NHS Digital Data Release Register - reformatted
University College London (UCL)
Opt outs honoured: Y
Basis: Health and Social Care Act 2012
Format: Anonymised - ICO code compliant Non Sensitive
How often: One-Off
When: unknown — 11/2016
HSCIC Id: DARS-NIC-00656-V0Z4C-v0.0
Data: Hospital Episode Statistics Accident and Emergency
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Outpatients
Data: Office for National Statistics Mortality Data
Data: Hospital Episode Statistics Accident and Emergency
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Outpatients
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Members and Postings Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - List Cleaning Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Flagging Current Status Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Cause of Death Report
Data: MRIS - Flagging Current Status Report
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Cause of Death Report
Data: MRIS - Scottish NHS / Registration
Data: Hospital Episode Statistics Accident and Emergency
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Outpatients
Data: Hospital Episode Statistics Accident and Emergency
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Outpatients
Data: Hospital Episode Statistics Outpatients
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Accident and Emergency
Data: Office for National Statistics Mortality Data (linkable to HES)
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Accident and Emergency
Data: Hospital Episode Statistics Admitted Patient Care
Data: Hospital Episode Statistics Outpatients
Data: MRIS - Bespoke
Data: MRIS - Cohort Event Notification Report
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Cause of Death Report
Data: MRIS - Cohort Event Notification Report
Data: Office for National Statistics Mortality Data (linkable to HES)
Data: Hospital Episode Statistics Critical Care
Data: Hospital Episode Statistics Admitted Patient Care
Data: MRIS - Scottish NHS / Registration
Data: MRIS - Members and Postings Report
Data: Hospital Episode Statistics Admitted Patient Care
Output: The output of the research will be presented at seminars and conferences (such as AAG - American Association of Geographers – Annual Conference: AAG Health and Medical Geography Specialty Group; International Population Data Linkage Conference – Farr Institute; REVES International Network on Health Expectancy Annual conferences), and included in published research papers in peer-reviewed journals (such as Social Science & Medicine; Health & Place; Int Journal of Epidemiology and Community Health). Intermediate findings are currently being prepared for publication in Health & Place and presentation at the above-mentioned AAG conference as part of the International Geography, GIScience, and Urban Health featured theme.
Once the data has been disseminated from NHS Digital to UCL , other conferences and papers may also be included but cannot be confirmed at this time.
The final classification of LSOAs along with recommendations for policy will also be published and made available for download on a website that is linked to the ESRC-funded Consumer Data Research Centre (CDRC – data.cdrc.ac.uk). The CDRC websites attracts an increasing user base from a range of NHS organisations (Public Health England, CCGs, NHS trusts), local government, third sector organisations. New data products are announced and linked in email notifications, quarterly newsletters and special features, such as the CDRC Map of the Month, and Twitter feeds. On average, there are about 150,000 page views and 50,000 data downloads per year.
Outside the CDRC community, UCL are keen to engage with Public Health England and are in contact with them to help identify suitable methods of dissemination, possibilities of promoting the CDRC website and products to the health care community, notably CCGs.
In summary, the outputs of the research comprises
• a novel geodemographic classification of health needs at LSOA level
• a detailed, contextualised characterisation of health needs and challenges associated with each area profile
• recommendations for policy and health and social care
The results will be disseminated in the following ways:
• presentation and publications within the academic community
• provision on data.cdrc.ac.uk, the CDRC data catalogue
• visualisation on maps.cdrc.ac.uk, the widely accessed map service of CDRC
• targeted engagement with Public Health England using existing links with this research group
• presentation of outputs to CCG Ealing using existing formal links between the CCG and CDRC
All outputs will use aggregate data with small numbers suppressed in line with the HES analysis guide.
Activities: Data minimisation
The following data minimisation strategies have been considered:
(1) Temporal censoring is deemed impractical for the following two reasons. First, health care benefit of the output (geodemographic profiles of health needs) is strongly limited, if the question of temporal stability remains unaddressed. Are the health needs only short-lived phenomena or a long-term challenge that require strategic attention? Second, the research needs to capture at least two census periods in the analysis so as to derive age-and sex standardisation. Currently, the existing work with data pertaining to years 2003-2009 is compromised due to potentially inaccurate age- and sex standardisation.
(2) Demographic censoring (restricting the analysis to certain age, sex or ethnic groups) would limit the validity of the work, since the work aims at developing profiles for the entire population that comprises all ethnicities, age and sex groups. Only with the full population, will the research be able to display the full range of challenges and priorities facing local health care.
(3) Geographic censoring would equally reduce the possibility to identify specifically local challenges and hence limit the utility of the geodemographic profiles for the health care sector. Specific regional and local challenges can only be identified by viewing local patterns against country-wide patterns (e.g. averages). For example, in previous analysis, it was found that London faces greater incidence of sense organs and nerves-related conditions than other city regions in England. This specific challenge would have remained masked in an analysis that focussed on London only.
(4) A sample would undermine the validity, granularity and robustness of the work. In addition, sampling would be extremely difficult to implement within the context of this study. The estimation of geographically varying challenges need to be performed at a sufficient level of geographical granularity. Geodemographic indicators are typically developed at postcode or Census Output area level. The most granular level available in HES is LSOA (Lower Layer Super Output Area). There are approx. 27,000 LSOAs in England, and in previous work with HES data the research has confirmed there is a need for a sufficient number of cases to develop robust small area estimates and be capture temporal trends. A sample would need to cover each LSOA, be proportionate to local admissions within each LSOA, be demographically representative of all patients in each LSOA and be repeated in this way each year. As such, this would be extremely difficult to implement and significantly compromise the objective of the study to provide robust, full-fledged and sufficiently granular health profiles.
This data are obtained for research purposes in medical geography and will be processed fairly according to regulations and standards of the Data Protection Act and corresponding UCL policies.
The processing of the information is carried out for non-commercial research and educational purposes at a higher educational institution (UCL) in exercise of its legitimate functions of training and research.
The data will be accessed only by substantive employees of UCL and only for the purposes described in this document.
All relevant individuals (Data Protection Officer, Departmental IT Representative, Computer Security Team, Data Protection Coordinator) are informed about the research proposed and are able to monitor proper conduct in all procedures.
The data are used as direct input in the analysis and some data will be processed for the purpose of names classification and geographical classification. Items that are not used as input in the way set out in this application are not of interest and therefore not requested. In order to reduce the risk of identification to an acceptable minimum, a special procedure to names extraction has been elaborated jointly with NHS Digital.
At the end of the study, the data will be destroyed in accordance with UCL’s retention policy. The UCL Computer Security Team has developed guidelines of safe removal, and a retention schedule that is developed with the UCL Records Manager will ensure that timely removal can be monitored.
No data will be transferred to third parties, EU countries or countries outside the EU at any point of time in the research.
The project has successfully undergone Ethical Review and review by HRA's Confidentiality Advisory Group.
The DH Information Governance toolkit has been completed and is reviewed regularly. The project has achieved level 2 of requirements for Hosted Secondary Use Team/Project (IG toolkit version 13).
The data will be stored in a database at UCL's School of Life Science and Medical Studies’ safe haven environment and undergo statistical, multivariate analysis. The data will be processed from an authorised PC client located within UCL and queries will be performed through secure data access. Secure output files (e.g. statistical results) may be transferred through secure file transfer subject to standards of disclosure control. The data will also be related to census datasets (UK Census 2001 and 2011) using information on LSOA (Lower Layer Super Output Areas) level. The LSOA code held in HES extract will be used to match HES records to residential context.
In terms of patient classification, UCL will use demographic data (age, sex), primary diagnosis and admission and discharge information. UCL will derive ethnicity from a patient by linking HES records to the PDS and classify patients’ names. This linkage will be performed by NHS Digital. In previous communication with NHS Digital (ref NIC-216528-N0N5Q), UCL established the technical feasibility of a procedure to link HES extracts and use names stored in PDS to create a bespoke patient classification. CAG has reviewed this procedure and confirmed that section 251 support is not required as no confidential data will be disclosed to the researcher.
The data processing is a five-point process which is as follows;
1. NHS Digital extracts the patient identifiers of all patients in the HES index for the years 1998/99 to 2012/14
2. NHS Digital links the identifiers to the PDS (Personal Demographics Services) data which is held on the MIDAS system
3. NHS Digital applies the names classification algorithm provided by UCL. No names are disclosed to the researcher.
4. NHS Digital adds to each row of HESID to the name class
5. NHS Digital supply a file of HES records with requested fields including pseudonymised HESIDs and linked classes
UCL expect to develop such a method by grouping and aggregating health diagnosis for each small area by different patient categories over a study period from 1999/00 to 2013/14, which covers 2001 and 2011 Census periods. This allows area linkage to contextual variables, including the prevalence of long-term limiting illness and aggregate demographic characteristics of Lower-Layer Super Output Areas (LSOAs), which is also required for age-and-sex standardisation. The Census neighbourhood statistics, which the data will be linked to, are Office for National Statistics-cleared aggregate statistics; they do not contain information on individuals.
The work will be carried out by specified users at UCL at the Department of Geography at University College London, who are substantive employees of UCL and only for the purposes described in this document. The project objectives and plan have been reviewed by academic staff from UCL Epidemiology and Public Health, and the interaction will continue throughout the project to ensure scientific rigour and maximum impact to relevant interest or user groups, notably CCGs.
Objective: The objective of this research project is to create geodemographic, small area profiling of Health Needs, which takes into account a range of patient and contextual characteristics.
Comprehensive and precise assessment of health needs, as required by The Local Government and Public Involvement in Health Act 2007 (see section 116 on JHWBs and JSNAs) remains a significant challenge due to the existence of complex pathways and multi-level processes. Because different groups of people have been shown to have different vulnerabilities to ill-health in different circumstances, this study seeks to identify;
(1) general health needs/vulnerability at the small area level and
(2) the geographical circumstances in which certain types of patients (classified by ethnicity) appear to be vulnerable/express different health needs.
This is novel research that will account for health needs at multiple levels (patient group, small areas) by taking a distinct geomedical angle that draws on latest advances in spatio-epidemiological modelling, geodemographics, analysis of surname geographies and population genetics.
The work that has been completed under the current Data Sharing Agreement has found marked differences in health needs between geographical areas. However, the work was limited by the available years of HES extracts (2003-2009). This time period does not allow for appropriate and robust age-and-sex standardisation based on reliable data sources (i.e. Census 2001 and 2011). In addition, the short time period does not permit investigation of temporal trends and thereby the stability of area health profiles. Stability of small areas health profiles is an essential dimension of health needs assessment and crucial for informed policy decisions, such as resource allocation.
UCL therefore wish to refine these objectives in two ways:
a) to break down small area health profiles by patient category; and
b) to assess the stability of health profiles by using more years of HES extracts covering two Census periods.
As soon as 2001 and 2011 HES extracts become available, the work completed so far can be updated and put forward to peer review and academic publication as well as dissemination to the health care community (see section 5c below).
In short, the follow-up research will develop a dynamic model to predict long term health needs refined for groups of patients and small areas.
Reasons for this study:
Geo-temporal small-area profiling is useful in identifying need for intervention, assessing causes of health challenges (specifically different profiles of disease burdens) by characterising their nature and spatial extent as well as the geographic and demographic context in which they are manifest. Small area profiling thus supports the definition of policy priorities at the strategic level as well as more tactical decisions by care providers.
For example, a temporally persistent disadvantaged profile in a number of neighbourhoods in a city can support evidence-based policy making by defining local strategies to address these challenges. On the provider level, awareness about locally specific health challenges and their contexts can support operational decisions, such as treatment or screening choices.
Small-area profiling are suitable methods to support strategies and decisions, because of their capacity to estimate health challenges in robust ways (or at least with a number of confidence measures about the certainty of the estimate) and in accounting for geographical and demographic context.
Similar approaches have been applied on ONS mortality data, and geographically and demographically varying challenges could be identified (see e.g. Green et al 2014 Health & Place 30C, Shelton et al 2006 Health & Place 12.4). Yet, mortality data are limited in a sense that they are retrospective and less useful for defining care priorities (since a death has already occurred) and only considers the cause of death, leaving out non-fatal and temporary conditions.
There is no study that uses HES data in this way, but work that has been carried out under the previous agreement (NIC-33864-6226N) suggests that resulting classifications can provide significant benefit in summarising and contextualising health needs with direct implications for care.
A similar product that already exists is healthACORN, but this product is limited in terms of health conditions it focusses on and its transparency: as a commercial product, the methods are not revealed, only the resulting classifications of fairly broad categories.
More generic classifications have been employed in health studies, particularly focussing on health screening (e.g. Sheringham et al 2009 Sexual Health 6.1, Noaham et al 2010 Journal of Public Health 32.4). But these studies, too, had to rely on generic and partly commercial products, and it is expected that the utility in assessing need and health screening can be improved if classifications are devised more scientifically and focussed on observed morbidity.
Therefore, a transparent, robust, dynamic and contextualised small area classification would be a valuable resource of policy makers and care providers alike who seek to define priorities and take decisions that are informed by a detailed and local understanding of health needs in the spirit of joined-up health care (see NHS Institute for Innovation and Improvement 2010 ‘Joined-up Care’).
Benefits: The outcome of this research is envisioned to support local Clinical Commissioning Groups (CCGs), local authorities and other public partners including third sector organisations in determining their budgeting and commissioning priorities, in which a long-term view and strategic need orientation is crucial. CCGs are expected to lead Joint Strategic Health Needs Assessments (JSNAs) and Joint Health and Well-being Strategies (JHWBs), and there is a need to find appropriate methods and data sources to do this.
By investigating and contextualising specific health needs by different group of patients and areas, UCL intend to support these strategic players in assessing health needs as part of JSNAs and developing JHWBs by taking a novel long-term, geographic and temporal view of local health needs and preventive health care. At the strategic level, these will be the primary beneficiaries through the provision of a detailed, transparent and robust small area profiling. At an operational level, care providers (such as GP practices or hospital trusts) can use the classification in delivering personalised patient care, on which there is currently an important emphasis (see e.g. latest report of London Health Commission, 2014. ‘Better health for London’), including treatments and targeted health screening initiatives.
The classification including metadata and policy recommendations will be available for download at the aforementioned website and is thus likely to reach a wide audience. In the long-run, a common evidence base will enable various players of the health care system to deliver a higher level of joint-up care.
So far, analysis of the data has demonstrated the relevance of geographical variations in health needs and is being prepared for publication and sharing with CCG Ealing to link the findings to practice in North West London. It is intended to further and refine the investigation (with the inclusion of ethnicity) to deliver benefits using the existing formal connection to CCG of Ealing (which is their representation on the advisory board of the UCL research group of which the applicants are part). The data will be made available for download when classifications can be refined, i.e. by the end of the project at the latest, if not earlier for sufficiently robust, intermediate findings.
The target date for this research is December
2017
In summary, there are three main areas in which this study is intended to contribute to improving patient care.
1, At the strategic level, UCL seek to support Clinical Commissioning Groups (CCGs), local authorities and other public sector partners in defining budgeting and commissioning priorities as part of the compulsory Joint Strategic Health Assessments (JSNAs) and Joint Health and Well-Being Strategies (JHWBs) (as per section 116, The Local Government and Public Involvement in Health Act 2007).
2. At an more operational level, the study is also intended to improve personalised patient care by providing an evidence base (the profiling) for care providers, such as GP practices with a detailed contextualisation of local challenges including conditions by a refined measure of ethnicity. In addition, the classification is intended to facilitate targeted health screening in local areas, in which ethnicity has emerged to be a crucial factor in both screening uptake as well as condition onset.
3. The output of this research is intended to support preventive care by all strategic and operational organisations and agencies, by suggesting potential causes of observed health challenges through contextualisation, including links to a refined measure of ethnicity.
Source: NHS Digital.