationally to ascertain the scale of the issue and how the healthcare system interacts at different levels in relation to the type of incident investigated. A system-based investigation analyses the numerous levels of influence in healthcare such as government policy, regulation, commissioning, local-level management and infrastructure, frontline care providers and the patient perspective. HSIB also undertakes a maternity investigation programme, which has specific criteria on which it bases its decision to investigate. Observational work across the country is a core part of the national investigation process of the HSIB. This is a way of understanding the trade-offs people apply in order to resolve goal conflicts and to cope with the complexity of the system and the uncertainty of the environment. Investigators work closely with expert subject matter advisors who help with understanding work system behaviour and work-as-done from people who are part of the system but removed from the incident itself. At the end of an investigation, system-level recommendations are made to regulators, government departments and professional bodies. Both Norway and South Korea have also established national independent bodies for healthcare safety investigation. These are the Norwegian Healthcare Investigation Board (UKOM) and the Patient Safety Headquarters, Korea Institute for Healthcare Accreditation respectively.26The Norwegian healthcare investigation board. https://www.ukom.no/(accessed 29 December 2020).Google Scholar,27The patient safety Headquarters, Korea Institute for Healthcare Accreditation https://www.koiha.or.kr/web/en/index.do (accessed 29 December 2020).Google Scholar An individual hospital or department does not have the resources or powers to make national system-level recommendations. However, within its new PSIRF, NHS England and NHS Improvement, in March 2020, published its national standards for patient safety incident investigation (PSII) in which it advocates ‘strong/effective systems-based improvements to prevent or significantly reduce the risk of a repeat incident’.28NHS England and NHS ImprovementNational standards for patient safety investigation guiding principles and standards for a local, systems approach to patient safety investigation in NHS-funded care.2020Google Scholar The document also recommends a selective approach to PSIIs. This selection will be based on risk and learning potential and not on a particular incident severity outcome. No particular patient safety investigation method has been advocated; however, it promotes ‘analysis techniques that facilitate a systems approach’ and a methodology that ‘identifies system strengths as well as problems’. The approaches and methods described above are all grounded in these core principles and potentially may form the basis of local patient safety incident responses. The authors thank Dr Laura Pickup for her input into the FRAM section of this article. At the time of writing, all three authors were employed by the Healthcare Safety Investigation Branch. The associated MCQs (to support CME/CPD activity) will be accessible at www.bjaed.org/cme/home by subscribers to BJA Education. The following are the Supplementary data to this article: Download .docx (.07 MB) Help with docx files Multimedia component 1 Download .docx (.02 MB) Help with docx files Multimedia component 2 Paul Sampson FRCA is a specialty registrar in anaesthesia at University Hospitals Plymouth and spent a year as a clinical fellow with the Healthcare Safety Investigation Branch. Jonathan Back PhD is an intelligence analyst with the Healthcare Safety Investigation Branch. He has a background in safety science and systems thinking in healthcare. Stephen Drage FRCA FFICM DipICM (UK) MSc is a consultant in anaesthesia and intensive care medicine at Brighton and Sussex University Hospitals NHS Trust. He is also director of investigations at the Healthcare Safety Investigation Branch.

Background and Aims: Antibodies targeting tumor necrosis factor-alpha [TNF-alpha] are a mainstay in the treatment of inflammatory bowel disease. However, they fail to demonstrate efficacy in a considerable proportion of patients. On the other hand, glycosylation of antibodies might influence not only their immunogenicity but also their structure and function. We investigated whether specific glycosylation patterns of the Fc-fragment would affect the immunogenicity of anti-TNF-alpha antibody in monocyte-derived dendritic cells. Methods: The effect of a specific Fc-glycosylation pattern on antibody uptake by monocyte-derived dendritic cells [mo-DCs] and how this process shapes the immunologic profile of mo-DCs was investigated. Three N-glycoforms of the anti-TNF-alpha antibody adalimumab, that differed in the content of fucose or sialic acid, were tested: [1] mock treated Humira, abbreviated 'Fuc-G0', where the N-glycan mainly consist of fucose and N-acetylglucosamine [GlcNAc], without sialic acid; [2] 'Fuc-G2S1/G2S2' with fucose and alpha 2,6 linked sialic acid; and [3] 'G2S1/G2S2' with alpha 2,6 linked sialic acid, without fucose. Results: Our data demonstrated that neither fucosylation nor sialylation of anti-TNF-Abs [Fuc-G0, FucG2S1/G2S2, G2S1/G2S2] influence their uptake by mo-DCs. Additionally, none of the differentially glycosylated antibodies altered CD80, CD86, CD273, CD274 levels on mo-DCs stimulated in with lipopolysaccharide in the presence of antibodies. Next, we evaluated the levels of cytokines in the supernatant of mo-DCs stimulated with lipopolysaccharide in the presence of Fuc-G0, Fuc-G2S1/G2S2 or G2S1/G2S2-glycosylated anti-TNF antibodies. Only IL-2 and IL-17 levels were downregulated, and IL-5 production was upregulated by uptake of Fuc-G0 antibodies, as compared to control without antibodies. Conclusions: The specific modification in the Fc-glycosylation pattern of anti-TNF-alpha Abs does not affect their immunogenicity under the tested conditions. As this study was limited to mo-DCs, further investigation is required to clarify whether Ab uptake into mo-DCs might change the immunological profile of T- and B-cells, in order to ultimately reduce the formation of anti-drug antibodies and to improve the patient care.

INTRODUCTION:We sought to provide the first report of the use of NEWS2 monitoring to pre-emptively identify clinical deterioration within hospitalised COVID-19 patients.METHODS:Consecutive adult admissions with PCR-confirmed COVID-19 were included in this single-centre retrospective UK cohort study. We analysed all electronic clinical observations recorded within 28 days of admission until discharge or occurrence of a serious event, defined as any of the following: initiation of respiratory support, admission to intensive care, initiation of end of life care, or in-hospital death.RESULTS:133/296 (44.9%) patients experienced at least one serious event. NEWS2 ≥ 5 heralded the first occurrence of a serious event with sensitivity 0.98 (95% CI 0.96-1.00), specificity 0.28 (0.21-0.35), positive predictive value (PPV) 0.53 (0.47-0.59), and negative predictive value (NPV) 0.96 (0.90-1.00). The NPV (but not PPV) of NEWS2 monitoring exceeded that of other early warning scores including the Modified Early Warning Score (MEWS) (0.59 [0.52-0.66], p<0.001) and quick Sepsis Related Organ Failure Assessment (qSOFA) score (0.58 [0.51-0.65], p<0.001).CONCLUSION:Our results support the use of NEWS2 monitoring as a sensitive method to identify deterioration of hospitalised COVID-19 patients, albeit at the expense of a relatively high false-trigger rate.

Objective: The aim was to develop a method based on resilient healthcare principles to proactively identify system vulnerabilities and quality improvement interventions. Design: Ethnographic methods to understand work as it is done in practice using concepts from resilient healthcare, the Concepts for Applying Resilience Engineering model and the four key activities that are proposed to underpin resilient performance-anticipating, monitoring, responding and learning. Setting: Accident and Emergency Department (ED) and the Older People's Unit (OPU) of a large teaching hospital in central London. Participants: ED-observations 104 h, and 14 staff interviews. OPU-observations 60 h, and 15 staff interviews. Results: Data were analysed to identify targets for quality improvement. In the OPU, discharge was a complex and variable process that was difficult to monitor. A system to integrate information and clearly show progress towards discharge was needed. In the ED, patient flow was identified as a complex high-intensity activity that was not supported by the existing data systems. The need for a system to integrate and display information about both patient and organizational factors was identified. In both settings, adaptive capacity was limited by the absence of systems to monitor the work environment. Conclusions: The study showed that using resilient healthcare principles to inform quality improvement was feasible and focused attention on challenges that had not been addressed by traditional quality improvement practices. Monitoring patient and workflow in both the ED and the OPU was identified as a priority for supporting staff to manage the complexity of the work.

In this chapter, we outline the prerequisites for an effective response to patient flow pressures and discuss two case studies of work practices that have evolved to manage flow across organisational boundaries. An argument is presented calling for a better coordinated, more effective and sustainable approach to patient flow transitions across boundaries.

Background: GBR 830 is a humanized mAb against OX40, a costimulatory receptor on activated T cells. OX40 inhibition might have a therapeutic role in T cell-mediated diseases, including atopic dermatitis (AD). Objective: This exploratory phase 2a study investigated the safety, efficacy, and tissue effects of GBR 830 in patients with AD. Methods: Patients with moderate-to-severe AD (affected body surface area, >= 10%; Eczema Area and Severity Index score, >= 12; and inadequate response to topical treatments) were randomized 3:1 to 10 mg/kg intravenous GBR 830 or placebo on day 1 (baseline) and day 29. Biopsy specimens were collected (n = 40) at days 1, 29, and 71. Primary end points included treatment-emergent adverse events (TEAEs) and changes from baseline in biomarkers (epidermal hyperplasia/cytokines) at days 29 and 71. Results: GBR 830 was well tolerated, with equal TEAE distribution (GBR 830, 63.0% [29/46]; placebo, 63.0% [10/16]). One serious TEAE in the GBR 830 group was deemed unrelated to study drug. At day 71, the proportion of intent-to-treat subjects achieving 50% or greater improvement in Eczema Area and Severity Index score was greater with GBR 830 (76.9% [20/26]) versus placebo (37.5% [3/8]). GBR 830 induced significant progressive reductions in T(H)1 (IFN-gamma/CXCL10), T(H)2 (IL-31/CCL11/CCL17), and T(H)17/T(H)22 (IL-23p19/IL-8/S100A12) mRNA expression in lesional skin. Significant progressive reductions until day 71 in the drug group were seen in OX40(+) T cells and OX40L(+) dendritic cells (P < .001). Hyperplasia measures (thickness/keratin 16/Ki67) showed greater reductions with GBR 830 (P < .001). Conclusions: Two doses of GBR 830 administered 4 weeks apart were well tolerated and induced significant progressive tissue and clinical changes until day 71 (42 days after the last dose), highlighting the potential of OX40 targeting in patients with AD.

Current approaches to quality improvement rely on the identification of past problems through incident reporting and audits or the use of Lean principles to eliminate waste, to identify how to improve quality. In contrast, Resilience Engineering (RE) is based on insights from complexity science, and quality results from clinicians' ability to adapt safely to difficult situations, such as a surge in patient numbers, missing equipment or difficult unforeseen physiological problems. Progress in applying these insights to improve quality has been slow, despite the theoretical developments. In this chapter we describe a study in the Emergency Department of a large hospital in which we used RE principles to identify opportunities for quality improvement interventions. In depth observational fieldwork and interviews with clinicians were used to gather data about the key challenges faced, the misalignments between demand and capacity, adaptations that were required, and the four resilience abilities: responding, monitoring, anticipating and learning. Data were transcribed and used to write extended resilience narratives describing the work system. The narratives were analysed thematically using a combined deductive/inductive approach. A structured process was then used to identify potential interventions to improve quality. We describe one intervention to improve monitoring of patient flow and organisational learning about patient flow interventions. The approach we describe is challenging and requires close collaboration with clinicians to ensure accurate results. We found that using RE principles to improve quality is feasible and results in a focus on strengthening processes and supporting the challenges that clinicians face in their daily work.

Background: GBR 1302 is a HER2xCD3 bispecific antibody engineered to direct T-cells to HER2expressing tumor cells. This ongoing firstinhuman study (NCT02829372) in subjects with HER2positive cancers aims to evaluate the safety, tolerability, and preliminary efficacy of GBR 1302. Methods: Adults with HER2-positive (immunohistochemistry 2+ or 3+) solid tumors with no available standard treatment receive GBR 1302 on Day 1 and Day 15 in 28-day treatment cycles at escalating dose levels, starting at 1 ng/kg. The primary endpoint includes determination of the maximum tolerable dose and safety profile of GBR 1302. Secondary and exploratory endpoints include pharmacodynamic (PD) testing for modulation of cellular and cytokine biomarkers. Results: To date, 19 evaluable subjects for dose limiting toxicity (DLT) have been treated up to a dose of 750 ng/kg; dose escalation is ongoing. Grade (G) 1 to 2 infusion related reaction (IRR)/cytokine release syndrome (CRS) is the most comment treatment emergent adverse event that has been observed in subjects treated at doses ≥100 ng/kg. The majority of subjects were managed with conservative treatment. 2 subjects experienced DLT events: one asymptomatic subject (100 ng/kg) was noted to have reduced left ventricular ejection fraction on routine echocardiogram at 4 weeks, which resolved spontaneously after treatment discontinuation; the second subject (500 ng/kg) experienced G4 IRR/CRS which required ICU care but resolved within 36 hours. Beginning at 30 ng/kg, CD3, CD4, and CD8 positive T-cell populations decreased within 6 hours of administration and recovered to levels at or above baseline by 48 hours. Dose-proportional, transient increases in cytokines (IL-2, IL-6, IL-10, IFN-γ, TNF-α), which peaked at 6 hours and began to normalize within 48 hours, were observed. No subjects have documented radiological response, but 2 subjects (HER2 3+ gastroesophageal adenocarcinoma and HER2 2+ breast adenocarcinoma) have prolonged disease stabilization lasting ≥4 months. Conclusions: The combination of clinical findings and PD changes suggests T-cell activation with higher dosages of GBR 1302. Dose escalation is continuing and updated results will be presented. Clinical trial identification: NCT02829372. Editorial acknowledgement: Editorial assistance provided by Ashley Skorusa, PhD of Prescott Medical Communications Group (Chicago, IL) and funded by Glenmark Pharmaceuticals S.A., Switzerland. Legal entity responsible for the study: Glenmark Pharmaceuticals, SA. Funding: Glenmark Pharmaceuticals, SA. Disclosure: M. Wermke: Honoraria: BMS, Novartis, Roche, Bayer, Glenmark, AstraZeneca; Travel reimbursements: AstraZeneca, BMS, MSD, Novartis, Glenmark; Research funding: Novartis, Pfizer. J. Kauh, Y. Salhi, V. Reddy: Employee: Glenmark Pharmaceuticals, Inc. J. Back: Employee: Glenmark Pharmaceuticals, SA. S. Ochsenreither: Payment for scientific advice: Glenmark. All other authors have declared no conflicts of interest.

GBR 830 is a first-in-class, humanized, monoclonal IgG1 antibody that specifically inhibits OX40, a costimulatory receptor on activated T cells. Three studies have been completed to evaluate the pharmacokinetics (PK) and immunogenicity of GBR 830. A single ascending dose study with GBR 830 (0.3, 1, 3, and 10 mg/kg) by intravenous (IV) infusion was conducted in healthy volunteers. GBR 830 was well tolerated with no clinically significant findings. GBR 830 showed linear PK with dose proportional increases in Cmax and AUC. An early Cmax (median Tmax 1.5-4 hours) and a bi-exponential decline with a long terminal elimination phase (T1/2 10-15 days) was observed without discernable influence of target-mediated clearance. Six of 34 GBR 830-treated subjects were positive for anti-drug antibody (ADA), 2 of whom had neutralizing ADA. An absolute bioavailability (%F) study was conducted in healthy adults with a single dose of GBR 830 by IV (600 mg/subject) or subcutaneous (SC) administration (75 or 600 mg/subject). The %F of GBR 830 by the SC route was ∼65%, with Cmax achieved around 5 days post-dose. A lower incidence of ADA was observed with higher doses (600 mg SC: 1/15 subjects; 600 mg IV: 1/10 subjects) compared to the lower dose (75 mg SC: 10/15 subjects). PK of GBR 830 was also evaluated in subjects with moderate-to-severe atopic dermatitis (AD) (NCT02683928). Two IV doses of GBR 830 (10 mg/kg; 4 weeks apart) in subjects with AD showed minimal accumulation in AUC0-tau (1.22-fold). Six of 46 GBR 830-treated AD subjects were ADA positive. In conclusion, GBR 830 was well tolerated and showed a similar PK profile in healthy volunteers and subjects with AD. A favorable linear PK profile with a long half-life, high bioavailability, and no evidence of target-mediated disposition was observed. ADA generation had no discernable effect on PK and safety.

TPS81 Background: Available therapies have improved outcomes in multiple myeloma but patients eventually relapse, requiring treatment with agents that are active in refractory disease. CD38, a transmembrane glycoprotein, is upregulated on myeloma cells and is a validated disease target, evidenced by the anti-myeloma activity of daratumumab, an anti-CD38 human IgG1κ monoclonal antibody. GBR 1342 is a CD3xCD38 bispecific antibody engineered (using Glenmark’s BEAT® platform) to direct T-cells to CD38-expressing myeloma cells. In preclinical studies, GBR 1342 redirected the cytotoxic potential of T-cells to human myeloma cell lines in vitro and in mouse xenograft models. This ongoing, 2-part, first-in-human study aims to: (1) evaluate the safety profile and maximum tolerable dose (MTD) of GBR 1342 monotherapy in subjects with relapsed/refractory multiple myeloma ( > 3 prior therapies); and (2) further elucidate the safety, tolerability, and preliminary clinical activity of GBR 1342 at the MTD. The study is also evaluating the mechanisms by which GBR 1342 redirects T-cells to tumor and enhances cytolytic activity of cytotoxic T-cells. Methods: In Part 1, intravenous GBR 1342 is administered on Day 1 and Day 15 in 28-day treatment cycles at escalating dose levels (Table). The first 4 cohorts consist of a single subject. Subsequent cohorts will enroll using a 3+3 design. In Part 2, 65 evaluable subjects will be treated at the MTD identified in Part 1 until disease progression or unacceptable toxicity occurs. Primary endpoints include frequency and severity of AEs, number of dose-limiting toxicities during Cycle 1 (Part 1 only), and objective response to GBR 1342 (Part 2 only). Secondary endpoints include pharmacokinetics and anti-tumor activity of GBR 1342 (objective response, progression-free and overall survival). [Table: see text]

This chapter describes a study of resilience in the Older Persons' Unit (OPU) of a large London teaching hospital in which they developed practical tools to study resilience and identify potential quality improvement initiatives. It reports initial results from the OPU site to illustrate how the author have used Resilience Engineering (RE) principles to inform quality improvement. Despite increasing interest in the principles of RE there is little guidance available for applying the ideas in practice in health care. The Concepts for Applying RE theoretical model was developed and used to design data collection instruments, analysis methods and interpretation of the data. Concepts from the RE literature were also identified: goal trade-offs, learning from what goes right and the four resilience abilities of responding, monitoring, anticipating and learning. The narratives coded with RE theoretical concepts were analysed to identify opportunities for improvement.

OX40 is a costimulatory receptor member of the NGFR/TNFR superfamily expressed predominantly on activated T cells. Ligation of OX40 by its ligand OX40L leads to enhanced T cell survival, proliferation, and effector functions. Blocking the OX40/OX40L pathway is therefore highly attractive to treat a broad range of T cell-mediated autoimmune diseases. While several OX40 agonist antibodies are under development in oncology, generating an OX40 antagonist devoid of agonist activity remains a challenge. GBR 830 is a humanized IgG1 targeting OX40 with a monovalent affinity for human OX40 (∼90 nM as measured by surface plasmon resonance) and cross-reactivity to macaque OX40 albeit with a lower affinity. However, its apparent affinity drastically increases when GBR 830 binds bivalently. GBR 830 blocks OX40L binding and inhibits OX40L-mediated T cell proliferation at a low nM concentration. It also mediates low levels of antibody dependent cellular cytotoxicity and complement dependent cytotoxicity. Importantly, GBR 830 was evaluated for residual agonism by assessing its costimulatory effect on the proliferation of purified T cells from multiple donors. Compared to OX40L or anti-CD28 positive controls, GBR 830 did not stimulate T cells with or without addition of a crosslinking antibody. In a more sensitive experimental setup in which anti-OX40 antibodies were co-coated with an anti-CD3 antibody, no agonism was detected with GBR 830, whereas all other anti-OX40 antibodies tested showed agonism. When administered to cynomolgus monkeys in the repeat-dose intravenous/subcutaneous toxicity studies (6-weeks or 6-months duration), GBR 830 was well tolerated without any adverse findings. The no-observed-adverse-effect-level was 100 mg/kg/week. These data show that GBR 830 is able to block OX40L-induced proliferation without inducing receptor agonism, in contrast to other anti-OX40 antibodies.

Abstract not available. Disclosures: Study sponsored by Glenmark Pharmaceuticals. Copyright 2018 SKIN

Background: HER2 is dysregulated in a wide range of solid tumors, including breast cancer, and is an attractive target for tailored oncologic treatment. GBR 1302 is a HER2xCD3 bispecific antibody that redirects cytotoxic T-cells to kill HER2 overexpressing cancer cells. This unique mode of action is anticipated to result in superior antitumor activity in HER2-positive tumors by harnessing the cytotoxic capabilities of patients’ existing T-cells. Methods: This ongoing, phase 1, first-in-human, open-label, multicenter, dose-escalation study is evaluating GBR 1302 in adults with progressive HER2-positive solid tumors for which no standard or curative treatment is available. Subjects receive intravenous GBR 1302 on Day 1 and Day 15 in 28-day treatment cycles at escalating dose levels, starting at 1 ng/kg. The first 4 cohorts consisted of a single subject; subsequent cohorts are being enrolled using a 3 + 3 design. Blood samples were collected for pharmacokinetic (PK) and anti-drug antibody (ADA) analyses (secondary endpoints). Quantification of GBR 1302 serum concentrations (for PK) and detection/confirmation of anti GBR 1302 antibodies (for immunogenicity) were performed using validated LC/MS/MS and ELISA methods, respectively. PK parameters were evaluated using standard non-compartmental methods. Results: As of 21 August 2018, PK data were available from 31 subjects over dose range of 1 ng/kg to 750 ng/kg. Serum concentrations were less than the lower limit of quantification of 50 pg/mL at the first dose (1 ng/kg), and only transient concentrations were observed at 3 and 10 ng/kg dose levels. Evaluable PK profiles were observed from 30 ng/kg onwards. GBR 1302 showed maximum serum concentration (Cmax) around the end of infusion, after which serum concentrations declined bi-exponentially with a mean terminal half-life of around 4 to 7 days. Both Cmax and area under the curve (AUC0-t) showed a near dose-proportional increase up to 750 ng/kg (maximum evaluated dose). None of the samples collected from subjects up to cohort 5 showed positive ADA response. Conclusions: Per ongoing analysis, GBR 1302 showed a favorable, linear PK. None of the subjects evaluated so far showed positive ADA response. Editorial acknowledgement: Editorial assistance was provided by Jacqueline Benjamin, PhD of Prescott Medical Communications Group, Chicago, IL. Clinical trial identification: NCT02829372. Legal entity responsible for the study: Glenmark Pharmaceuticals SA. Funding: Glenmark Pharmaceuticals SA. Disclosure: G. Gudi, E. Fluhler: Employee of Glenmark Pharmaceuticals Inc., V. Ca, S. Gn: Employee of Glenmark Pharmaceuticals, Ltd., C. von Gunten, J. Back: Employee of Glenmark Pharmaceuticals SA.