The Administrative Data Research Centre Scotland: A Scoping Report on the Legal & Ethical Issues Arising from Access & Linkage of Administrative Data

IntroductionLarge administrative datasets are now being used for secondary purposes across a wide range of public sector organisations, including in health and higher education. However, governance, regulation and policy surrounding the use of these datasets are at different stages of development in these sectors. Our aim was to explore similarities and differences in the use of administrative data between the health and higher education sectors to inform policy development. Objectives and ApproachWe investigated views on the use of administrative data in both the health and higher education sectors. We conducted 18 qualitative in-depth interviews with key stakeholders, to provide insight into the ethical, social and legal issues associated with the use of big data in these settings. The interviews were transcribed and thematically coded. ResultsParticipants indicated the rapid pace of technological change and large volume of potentially sensitive data collected raises governance, infrastructure and ethical issues in both settings. Common challenges include communication, staff capabilities, delays in access, multiple policies and governance committees, and technical and operational issues. In the health sector, there was clear understanding of the issues and governance structures to address these issues, whereas this understanding was more variable in the higher education sector. Trust in government (to use responsibly and store securely) was raised in the health sector but not in universities. Conclusion / ImplicationsUnderstanding and use of administrative data are at quite different levels of development in the higher education and health sectors. Higher education needs policy and ethical guidance and higher level governance and greater consultation across the sector. Both sectors would benefit from a national approach to data governance.

Download Full-text

Establishment of a National Homelessness Prevention Programme dataset to enable an anonymised longitudinal dynamic cohort based study

International Journal for Population Data Science ◽

10.23889/ijpds.v3i4.905 ◽

2018 ◽

Vol 3 (4) ◽

Author(s):

Sarah Lowe ◽

Rhodri Johnson ◽

Ian Jones ◽

Sian Morrison-Rees

Keyword(s):

Statistical Analysis ◽

Administrative Data ◽

Data Linkage ◽

Healthcare Utilisation ◽

Health Data ◽

Data Sources ◽

Research Centre ◽

Control Group ◽

External Data ◽

Group Methods

IntroductionWelsh Government invests over £120m annually in housing related support to help prevent and tackle homelessness under the ‘Supporting People Programme’. A 2016 data-linkage Feasibility Study indicated health-service utilisation reductions post-intervention, and led to a four year project to create a national, all-Wales dataset to provide robust statistical results. Objectives and ApproachEstablish data sharing agreements, acquire and import anonymised individual-level data into the SAIL Databank. Create a research ready dataset, designed to permit annual administrative data updates to form dynamic cohort and control groups. Create several control group methods: 1) Internal Programme Data; 2) Matched controls; 3) Healthcare-Utilisation Patterns; 4) External Data Sources. Link to routine health data, obtain and link to other public service data to gain a deeper understanding of the Programme; how it affects use of other public services, and whether it helps people live independently. Complete statistical analysis using a Generalised Linear Mixed Modelling approach. ResultsData sharing agreements, data acquisition and standardisation complete for nineteen of twenty-two Unitary Authorities in Wales. Temporal coverage varies by Unitary Authority (2003-2017). 2016 data measures: match rates >85%; 57% female; lead reason for support (top 5) : ‘General’ 20%, ‘Mental Health’ 15%, ‘Older People’ 14%, ‘Domestic Abuse’ 9%, ‘Young People’ 7%. Various control group methods employed: 1) Internal ‘Programme’ Data – no support taken up; 2) Matched controls; 3) Healthcare-Utilisation Patterns – rejected due to sparse outcome data; 4) External Data Sources being further explored. Health data-linkage (emergency admissions, emergency department attendance and primary care events) complete. Ongoing discussions to obtain sample social care, and police call data during 2018. Statistical analysis underway with results planned to be published during the summer of 2018. Conclusion/ImplicationsDespite many challenges, creation of a national linked dataset for people at risk of homelessness is possible with collaborative working between central government, academic and local government bodies. This ‘Administrative Data Research Centre Wales’ project has created a rich research resource enabling statistical analysis to answer research questions around homelessness.

Download Full-text

Rates and predictors of anti-depressant prescribing in Northern Ireland 2011–2015: a data linkage study using the Administrative Data Research Centre (NI)

Irish Journal of Psychological Medicine ◽

10.1017/ipm.2019.14 ◽

2019 ◽

Vol 37 (1) ◽

pp. 32-38 ◽

Cited By ~ 1

Author(s):

Mark Shevlin ◽

Michael Rosato ◽

Stephanie Boyle ◽

Daniel Boduszek ◽

Jamie Murphy

Keyword(s):

Northern Ireland ◽

Administrative Data ◽

Religious Beliefs ◽

Depressive Disorders ◽

Data Linkage ◽

Mental Health Problems ◽

Population Based ◽

Linkage Study ◽

Research Centre ◽

Logistic Analysis

Objectives:Research indicates that anti-depressant prescribing is higher in Northern Ireland (NI) than in the rest of the UK, and that socio-economic and area-level factors may contribute to this. The current study provides comprehensive population-based estimates of the prevalence of anti-depressant prescription prescribing in NI from 2011 to 2015, and examined the associations between socio-demographic, socio-economic, self-reported health and area-level factors and anti-depressant prescription.Methods:Data were derived from the 2011 NI Census (N = 1 588 355) and the Enhanced Prescribing Database. Data linkage techniques were utilised through the Administrative Data Research Centre in NI. Prevalence rates were calculated and binary logistic analysis assessed the associations between contextual factors and anti-depressant prescription.Results:From 2011 to 2015, the percentages of the population in NI aged 16 or more receiving anti-depressant prescriptions were 12.3%, 12.9%, 13.4%, 13.9% and 14.3%, respectively, and over the 5-year period was 24.3%. The strongest predictors of anti-depressant prescription in the multivariate model specified were ‘very bad’ (OR = 4.02) or ‘Bad’ general health (OR = 3.98), and self-reported mental health problems (OR = 3.57). Other significant predictors included social renting (OR = 1.67) and unemployment (OR = 1.25). Protective factors included Catholic religious beliefs, other faith/philosophic beliefs and no faith/philosophic beliefs in comparison to reporting Protestant/other Christian religious beliefs (ORs = 0.78–0.91).Conclusion:The prevalence of anti-depressant prescription in NI appears to be higher than the prevalence of depressive disorders, although this may not necessarily be attributable to over-prescribing as anti-depressants are also prescribed for conditions other than depression. Anti-depressant prescription was linked to several factors that represent socio-economic disadvantage.

Download Full-text

Protecting children during child protection research using administrative data

International Journal for Population Data Science ◽

10.23889/ijpds.v4i3.1241 ◽

2019 ◽

Vol 4 (3) ◽

Author(s):

Jade Hooper ◽

Linda Cusworth ◽

Helen Whincup

Keyword(s):

At Risk ◽

Administrative Data ◽

Child Protection ◽

Research Study ◽

Ethical Issues ◽

Research Process ◽

Lessons Learned ◽

Children And Families ◽

Effective Protection ◽

Protection Systems

Background with rationaleChild protection systems aim to protect children from harm. Child protection research shares the same goal. However, the process of undertaking this research can present a risk. For child protection research using administrative data, children and families are at risk through breaches of privacy and identification, and further ethical issues of fair and necessary processing. This can be particularly challenging for countries such as Scotland, where smaller communities may mean children and families are more easily identifiable. For researchers and data gatekeepers alike, it is our duty to protect children in every way that we can throughout the process. Main AimThis paper aims to discuss with delegates the importance of protecting children during child protection research using administrative data, and some of the ways this can be done. Methods/ApproachThe paper will use an example of a recent research study which used administrative data from child protection systems in Scotland – Permanently Progressing? Building secure futures for children in Scotland. The methods used to protect children within this research and the lessons learned will be discussed. Input from delegates and the sharing of experiences will be most welcome throughout. Results and ConclusionChild protection research using administrative data often aim to inform and improve child protection systems. Such research can thereby enable more effective protection. However, such a research process raises its own child protection challenges, both to researchers and to the gatekeepers of this data. Despite this, although tricky at times to navigate, risks can be reduced through thoughtful care and consideration of these issues throughout the research process, from application to publication.

Download Full-text

Challenges of administrative data linkages: experiences of Administrative Data Research Centre for England (ADRC-E) researchers

International Journal for Population Data Science ◽

10.23889/ijpds.v3i2.566 ◽

2018 ◽

Vol 3 (2) ◽

Author(s):

Hannah Morris ◽

Silvia Lanati ◽

Ruth Gilbert

Keyword(s):

Administrative Data ◽

Data Linkage ◽

Personal Data ◽

Research Centre ◽

Birth Cohorts ◽

Face To Face ◽

Education Data ◽

Outcomes For Children ◽

One Year ◽

Data Linkages

BackgroundLinkage between administrative datasets routinely collected by government departments and other statutory bodies create rich resources for policy-relevant research. We describe how ADRC-E has achieved linkage between health and education data for England, and the challenges presented by this process. MethodsA key task of ADRC-E is to progress exemplar studies of novel data linkages, and build relationships with data providers. While navigating untested data application and permission processes, ADRC-E researchers have maintained detailed timelines and developed a framework to improve the efficiency of future applications. ResultsThe ADRC-E has approval to link four one-year birth cohorts of the National Pupil Database and Hospital Episode Statistics to facilitate research into outcomes for children with chronic conditions. The timeline of 3 years and 4 months represents 6 face-to-face meetings and 108 email and telephone correspondences. The technical challenges of data linkage have yet to be overcome before we receive data for research. Other examples of data timelines for cross-sectoral data linkage requests will be reported. ConclusionsThe shifting legal landscape governing the use of personal data, and a lack of precedent mean that unlocking administrative data for research requires substantial time, diligence and expertise on the part of both researcher and data provider.

Download Full-text

Impact through Engagement

International Journal for Population Data Science ◽

10.23889/ijpds.v5i3.1369 ◽

2020 ◽

Vol 5 (3) ◽

Author(s):

Elizabeth Nelson ◽

Frances Burns

Keyword(s):

Administrative Data ◽

Policy Development ◽

Positive Impact ◽

Research Impact ◽

Research Centre ◽

Evidence Based ◽

Research Partnership ◽

Public Benefit ◽

Academic Knowledge ◽

Key Stakeholders

BackgroundThe Administrative Data Research Centre Northern Ireland (ADRC NI) is a research partnership between Queen’s University Belfast and Ulster University to facilitate access to linked administrative data for research purposes for public benefit and for evidence-based policy development. This requires a social licence extended by publics which is maintained by a robust approach to engagement and involvement. ApproachPublic engagement is central to the ADRC NI’s approach to research. Research impact is pursued and secured through robust engagement and co-production of research with publics and key stakeholders. This is done by focusing on data subjects (the cohort of people whose lives make up the datasets, placing value on experts by experience outside of academic knowledge, and working with public(s) as key data advocates, through project steering committees and targeted events with stakeholders. The work is led by a dedicated Public Engagement, Communications and Impact Manager. DiscussionWhile there are strengths and weaknesses to the ADRC NI approach, examples of successful partnerships and clear pathways to impact demonstrate its utility and ability to amplify the positive impact of administrative data research. Working with publics as data use becomes more ubiquitous in a post-COVID-19 world will become more critical. ADRC NI’s model is a potential way forward.

Download Full-text

It’s a big, BAD World- the UK picture of Big and Administrative Data

International Journal for Population Data Science ◽

10.23889/ijpds.v1i1.250 ◽

2017 ◽

Vol 1 (1) ◽

Author(s):

Joanne Webb

Keyword(s):

Social Sciences ◽

Big Data ◽

Civil Society ◽

Local Government ◽

Administrative Data ◽

Research Centre ◽

Data Centre ◽

Civil Society Organisations ◽

Government Data ◽

The Uk

ABSTRACT Background & AimsThe Economic and Social Research Council (ESRC) have established a three strand programme to encourage the use of Administrative and Big Data in research. The first strand funded the Administrative Data Network, a UK-wide partnership between universities, government bodies, national statistics authorities and the wider research community. www.adrn.ac.uk. The Network facilitates secure research access to linked, de-identified administrative data to enable real-world analysis that can benefit society. The second strand covers the Big Data Centres: the Business and Local Government Data Centre, the Urban Big Data Centre and the Consumer Data Research Centre. These concentrate on making data routinely collected by business and local government organisations accessible for academics in order to undertake research in the social sciences. The third strand enables partnerships between academic institutions and citizen and voluntary sector organisations, to establish or build on relationships between academic researchers and civil society organisations. The aim is to demonstrate the value of improved data infrastructure, enabling collection and analysis of data which is of interest to civil society organisations and empowering the sector to better use its own data. All three strands aim to enable research in social sciences while safeguarding individuals’ identities. Working in a diverse changing landscape brings its own challenges and possibilities. The University of Essex hosts the Administrative Data Service of the ADRN, the Business and Local Government Data Centre and the Human Rights, Big Data and Technology Project. The challenge is to describe the boundaries, overlaps and synergies in the landscape. This paper will review the UK network of these partnerships, a description of some of the challenges of taking part in such a diverse growing landscape and some of the benefits to researchers and society.

Download Full-text

Risk identification and management for the research use of government administrative data

Records Management Journal ◽

10.1108/rmj-03-2019-0016 ◽

2019 ◽

Vol 30 (1) ◽

pp. 101-123

Author(s):

Elizabeth Shepherd ◽

Anna Sexton ◽

Oliver Duke-Williams ◽

Alexandra Eveleigh

Keyword(s):

Risk Management ◽

Administrative Data ◽

Risk Identification ◽

Research Centre ◽

Research Use ◽

Content Type ◽

Secondary Use ◽

Management Actions ◽

Risk Framework ◽

The Government

Purpose Government administrative data have enormous potential for public and individual benefit through improved educational and health services to citizens, medical research, environmental and climate interventions and better use of scarce energy resources. The purpose of this study (part of the Administrative Data Research Centre in England, ADRC-E) was to examine perspectives about the sharing, linking and re-use (secondary use) of government administrative data. This study seeks to establish an analytical understanding of risk with regard to administrative data. Design/methodology/approach This qualitative study focused on the secondary use of government administrative data by academic researchers. Data collection was through 44 semi-structured interviews plus one focus group, and was supported by documentary analysis and a literature review. The study draws on the views of expert data researchers, data providers, regulatory bodies, research funders, lobby groups, information practitioners and data subjects. Findings This study discusses the identification and management of risk in the use of government administrative data and presents a risk framework. Practical implications This study will have resonance with records managers, risk managers, data specialists, information policy and compliance managers, citizens groups that engage with data, as well as all those responsible for the creation and management of government administrative data. Originality/value First, this study identifies and categorizes the risks arising from the research use of government administrative data, based on policy, practice and experience of those involved. Second, it identifies mitigating risk management activities, linked to five key stakeholder communities, and it discusses the locus of responsibility for risk management actions. The conclusion presents the elements of a new risk framework to inform future actions by the government data community and enable researchers to exploit the power of administrative data for public good.

Download Full-text

The role and nature of consent in government administrative data

Big Data & Society ◽

10.1177/2053951718819560 ◽

2018 ◽

Vol 5 (2) ◽

pp. 205395171881956 ◽

Cited By ~ 2

Author(s):

Anna Sexton ◽

Elizabeth Shepherd ◽

Oliver Duke-Williams ◽

Alexandra Eveleigh

Keyword(s):

Administrative Data ◽

Social Attitudes ◽

Public Perceptions ◽

Research Centre ◽

Relational Context ◽

Secondary Use ◽

Stakeholder Perspectives ◽

Information Governance ◽

Legislative Framework ◽

Study Setting

This article draws on research undertaken by the authors as part of the Administrative Data Research Centre in England (ADRC-E). Between 2014 and 2017, we conducted four case studies on government administrative data for education, transport, energy and health. The purpose of the research was to examine stakeholder perspectives about the sharing, linking and re-use (secondary use) of government administrative data. In relation to the role and nature of consent given by data subjects for re-use, our study revealed significant variations in data provider and researcher attitudes. Although our study setting was England, we believe that the findings have wider resonance. Our analysis identified six factors which might account for the variations around consent: the specificities of the legislative framework governing the collection and processing of particular data; the type of data being collected and the relational context in which it is created; the broader information governance framework in which the data resides; the creating organization's approach to data release; the relative levels of risk aversity within the creating organization; and public perceptions and social attitudes. In conclusion, we consider whether consent is still the best mechanism available for data re-use, or whether a social contract model of data sharing should be developed.

Download Full-text

Common Mental Disorder across Standard Occupational Classifications in Northern Ireland: an administrative data study

International Journal for Population Data Science ◽

10.23889/ijpds.v4i3.1240 ◽

2019 ◽

Vol 4 (3) ◽

Author(s):

Finola Ferry ◽

Michael Roasto ◽

Emma Curran ◽

Gerard Leavey

Keyword(s):

Mental Health ◽

Northern Ireland ◽

Mental Disorder ◽

Administrative Data ◽

Common Mental Disorder ◽

Large Population ◽

Health Condition ◽

Research Centre ◽

Prescription Data ◽

Mental Health Condition

Background While employment in general promotes positive health and wellbeing, a number of studies show disparities in mental health across occupational groupings (Bell et al., 2014; Stansfield et al. 2011). However, definitive studies to allow estimation of prevalence and risk across a range of occupation types require large population sub-groups and common measures of mental health. AimWe hypothesise: (1) we will detect differences in the prevalence of Common Mental Disorder (CMD) across occupational groupings; and (2) risk of CMD across occupation types is moderated by family demands. MethodsUsing linked administrative data we examine CMD among the Northern Ireland (NI) adult population across Standard Occupational Classifications (SOCs). We also examine the influence of socio-demographic, socio-economic and health-related variables. Data is accessed through the Administrative Data Research Centre NI (ADRC-NI) and comprises the individual and household 2011 Census; Land and Properties Services data; Enhanced Prescribing Database and Multiple Deprivation Measures. ResultsExamination of nine major SOC groups shows the prevalence of any self-reported (SR) mental health condition was highest among ‘elementary trades and related occupations’ (24.63%) followed by ‘process, plant and machine operatives’ (23.65%). The risk of mental ill health among these groups remained elevated for both men and women after controlling for age, sex, marital status and educational attainment. Analyses based on the 90 sub-minor SOC groups shows that ‘textiles and garments trades’ workers had the highest prevalence of a SR mental health condition (36.15%). Logistic regression analyses shows that individuals with dependent children and informal caregiving responsibilities had lower risk of a SR mental health condition (OR=0.75, CI=0.73, 0.76 and OR=0.81, CI=0.79, 0.82 respectively). ConclusionResults presented in this abstract are based on preliminary analyses using self-reported mental health. Further analyses based on mental health prescription data will subsequently be undertaken and presented at the ADR conference.

Download Full-text