The Power Of Connections: Mapping the Behaviour of Health Care Networks

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Datasets: + Emergency Care Data Set (ECDS)

Objective for processing

This Data Sharing Agreement permits the retention and processing of the data provided under previous iterations of this Agreement for an interim period. The purpose of this Agreement is to extend the scope of approval and to use data previously disseminated alongside newly requested data, to create new databases that aim to provide information on the flow of patients between hospitals. Specifically, Imperial College London are requesting more data (Hospital Episode Statistics (HES) Admitted Patient Care (APC), HES Outpatients (OP) and HES Accident and Emergency (A&E) from 2018/19 to 2019/20 and the inclusion of the Emergency Care Dataset (ECDS) from 2020/21 to 2021/22 inclusive) as well as the retention and processing of the data provided under previous iterations of this Agreement (HES APC, HES OP and HES A&E from 2011/12 to 2017/18). The following describes the purposes for which the data was disseminated under the previous DSA. The purpose of this Agreement is to investigate the extent to which providers of hospital care in England are connected to one another, and the extent to which clinical care for patients is fragmented between providers. This proposed project is an extension of the work conducted under previous iterations of this Agreement and builds extensively upon the findings of this prior work. The Department of Surgery and Cancer, based at Imperial College London, is requesting data for use in the following research project: The Power of Connections: Mapping the Behaviour of Health Care Networks. What data is required: The purpose of this study is to examine how care providers in England are connected by virtue of the patients that flow between them. This request for data will, through the application of network analysis, provide insights into the factors determining how patients flow through the network and where the network may be particularly vulnerable will be identified. All data in this study will be record level pseudonymised data. Strategies to improve the efficiency, equity and safety of the network may be developed and tested using predictive modelling in order to identify to optimal routes for investment and restructuring of the health care providers. - HES OP 2011/12 to 2021/22 inclusive This project will use the following data: HES OP 2011/12-2017/18, HES A&E 2011/12-2017/18 and HES APC 2011/12-2017/18. These years of data are necessary to provide an adequate picture of health care utilization and capture less common events. - HES A&E 2011/12 to 2019/20 The use of HES data from 2011/12 to 2017/18 will facilitate an accurate capture of trends in patient-sharing networks and a temporal analysis of changes to patterns. - ECDS from 2019/20 to 2021/22 inclusive Data is also requested to identify recent provider care networks within and between hospitals. - HES APC 2011/12 to 2021/22 inclusive This data will allow Imperial College to ensure outputs from the programme of work closely reflects current practice, in a healthcare environment where care pathways and networks are constantly evolving. ECDS data are requested in addition to HES A&E data owing to HES A&E data being discontinued in March 2020. This is a retrospective observational study and does not consist of control or cohort groups. The dataset may be considered a single longitudinal cohort of patients. No cohort will be submitted to NHS England by Imperial College London for the purpose of this Agreement. The cohort for this study is all patients receiving inpatient, outpatient or A&E care in England from 2011 to 2021/22 data period. HES data will also allow the study to analyse how health networks have evolved over the period of data received, and explore reasons for this. This will also allow the study to better understand current trends in health networks and predict future trends more accurately which is important for health policy. Why the data is required for this purpose: Data held for this is all part of one overarching project regarding networks of care and all researchers involved are in the same study theme. - As this study aims to construct patient-sharing networks and to calculate the extent to which healthcare is fragmented at the level of individual patients, is it necessary that record level data is used. - The study specifically aims to understand the extent to which care fragmentation and patient-sharing networks have changed over time. Some changes during the period of requested data, such as the COVID-19 pandemic, are expected to have immediate impact on this, while other changes, such as the gradual shift towards greater integration between primary and secondary care are expected to influence fragmentation and patient networks at a slower pace. As such, data from 2011/12 to 2021/22 are requested to determine temporal changes in these features. - The use of HES data will hopefully facilitate an accurate capture of trends in patient-sharing networks and a temporal analysis of changes to patterns. The data will allow investigation of longitudinal trends in continuity of patient care and patient sharing networks and will ensure that findings are as relevant as possible to current and future patterns of patient-sharing. - Using the latest available data will ensure the output is as relevant as possible to current and future patterns of care. - The NHS in England consists of many provider organisations who provide care to patients irrespective of where they live. Previous research has shown that patient-sharing networks in the NHS cannot be reliably constructed using data from only part of the country. It is for this reason that data for all patients is requested, rather than just those living in a particular part of the country or attending a particular subset of provider organisations. Efforts taken to minimise the data required: - The data fields requested in this Agreement has been reduced significantly (approximately 40%) compared to the previous iteration as a result of experience working with the data extracts. - The variables requested are largely limited to patient demographic data, clinical organisation data and the dates of clinical attendances as well as any associated diagnoses or procedures. This is the minimum dataset required to enable the proposed analyses described above. - As the study requires record level longitudinal records to construct patient-sharing networks and measures of care fragmentation, it is not possible to restrict the analysis to a random subgroup of patients or locations, as this will significantly bias results to the extent they are likely to be unreliable. Organisations involved: - Imperial College London is solely responsible for decisions made to determine how and why the data is used for this study, hence is the sole Data Controller who also process the data. - The Wellcome Trust provide funding but are not involved in the study in any way - No other organisations are involved in the wider project. The databases created using the NHS England data are detailed below. This data will allow Imperial College London to ensure outputs from the programme of work closely reflects current practice, in a healthcare environment where care pathways and networks are constantly evolving. HES data will also allow the study to analyse how health networks have evolved over the periods of data received, and explore reasons for this. This will also allow the study to better understand current trends in health networks and predict future trends more accurately which is important for health policy. The data disseminated under this Agreement will be used to form the following multiple databases that will be used to achieve the objectives of this study: 1. A patient level database for inter hospital transfers of patients, in addition to a range of utilisation and outcome variables (e.g. length of stay, additional procedures, readmissions). Directed networks of care transitions from one provider to another will be constructed and the characteristics of the network, providers and patients will be analysed using linear and logistic regression. 2. A patient level database of presentations to acute hospitals. This database will be used to identify the probability of presentation of a patient in a particular geographic location to a particular centre with a particular diagnosis. These values will be used to undertake computational community detection algorithms to identify geographically nested networks of care providers to compare to existing predetermined care networks and guide the implementation of novel, more efficient networks of care. 3. A patient level database of outpatient, inpatient and A&E presentations. The aggregate interaction across datasets between a particular geographic region (e.g. postcode district, lower super output area [LSOA] (or primary care provider) and a secondary care institution will be used to generate a unipartite network of acute hospitals linked to one another by the strength of their shared patient activities. The network will then be interrogated to identify how patterns of patient flow will change in response to increased patient demand and altered provider capacity. Clusters of vulnerability in the network will be identified and optimal avenues for intervention will be suggested. 4. A patient level database of multi-morbidity and complex disease processes will be constructed to examine care fragmentation between hospitals. Networks of care transitions will be constructed and the characteristics of the network, providers and patients will be analysed. 5. A provider level database of presentations to acute hospitals will be constructed using the PCONSULT HES data field. This is a pseudonymised version of the consultant code. Aggregate presentations will be used to identify differences in admission patterns and transitions of care between hospitals. This will be interrogated to identify resultant differences in patient outcomes, including morbidity, mortality and length of stay between hospitals with different provider admission patterns. In response to the findings of this study to date, further research questions have arisen which the study team seek to address through the additional dissemination of pseudonymised record-level data requested under this Agreement. Imperial College London also propose that additional databases will be created using updated HES data as detailed below: 6. A longitudinal database of the extent of care fragmentation for individual patients over time from 2011 to 2021/22 data. Temporal variation will be examined at the level of hospitals, Lower Layer Super Output Areas and with respect to the clinical and demographic features of patients. 7. The database described in point 1 will be extended to incorporate the period from April 2018 to 2021/22 data and thereby examine the fragmentation of clinical care for patients with multimorbidity before and during the Covid-19 pandemic. Networks of care transitions will be constructed, and characteristics of the network, providers and patients will be analysed. 8. A longitudinal database of outpatient clinical appointments will be constructed to examine changes in the fragmentation of outpatient care following the implementation of Primary Care Networks. Under GDPR, the lawful basis on which processing of data from NHS England concerning the proposed study is laid out by Articles 6(1)(e) and 9(2)(j) namely that processing is necessary for the performance of a task carried out in the exercise of official authority, such authority to make provision for research being vested in the University by virtue of the Royal Charter establishing the University dated 1907, and, that in respect of special category personal data, processing is necessary for research purposes carried out in accordance with Article 89(1) of the GDPR. Imperial College London are a public authority and the outcomes of this research are for the benefit of public interest. Two PhDs have been completed using data from this Data Sharing Agreement however the PhD studies were not the primary purpose under this Agreement. The first PhD thesis was titled "The power of connections: mapping the behaviour of healthcare networks”. It focussed on using network analysis methods to identify how patients flow move between hospitals in the NHS and what that would mean for how services could be organised differently. The PhD was awarded in August 2020 (https://doi.org/10.25560/82123). The second thesis was titled "Transitions of care in the digital age: measuring and improving transitions of care using big data and digital technology” and looked primarily from a qualitative perspective at what transitions of care might mean for patients and healthcare providers. The second PhD was awarded in February 2022 (https://doi.org/10.25560/97018).

Processing activities

This Data Sharing Agreement permits the retention and processing of the data provided under previous iterations of this Agreement for an interim period. No additional data will flow under v.3 of this Agreement. [1 paragraph unchanged] The Department of Surgery and Cancer confirms that the data under this application would only be used for the project described in this data sharing agreement. Individuals working on this project would only be permitted to access data relating to that project, as identified within the application. Access is granted to the data only to named individuals working on the project under authorized user names. Such access is password controlled (with a password reset required on a regular refresh). Only substantive employees of Imperial College London will use the data disseminated by NHS Digital and only for the purposes described in this document. No data will flow from Imperial College London to NHS England. The raw data will be handled only within the Department of Surgery and Cancer to support academic research. Pseudonymised, record-level HES APC, HES OP, HES A&E and ECDS data will be disseminated from NHS England to Imperial College London. The variables requested are largely limited to patient demographic data, clinical organisation data and the dates of clinical attendances as well as any associated diagnoses or procedures. The data will be received from NHS Digital and stored on Imperial's Big Data and Analytical Unit's Secure Environment (BDAU SE) hosted at Imperial College London's data centre in Slough. Access to data on this server is restricted to authorized individuals only. The BDAU SE is ISO 27001 certified and compliant with NHS Digital Data Security and Protection Toolkit. It can be accessed remotely by users using multi-factor authentication once the user registration process is completed. Data will only be provided once the appropriate dataset registration process is completed. To access the data in the BDAU SE, researchers will require approval from the study chief investigator (in their role as Asset Owner). Once approval is given, researchers are required to: 1) pass the Imperial Data Protection Awareness, including GDPR, and Information Security Awareness training 2) agree, with possible disciplinary action for noncompliance, to BDAU ISO 27001 certified standard operating procedures by signing the BDAU SE User and Dataset Registration forms. All training and forms are required to be renewed on an annual basis. Users who do not pass the annual renewal of training and forms will be automatically removed from access. The de-identified data transferred to the BDAU SE from NHS Digital will be held in the BDAU SE as a standalone dataset. The research team will analyse data within the BDAU SE in the allocated BDAU SE dataset. Data will not be transferred outside of the BDAU SE in raw format for any reason. Only summary research outputs that are aggregated with small numbers suppressed in line with the HES Analysis Guide can be transferred outside of the BDAU SE. The data will be stored on Imperial's Big Data and Analytical Unit's Secure Environment (BDAU SE) hosted at Imperial College London's data centre in Slough. Record level data will not be distributed to different parts of the organization. The data will not be made available to third party individuals, institutions, or companies. No other data will be linked to this data though data will be compared at aggregate levels if required. The data will be processed as part of the above-mentioned research project within the Department of Surgery and Cancer. It will be queried using data analytical tools and software to aid in answering specific research questions. Data visualizations will be done to present insights gained using suitable tools. Access to data on this server is restricted to authorized individuals only. The analysis will only be conducted by substantive employees of Imperial College London who are trained in Data Protection including GDPR and Information Security Awareness before they can access and analyse the data. Due to the COVID-19 pandemic and changes in researchers’ working circumstances, data will be accessed and analysed remotely (via a screen view to the Imperial’s BDAU SE from within England and Wales. This is via a VPN connection by users using Multi-Factor Authentication (MFA) once the user registration process is completed. Data will only be provided once the appropriate dataset registration process is completed. The specific processing activities will be as follows: To access the data in the BDAU SE, researchers will require approval from the study Chief Investigator (in their role as Asset Owner). Once approval is given, researchers are required to: 1) pass the Imperial Data Protection Awareness, including GDPR, and Information Security Awareness training 2) agree, with possible disciplinary action for noncompliance, to BDAU ISO 27001 certified standard operating procedures by signing the BDAU SE User and Dataset Registration forms. All training and forms are required to be renewed on an annual basis. Users who do not pass the annual renewal of training and forms will have access revoked automatically. The BDAU SE is ISO 27001 certified and compliant with NHS England Data Security and Protection Toolkit. 1. A patient level database for inter hospital transfers of patients will be constructed, in addition to a range of utilisation and outcome variables (e.g. length of stay, additional procedures, readmissions). A patient level database of maternity care will be constructed: to examine patient choice in relation to delivery location. In both cases, directed unipartite networks of care transitions from one provider to another will be constructed and the characteristics of the network, providers and patients will be analysed using linear and logistic regression. The pseudonymised HES/ECDS data transferred to the BDAU SE from NHS England will be held in the BDAU SE as a standalone dataset. The research team will analyse data within the BDAU SE in the allocated BDAU SE dataset. Data will not be transferred outside of the BDAU SE in raw format for any reason. Only summary research outputs that are aggregated with small numbers suppressed in line with the HES Analysis Guide can be transferred outside of the BDAU SE. Record level data will not be distributed to different parts of the organisation. The NHS England data will not be made available to third party individuals, institutions, or companies not disclosed in this Agreement. 2. A patient level database of presentations to acute hospitals will be constructed. This database will be used to identify the probability of presentation of a patient in a particular geographic location to a particular centre with a particular diagnosis. These values will be used to undertake computational community detection algorithms to identify geographically nested networks of care providers to compare to existing predetermined care networks and guide the implementation of novel, more efficient networks of care. Within the BDAU SE secure environment, data transferred from NHS England will be analysed using a range of different statistical packages and computing languages, including but not limited to python, R and STATA. 3. A patient level database of outpatient, inpatient and A&E presentations will be created. The aggregate interaction across datasets between a particular geographic region (e.g. postcode, lower super output area [LSOA] (or primary care provider) and a secondary care institution will be used to generate a unipartite network of acute hospitals linked to one another by the strength of their shared patient activities. To achieve the stated purpose, data received from NHS England will be processed in the following general manner: The network will then interrogated to identify how patterns of patient flow will change in response to increased patient demand and altered provider capacity. Clusters of vulnerability in the network will be identified and optimal avenues for intervention will be suggested. - Record level data will be filtered with respect to time periods, geography, patient characteristics and clinical processes specific to the study. This may include for example identifying only those records pertaining to patients with multiple long-term conditions. 4. A patient level database of multi-morbidity and complex disease processes will be constructed to examine care fragmentation between hospitals. Networks of care transitions will be constructed and characteristics of the network, providers and patients will be analysed. - Record level data will be manipulated to perform the calculations specific to the study. This may include for example the creation of a patient-sharing network from aggregated record level data, or the calculation of care fragmentation indices. 5. A provider level database of presentations to acute hospitals will be constructed using the PCONSULT HES data field this is a pseudonymised version of the consultant code. Aggregate presentations will be used to identify heterogeneity in admission patterns and transitions of care between hospitals. This will be interrogated to identify resultant differences in patient outcomes, including morbidity, mortality and length of stay between hospitals with different provider admission patterns. - Statistical analysis of the results of these calculations will be performed. This may include the use of multivariable linear regression to estimate the relationship between care fragmentation indices and patient factors include for example age, and clinical comorbidities. The maternity data has been requested as maternity care has been defined by researchers, patients and policy makers as an area where research into networks of care may have specific benefits to care delivery and patient safety which is why it has been specifically selected as a patient cohort to study. - Findings will be visually represented in a manner appropriate to the study. This may include producing maps, charts or tables of results. In all cases outputs will be produced using aggregate data with small numbers suppressed and in accordance with the HES Analysis Guide. - Findings will then be shared with the public and relevant stakeholders in the form of reports, peer reviewed academic journal publications and conference presentations as further described in the Outputs section of this Agreement. No linkage of record level data will be performed. Aggregate data, with small numbers suppressed, at the level of Lower Layer Super Output Areas may be matched to publicly available data in order to, for example, include updated Index of Multiple Deprivation scores or population density estimates as they are released. There will be no attempt to re-identify individuals. Only aggregate outputs with small numbers suppressed will leave the secure environment. Imperial College London will use the data disseminated under this agreement to support academic research, only for the purpose described in this data sharing agreement. Individuals working on this project would only be permitted to access data relating to that project, as identified within the application. Access to the data is granted only to named individuals working on the project under authorised usernames. Only substantive employees of Imperial College London will use the data disseminated by NHS England and only for the purposes described in this Agreement.

Expected output

[1 paragraph unchanged] The study will yield a PhD Thesis between October 2019 and October 2020 in addition to published academic papers as follows: A number of in progress projects relating to the data previously disseminated under this Agreement were delayed due to clinical and academic redeployment as part of the COVID-19 response by Imperial College London and Imperial College Healthcare NHS Trust. Work on these projects has now re-commenced and the Study Team anticipate that these projects will be completed by 2023, with the hope of completion before this date. 1. The structure of interhospital transfers in the NHS. The projects are as follows: 2. The impact of patient choice in maternity care on local service supply and demand. • The structure of care networks for patients with multi-morbidity and complex disease processes 3. The structure of care networks for patients following trauma, stroke and cardiovascular • Quantifying fragmentation of hospital care for patients with multi-morbidity events, comparing regions with established care networks to those without. • Quantifying fragmentation of hospital care for patients following renal transplantation 4. A network analysis demonstrating the interdependence of secondary care providers in • Characteristics of intra and inter-hospital transition of care networks and their impact on patient outcomes the NHS. • Predicting redistribution of emergency patient flow following critical incidents including cyber-attacks. 5. Predictive modelling of patient flows to secondary care providers in response to changing It is hoped the study will yield published academic papers as follows: organizational capacity. • "The structure of care networks for patients with multi-morbidity and complex disease processes". Target date for submission Q4 2022 Each of these papers will be targeted at health policy or health informatics journals including; • "Characteristics of intra and inter-hospital transition of care networks and their impact on patient outcomes". Target date for submission in Q1 2023 • “Temporal changes in the fragmentation of secondary care for patients with multimorbidity in England”. . Target date for submission in Q3 2023 • “The impact of the Covid-19 pandemic on the fragmentation of secondary care for patients with multimorbidity in England”. . Target date for submission in Q1 2023 • “The impact of Primary Care Networks on the coordination of outpatient clinical services in England” . Target date for submission in Q1 2023 The study intends to disseminate the outputs from this research through several avenues including publication and presentation at local and international meetings. It is intended that all work will be presented at academic conferences once completed and prior to publication (likely in 2023-2024), including: • Academy Health Annual Research Meeting - USA • Health Services Research UK Conference - UK • International Conference on Complex Networks & Their Applications – International • British Journal of General Practice Research Conference - UK Each of these papers will be targeted at health policy or health informatics journals including: [8 paragraphs unchanged] International Health Policy Conference – UK. UK [1 paragraph unchanged] The study intend to disseminate the outputs from this research through several avenues including publication and presentation at local and international meetings. The addition of more recent HES data will ensure that the syudy can include the most recent and relevant data to reflect current activity and trends. Some of the current output from the previous data (up to 2015) includes: Publications accepted/in press: • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2018). Guiding interoperable electronic health records through patient sharing networks. Accepted by Nature Partner Journals Digital Medicine. • Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Caring About Sharing: A Review of Transitions of Care Across Secondary-Care Settings in Patients with Inflammatory Bowel Disease in England. Under review by World Journal of Gastroenterology • Clarke J, Warren L, Darzi A (2018). Care Fragmentation and Organisational Performance in the NHS in England: Results of a Retrospective Observational Review of Hospital Episode Statistics. Under review by the British Medical Journal. • Clarke J, Warren L, Barahona M, Darzi A. Interoperability of Electronic Health Records in England: A Simulated Optimisation Process Operating Over Patient Sharing Networks in England. Target Journal: Nature Digital Medicine • Clarke J, Warren L, Darzi A (2018). Interhospital Transfers: The Fragile Blindspot in Health Systems Research. Target journal: Health Services and Outcomes Research Methodology • Clarke J, Darzi A, Barahona M (2018). Defining Hospital Catchment Areas for Planned Orthopaedic Care in England Using Multiscale Community Detection. Target journal: Nature Digital Medicine. Presentations given: • Clarke J (2018). Understanding Demand Redistribution in Emergency Care in England. Presentation to the Farr Institute, University College London. • J Clarke (2018). The Power of Connections – Mapping the Behaviour of Healthcare Networks. Presentation to the Department of Biostatistics at Harvard University. • Clarke J (2018). The Power of Connections – Mapping the Behaviour of Healthcare Networks. Presentation to the midterm review of the EPSRC Centre for Mathematics of Precision Healthcare, Imperial College London. • Clarke J, Marti J, Barahona M, Darzi A (2017). Predicting Organizational Interdependence in Emergency Care in England. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Inter-Organisational Patient Sharing in England Through Neywork Analysis. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J (2017). An Ecological Analysis of Seasonal Equity in Access to Emergency Care in England. International Health Economics Association Congress, Boston, USA. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Characteristics of Inter-Organisational Patient Sharing in England Through Network Analysis. Sowerby Symposium Imperial College London. • Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Fragmentation of care in patients with inflammatory bowel disease in England. London Surgical Symposium, Imperial College London. [1 paragraph unchanged] The study have has ongoing patient and public engagement which has and continues to, guide the study transitions of care research through the following initiatives: [3 paragraphs unchanged] - Public engagement workshop held in April 2022 with members of the public aged 65 years and above in North-West London exploring the impact of care fragmentation on their perceptions of the quality of their healthcare. Collectively, this public engagement work has directly informed the direction of future work in several ways. Firstly, it has demonstrated the importance patients place on doctors having their clinical information readily available to them. From the perspective of the methods used in the proposed work, it supports the use of measures of informational continuity rather than other widely used measures of continuity of care. Secondly it has shown the difficulties posed by the transfer of patients between hospitals, and has shown how patients perceive these moments to be particularly risky parts of their hospital stay. This finding has driven efforts to specifically identify patient groups, hospitals and clinical conditions are most likely to have a transfer between hospitals. Thirdly, the most recent public engagement session with older residents of North-West London showed that many hospital appointments were being offered online and sometimes by hospitals other than their usual provider of care. This has prompted work proposed here to understand how the COVID-19 pandemic has influenced continuity of hospital care. Previous Outputs produced form data disseminated under this Agreement include: Clarke J, Beaney T, Majeed A et al. Defining Integrated Care Systems Through Patient Data From Referral Networks in the English National Health Service: A Graph-Based Clustering Study. Under review with BMC Health Services Research. In this paper patient sharing networks were constructed to identify the extent to which proposed Integrated Care System (ICS) boundaries aligned with historic patterns of presentation to patients to hospitals. This study finds that across the country some Integrated Care System footprints closely agree with existing patterns of patient referral, while in other areas patients frequently cross new ICS boundaries to receive their care. This suggests that some areas may face greater struggles to provide integrated care as intended by the ICS policy and may require either greater support from NHSE or redrawing of current boundaries. The findings from this study have been shared online in the form of a preprint publication while the article is under review. This project applied Markov Multiscale Community Detection methods and new methods to define hospital catchment areas developed during the first PhD project. Beaney T, Clarke J, Grundy E, Coronini-Cronberg S. A Picture of Health: determining the core population served by an urban NHS hospital trust and understanding the key health needs. BMC Public Health (2022). In these papers, catchment areas for the Chelsea and Westminster Foundation Trust were defined using a range of different network techniques. This work led to the publication of Chelsea and Westminster Foundation Trust's first public health report in which they were able to, for the first time, describe the health needs of their population of patients. This project applied new methods to define hospital catchment areas developed during the first PhD project. Clarke J, Murray A, Markar S, Barahona M, Kinross J et al., 2020, A new geographic model of care to manage the post-COVID-19 elective surgery aftershock in England: a retrospective observational study, BMJ Open, Vol: 10, Pages: 1-9, ISSN: 2044-6055. In this paper, patient sharing networks for hospitals providing surgical care were constructed. This project applied Markov Multiscale Community Detection methods developed during the first PhD project. Warren L, Clarke J, Arora S, Darzi A et al., 2019, Improving data sharing between acute hospitals in England: An overview of health record system distribution and retrospective observational analysis of inter-hospital transitions of care, BMJ Open, Vol: 9, ISSN: 2044-6055. In this article, patient-sharing networks were constructed to demonstrate the extent to which a lack of health record interoperability may impede the delivery of hospital care. This project applied methods to construct patient-sharing networks developed during the first PhD project.

Expected measurable benefits

[2 paragraphs unchanged] The transfer of a patient from one hospital to another often occurs at critical periods in a patient’s journey where they can no longer be optimally cared for by their current hospital. Recent centralization of specialist services has increased the need for transfer to another hospital to receive specialist care. The transfer of a patient from one hospital to another often occurs at critical periods in a patient’s journey where they can no longer be optimally cared for by their current hospital. Recent centralization of specialist services has increased the need for transfer to another hospital to receive specialist care. This transfer process is a period of increased patient risk, where an often critically ill patient is transferred by ambulance over significant distances and whose care is handed over to an entirely new team of clinicians. Which patients need to be transferred, when and to which hospital remains poorly understood, as do their health outcomes relative to those who do not need to be transferred to receive the same specialist care. Through the publication of this work in relevant academic journals, insights into the movement of patients from one hospital to another may be achieved by clinicians and commissioners. This knowledge may be used to both understand the factors which influence patient and physician choice, and also incorporate these factors into future service design. By understanding the circumstances that lead to patient transfer the aim is to identify patient groups that are particularly likely to undergo interhospital transfer and to focus on the development of local and national strategies to ensure optimal transfer of care for these specific groups. This transfer process is a period of increased patient risk, where an often critically ill patient is transferred by ambulance over significant distances and whose care is handed over to an entirely new team of clinicians. Which patients need to be transferred, when and to which hospital remains poorly understood, as do their health outcomes relative to those who do not need to be transferred to receive the same specialist care. 2. A network analysis demonstrating the interdependence of secondary care providers in the NHS followed by predictive modelling of patient flows to secondary care providers in response to changing organisational capacity. Through the publication of this work in relevant academic journals, insights into the movement of patients from one hospital to another may be achieved by clinicians and commissioners. This knowledge may be used to both understand the factors which influence patient and physician choice, and also incorporate these factors into future service design. By understanding the circumstances that lead to patient transfer the aim is to identify patient groups that are particularly likely to undergo interhospital transfer and to focus on the development of local and national strategies to ensure optimal transfer of care for these specific groups. Demand for health care within the National Health Service continues to rise. Each hospital has a finite capacity to provide safe care, therefore there is a threshold over which harm is more likely to result. The likelihood of the demand being placed on a hospital exceeding the care it can safely provide is dependent upon the local incidence of disease and its intrinsic capacity to provide care, but is also critically dependent on the performance of its neighbouring hospitals. If a hospital is unable to meet the demands placed on it, the burden of care provision falls to its neighbouring hospitals, which therefore see an increase in the demands placed on their services. Hospitals with many nearby hospitals may be less vulnerable to this pattern of behaviour and would therefore be said to have a low degree of interdependence, while a pair of hospitals with no nearby neighbours would be highly interdependent on the behaviour of one another. 2. The impact of patient choice in maternity care on local service supply and demand. This principle when applied across hospitals the National Health Service will identify areas of high interdependence within the health care network. Areas of high interdependence of care providers are expected to be less resilient to increases in demand for care or reduction in the capacity to provide care. Identifying these vulnerabilities will assist NHS England in identifying hospitals who require additional investment to ensure the ongoing delivery of high quality patient care. This is particularly the case when estimating the vulnerability of hospitals within the NHS to sudden loss of service in the form of cyber-attacks similar to the WannaCry attack of 2017 or the more recent NHS 111 cyber-attack of August 2022. Patient choice is an increasingly important factor of care delivery in the NHS. The factors underlying patient choice remain poorly understood, in part because of the many patient and provider factors that influence decision making. Expectant mothers can freely choose which hospital they would like to deliver their maternity care. Maternity care is delivered frequently across the country and as it is generally focused solely on the process of giving birth, the variability in patient and provider factors is far less than for other clinical scenarios. This therefore serves as an excellent setting to model the factors which underlie patient choice. In the context of maternity care, where patients can freely choose where their care is delivered, certain providers may be repeatedly favoured or avoided by expectant mothers in response to a range of factors including individual previous experience, geography or waiting times. This may lead to demand for certain providers becoming too great to be met, while others have unused capacity. Identifying and predicting these factors allows providers locally and nationally to correct imbalance in the supply and demand relationship for maternity care, thereby optimising the effectiveness of maternity provision nationally. 3. The structure of care networks for patients following trauma, stroke and cardiovascular events, comparing regions with established care networks to those without. The introduction of defined care networks for the treatment of trauma, stroke and cardiovascular disease in parts of the NHS have demonstrated significant improvements in patient outcomes where they have been implemented. In the case of stroke care, networks have been extremely successful in London and Manchester where they have been introduced. The rest of the country currently does not have the same effective network structure. Using the principles of community detection analysis and Markov models is would be possible to identify for the London and Manchester stroke networks whether their structure optimally reflects the distribution of disease and pattern of clinical practice in the geographic areas they cover. Outside of these two networks it would be possible to examine whether similar network structures already informally exist elsewhere in the country, and develop a nationwide stroke network, in a manner like that which was created for the highly successful national trauma network. This knowledge would inform the development of a national stroke network so that the benefits already obtained from its implementation in London and Manchester may be available nationally. Publication of these findings in high impact health policy journals will bring this work to the attention of key stakeholders nationally and locally. It is expected that this work will be completed within nine months of receipt of the data. 4. A network analysis demonstrating the interdependence of secondary care providers in the NHS followed by predictive modelling of patient flows to secondary care providers in response to changing organizational capacity. Demand for health care within the National Health Service continues to rise, and does so in a stochastic fashion. Each hospital has a finite capacity to provide safe care, and therefore a threshold over which harm is more likely to result. The likelihood of the demand being placed on a hospital exceeding the care it can safely provide is dependent upon the local incidence of disease and its intrinsic capacity to provide care, but is also critically dependent on the performance of its neighbouring hospitals. If a hospital is unable to meet the demands placed on it, the burden of care provision falls to its neighbouring hospitals, which therefore see an increase in the demands placed on their services. Hospitals with many nearby hospitals may be less vulnerable to this pattern of behaviour and would therefore be said to have a low degree of interdependence, while a pair of hospitals with no nearby neighbours would be highly interdependent on the behaviour of one another. This principle when applied across hospitals the National Health Service will identify areas of high interdependence within the health care network. Areas of high interdependence of care providers are expected to be less resilient to increases in demand for care or reduction in the capacity to provide care. Identifying these vulnerabilities will assist NHS England in identifying hospitals who require additional investment to ensure the ongoing delivery of high quality patient care. [2 paragraphs unchanged] It is anticipated that amendment submitted to create creating two additional databases will bring additional benefits. This will expand the study [110 words unchanged] understanding of how patients with complex-care needs interact with health care services. The amendment to create creation of the additional databases will also expand the study to cover the characteristics and implications of transition of care heterogeneity differences between care teams and secondary care providers. There are currently differences between [17 words unchanged] to identify characteristics of consultant-led care networks within hospitals in NHS England. Heterogeneity Differences in admission patterns between hospitals will be identified and analysed to identify [52 words unchanged] transitions of care networks between providers, teams and hospitals influence patient outcomes. This Agreement will also create additional longitudinal databases to examine changes in the fragmentation of clinical care and the evolution of interhospital networks over the course of a decade from 2011 to 2021. This will provide both a contemporary insight as to the extent of care fragmentation for different patient groups and different healthcare providers across the NHS and will also enable an understanding of how fragmentation of clinical care has changed in response to a decade of NHS policy aiming to improve the integration of hospital care and the impact of the Covid-19 pandemic.

Benefits reported

Yield benefits will be provided under a subsequent version of this Agreement. The study yielded a PhD Thesis in addition to published academic papers. There have been several publications from Imperial College London created using HES data disseminated under this Agreement. The benefit to healthcare provision, adult social care, or the promotion of health in some of these publications are described in more detail below: A paper titled: 'Defining Integrated Care Systems Through Patient Data From Referral Networks in the English National Health Service: A Graph-Based Clustering Study.' has been shared online in the form of a preprint publication while the article is under review, and have also been directly shared with representatives of the integrated care systems of North West London to inform their future planning of care provision in the area. . This area covers a population of around 2.5 million people, where the presence of multiple hospital trusts in close proximity to one another has created difficulties for the planning of services in the area. These analyses have helped to quantify this problem for NHS strategy teams in North West London for the first time. A paper titled: 'A Picture of Health: determining the core population served by an urban NHS hospital trust and understanding the key health needs' has been published in BMC Public Health and was identified by NHS Providers as an example of good practice for hospital trusts in 2021 and was nominated for an award that the Health Services Journal ‘Value in Healthcare’ awards in 2021. This study has been read 1300 times and its methods have been used by primary care researchers to quantify integration of primary care in England. A paper titled: 'A new geographic model of care to manage the post-COVID-19 elective surgery aftershock in England: a retrospective observational study', has been published in BMJ Open and was shared with the Royal College of Surgeons of England as part of their response to the COVID-19 pandemic and went on, in part to support the notion of providing surgical care through a series of regional surgical hubs. This paper has been read 3000 times and cited by 13 other studies in the UK and internationally. Over the course of the PhD and postdoctoral study, the researcher developed and applied network analysis methods to quantify relationships between secondary care providers and their patient populations. These methods included applying a network-based graph clustering technique known as Markov Stability to healthcare data for the first time, and optimising the implementation of the algorithm for this purpose. Additionally, a network analysis approach was taken to understand hospital catchment areas in a new way. These methods were applied in a range of outputs produced from data disseminated under this agreement.

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Objective for processing

This Data Sharing Agreement permits the retention and processing of the data provided under previous iterations of this Agreement for an interim period. The following describes the purposes for which the data was disseminated under the previous DSA. [4 paragraphs unchanged] The use of HES data from 2011/12 to 2017/18 will facilitate an accurate capture of trends in patient-sharing networks and a temporal analysis of changes to patterns. The inclusion of 2017/18 data will ensure that findings are as relevant as possible to current and future patterns of patient-sharing. Data is also requested to identify recent provider care networks within and between hospitals. Using recent data from 2017-2018 will ensure the output is as relevant as possible to current and future patterns of care. This request also seeks to obtain permission to create 2 additional databases, as detailed below in the processing activities. This data will allow Imperial College to ensure outputs from the programme of work closely reflects current practice, in a healthcare environment where care pathways and networks are constantly evolving. This health network project will benefit from the addition of recent data to the data sets already held for several reasons. This recent data will allow Imperial College to ensure outputs from the programme of work closely reflects current practice, in a healthcare environment where care pathways and networks are constantly evolving. Analysing this network data longitudinally HES data will also allow the study to analyse how health networks have evolved over the period of data received, and explore reasons for this. This will also allow the study to better understand current trends in health networks and predict future trends more accurately which is important for health policy. From the start of the original data to the most recent available HES data will also allow the study to analyse how health networks have evolved over this period and explore reasons for this. This will also allow the study to better understand current trends in health networks and predict future trends more accurately which is important for health policy. [1 paragraph unchanged]

Processing activities

This Data Sharing Agreement permits the retention and processing of the data provided under previous iterations of this Agreement for an interim period. No additional data will flow under v.3 of this Agreement. [1 paragraph unchanged] The Department of Surgery and Cancer confirms that the data under this application would only be used for the project described in this data sharing agreement agreement. Individuals working on this project would only be permitted to access data [53 words unchanged] by NHS Digital and only for the purposes described in this document. [1 paragraph unchanged] The data will be received from NHS Digital and stored on a secure server hosted at the South Kensington campus of Imperial College. Access to data on this server is restricted to authorized individuals only. The data is accessed and processed by researchers who are based in rooms with keyless combination locks that are always locked when not in use. The data will be received from NHS Digital and stored on Imperial's Big Data and Analytical Unit's Secure Environment (BDAU SE) hosted at Imperial College London's data centre in Slough. Access to data on this server is restricted to authorized individuals only. The BDAU SE is ISO 27001 certified and compliant with NHS Digital Data Security and Protection Toolkit. It can be accessed remotely by users using multi-factor authentication once the user registration process is completed. Data will only be provided once the appropriate dataset registration process is completed. To access the data in the BDAU SE, researchers will require approval from the study chief investigator (in their role as Asset Owner). Once approval is given, researchers are required to: 1) pass the Imperial Data Protection Awareness, including GDPR, and Information Security Awareness training 2) agree, with possible disciplinary action for noncompliance, to BDAU ISO 27001 certified standard operating procedures by signing the BDAU SE User and Dataset Registration forms. All training and forms are required to be renewed on an annual basis. Users who do not pass the annual renewal of training and forms will be automatically removed from access. The de-identified data transferred to the BDAU SE from NHS Digital will be held in the BDAU SE as a standalone dataset. The research team will analyse data within the BDAU SE in the allocated BDAU SE dataset. Data will not be transferred outside of the BDAU SE in raw format for any reason. Only summary research outputs that are aggregated with small numbers suppressed in line with the HES Analysis Guide can be transferred outside of the BDAU SE. This access is password-based and permitted solely to registered users logging on via permitted IP addresses. Record level data will not be distributed to different parts of the organization. The data will not be made available to third party individuals, institutions institutions, or companies. No other data will be linked to this data though data will be compared at aggregate levels if required. The data will be processed as part of the above-mentioned research project within the Department of Surgery and Cancer. It will be queried using data analytical tools and software to aid in answering specific research questions. Data visualizations will be done to present insights gained using suitable tools. No other data will be linked to this data though data will be compared at aggregate levels if required. The data will be processed as part of the above mentioned research project within the Department of Surgery and Cancer. It will be queried using data analytical tools and software. to aid in answering specific research questions. Data visualizations will be done to present insights gained using suitable tools. [5 paragraphs unchanged] Imperial College London now also propose (as of August 2018) that 2 additional databases will be created using updated HES data, as detailed below; 4. A patient level database of multi-morbidity and complex disease processes will be constructed to examine care fragmentation between hospitals. Networks of care transitions will be constructed and characteristics of the network, providers and patients will be analysed. A1. 5. A patient provider level database of multi-morbidity and complex disease processes presentations to acute hospitals will be constructed using the PCONSULT HES data field this is a pseudonymised version of the consultant code. Aggregate presentations will be used to examine identify heterogeneity in admission patterns and transitions of care fragmentation between hospitals. Networks of care transitions This will be constructed interrogated to identify resultant differences in patient outcomes, including morbidity, mortality and characteristics length of the network, providers and patients will be analysed. stay between hospitals with different provider admission patterns. A 2. A provider level database of presentations to acute hospitals will be constructed using the PCONSULT HES data field this is a pseudonymised version of the consultant code. Aggregate presentations will be used to identify heterogeneity in admission patterns and transitions of care between hospitals. This will be interrogated to identify resultant differences in patient outcomes, including morbidity, mortality and length of stay between hospitals with different provider admission patterns. [1 paragraph unchanged]

Expected output

[21 paragraphs unchanged] Imperial College London now also propose (as of August 2018) that additional outputs will be created using updated HES data, as detailed below; A1. This study will contribute towards more than one PhD thesis between October 2019 and October 2022 and additional academic papers as follows: 1. "The structure of care networks for patients with multi-morbidity and complex disease processes". This paper will be targeted at the following journals with in intended submission timeframe of Q4 2019: - British Medical Journal - British Medical Journal Quality and Safety - Health Affairs A2. This study will contribute towards more than one PhD thesis between October 2019 and October 2022 and additional academic papers as follows: 1. "Characteristics of intra and inter-hospital transition of care networks and their impact on patient outcomes". This paper will be targeted at the following journals with an intended submission timeframe of Q2 2019: - British Medical Journal - British Medical Journal Quality and Safety - Health Affairs - Annals of Surgery [3 paragraphs unchanged] Publications under review: [2 paragraphs unchanged] Publications in preparation: [17 paragraphs unchanged]

Benefits reported

There have been several publications from Imperial College London created with the use of the HES data NHS Digital have provided prevoiusly. As detailed below these include; Yield benefits will be provided under a subsequent version of this Agreement. Publications under review: • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2018). A Retrospective Observational Review of Inter-Organisational Patient-Sharing in England. Under review by Health Affairs. • Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Caring About Sharing: A Review of Transitions of Care Across Secondary-Care Settings in Patients with Inflammatory Bowel Disease in England. Under review by Colorectal Disease • Clarke J, Warren L, Darzi A (2018). Care Fragmentation and Organisational Performance in the NHS in England: Results of a Retrospective Observational Review of Hospital Episode Statistics. Under review by the Journal of the American Medical Association. Presentations given: • Clarke J (2018). The Power of Connections – Mapping the Behaviour of Healthcare Networks. Presentation to the midterm review of the EPSRC Centre for Mathematics of Precision Healthcare, Imperial College London. • Clarke J, Marti J, Barahona M, Darzi A (2017). Predicting Organizational Interdependence in Emergency Care in England. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Inter-Organisational Patient Sharing in England Through Network Analysis. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J (2017). An Ecological Analysis of Seasonal Equity in Access to Emergency Care in England. International Health Economics Association Congress, Boston, USA. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Characteristics of Inter-Organisational Patient Sharing in England Through Network Analysis. Sowerby Symposium Imperial College London.

Changed only in punctuation, spacing or capitalisation: Expected measurable benefits.

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Objective for processing

The Department of Surgery and Cancer, based at Imperial College London, is requesting data for use in the following research project: The Power of Connections: Mapping the Behaviour of Health Care Networks. The Power of Connections: Mapping the Behaviour of Health Care Networks [2 paragraphs unchanged] This project will use the following data: HES OP 2011/12-2014/15, 2011/12-2017/18, HES A&E 2011/12-2014/15 2011/12-2017/18 and HES APC 2011/12-2014/15. 2011/12-2017/18. These four years of data are necessary to provide an adequate picture of health care utilization and capture less common events. The use of HES data from 2011/12 to 2017/18 will facilitate an accurate capture of trends in patient-sharing networks and a temporal analysis of changes to patterns. The inclusion of 2017/18 data will ensure that findings are as relevant as possible to current and future patterns of patient-sharing. Data is also requested to identify recent provider care networks within and between hospitals. Using recent data from 2017-2018 will ensure the output is as relevant as possible to current and future patterns of care. This request also seeks to obtain permission to create 2 additional databases, as detailed below in the processing activities. This health network project will benefit from the addition of recent data to the data sets already held for several reasons. This recent data will allow Imperial College to ensure outputs from the programme of work closely reflects current practice, in a healthcare environment where care pathways and networks are constantly evolving. Analysing this network data longitudinally From the start of the original data to the most recent available HES data will also allow the study to analyse how health networks have evolved over this period and explore reasons for this. This will also allow the study to better understand current trends in health networks and predict future trends more accurately which is important for health policy. Data held for this is all part of one overarching project regarding networks of care and all researchers involved are in the same study theme.

Processing activities

The Department of Surgery and Cancer confirms that the data under this application would only be used for the project described in this document. Individuals working on this project would only be permitted to access data relating to that project, as identified within the application. Access is granted to the data only to named individuals working on the project under authorized user names. Such access is password controlled (with a password reset required on a regular refresh). Only substantive employees of Imperial College London will use the disseminated data and only for the purposes described in this document. All organisations party to this agreement must comply with the Data Sharing Framework Contract, including requirements on the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract i.e.: employees, agents and contractors of the Data Recipient who may have access to that data). The Department of Surgery and Cancer confirms that the data under this application would only be used for the project described in this data sharing agreement Individuals working on this project would only be permitted to access data relating to that project, as identified within the application. Access is granted to the data only to named individuals working on the project under authorized user names. Such access is password controlled (with a password reset required on a regular refresh). Only substantive employees of Imperial College London will use the data disseminated by NHS Digital and only for the purposes described in this document. [4 paragraphs unchanged] The data will be processed as part of the above mentioned research project within the Department of Surgery and Cancer. It will be queried using data analytical tools such as SPSS, STATA, SAS, Microsoft Excel, Matlab, Python etc. and software. to aid in answering specific research questions. Data visualizations will be done to present insights gained using suitable tools including Tableau, Inkscape and others. tools. [1 paragraph unchanged] 1. A patient level database for interhospital inter hospital transfers of patients will be constructed, in addition to a range of utilization utilisation and outcome variables (e.g. length of stay, additional procedures, readmissions). A patient level database of maternity care will be constructed constructed: to examine patient choice in relation to delivery location. In both cases, [19 words unchanged] network, providers and patients will be analysed using linear and logistic regression. [1 paragraph unchanged] 3. A patient level database of outpatient, inpatient and A&E presentations will be created. The aggregate interaction across datasets between a particular geographic region (e.g. postcode, LSOA or lower super output area [LSOA] (or primary care provider) and a secondary care institution will be used to [7 words unchanged] linked to one another by the strength of their shared patient activities. The network will then be interrogated to identify how patterns of patient flow will change in response to increased patient demand and altered provider capacity. Clusters of vulnerability in the network will be identified and optimal avenues for intervention will be suggested. The network will then interrogated to identify how patterns of patient flow will change in response to increased patient demand and altered provider capacity. Clusters of vulnerability in the network will be identified and optimal avenues for intervention will be suggested. Imperial College London now also propose (as of August 2018) that 2 additional databases will be created using updated HES data, as detailed below; A1. A patient level database of multi-morbidity and complex disease processes will be constructed to examine care fragmentation between hospitals. Networks of care transitions will be constructed and characteristics of the network, providers and patients will be analysed. A 2. A provider level database of presentations to acute hospitals will be constructed using the PCONSULT HES data field this is a pseudonymised version of the consultant code. Aggregate presentations will be used to identify heterogeneity in admission patterns and transitions of care between hospitals. This will be interrogated to identify resultant differences in patient outcomes, including morbidity, mortality and length of stay between hospitals with different provider admission patterns. The maternity data has been requested as maternity care has been defined by researchers, patients and policy makers as an area where research into networks of care may have specific benefits to care delivery and patient safety which is why it has been specifically selected as a patient cohort to study.

Expected output

-Specific outputs expected, including target dates: [4 paragraphs unchanged] 3. The structure of care networks for patients following trauma, stroke and cardiovascular events, comparing regions with established care networks to those without. 4. A network analysis demonstrating the interdependence of secondary care providers in the NHS. events, comparing regions with established care networks to those without. 5. Predictive modelling of patient flows to secondary care providers in response to changing organizational capacity. 4. A network analysis demonstrating the interdependence of secondary care providers in the NHS. 5. Predictive modelling of patient flows to secondary care providers in response to changing organizational capacity. [11 paragraphs unchanged] Imperial College London now also propose (as of August 2018) that additional outputs will be created using updated HES data, as detailed below; A1. This study will contribute towards more than one PhD thesis between October 2019 and October 2022 and additional academic papers as follows: 1. "The structure of care networks for patients with multi-morbidity and complex disease processes". This paper will be targeted at the following journals with in intended submission timeframe of Q4 2019: - British Medical Journal - British Medical Journal Quality and Safety - Health Affairs A2. This study will contribute towards more than one PhD thesis between October 2019 and October 2022 and additional academic papers as follows: 1. "Characteristics of intra and inter-hospital transition of care networks and their impact on patient outcomes". This paper will be targeted at the following journals with an intended submission timeframe of Q2 2019: - British Medical Journal - British Medical Journal Quality and Safety - Health Affairs - Annals of Surgery The study intend to disseminate the outputs from this research through several avenues including publication and presentation at local and international meetings. The addition of more recent HES data will ensure that the syudy can include the most recent and relevant data to reflect current activity and trends. Some of the current output from the previous data (up to 2015) includes: Publications accepted/in press: • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2018). Guiding interoperable electronic health records through patient sharing networks. Accepted by Nature Partner Journals Digital Medicine. Publications under review: • Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Caring About Sharing: A Review of Transitions of Care Across Secondary-Care Settings in Patients with Inflammatory Bowel Disease in England. Under review by World Journal of Gastroenterology • Clarke J, Warren L, Darzi A (2018). Care Fragmentation and Organisational Performance in the NHS in England: Results of a Retrospective Observational Review of Hospital Episode Statistics. Under review by the British Medical Journal. Publications in preparation: • Clarke J, Warren L, Barahona M, Darzi A. Interoperability of Electronic Health Records in England: A Simulated Optimisation Process Operating Over Patient Sharing Networks in England. Target Journal: Nature Digital Medicine • Clarke J, Warren L, Darzi A (2018). Interhospital Transfers: The Fragile Blindspot in Health Systems Research. Target journal: Health Services and Outcomes Research Methodology • Clarke J, Darzi A, Barahona M (2018). Defining Hospital Catchment Areas for Planned Orthopaedic Care in England Using Multiscale Community Detection. Target journal: Nature Digital Medicine. Presentations given: • Clarke J (2018). Understanding Demand Redistribution in Emergency Care in England. Presentation to the Farr Institute, University College London. • J Clarke (2018). The Power of Connections – Mapping the Behaviour of Healthcare Networks. Presentation to the Department of Biostatistics at Harvard University. • Clarke J (2018). The Power of Connections – Mapping the Behaviour of Healthcare Networks. Presentation to the midterm review of the EPSRC Centre for Mathematics of Precision Healthcare, Imperial College London. • Clarke J, Marti J, Barahona M, Darzi A (2017). Predicting Organizational Interdependence in Emergency Care in England. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Inter-Organisational Patient Sharing in England Through Neywork Analysis. Department of Surgery Annual Research Meeting, Imperial College London. • Clarke J (2017). An Ecological Analysis of Seasonal Equity in Access to Emergency Care in England. International Health Economics Association Congress, Boston, USA. • Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Characteristics of Inter-Organisational Patient Sharing in England Through Network Analysis. Sowerby Symposium Imperial College London. • Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Fragmentation of care in patients with inflammatory bowel disease in England. London Surgical Symposium, Imperial College London. Public/patient engagement: The study have ongoing patient and public engagement which has and continues to, guide the study transitions of care research through the following initiatives: - Regular patient input on all patient safety research through the Patient Safety Translational Research Centre (PSTRC) Patient Public Involvement and Engagement (PPIE) program - Ongoing patient interview study exploring the patient experience of inter-hospital transfers to identify priorities from a patient perspective - Public engagement workshop held in conjunction with the Helix Centre at the Wellcome Collection in November 2017, with a focus on redesigning transitions of care from a patient perspective

Expected measurable benefits

[5 paragraphs unchanged] It is expected that this work will be completed within three months of receipt of the data. [4 paragraphs unchanged] It is expected that this work will be completed within three months of receipt of the data. [10 paragraphs unchanged] It is anticipated that amendment submitted to create two additional databases will bring additional benefits. This will expand the study to cover care networks for patients with complex-care needs, including patients with multimorbidity and common complex disease processes. These patients often require care by several providers in multiple secondary care settings which can lead to care fragmentation. Identifying and measuring care fragmentation for these complex-care patients can guide improvements to inter-hospital data-sharing and continuity of care. The identification of common patterns of patient-sharing between secondary care providers through this work will assist in developing guidelines and interventions to reduce the patient safety risks associated with transitions of care. It is intended that the findings from this work will improve the quality and safety of care systems by providing a better understanding of how patients with complex-care needs interact with health care services. The amendment to create the additional databases will also expand the study to cover the characteristics and implications of transition of care heterogeneity between care teams and secondary care providers. There are currently differences between hospital networks in the structure of handovers and transitions of care between care teams. This work seeks to identify characteristics of consultant-led care networks within hospitals in NHS England. Heterogeneity in admission patterns between hospitals will be identified and analysed to identify resultant differences in patient outcomes, including morbidity, mortality and length of stay. Modifiable factors that may reduce the burden of transitions of care on patient safety will be highlighted. It is intended that the findings from this work will improve the quality and safety of care by improving the understanding of how transitions of care networks between providers, teams and hospitals influence patient outcomes.

Benefits reported

Not stated in the previous version; added here.

There have been several publications from Imperial College London created with the use of the HES data NHS Digital have provided prevoiusly. As detailed below these include;

Publications under review:

• Clarke J, Warren L, Arora S, Barahona M, Darzi A (2018). A Retrospective Observational

Review of Inter-Organisational Patient-Sharing in England. Under review by Health Affairs.

• Warren L, Clarke J, Arora S, Barahona M, Arebi N, Darzi A (2018). Caring About Sharing: A

Review of Transitions of Care Across Secondary-Care Settings in Patients with

Inflammatory Bowel Disease in England. Under review by Colorectal Disease

• Clarke J, Warren L, Darzi A (2018). Care Fragmentation and Organisational Performance in

the NHS in England: Results of a Retrospective Observational Review of Hospital Episode

Statistics. Under review by the Journal of the American Medical Association.

Presentations given:

• Clarke J (2018). The Power of Connections – Mapping the Behaviour of Healthcare

Networks. Presentation to the midterm review of the EPSRC Centre for Mathematics of

Precision Healthcare, Imperial College London.

• Clarke J, Marti J, Barahona M, Darzi A (2017). Predicting Organizational Interdependence

in Emergency Care in England. Department of Surgery Annual Research Meeting, Imperial

College London.

• Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Inter-Organisational

Patient Sharing in England Through Network Analysis. Department of Surgery Annual

Research Meeting, Imperial College London.

• Clarke J (2017). An Ecological Analysis of Seasonal Equity in Access to Emergency Care in

England. International Health Economics Association Congress, Boston, USA.

• Clarke J, Warren L, Arora S, Barahona M, Darzi A (2017). Identifying Characteristics of

Inter-Organisational Patient Sharing in England Through Network Analysis. Sowerby

Symposium Imperial College London.

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Benefits reported

Stated in the previous version and removed here.

Yielded Benefits is not a requirement for new applications.

Unchanged: Objective for processing, Processing activities, Expected output, Expected measurable benefits.