Data for Public Benefit

Balancing the risks and benefits of data sharing

B Data for Public Benefit

ABOUT THE PARTNERS:

Involve is the UK’s leading public participation charity. We believe that people should be at the heart of
decision-making. Involve gives people the power to take and influence the decisions that affect their lives;
helps decision-makers to harness the potential of public participation; and supports people and decision-
makers to collaborate to achieve social change.

Previously, Involve co-ordinated a two-year open policymaking process for the Cabinet Office to try to
make progress in data sharing in three areas of proposed government legislation. The Digital Economy Act
contains the provisions to turn these proposals into law. An important strand of the discussions during this
process was around how best to define public benefit and balance it against privacy and wider risks. While
significant progress was made, the discussion involved a small number of civil society organisations and
focused around three relatively narrow areas of public policy. In addition, we believe that although the Act
provides for some data sharing between government departments, it will not resolve the complex set of
trade-offs between public benefit, privacy and security by itelf.

The Carnegie UK Trust’s Digital Futures theme seeks to maximise the ability for new technologies to
improve wellbeing and mitigate the risks of the digital world for citizens. As part of this the Trust wanted to
explore the trade offs we make between our personal privacy and giving others access to our data in order
to provide us with goods and services.

The Carnegie UK Trust works to improve the lives of people throughout the UK and Ireland, by changing
minds through influencing policy, and by changing lives through innovative practice and partnership work.
The Carnegie UK Trust was established by Scots-American philanthropist Andrew Carnegie in 1913.

Understanding Patient Data supports better conversations about the uses of health information.
Our aim is to explain how and why data can be used for care and research, what’s allowed and what’s not,
and how personal information is kept safe. We work with patients, charities and healthcare professionals to
champion responsible uses of data.

Lead Author: Kaela Scott.

Contributors: Simon Burall, Nicola Perrin, Philippa Shelton, Douglas White, Gail Irvine and Anna Grant.

April 2018

The text of this work is licensed under Digital Futures

the Creative Commons Attribution-
ShareAlike 3.0 Unported License. Policy
To view a copy of this license visit,
http://creativecommons.org/licenses Privacy and Public
by-sa/3.0/ or send a letter to Creative Benefit 2018
Commons, 444 Castro Street, Suite 900,
Mountain View, California, 94041, USA.
 Data for Public Benefit 1

Contents
Foreword 2

Data for Public Benefit Summary 4

Executive Summary 6

1. Introduction 12

2. Data Sharing and Public Service Delivery 14

Drivers for greater data sharing 14
Understanding data sharing 14
A focus on personal data 15
Data sharing and privacy 15

3. Understanding ‘Public Benefit’ in the context of data sharing 17

Identifying the benefits of data sharing 17
Evaluating the relative value of different benefits 22
Conclusions following the workshop process 25

4. Identifying Dimensions of Risk 26

Conclusions following the workshop process 31

5. Determining acceptable uses of data to deliver Public Benefits 32

Principles for assessing public benefit 32
Conclusions following the workshop process 38

6. Moving Forward.... balancing the risks of using data against ambitions

to deliver ‘Public Benefit’ 39
Framework for Evaluating Public Benefit 39
Framework for assessing the merit of a data sharing activity to deliver ‘public benefit’ 40
How this framework can be used within (and across) organisations 43
Involving the public in the discussion 43

Annexes 45
Annex A: Workshop Methodology 45
Annex B: Spectrum of Identifiability 47
Annex C: Legal Context for Data Capture and Sharing 48
Annex D: Defining Personal Data 50
Annex E: The Right to Privacy 51
2 Data for Public Benefit

Foreword
Involve, Understanding Patient Data and the Carnegie UK Trust all share a
collective interest in how local and national governments, civil society and the
public can have more effective conversations around data sharing, to enable
better informed, consistent and transparent data decisions to ultimately
create improvements in wellbeing for citizens.

Data – how it’s acquired, stored, and used – is one of the great social, economic and cultural issues of our
age. It has never been a more in-demand commodity. The ever-growing volume of data that we generate
about ourselves, and which is generated about us, has profound effects on how we consume, work and live.
The deployment of this data has produced a great many benefits, in particular the availability of content,
goods and services which are unprecedented in their level of personalisation, speed and responsiveness. At
the same time, the proliferation of vast amounts of data about who we are and how we live our lives brings
significant risk, to our privacy, our autonomy and our sense of identity. Addressing these challenges and
identifying where the acceptable trade-offs lie, appears certain to be a complex, high profile public policy
priority for many years to come.

For public services, the collection, use and sharing of data is implicitly understood as vital to the provision
of almost any effective service. However, despite the quantity of data that many individual public service
departments hold and have access to, the landscape to utilise this data to best effect across different
services remains complex and fragmented. Attempts at both national and local level to better share
personal data in order to improve public service delivery for citizens’ benefit have been challenged on the
grounds of data security and individual privacy concerns.

In this context, Involve, Understanding Patient Data and the Carnegie UK Trust, organised workshops in
six local authority areas to examine how the risks and public benefits associated with data sharing are
recognised, quantified and evaluated by stakeholders. Ultimately we wanted to understand the data trade-
offs that are currently being made every day across local government and civil society, and the reasons for
these decisions.

We found through our workshops a great deal of variation across services and localities in how public
benefit and risk in data sharing are defined and balanced. We also found a real commitment and desire
amongst public service providers to reach the best possible arrangements in each instance for citizens
and communities, as well as a real appetite to engage in the debate in order to shape future change and
improvement.

The purpose of our work was not to determine the absolutes of what is and what is not acceptable in terms
of data sharing risks and benefits. Indeed, the workshops demonstrated just how difficult it is to reach a
single common view. Our report is therefore designed to provide a framework to support government and
civil society organisations at both national and local level, to define, assess and evaluate the public benefits
that use of data may be able to deliver. This in turn will enable these organisations to more effectively,
consistently and transparently balance the full opportunities that data sharing presents against the risks
that it may incur.

In order for public service providers to be truly confident in data sharing decisions, it is vital that they
are able to engage the public in an informed and meaningful dialogue, and to give communities the
opportunity to shape the processes and systems through which decisions are made. We hope that our
report can also provide a platform for the advancement of these conversations.
 Data for Public Benefit 3

We are hugely grateful to all of our partners who supported us in this project. The input of Open Rights
Group, the Office of the National Data Guardian and The Centre of Excellence for Information Sharing was
critical in framing the project and supporting the workshop process. Our local partners Greater Manchester
Connect, Sheffield City Council, West Midlands Combined Authority, Melton Borough Council, Essex County
Council, Leeds City Council and Data Mill North generously hosted workshops and engaged public service
providers from across a range of services in their area. We are thankful to them all.

We look forward to working with a wide range of stakeholders, at national and local level, from across the
public service landscape, to help advance the issues and ideas set out in this report.

Douglas White Simon Burall Nicola Perrin

Head of Advocacy Senior Associate Head of Understanding Patient Data
Carnegie UK Trust Involve Understanding Patient Data
4 Data for Public Benefit

Data for Public Benefit Summary

Individuals
• Safeguarding
• Ensuring continuity of care and support
• Enabling the right services to be offered to individuals at the right time
• Co-ordinate of interventions and providing an integrated support service
• Connecting the administrative information held about citizens across departments to
simplify and streamline consumer interactions
POTENTIAL BENEFITS FROM DATA SHARING

• Identification of people who might benefit from a new, or different, service offer
• Automatically provide citizens with benefits they are entitled to

Service Providers (Short term) Service Providers (Longer term)

• Reducing duplication and waste • Producing local statistics

• Supporting effective, better targeted front- • Enabling the better monitoring and
line service delivery evaluation of service impacts
• Directing resource allocation • Facilitating evidence based policy
• Identifying instances of fraud and/or debt and decision making
• Monitoring demand and delivery patterns • Identifying the root causes of
• Improving levels of customer satisfaction problems
• Tracking a service user’s journey
• Predicting future service needs
Communities • Testing strategic and service
• Providing intrinsic benefits to society planning hypotheses
through access to better public services • Developing shared outcomes and
• Delivering improved outcomes for indicators across service providers
communities • Increasing public trust in
• Enabling Research government

Individuals
• Incursions into privacy from the use of identifiable personal information
POTENTIAL RISKS FROM DATA SHARING

• Risks from insecure data management and storage

• Risks from unintended re-identification
• Exposure to unwanted attention / service offers
• Punitive impacts

Service Providers
• Legal risks associated with data loss or misuse
• Loss of public trust from using data without consent
• Reputational risks from using sensitive personal data
• Risks from sharing data with organisations outside the public sector
• Using data for purposes that are not publically acceptable
• Risks to service provision from the use of unreliable data

Communities
• Stigma and discrimination
• The potential for negative impacts on communities from the selective use of data
• Loss of services
 Data for Public Benefit 5

The effective use of data needs

to meet three conditions:

Purposeful
Involving the Public
• provides direct and tangible benefits to individuals
• delivers positive social outcomes
• impacts on multiple beneficiaries aligns with public Underlying all of this work is the
expectations and does not over-step the boundaries of need for informed and meaningful
reasonable expectation conversations with the public.
• achieves long-term impacts by addressing the root-causes
of problems (for individuals or wider social issues)
• addresses a significant social problem Evaluating Public Benefit
• minimises negative effects
(Full Framework on Page 40)
5 key features that a data sharing
initiative designed to deliver
Proportionate public benefits should be able to
demonstrate:
• actively minimises the amount of data needing to be
shared 1. Enables high quality service
• considers whether personally identifiable data is delivery which produces better
necessary to achieve the goal outcomes for people, enhancing
• has clear parameters in order to protect against ‘mission their wellbeing.
creep’ and the risk of data being used for purposes other than
2. Delivers positive outcomes for the
which it was provided or shared
wider public, not just individuals.
• considers the likelihood of risks being realised and balances
the severity of a potential negative impact against this 3. Uses data in ways that respect the
• considers the sensitivity of the data being used and individual, not just in the method
whether sensitive personal data is needed of sharing but also in principle.
4. Represents, and supports, the
effective use of public resources
Responsible (money, time, staff) to enables
the delivery of what people need/
• is a ‘good’ use of data i.e. is a more efficient and effective want from public services.
way to add significant value to decision making or policy
implementation than other approaches 5. Benefits that are tangible,
• has effective measures in place to ensure that the data can recognised and valued by service
be used and shared securely providers and the wider public.
• will deliver the intended outcomes
• was justifiable
• is defensible
6 Data for Public Benefit

Executive Summary
Data is frequently touted as the new oil; value is to be extracted from it in
order to deliver services that the public wants and unlock new applications that
will make their lives easier. This is as true for personal data held by the public
sector as it is commercial data extracted from consumers.

Yet despite the many claims made for the public This report establishes a framework for those
benefit which can be derived from the use of providing public services to assess and evaluate
such personal data – and attitudes research the public benefits that the better use of data
that shows the public is much more likely to view may be able to deliver and attempt to balance
data sharing as acceptable if there’s a public this against the risks sharing data may entail.
benefit – the term ‘public benefit’ is rarely, if ever,
clearly defined. In a context of significant concern
about the public acceptability of much of the Determining acceptable uses of
data currently shared, this presents public service data to deliver Public Benefits
providers with challenges in deciding when they
should share data and for what purposes. There are clear tensions between reaping
particular benefits on one hand and mitigating
This report presents the findings from a series the range of risks associated with sharing
of six workshops in different local authority personal data on the other. Finding an acceptable
areas (Melton Borough Council, Essex County settlement between these benefits and risks
Council, Leeds and Sheffield City Councils, and remains a key challenge for policymakers,
Manchester and West Midlands Combined frontline staff, advocacy groups and the public
Authorities). These workshops brought together at large if the ambitions for data to be used as a
over 120 professionals from the public and tool for delivering public benefit are to be realised.
voluntary sectors (working in the fields of housing,
criminal justice, health, social care and welfare) Three clear tests emerged throughout the
to explore how they understand, define and value workshops as being necessary for public service
the public benefits which could be derived from providers to gain the social licence to share and
the use of personal data. use data more widely.
 Data for Public Benefit 7

a) Purposeful
When the purpose of a proposed use or sharing of data is clearly and transparently defined, participants were
readier to accept a proposed use of data. It was also felt this helped provide protection (for both the data
subject and the organisations involved) against data being used for purposes beyond that for which it was
initially shared.
A number of additional factors impacted on assessments of purposes geared towards delivering public benefit,
and the level of risk tolerable to achieve them, including that the use of data:
• provides direct and tangible benefits to individuals;
• delivers positive social outcomes (eg reducing social isolation, reducing inequalities or supporting local
area regeneration);
• impacts on multiple beneficiaries eg on individuals, services and the wider public;
• aligns with public expectations and does not overstep the boundaries of what the public could reasonably
expect data they had provided to be used for;
• achieves long-term impacts by addressing the root-causes of problems (for individuals or wider social
issues);
• addresses a significant social problem ie when the initiative tackled important and complex social
problems that other methods had failed to resolve;
• minimises negative effects (intended or unintended) for individuals and groups eg through excessive
incursions into privacy; punitive action; stigmatisation/discrimination; or the diversion of resources from one
area/sector of the population.

b) Proportionate
A number of factors contributed to assessments of proportionality in the workshops, including that the
proposed use of data:
• actively minimises the amount of data needing to be shared;
• considers whether personally identifiable data is necessary to achieve the goal;
• has clear parameters in order to protect against ‘mission creep’ and the risk of data being used for
purposes other than which it was provided or shared;
• considers the likelihood of risks being realised and balances the severity of a potential negative impact
against this ie that the existence of even a significant risk should not automatically discount an opportunity
being pursued;
• considers the sensitivity of the data being used and whether sensitive personal data is needed.

c) Responsible
Participants explored the idea that to be a justifiable use of public resources, the benefits likely to be achieved
by a particular use of data have to be balanced not just against the risks, but against delivering the intended
outcome.
A variety of factors were identified as contributing to whether a particular opportunity for data sharing could be
considered an efficacious use of public resources, including that a proposed use of data:
• is a ‘good’ use of data ie is a more efficient and effective way to add significant value to decision making
or policy implementation than other approaches;
• has effective measures in place to ensure that the data can be used and shared securely;
• will deliver the intended outcomes;
• was justifiable ie that service providers could demonstrate that they had considered the potential impacts,
done what they could to mitigate negative consequences, and ultimately determined that the benefits
outweighed the potential for harm;
• is defensible ie that a service provider would be able, and willing, to make a publically acceptable defence
of their decision to use data in a particular way if challenged.
8 Data for Public Benefit

Understanding ‘Public Benefit’ in the context of data sharing

Participants across the workshops focused on three groups which could directly benefit from the better use
of data:

Individuals. There was significant support for using data to provide benefits directly to
individuals, including for the purposes of, for example: safeguarding; ensuring continuity of
care; enabling service offers to individuals at the right time; simplifying user interactions with
services; automatically providing citizens with benefits they are entitled to.
Participants identified a number of criteria for determining whether such data sharing delivers public
benefits: the number of people able to benefit; the perceived level of need for example, having direct
impact on vulnerable groups (eg the homeless) and/or address key social problems (eg social isolation);
and/or having long-term impacts on individuals and the services available to people.

Public Service providers. There was a high level of support for using data to enable services to
be delivered more efficiently and therefore to support them to ‘do more with less’, for example,
by reducing duplication and waste, supporting better targeted front-line service delivery and/or
directing resource allocation.
Participants also identified the benefits that a more open flow of data between service providers could
bring for developing joint work focussed on outcomes, evaluating service impacts, identifying
the root causes of problems and predicting futures service needs.

Communities. Effective use of data should be able to deliver wider, positive social outcomes
beyond the benefits delivered to individuals and services. Dimensions of this type of benefit
include: providing intrinsic benefits to society through access to better public services;
delivering improved outcomes for communities; and enabling research, even if those benefits may
be less tangible and not immediately apparent.

Overall, the discussions at the workshops clearly demonstrated that each decision to share personal data
needs to be assessed on its own merits; there is no simple definition of public benefit which can be applied;
and that, at present, approaches to identifying and classifying benefits are inconsistent. Despite lacking a
common framework to follow, the three dimensions of public benefit are clear.
 Data for Public Benefit 9

Identifying Dimensions of R isk

The majority of the debate and concern about risks from sharing personal data focuses on the privacy of the
individual whose data is being shared. Throughout the workshops, however, participants raised a far wider range
of risks to individuals, organisations and the wider community. This suggests that there is a need for a renewed
and deeper societal debate to understand the extent to which the public share the same concerns and where
the deepest public disquiet lies.

The risks identified at the workshops can be summarised as being to three groups:

Individuals. Risks relating to individual privacy ranged from concerns about insecure data
management and storage and unintended reidentification to the wider concern that to
deliver the types of benefits outlined above, would require the use of types information that many
people would consider private, and may simply want to keep that way.
Participants were also concerned sharing personally identifiable information (particularly without the
direct consent of the individual) could expose individuals to unwanted or inappropriate offers of
service. Data sharing initiatives that resulted in punitive impacts for individuals also received a mixed
response from participants, particularly when data obtained for a different purpose was used to deliver
these consequences.

Communities. Concerns were raised that sharing and linking data for the purposes of targeting
public services, even if the information used was not personally identifiable, could result in the
production of generalisations that categorise individuals, social groups or geographic areas in ways
that could result in stigma, lead to discriminatory treatment, or inappropriate targeting.
Participants were also concerned about poor data quality leading to false conclusions being drawn
leading to, for example, evidence being developed for advocacy or targeting resources differently.
Further, as data may be able to support agencies to more effectively allocate resources, it can also
be used to reduce or remove services from areas or communities. While these decisions may be
evidence based, they can still lead to feelings of loss among communities.

Public Service Providers. Legal risks associated with data loss or misuse are perennial
concerns for service providers. However, even if a data use may be legally compliant, that may
often not be enough to assess it as being of low risk; service providers also identified risks associated
with loss of public trust from using data without consent (even if legally permitted) or for purposes
unacceptable to the public, for example.
10 Data for Public Benefit

Developing a framework to balance identified points of common ground that,

risk and benefit together, encapsulate the elements necessary
for a data sharing initiative to be described as
Choices about what data public service providers producing ‘public benefit’.
use and share, and for what purposes, requires
them to balance and assess the relative benefits The framework we have developed sets out a
and risks to individuals, the wider public, and series of evaluative questions, grouped around
services themselves. While there are a variety of each of these features, to help service providers
privacy impact assessment tools and planning and data controllers (in the first instance) clarify
frameworks that have been produced to assist the potential public benefits of using and sharing
organisations in assessing opportunities for data data to deliver public services and consider the
sharing within and between organisations most risks involved in doing so. The questions have
of these tend to focus on aspects of the data been selected in order to initiate discussions
itself. To date however, very little focus has been about the various dimensions that data sharing
given to supporting organisations to evaluate proposals can be measured against and, while
dimensions of benefit, assess the dimensions not exhaustive, our research suggests that
of risk and weigh up effectively the types of collectively they provide an effective framework
purposes that are valued most against the for assessing the acceptability and relative merit
potential for harm. of data sharing initiatives intended to deliver
‘public benefit’.
Our analysis of the debates and deliberations
that took place across the workshops has

Evaluating Public Benefit (Full Framework on Page 40)

We conclude that there are five key features that a data sharing initiative designed to deliver public
benefits should be able to demonstrate:

1. That it enables high-quality service delivery which produces better outcomes for people, enhancing
their wellbeing;
2. That it delivers positive outcomes for the wider public, not just individuals;
3. That it uses data in ways that respect the individual, not just in the method of sharing but also in
principle;
4. That it represents and supports the effective use of public resources (money, time, staff) to enables
the delivery of what people need/want from public services;
5. That the benefits are tangible, recognised and valued by service providers and the wider public.

The framework we have developed presents each of these features alongside a set of evaluation
questions that can be applied assess the potential of a data use to deliver public benefit.
 Data for Public Benefit 11

Moving forward... then the public also need to have the opportunity
to be involved in the discussions. This needs to
For organisations and partnerships that want to be undertaken in ways that enable the public to
unlock the potential of the data they and other help shape the future of data use by engaging
services hold about citizens, developing a shared in informed and meaningful dialogue with
understanding of the public benefits people service providers regarding their aspirations for
want and expect from data use, and a consistent how public services should be provided and their
language with which to talk about them, is vital. concerns about how data about them should be
used. This framework will also provide a useful
Working through the questions presented in tool for initiating these conversations with service
this framework with staff, partners and other users, community representatives, and ultimately
stakeholders will help service providers clarify the wider public, in ways that will increase their
their own understanding of the benefits and understanding of the complexities involved.
risks associated with the wider use of data and
become better able to articulate, and/or justify, While in the short term, the efforts required to
the decisions they may make. This, in turn, will begin this process are not insignificant, in the long
help set the stage for a wider, national strategic term, the costs to service providers of not taking
discussion that could help nurture a more the public with them on this journey are likely to
coherent approach to balancing the risks of be much higher. To not begin these conversations
using data against the goal of delivering ‘public now may undermine the reputation of public
benefit’. service agencies, hamper their ability to resolve
disputes, and ultimately constrain their ability to
If the social licence for greater data sharing is to use data in modern, beneficial and potentially
be realised the basis of delivering public benefit, transformative ways.
12 Data for Public Benefit

1. Introduction
Government and other agencies providing public services increasingly collect,
store and use personal data about citizens as a standard part of the business
of delivering services.

It is widely held that more effective use of this review of relevant literature on public attitudes
data could support agencies to provide better to data sharing. It establishes a proposed
targeted, more efficient services in ways that framework for service providers, across the public
deliver clear ‘public benefits’. This concept of and voluntary sectors, to both evaluate the public
‘public benefit’, however, is rarely defined, and it benefits that the better use of data may support;
can often mean very different things to different and assess this potential benefit against the risks
people in different contexts. that sharing data may entail.

This prompts fundamental questions about what Overview of the workshops

the public views as acceptable use of the data
held about them by public service providers, and The Better Use of Data: Balancing Privacy and
for what purposes these providers should have a Public Benefit workshops were designed to bring
social licence1 to use and share this data in the together a wide range of professionals working
pursuit of improved service delivery. Without a in the housing, criminal justice, health, social care
clear understanding of the public benefits that and welfare sectors2 to define and evaluate the
may be possible to achieve through the better ‘public benefits’ that may be delivered by the
use of data, and a shared language with which better use of data; consider the risks involved in
to discuss these benefits, these questions are data sharing; and begin to make sense of where
difficult to answer. an acceptable balance may lie between these
risks and benefits.
In order to establish a better understanding of
the issues at the heart of this impasse, Involve, In each area, the workshops were hosted by a
the Carnegie UK Trust and Understanding Patient local authority partner. These were:
Data organised a series of workshops in diverse
local authority areas across England during • Greater Manchester Connect;
the summer of 2017. These workshops brought • Sheffield City Council;
together professionals from the public and • West Midlands Combined Authority;
voluntary sectors to explore how they collectively • Melton Borough Council;
understood, defined and valued the public • Essex County Council;
benefits that may be delivered by the use of • Leeds City Council / Data Mill North.
personal data about service users and the wider
public. The purpose was to begin to make sense In the workshops, a series of examples and
of where an acceptable balance may lie between case studies were used to prompt discussions.3
the risks and benefits of data sharing and use in These case studies illustrated how data is
the context of public service provision.
2 While recognising that the issues and debates surrounding data sharing
cut across all aspects of public service delivery, the decision to focus
This report is informed by the findings from these this project on these areas of public service delivery was based on clear
workshops, alongside information drawn from a evidence of increasing demand within these sectors for the sharing of
personal data to support more effective multi-agency working at a local
level. Further, in all of these fields, decisions about what data to share,
1 Social Licence is a term emerging from debates about data sharing that when to share it, and who to share it with have potentially significant
are taking place in New Zealand and rests on the assumption that ‘when impacts for individuals and create challenging ethical dilemmas for
people trust that their data will be used as they have agreed, and accept professionals.
that enough value will be created, they are likely to be more comfortable 3 The examples prepared by Involve for use during the workshops can be
with its use.’ Data Futures Partnership (2015) What is Social Licence? viewed at https://www.involve.org.uk/wp-content/uploads/2017/07/
http://datafutures.co.nz/our-work-2/talking-to-new-zealanders/social- workshop-examples.pdf and the case studies at https://www.involve.org.
licence/ (accessed 2/6/2017) uk/wp-content/uploads/2017/07/3-case-studies-data-sharing.pdf
 Data for Public Benefit 13

currently being shared and used by public service Structure of this report
providers across the country for a variety of
purposes. Building in complexity throughout This report is intended to present an overview
each workshop, these case studies were used to of the findings from the workshops in order to
challenge participants to develop shared criteria build an over-arching picture of how service
to assess the relative acceptability of using providers from across these sectors understand
different types of data for different purposes. and evaluate the potential benefits and risks
associated with data sharing. It also looks at the
Throughout the workshops, participants also impact this has on assessments of acceptable
benefited from expert input from Understanding uses of data in the pursuit of public benefit.
Patient Data’s research into the best language to
use when discussing the different forms in which While the findings from the six local workshops
personal data can be shared4; members of the have informed the content of this report, the
National Data Guardian’s Panel highlighting how focus here is on the common and over-arching
challenges in relation to data sharing are being considerations, concerns and questions that
navigated in a health and social care context; emerged throughout the discussions.
and the Open Rights Group discussing the risks
and opportunities associated with data sharing. A Chapter 2 outlines the context of this
more detailed description of the rationale for the research and the drivers for increased data
workshops, an outline of the methodology used sharing in the context of public service
and an explanation of how participants were provision. It provides an explanation of key
selected can be found in Annex A. concepts and terms relevant to the debate in
order to establish a shared understanding of
In total, over 120 participants took part in these themes explored in the body of the report.
workshops – from city, borough and county
councils, the Police, the Fire Service, the NHS,
Chapter 3 examines how workshop
Housing Associations, Universities and voluntary
participants understood and evaluated the
sector organisations working in health, care,
public benefit associated with data sharing
consumer advocacy and welfare – each bringing
and compares how this relates to public
their own perspectives, experiences, reservations
views on these benefits as evidenced through
and aspirations to the discussion.
previous research.
At the conclusion of the workshops, a series of
standalone area reports were prepared and Chapter 4 explores the risks participants
distributed to workshop participants to support identified in relation to the use of data by
the continuation of the local dialogues that the public service providers.
process had initiated.5
Chapter 5 focuses on how participants
across the workshops assessed the overall
acceptability of potential data sharing
initiatives, balancing the trade-offs between
benefits and risks to determine appropriate
uses of personal data within the context of
public service provision.
4 Understanding Patient Data’s recent work on the best language to use
when discussing identifiability and anonymisation with a non-expert
audience, based on extensive testing with healthcare workers and
the public, was used to frame the discussion of these issues during The final chapter reflects on what the
the workshops. This language will also be used throughout this report
when referring to differing levels of identifiability and a summary is
findings mean for ongoing discussions about
included with this report as Annex B. Understanding Patient Data data sharing, public benefit and privacy in
(2017) What are the best words to use when talking about data https://
understandingpatientdata.org.uk/what-are-best-words-use-when- the UK and propose a framework for how
talking-about-data assessments of public benefit might be
5 The 6 area reports can be accessed at https://www.involve.org.
uk/2017/07/17/theres-benefits-talking-data-sharing/ advanced.
14 Data for Public Benefit

2. Data Sharing and

Public Service Delivery
This section of the report establishes the wider context in which the research
took place by outlining the drivers for data sharing in the context of public
service provision. It also provides an explanation of key terms and concepts, and
as such may be of particular value to readers less familiar with the topic.

Drivers for greater data sharing term data sharing is used to cover both systematic,
routine forms of sharing ie where the same data
Increased data sharing across and between public sets are regularly shared between the same
service providers is a cornerstone of ambitions to organisations for an established purpose, as well
modernise government and transform the way as exceptional, one-off processes of data sharing
public services across the UK are delivered. The for different purposes. It also includes data linking,
overarching rationale behind the drive to extend where identifiable information from two or more
the way data is used in public service provision is data sets, from two or more sources is matched
that the better use of data has the potential to: together.

produce direct benefits to individual service Advances in information technology, alongside

users by allowing more personalised and the rise in the number and type of organisations
targeted services to be provided; involved in delivering public services, have led to
significant increases in the amount of data being
create wider public benefit by increasing produced across governments and other agencies
the efficiency and effectiveness of public in recent years. Moves towards e-government
services overall; and have also resulted in the automatic capture of vast
quantities of data by public authorities as part of
deliver social outcomes that enhance the their routine business, for example for the purposes
wellbeing of people and communities. of registration, financial and service transactions
or record keeping. For the most part, however,
data generated across government and public
‘The ability to make easy data- service providers, is held and used solely within
driven decisions is becoming the organisation or department that collected
vital to the way that we all live and work. it. This creates a situation where there can be
This should be the way that government both duplication and, potentially, contradiction
provides services.’6 in the datasets agencies hold, and effective data
management is an ongoing challenge.

Understanding data sharing These same developments in information

management and communications technology
The process of data sharing is, fundamentally, the have also given organisations the ability to link
disclosure of data from one or more organisations and process large amounts of additional data. This
to a third-party organisation or organisations, or enables information about a specific individual or
the sharing of data between different parts of an event to be linked and used in ways that are not
an organisation. In the context of this report, the possible using any single set of records separately
and, if used effectually, can provide insight into
6 Cabinet Office (2017) Government Transformation Strategy. UK
how services can deliver better outcomes. However,
Government https://www.gov.uk/government/publications/government- it is not always easy for government and service
transformation-strategy-2017-to-2020 (accessed 28/5/2017)
 Data for Public Benefit 15

providers to gain access to data held by other public Personal data can be shared and used in a variety
service providers, particularly in a timely manner. of different forms, both anonymised and personally
identifiable. The technical language of identifiability
The legal context for data sharing between public is complex and many different words are used to
agencies is complex and continually evolving.7 Until describe the same thing in different contexts (for
recently the ability to share data has tended to example pseudonymised, key-coded, de-identified
involve bespoke, bilateral legal gateways established for limited disclosure). Throughout this report, the
for a specific purpose. The 2017 Digital Economy terms presented in Understanding Patient Data’s
Act8 aims to streamline these processes for a Spectrum of Identifiability – personally identifiable,
limited number of areas of public service delivery by de-personalised and anonymous – will be used to
creating a more permissive environment for data distinguish between different levels of identifiability
sharing across and between agencies to support inherent in data use.10
the better use of the information to inform policy,
planning and service delivery.
Data sharing and privacy

A focus on personal data A recent study into public attitudes to data sharing
found that one of the top reasons for the public
The range of data collected by public service opposing the increased use of the data held about
providers which could be shared to support public them was that: ‘People have a right to privacy’
service delivery covers a wide range of operational (32%).11 Individual privacy is defined by Privacy
and administrative information, for example International as the conditions that enable people
service level performance and budget records, user ‘to create barriers and manage boundaries to
satisfaction levels, impact monitoring statistics (eg protect ourselves from unwarranted interference in
reductions in ASB reports or homelessness figures) our lives, which allows us to negotiate who we are
etc. This study, however, explicitly focuses on the and how we want to interact with the world around
sharing of personal information about individuals. us. Privacy helps us establish boundaries to limit who
has access to our bodies, places and things, as well
Using the definition of personal data included in as our communications and our information’.12
the Data Protection Act,9 it is clear that personal
information held by governments and public The main threats to privacy from data sharing
services could take many forms: from directly can be summarised as:
identifiable fields like name, address, biometric
• the risk of data loss (through accident or
information or national insurance number, through
malice)
to information like ethnic background, health
• statistical disclosure (the potential to identify
records, disability, criminal record, income or credit
an individual within a dataset by their unique
history which, although not able to identify an
or rare combination of characteristics)
individual in isolation, could render them identifiable
• the potentially negative impacts of
if it was part of a collection of data held by an
secondary usage of data (through the
agency, or linked with data held by another agency.
disclosure or linking of information about
a person that they would prefer to have
remained private in a given context).

7 Annex C provides a more detailed summary of the legal context for 10 Understanding Patient Data (2017) What are the best words to use when
data capture and sharing, as provided to participants in advance of the talking about data https://understandingpatientdata.org.uk/what-are-
workshops to help inform their discussions. best-words-use-when-talking-about-data Further definitions of these
8 UK Government (2017) Digital Economy Act http://www.legislation.gov. terms can be found in Annex B.
uk/ukpga/2017/30/part/5/enacted (accessed 28/5/2017) 11 Ipsos Mori (2014) Public attitudes to the use and sharing of their data.
9 UK Government (1998) Data Protection Act https://www.legislation.gov. Royal Statistical Society https://www.statslife.org.uk/files/perceptions_
uk/ukpga/1998/29/contents (accessed 28/5/2017) Further information of_data_privacy_charts_slides.pdf
about how the Data Protection Act defines personal data, and sensitive 12 Privacy International (2017) What is Privacy?
personal data, can be found in Annex D. https://www.privacyinternational.org/node/54 (accessed 28/5/2017)
16 Data for Public Benefit

The challenge of ensuring that data sharing right to privacy is not an absolute right, and public
practices within public service provision protect authorities are permitted to share information
individuals’ right to privacy, however, is not a without consent if there are lawful gateways and
new one. The UK is a signatory to the European clear and proportionate reasons for sharing.13
Convention of Human Rights, so has incorporated
13 More information about an individual’s right to privacy can be found in
the right to privacy into national laws. However, the Annex E.
 Data for Public Benefit 17

3. Understanding ‘Public Benefit’

in the context of data sharing
The idea of delivering ‘public benefit’ is extensively used in policy documents
and research reports to describe the purpose of data sharing. To date,
however, there has been little examination of how either the public, or those
involved in delivering public services, actually understand and evaluate the
‘public benefits’ that data sharing may be able to deliver.

The only formal definition of ‘public benefit’ that Identifying the benefits of data
we were able to find during a review of literature sharing
undertaken for this project came from Statistics
and Registration Services Act (s7(2)). It states Throughout all of the workshops, there was a
that ’public benefit includes in particular (a) general consensus among participants that
informing the public about social and economic service providers need data, and that there are
matters, and (b) assisting in the development and many circumstances in which its collection and
evaluation of public policy’.14 While this definition use to improve the delivery of public services is
is functional, when interrogated in practice, constructive and justified.15
it emerges as being both overly broad and,
simultaneously, prescriptively limiting. Three groups of beneficiaries were focussed on by
participants when identifying the different types
In order to move discussions forward about the of benefits that the better use of data could bring:
potential benefits of data sharing, it is vital that individuals benefiting directly from the use of data
proponents are able to clearly articulate the collected about them; service providers deriving
different dimensions of public benefit in ways benefits from the use of personal data about service
that are easily understood and resonate with users or the wider public; and uses of data providing
data controllers, service providers and the wider wider benefits for people and communities. The
public. A key aim of this research was to develop a potential dimensions of benefits identified for each
more nuanced understanding of how the concept of these groups are outlined below.
of public benefit is understood by professionals
involved in the delivery of public services. a) Sharing data to provide
direct benefits to the
individual
‘If we can’t explain and
demonstrate the benefits of There are a range of well-established practices
sharing data, should we be doing it?’ for sharing personally identifiable data to support
WORKSHOP PARTICIPANT, ESSEX the provision of tailored and responsive services
to individuals: between GPs and other branches
of the NHS to ensure continuity of care; between
social work services and the police to protect
vulnerable adults; or between parole boards
and housing authorities to ensure appropriate
accommodation for offenders leaving custody.
15 It does, however, have to be noted that the participants in the workshops were
not recruited to be representative cross-section of practitioners in these fields.
While every endeavour was made to ensure that in each area the group was
as diverse as possible, the participants were invited from a self-selecting group
14 UK Government (2007) Statistics and Registration Service Act http:// who had expressed interest in spending a day exploring opportunities and
www.legislation.gov.uk/ukpga/2007/18/contents (accessed 28/5/2017) challenges relating to increased data sharing in their local area.
18 Data for Public Benefit

Traditionally this type of information sharing has However, as the Care Act places a fixed, non-
taken place directly between staff involved in discretionary obligation on staff to do so in
providing services to the individual eg through certain circumstances, data shared for the
discharge notes, case conferences or in response purposes of safeguarding has been excluded
to a specific request. In each case, the ability to from the remainder of these discussions as,
share information is permitted by legal gateways given it is a defined professional duty, it is
for data sharing and bespoke cross-sectoral outside the debate about when data should
agreements, which typically place constraints on be shared to deliver other public benefits.
information being shared on a ‘need-to-know’
basis and on what the information can be used for. • Sharing case information between services
to ensure continuity of care and support
As we have already noted, the context of public for an individual at points of transition eg
service delivery is changing and becoming patient care plans shared between the NHS
increasingly integrated: between government and community care providers upon discharge
departments; through partnership work from hospital;
across the wider the public sector; and with
commissioned agencies from the voluntary and • Sharing information to enable the right
private sectors. This is leading to new demands services to be offered to individuals at the
and expectations for data sharing in order to right time eg when the Department of Work
streamline the way services are delivered. The and Pensions notifies local authorities and
Digital Economy Act has responded to this by other social landlords when tenants receive
providing a new legal mechanism to enable Universal Credit so as to allow for the offer of
greater data sharing between specified public- Universal Support;
sector bodies and support the better use of
data for defined purposes, including targeted • Sharing information to co-ordinate
interventions ‘where its purpose is to improve the interventions and provide an integrated
welfare of the individual in question’.16 support service – Integrated Offender
Management programmes wherein
Findings from the workshops probation, police, local authorities, drugs and
Within this evolving context, participants in alcohol services and health providers take a
the workshops identified a number of different multi-agency approach to supporting and
rationales for data sharing in order to provide supervising persistent offenders on release
direct benefits to individuals, including: from custody to address issues which may
contribute to the risk of re-offending, such as
• Safeguarding – protecting an individual’s drug and alcohol addiction, homelessness,
rights to live in safety, free from abuse and unemployment, health problems and access
neglect.17 The need to share data for purposes to state benefits.
of safeguarding was raised as a key reason
for sharing data in all of the workshops. • Connecting the administrative information
held about citizens across government
16 UK Government (2017) Digital Economy Act http://www.legislation.gov.
uk/ukpga/2017/30/part/5/enacted (accessed 28/5/2017) departments to simplify and streamline
17 The ability to share the right information, at the right time, with the right consumer interactions – reducing the
people, is fundamental to good practice in safeguarding. The Care Act
2014 therefore permits sensitive or personal information about adults need for citizens to input address and other
to be shared for safeguarding purposes between the local authority and
its safeguarding partners (including GPs and health, the police, service
identifying data multiple times when, for
providers, housing, regulators and the Office of the Public Guardian), and example, applying for a passport, a driving
allows information to be shared without consent if: the person lacks the
mental capacity to make that decision; other people are, or may be, at licence and a Blue Badge.
risk, including children; sharing the information could prevent a crime;
the alleged abuser has care and support needs and may also be at risk; a
serious crime has been committed; staff are implicated; the person has • Sharing information held by a service to
the mental capacity to make that decision but they may be under duress
or being coerced; the risk is unreasonably high and meets the criteria for identify people who might benefit from
a multi-agency risk assessment conference referral; a court order or other
legal authority has requested the information. Care Quality Commission
a new, or different, service offer – a local
(2017) Safeguarding People http://www.cqc.org.uk/what-we-do/how-we- authority sharing details of those in receipt of
do-our-job/safeguarding-people (accessed 20/9/2017)
 Data for Public Benefit 19

council tax reductions due to disability with relating to how people balance concerns about
their recreation department to enable them how their information may be used when they
to contact residents to promote a new keep fit provide details online. For instance, it showed
service for disabled people. that ‘71% of consumers would provide more
information online if it helped them save money;
• Sharing information between services 60% would be willing if the resultant service was
to automatically provide citizens with better tailored to their needs; and 56% would be
benefits they are entitled to, eliminating the willing if it helped them to make better decisions’.20
need for them to proactively apply for support
eg as part of the Government’s programme b) Sharing data to deliver
to address fuel poverty, information held benefits to Public Service
by HMRC and the Department of Work and providers
Pensions would be matched with records held
by the Department of Energy and Climate In a climate in which growing pressure is being
Change and the Valuation Office Agency to placed on limited public-sector resources,
identify those on the lowest incomes living in governments at all levels have embraced the
the coldest homes so a rebate on their energy idea that there is a need for more joined up
bills could be automatically applied. working to ‘modernise’ and ‘transform’ public
service provision. The better use of data across
In most cases, participants in the workshops and between service providers is increasingly
were very supportive of using data in these ways, positioned as a vital tool for achieving this goal,
viewing sharing arrangements like these as vital by maximising the value of the information being
tools for enabling public service providers to offer collected by individual services.
the best possible services to people. The belief was
also widely expressed that this is what the public
expects from modern services, reflecting evidence ‘Data is a critical resource
from dialogues with the public which showed that for enabling more efficient,
they ‘commonly assumed that governmental effective government and public
administrative data is already linked and shared services that respond to citizen’s needs.
across departments, and [that they] supported Data acts as the foundation upon which
this for operational uses’.18 Indeed, the research everything else rests.’ 21
literature suggests that although most members
of the public express opposition when directly
asked if they support the use of data held about Findings from the workshops
them being shared, the majority become more Across the six workshops, participants readily
ambivalent where they can see direct benefits to identified situations where having access to more,
themselves from the process.19 high-quality information would have made the
work of their organisation easier, better targeted
In several of the workshops, participants also went and ultimately able to deliver better outcomes
as far as suggesting that public services providers for the public. This was, in fact, the dimension
had a lot they could learn from private companies of benefit where workshop participants were
in relation to using data to deliver more bespoke consistently able to list the greatest range of both
and ‘frictionless’ services to the public that are short-term and long-term benefits.
‘fit for the 21st century’. This conclusion seems to
be reinforced by findings from consumer research Some of the short-term / direct benefits to service
providers identified included:
18 Ipsos Mori (2014) Dialogue on data: Exploring the public’s views on using
administrative data for research purposes. Economic and Social Research 20 Coll, L. (2015) Personal Data Empowerment: Time for a Fairer Data Deal,
Council (ESRC). http://www.esrc.ac.uk/files/public-engagement/public- Citizens Advice; https://www.citizensadvice.org.uk/about-us/policy/
dialogues/dialogue-on-data-exploring-the-public-s-views-on-using- policy-research-topics/consumer-policy-research/consumer-policy-
linked-administrative-data-for-research-purposes/ research/personal-data-empowerment-time-for-a-fairer-deal/
19 Davidson, S. McLean, C., Cunningham-Burley, S., and Pagliari, C. (2012) 21 Cabinet Office (2017) Government Transformation Strategy. UK
Public Acceptance of Cross-Sectoral Data Linkages, Scottish Government Government https://www.gov.uk/government/publications/government-
http://www.gov.scot/Publications/2012/08/9455/0 transformation-strategy-2017-to-2020 (accessed 28/5/2017
20 Data for Public Benefit

• Reducing duplication and waste – • Facilitating evidence-based policy and

eliminating instances where the same decision making regarding what services
information is being collected, processed and are developed and offered to the public, on
maintained in multiple databases, thereby the basis of rigorously established objective
freeing up staff resources for other activities; evidence of ‘what works’, and particularly
‘what works’ in a local context, speeding up
• Supporting effective, better targeted front- cycles of improvement;
line service delivery by providing staff with
information from multiple sources to help them • Identifying the root causes of problems
make decisions about the support an individual by combining multiple data sets about the
might need, the level of risk they might be same people or issues to provide a better
exposed to, or whether other organisations picture of the range of contributing factors
should be involved in their case, and thereby and improve the accuracy of the diagnosis.
better manage cases and caseloads; This, in turn, helps service providers to identify
the best way to both tackle the problem at its
• Directing resource allocation – by using source and institute preventative measures;
data from multiple sources to identify crime
hot-spots or areas of multiple deprivation to • Tracking a service user’s journey to provide
enable resources to be used in ways that will longitudinal insight into the experiences of
have the most impact; service users and identify opportunities for
improvement. For example, one of the case
• Identifying instances of fraud and/or debt studies used in the workshops illustrated how
– maximising revenue generation by using anonymised linked data had been used to
Council Tax records to identify households map the interactions a very complex social
claiming single person council tax discount where care client had with the council, and external
there is more than one registered resident; agencies, over a 10-year period. This provided
new insights around the system as a whole
• Monitoring demand and delivery patterns and flagged where interventions could have
for different services across agencies to been made earlier to improve care;
ensure existing models of service provision are
making the best use of resources; • Predicting future service needs by better
understanding the changing make-up and
• Improving levels of customer satisfaction needs of their client group;
by making the public’s interactions with services
easier, more streamlined and more likely to • Testing strategic and service planning
deliver a positive outcome quickly and effectively. hypotheses by modelling the impact of new
service configurations on resourcing, outcomes
and budgets to identify where and how resources
Some of the longer-term, strategic and can be targeted to have the biggest impact;
operational planning benefits identified included:
• Developing shared outcomes and
• Producing local statistics that maximise the indicators across service providers
value of existing data to identify, for example, to support more integrated service delivery
where there is unmet need in the area or where practices and increase collaboration.
services are not meeting people’s needs;
• Increasing public trust in government
• Enabling the better monitoring and through the use of data in ways that
evaluation of service impacts – providing demonstrated transparency, accountability
evidence that interventions are making a and wider aspects of good governance.
difference in order to appraise options and inform
future service planning and commissioning;
 Data for Public Benefit 21

Overall, participants in the workshops generally being’24. A key goal of the workshops therefore
expressed high levels of support for using data to was to dig deeper into what service providers
enable services to be delivered more efficiently understand these benefits to include.
and therefore to support them to ‘do more with
less’. This sits well alongside research with the Findings from the workshops
public which found one of the main benefits Across all of the workshops, participants
identified from the collection and use of personal expressed the belief that the effective use of
data is to assist the Government in ‘identifying data should be able to deliver wider, positive
needs, planning resources and services, and social outcomes for people and communities
allocating funds’.22 (beyond the benefits delivered to individuals and
service providers). What these should be, however,
There was also particular interest in the was something that participants generally found
opportunities created by a more open flow much harder to articulate.
of data between service providers to develop
integrated, cross-service performance Three themes did emerge throughout the
management frameworks focussed on end discussions as being central to the idea of wider
outcomes. This was seen as a vital step for public benefit:
demonstrating the impact of ‘joined-up’
working and evidencing the impacts of co- • Providing intrinsic benefits to society
ordinated activity. There was, however, a degree through access to better public services:
of hesitancy about whether the wider public Throughout the workshops it was generally
would see data being used in for some of these agreed by participants that improving public
purposes as providing ‘public benefits’. service delivery overall provided knock-on
benefits for all citizens, even if they did not use
c) Sharing data to provide the specific services. This was largely due to
wider social benefits for the belief that increasing overall effectiveness,
people and communities and delivering efficiency savings, would relieve
pressure on the public purse.
The idea that data sharing can deliver wider
‘public benefit’, while regularly cited as an • Delivering improved outcomes for
important factor in increasing the social licence communities: Stimulated by the examples
for data use23, is, as noted above, rarely defined presented in the workshops, participants
further. Instead what these benefits might be made reference to social outcomes such
in practice tends to be variously glossed over as as improved community safety, reduced
impacts that are ‘in the public interest’, ‘for the social isolation, reduced inequalities and
greater good’, or ‘leading to the improvement of local area regeneration. This aligns with the
health, education or economic and social well- findings from previous social research which
showed that, when directly asked about their
expectations relating to wider benefits, the
public tended to understand it primarily in
terms of improvements to local services, local
22 Wellcome Trust (2013) Qualitative Research into Public Attitudes to
Personal Data and Linking Personal Data. Wellcome Trust https:// areas or public health.25
wellcomelibrary.org/item/b20997358#?c=0&m=0&s=0&cv=0
23 Ipsos Mori (2016) The one-way mirror: public attitudes to commercial
access to health data. Wellcome Trust https://wellcome.ac.uk/sites/ 24 Cabinet Office (2016) Better Use of Data – Consultation Paper. UK
default/files/public-attitudes-to-commercial-access-to-health-data- Government https://www.gov.uk/government/uploads/system/
wellcome-mar16.pdf; Aitken, M., de St. Jorre, J., Pagliari, C., Jepson, R. uploads/attachment_data/file/503905/29-02-16_Data_Legislation_
and Cunningham-Burley, S. (2016) Public responses to the sharing and Proposals_-_Con_Doc_-_final__3_.pdf; Davidson, S. McLean, C.,
linkage of health data for research purposes: a systematic review and Cunningham-Burley, S., and Pagliari, C. (2012) Public Acceptance of
thematic synthesis of qualitative studies. BMC Medical Ethics 17:73 Cross-Sectoral Data Linkages, Scottish Government http://www.gov.
https://bmcmedethics.biomedcentral.com/articles/10.1186/s12910- scot/Publications/2012/08/9455/0; Open Policy Making Process (2016)
016-0153-x; Sciencewise (2014) Big Data Public views on the collection, Conclusions of civil society and public sector policy discussions on data use
sharing and use of personal data by government and companies in government http://www.datasharing.org.uk/conclusions/index.html
Sciencewise http://www.sciencewise-erc.org.uk/cms/assets/Uploads/ 25 Davidson, S., McLean, C., Treanor, S., Cunningham-Burley, S., Laurie, G.,
SocialIntelligenceBigData.pdf; Ipsos Mori (2014) Public attitudes to and Pagliari, C. and Sethi, N. (2013) Public Acceptability of Data Sharing
the use and sharing of their data. Royal Statistical Society https://www. Between Public, Private and Third Sectors for Research Purposes, Scottish
statslife.org.uk/files/perceptions_of_data_privacy_charts_slides.pdf Government http://www.gov.scot/Publications/2013/10/1304/0
22 Data for Public Benefit

• Enabling Research: Providing data for When individuals were the main
research purposes was also seen as an beneficiaries
important aspect of delivering public benefit,
even if those benefits may be less tangible Overall, when workshop participants came to
and/or not immediately apparent. Previous evaluating the comparative public benefits of
engagement with the public suggests that different forms of data sharing, a key criterion
they are also cognisant of the value of the in all locations was the ability of an initiative to
long-term impacts that wider social research deliver clear, demonstrable benefits to individual
can be enable. A systematic review of 25 service users. This corresponds with messages
studies examining public attitudes towards the drawn from research with the public which
sharing or linkage of health data for research, suggest that ‘personal benefit is the strongest
for example, showed that, while individuals incentive for being in favour of the collection
receiving direct benefits from research uses and use of personal data by government and
of data was important to the public, overall companies’ and further that the offer of a
societal benefits were also assessed as specific, tangible benefit has a significant impact
valuable, particularly when they were able on the public acceptability of using personal
to demonstrate a ‘real-world . . . practical data.27
application’.26

However, it was repeatedly noted that in order to ‘If we could see a benefit to an
improve public awareness of how data sharing individual or a life then we saw
can be deployed to deliver benefits like this, public this as most important.’
service providers need to get better at ‘telling WORKSHOP PARTICIPANT, MANCHESTER
the stories’ of how the effective use of data has
delivered outcomes for people and communities.
The differences across the workshops, however,
were in relation to what type of benefits
Evaluating the relative value of individual participants valued most highly.
different benefits
• For some groups, the number of people able
A core objective of this project was to investigate to benefit from a data sharing initiative was a
how stakeholders working in different aspects principle factor for attributing value. Examples
of public service delivery not only understood that used data to directly identify individuals
different aspects of public benefit, but how in need of a specific intervention to improve
they valued the different types of benefits their wellbeing, or to apply a direct benefit (a
the better use of data could deliver. In each rebate or discount), therefore tended to be
workshop therefore, participants were asked to evaluated highly.
comparatively evaluate a number of examples of
data sharing practice, purely in relation to their • The perceived level of need was another
ability to deliver public benefits. The participants significant factor for some participants.
were explicitly asked not to take the risks Examples that were seen to have direct
associated with the initiatives into account when impacts on vulnerable groups (eg the
weighing up the relative values of the different homeless) and/or which set out to address
benefits. key social problems (eg social isolation) were
therefore attributed greater value. The degree
of positive impact on individuals, and the
depth of that impact on their wellbeing, also
contributed to participants’ assessments. As
26 Aitken, M., de St. Jorre, J., Pagliari, C., Jepson, R. and Cunningham-Burley,
S. (2016) Public responses to the sharing and linkage of health data 27 Sciencewise (2014) Big Data Public views on the collection,
for research purposes: a systematic review and thematic synthesis of sharing and use of personal data by government and companies
qualitative studies. BMC Medical Ethics 17:73 https://bmcmedethics. Sciencewise http://www.sciencewise-erc.org.uk/cms/assets/Uploads/
biomedcentral.com/articles/10.1186/s12910-016-0153-x SocialIntelligenceBigData.pdf
 Data for Public Benefit 23

one group from the workshop in Manchester • The ability of data to be used to help inform
noted, they made their evaluation based on resource allocation. Examples that allowed
the ‘level of need of beneficiaries, the urgency service providers to make efficiency savings
of the problem and the level of positive impact by more effectively allocating staff and
it would have on beneficiaries’. resources were generally seen as producing
wider benefits, by freeing up scarce public-
• Long-term impacts on the services available sector resources for other uses, and thus
to people was also seen by some groups as a ranked highly. Examples that had the potential
critical criterion. In these cases, examples that to impact on long-term service planning and
demonstrated how large-scale, linked data efficiency, by using data to monitor patterns
sets could be used to both monitor service of service use, predict future demand and
performance and identify which services identify opportunities for innovation, however
would be most beneficial to individual users tended to be ranked even higher.
were particularly valued. Examples that used
large aggregate data sets to identify trends • When the use of data enabled service
(eg areas of multiple deprivation) and provide providers to produce evidence of ‘what
evidence for targeting resources differently, works’. For instance to evaluate the impact of
were also seen as able to provide benefits to interventions, demonstrate value for money
large numbers of people in the long term. and inform future service design in ways that
would ultimately improve outcomes for service
users then this was seen to deliver wider public
‘Using data well now can benefits. This led to examples like the Justice
mean future service users get a Data Lab being assessed as very beneficial.
better service – or even don’t need the In this example, organisations working with
service.’ offenders would share information about
WORKSHOP PARTICIPANT, SHEFFIELD their participants with the Ministry of Justice.
The MOJ then compares rates of re-offending
among participants in these programmes
with re-offending rates among a matched
When Public Service providers control group to provide the organisation
are the main beneficiaries with information they can use to evaluate the
effectiveness of a specific intervention.
When organisations and service providers
themselves were seen to be the sole, or primary, • Examples that were designed to use data to
beneficiary from a data sharing initiative, identify the underlying causes of social
participants tended to assess this scenario as problems and reduce the need for crisis
having lesser public benefit value, expressing services in the future were also highlighted.
concerns about whether the public would see These types of examples were viewed by
this as an acceptable use of the data held many as having the ability to inform strategic
about them. In contrast, when service providers service changes in ways that could have
were identified as being one among a number significant impacts on both public service
of potential beneficiaries from a data sharing providers and community wellbeing in the
initiative, support for the initiative’s ability to future. As participants in the workshop in
deliver public benefit significantly increased. Melton Mowbray noted: ‘If you reduce need,
you reduce costs!’
Of the different types of benefits that could be
accrued by public services, participants ranked • When data was used to enable a more
the following four most highly: holistic approach to service provision,
through integrated working and sharing
information across multiple service providers
then these initiatives were also favoured, for
24 Data for Public Benefit

example, using data to track service users’ benefits to individuals (by ensuring patients
journeys in ways that could inform long-term were offered appropriate services at the right
improvements in when and how services are time); to staff (by enabling more efficient and
offered and delivered. co-ordinated working); and to services as a
whole (by increasing their ability to monitor
performance, evaluate impact and use this
information to inform service planning).
When the wider public is the This type of approach to data sharing was
main beneficiary assessed as having clear public benefits in
the workshops, as it was seen to have the
As already noted, the importance of the wider potential to deliver cumulative impacts across
public being a beneficiary from the use of data services and wider society.
was seen as vital for obtaining social licence for
the greater use of data. This seems to resonate • In a number of the workshops, participants
with the research that shows that, while personal used the criterion of positive social
benefit is the strongest incentive for members outcomes to rank highly examples that
of the public being in favour of the collection focussed on addressing fundamental social
and use of personal data, it is closely followed by problems (like social isolation, violence in
‘public goods’ (health improvement, prevention communities and homelessness). When data
and detection of crime, and the detection was used to undertake social research, for
fraudulent behaviour).28 example into the impacts of poor housing on
long-term health, these were also rated highly
Examples that participants believed could for providing evidence that could be used to
generate positive social outcomes (rather than deliver wider social benefits if applied to policy
simply benefits to individuals and/or service making.
providers), therefore tended to be evaluated most
highly in terms of delivering public benefits. • Improving the effectiveness and efficiency
of public service provision overall was also
• A key criterion across most of the workshops seen as a ‘win-win,’ benefiting individuals,
was the ability of an initiative to have public agencies and the wider public. These
positive impacts on multiple types of types of benefits were variously described
beneficiaries, such as individuals, services across the workshops as the ‘knock-on’ or
and the wider public. The example of the ‘ripple effects’ of a data sharing initiative.
integrated database for Health and Social They included outcomes such as increased
Care providers in Hertfordshire was highlighted partnership working, efficiency savings
as archetypal in this regard: able to provide decreasing demands on the public purse
and reduced need for services due to early
28 Suherman-Bailey, J. (2015) Data policy and the public: shaping a deeper
conversation. Sciencewise http://www.sciencewise-erc.org.uk/cms/
intervention.
assets/Uploads/Data-policy-and-the-publicJan-2015.pdf
 Data for Public Benefit 25

Conclusions following the workshop process

Across all of the workshops, participants were easily able to identify a range of opportunities for
increased data sharing in their own work or organisation and recognise the positive impact that the
continued development of data use across their local area could make to service delivery. When it
came to being able to articulate the wider public benefits that the better use of data could deliver,
participants generally had more difficulty.

Importantly, it also became apparent very early in the process that public service providers do not have
anything approaching a common framework to draw upon in order to identify, classify or evaluate the
benefits that data sharing may be able to provide. This was found to be equally the case in areas where
there was already a strong leadership commitment and resource investment in improving how data is
being shared, as it was in areas where processes were less developed.

This lack of consistency in the way the potential benefits from using data are understood and
evaluated has a direct impact on what different stakeholders consider to be beneficial uses of data.
This also has clear implications for how service providers perceive and assess the risks associated with
greater data use, as discussed in the next chapter of this report.
26 Data for Public Benefit

4. Identifying Dimensions of Risk

Whenever personal data is collected, accessed, analysed, shared or linked
there is some risk to individual privacy. Much of the current debate around when
personal data about service users and the wider public should be used to support
the better delivery of public services has therefore focussed on whether data
can be used in ways that protect and enhance privacy, while still enabling service
providers to maximise value for service planning and delivery.

In framing this research, our initial focus was on between what counts as personally identifiable data
determining how professionals working across the and anonymised data are not that clear cut.
public and the voluntary sectors identified and
assessed the risks to privacy that the increased Fundamentally, in order to achieve the types of
use and sharing of data might entail: with a benefits outlined in the previous sections, the
particular focus on how issues such as the type information that agencies would need to share
of data shared; levels of informed consent; and about individuals is likely to include a wide range
the type and number of organisations data was of personal data, including potentially sensitive
shared with impacted upon assessments of risks personal data. It could include information
to privacy. Throughout the workshops, however, it supplied for the purposes of claiming benefits,
quickly became apparent that individual privacy personal medical or financial information, or
was only one dimension of risk that participants records demonstrating eligibility for support
considered when identifying the potential harm services like counselling or housing assistance.
that a data sharing initiative may cause – to In short, information that many people would
individuals, communities and services themselves. consider, and want to keep, private.

a) R isks to individuals from When services use this type of information,

the use of personal data it has the potential to create a range of risks,
and indeed perceived harms, to individual data
Previous research exploring public attitudes towards subjects. A range of dimensions of risk were
public sector data sharing suggests that a key identified across the workshops.
concern for most people is whether the information
would be personally identifiable.29 In most cases, • Incursions into privacy from the use
it seems that people intuitively understood this of identifiable personal information:
to mean whether their name, address or another Unsurprisingly, participants in the workshop
unique identifier like NHS number or National were most comfortable with personal data
Insurance number would be disclosed. However, being used when it was shared in anonymous,
as Understanding Patient Data’s work on the aggregated forms, assessing this as posing a
Spectrum of Identifiability30 demonstrates, the lines minimal risk to individual privacy. This aligns
clearly with findings by the Economic and Social
29 Aitken, M., de St. Jorre, J., Pagliari, C., Jepson, R. and Cunningham-Burley, Research Council which showed that 61% of
S. (2016) Public responses to the sharing and linkage of health data
for research purposes: a systematic review and thematic synthesis of the public did not care how their personal data
qualitative studies. BMC Medical Ethics 17:73 https://bmcmedethics.
biomedcentral.com/articles/10.1186/s12910-016-0153-x; Davidson, S., was used, as long as it was anonymised and
McLean, C., Treanor, S., Cunningham-Burley, S., Laurie, G., and Pagliari, C. could not be linked back to them.31
and Sethi, N. (2013) Public Acceptability of Data Sharing Between Public,
Private and Third Sectors for Research Purposes, Scottish Government
http://www.gov.scot/Publications/2013/10/1304/0; Wellcome Trust
(2013) Qualitative Research into Public Attitudes to Personal Data and Participants were also generally comfortable
Linking Personal Data. Wellcome Trust. https://wellcomelibrary.org/item/
b20997358#?c=0&m=0&s=0&cv=0 31 Ipsos Mori (2014) Public attitudes to science. Department for Business,
30 Annex B Understanding Patient Data (2017) What are the best words to Innovation and Skills (BIS) and the Economic and Social Research Council
use when talking about data https://understandingpatientdata.org.uk/ (ESRC). https://www.ipsos.com/ipsos-mori/en-uk/public-attitudes-
what-are-best-words-use-when-talking-about-data science-2014
 Data for Public Benefit 27

with de-personalised linked data being used Across the workshops, most participants
within a particular service, or as part formal agreed that compliance with regulation,
partnership arrangement, to enable service strong controls on who is able to access
improvements, monitor impact and/or inform personal information and good data sharing
organisational strategic planning. This position governance practices should be able to ensure
appears to align with public attitudes research that security risks associated with a data
that showed ‘there was near universal breach or loss, or the intentional misuse of
acceptance of public bodies’, including the data were minimised.
Government, the NHS, local authorities and
the police, having access to anonymised • Risks from unintended re-identification:
personal data from other organisations for While the use of de-personalised linked
research and planning purposes.32 data was recognised as a valuable tool for
service providers, and broadly supported
Views toward sharing personally identifiable by participants in the workshops, they
information were generally more mixed, were generally less comfortable with de-
with health information highlighted as being personalised data being shared when the
particularly private, sensitive and open to misuse data sets were small, as it was feared that this
should they be disclosed inappropriately. In increased the likelihood of re-identification.
several of the workshops, however, participants An example of an A&E department which
did question whether the potential negative routinely shares de-personalised patient
impacts of this were any more significant or information related to violent crime injuries
damaging to an individual’s wellbeing and life with the Community Safety Partnerships
chances than the disclosure of other forms (about the time, date and location of the
of personal data, particularly data relating to incident and the primary means of assault ie
interactions with the criminal justice system or weapon or body part used) in order to allow
financial and/or benefits records. police to identify and target violence hot-
spots, raised particular concerns in this regard.
Overall, however there was a general It was considered to be a situation in which
consensus among participants that public the potential re-identification of individuals
services should avoid sharing the information may be likely and also that the impact of
they held in personally identifiable forms re-identification would represent a significant
unless the objectives of the initiative could not potential cost to the individual by making
be met without it. them vulnerable to punitive consequences,
such as punishment from the police and/or
• Risks from insecure data management and from others within the community.
storage: Uncertainty about whether personal
data can, in practice, be securely collected, • Exposure to unwanted attention /
stored, shared and used invokes an inherent risk service offers: When personally identifiable
to individual privacy and has major implications information is shared between agencies,
for whether or not agencies and individuals particularly without the direct consent of the
are supportive of data sharing. While technical individual, this could expose individuals to
debates regarding whether it is possible to unwanted or inappropriate offers of service.
guarantee the security of information sharing Participants worried that this could result
technologies were explicitly placed outside the in harmful outcomes for individuals if they
scope of this project, the impact that these became hesitant about accessing necessary
concerns have on attitudes to data sharing are services due to concerns about how the
real and justified and were raised during all of information they provide may be used.
the discussions.

32 Davidson, S., McLean, C., Treanor, S., Cunningham-Burley, S., Laurie, G.,
and Pagliari, C. and Sethi, N. (2013) Public Acceptability of Data Sharing
Between Public, Private and Third Sectors for Research Purposes, Scottish
Government http://www.gov.scot/Publications/2013/10/1304/0
28 Data for Public Benefit

• Punitive impacts: Across the workshops, data social groups or geographic areas in ways
sharing initiatives that identified individual that could result in stigma. The example
cases of fraud, error or debt received a mixed referred to above, where information about
response from participants, particularly when violent assaults was shared between A&E
data about individuals was used in ways people departments and Community Safety
were unlikely to expect (for example matching Partnerships, proved to be particularly
properties receiving an Empty Homes Council contentious in this regard. While participants
Tax discount with credit card registrations recognised and valued the public safety
to identify lived-in properties). While civil benefits that an initiative like this could
society campaign organisations working in bring, many felt that the potential negative
this area tend to argue that ‘it is unethical for impacts on communities from being identified
improvements in technology for collecting as ‘violent hot-spots’ or as areas in need
and analysing data to lead to sanctions for of additional police attention could lead to
individual citizens, even if an argument around discriminatory treatment or inappropriate
the wider public benefit can be made’33, targeting. This reflects fears raised by the
conclusions drawn from research with the public public during consultations related to the
on this matter appear contradictory. Some Connected Health Cities initiative, where one
studies have shown that the public were broadly of the main concerns raised by the citizen
supportive of data being used to unearth jurors was that the proposed use of data
dishonesty (for example with 72% agreeing ‘may lead to an increase in geographic,
that tax and benefit records should be used to community-based, and social stereotyping
prevent fraud)34 while others have concluded and stigmatisation as well as an inequitable
that the public are very concerned about data distribution of resources’, in other words a
being used by the Government to punish or ‘postal code lottery’ in relation to service
withdraw a benefit or service from individuals.35 provision across different areas.36

• The potential for negative impacts on

b) R isks to communities and communities from the selective use of data:
the wider public from the use Concerns were raised during the workshops
of personal data that the selective use of data (particularly
without wider qualitative or contextual
Just as the better use of data may have the information) could result in erroneous or
potential to deliver benefits to communities and questionable conclusions being drawn. This
the public, there is also a risk that data can be used was particularly considered to be a risk where
in ways that have negative impacts on specific data was being used to develop evidence for
areas or groups. Opportunities for potential harm advocacy or targeting resources differently. The
identified in the workshop are summarised below. example where data from the index of multiple
deprivation was combined with records showing
• Stigma and discrimination: Concerns were patterns of GP use to argue for new models
raised that sharing and linking data for the of funding GP surgeries in deprived areas, was
purposes of targeting public services, even cited as a case where data might be selected to
if the information used was not personally confirm a pre-existing hypothesis. Participants
identifiable, could result in the production of suggested that if either different data was
generalisations that categorise individuals, used, or an alternative approach taken to
analysis, the data might ‘tell a different story’.
33 Open Rights Group (2016) Consultation Response Data Sharing https://
www.openrightsgroup.org/ourwork/reports/orgs-response-to-data- • Loss of services: just as the better use of
sharing-consultation
34 Ipsos Mori (2014) Public attitudes to the use and sharing of their data. data may be able to support agencies to
Royal Statistical Society https://www.statslife.org.uk/files/perceptions_
of_data_privacy_charts_slides.pdf
35 Wellcome Trust (2013) Qualitative Research into Public Attitudes to 36 Citizens Juries c.i.c (2017) Connected Health Cities Citizens’ Juries
Personal Data and Linking Personal Data. Wellcome Trust. https:// Report. NHSA https://www.connectedhealthcities.org/wp-content/
wellcomelibrary.org/item/b20997358#?c=0&m=0&s=0&cv=0 uploads/2016/08/CHC-juries-report-Feb-2017.pdf
 Data for Public Benefit 29

more effectively allocate resources on the • Reputational risks from using sensitive
basis of areas of greatest need and evaluate personal data: The Data Protection Act
the demand and performance of different recognises some forms of personal data as
services, it can also be used to reduce or being particularly sensitive, and of a particularly
remove services from areas or communities. private nature, because information about these
While these decisions may be evidence based, matters could be used in a discriminatory way.37
they can still lead to legitimate feelings of loss In the workshops, financial and health data were
among the communities affected. highlighted as being particularly sensitive forms
of data that service providers are often especially
cautious about using.
c) R isks to public service
providers from the use of One example discussed was the new powers
personal data granted to government in the Digital Economy
Act to share identifiable tax credit information
The participants in the workshops identified a about citizens with licenced energy suppliers
number of risks to service providers that they felt so that companies can automatically apply a
needed to be considered alongside risks to individual ‘Warm Home’ rebate or offer support under
privacy if ambitions for the increased use of data to Energy Company Obligation schemes. This was
inform service provision were to be realised. considered by many as a significant incursion
into privacy. Participants tended to agree that
• Legal risks associated with data loss or since it was unlikely that most people would
misuse: Participants in all of the workshops expect this type of financial information about
noted that this was a perennial concern them to be shared without their explicit consent,
for service providers when it came to using the risk associated to organisations through the
personal data. While, as noted above, there use of this type of data was heightened.
was general agreement that effective data
management controls could minimise the
risk of unintentional loss or disclosure, there “the standards for demonstrating
remained concern that lack of awareness or public benefit have to be seen
malicious acts still had the potential to expose as higher for some areas of public policy
an organisation to significant risk. where the possible harm caused by the
intrusion is higher.”
• Loss of public trust from using data without WORKSHOP PARTICIPANT, WEST MIDLANDS
consent: As noted in the introduction to this
report, public authorities are permitted to share
information without the consent of the data Although health and care data was also
subject through a wide range of legal gateways. identified as being particularly sensitive, it was
For many participants in the workshops, the also noted that the public generally expect
fact that a data sharing initiative may be legally this type of information to be shared between
compliant was very often not enough to assess it service providers who are involved in supporting
as being of low risk. Instead, public expectation their wider wellbeing (and often assume
was considered a significant factor, with that it is shared more widely than it usually
participants tending to make their judgement is in practice). The perceived acceptability of
based on whether a proposed use of data was sharing this information within a context of
something that the public was likely to be aware providing direct care, despite its sensitivity,
of or could reasonably expect. This led to some aligned clearly with findings from research with
of the examples considered that used data
without consent being assessed as high risk for 37 Sensitive Personal Data is denoted in the Act as being information related
to racial or ethnic origin, political opinions, religious beliefs, trade union
service providers, even if they were compliant membership, physical or mental health, sexuality and criminal justice
history, and there are provisions within the Act to ensure that personal
with the principles of the Data Protection Act. information of this type is treated with greater care than other types of
personal data. UK Government (1998) Data Protection Act https://www.
legislation.gov.uk/ukpga/1998/29/contents (accessed 28/5/2017)
30 Data for Public Benefit

the public which showed there is ‘a perceived • Using data for purposes that are not publicly
unquestionable benefit to people in terms acceptable: Concerns were also raised during
of experts having information about their the workshops that uses of data that primarily
health, in relation to illness or avoiding it’.38 benefited service providers – particularly uses
When workshop participants were discussing that had punitive impacts on individuals – may
examples of personally identifiable health data not be considered an acceptable use of personal
being shared outside a direct care context (for data by the public. This nervousness was
example with police or housing authorities) particularly apparent in response to the one of
they tended to express more caution, arguing the case studies used in some of the workshops
that the impacts of it being disclosed, lost or – the Camden Residents Index which brings
otherwise misused could be very significant for together data from 16 council business systems,
individuals, and result in significant reputational covering 123 fields of primarily demographic
damage to service providers. information, to create a complete picture of each
resident in order to streamline Council processes
• Risks from sharing data with organisations and residents’ interactions with services. Here,
outside the public sector: Across the participants were especially concerned about
workshops, a number of the examples presented how the public would evaluate this model of
involved data sharing between the public and data sharing as the benefits identified were
voluntary sectors and/or with private companies. predominantly for the service provider itself (eg
Overall, there was very little objection raised validating residency for accessing council services
to the principle of sharing data with the such as school places), and included punitive
voluntary sector and in fact, it was widely seen impacts for individual residents (eg flagging
as increasingly necessary in order to enable cases of illegal subletting). Overall in this case,
effective partnership working. However, it was while participants could recognise the benefits
acknowledged that the public may not expect to a local authority of this type of systematic
their information to be shared in this way. internal data sharing, many remained nervous
about how an initiative like this would be
The intuitive response to data sharing with the perceived by the public.
private sector from many workshop participants
was that it posed a significant risk, as they did • Risks to service provision from the use of
not trust commercial organisations to behave unreliable data: participants identified that
responsibly with data. However, participants there were significant risks to service providers,
tended to become more accepting when they and ultimately to the services provided to the
considered the reality that any data sharing public, if flawed or incomplete data was used to
arrangement between a public body and a inform policy decisions or resource allocations.
private company would likely be based on a Participants in many of the workshops
contractual arrangement to deliver services and recognised that public service organisations
have strict data use controls in place. do not always have effective and reliable
mechanisms in place to collect the right types
Overall, however, in relation to sharing public of information, categorise it consistently and
sector data with the voluntary or private update it regularly – noting that ‘any system is
sector, it appeared that workshop participants’ only as good as the data you put in’. This created
evaluation of risk rested more on the potential a nervousness that services using data to inform
for gain in terms of likely derived benefit, decisions might be relying on unreliable or
and on the process and controls in place to potentially out of date information.
mitigate data privacy risks, than the type of
organisations the data was being shared with.

38 Wellcome Trust (2013) Qualitative Research into Public Attitudes to

Personal Data and Linking Personal Data. Wellcome Trust https://
wellcomelibrary.org/item/b20997358#?c=0&m=0&s=0&cv=0
 Data for Public Benefit 31

Conclusions following the workshop process

Throughout the workshops, participants from both the public and voluntary sectors identified a wide range
of potential risks from sharing and using data to deliver improved public services. The risks they identified
were far wider than the risks to privacy which currently dominate the debate about data sharing. This
suggests that there is a need for a renewed, and deeper societal debate to understand the extent to which
the public share the same concerns and where the deepest public disquiet lies.

It was also very clear that as in the case of assessing and evaluating potential benefits, professionals
working across these field have no common framework for attributing risk when considering potential
data sharing initiatives. Instead, as many participants readily admitted, the default position they
adopted in practice was one of caution and restraint.

In the next chapter, we explore how participants went about balancing the risks and benefits related
to different examples of data sharing practice when challenged to assess the relative acceptability of a
potential data sharing initiative.
32 Data for Public Benefit

5. Determining acceptable uses of

data to deliver Public Benefits
It is clear from the previous chapters that there are tensions between
expanding opportunities for data sharing, protecting privacy and mitigating the
wider risks associated with sharing personal data. It cannot be assumed that
these are entirely compatible goals.

Finding an acceptable settlement between the That data use should be purposeful
use of data to deliver services that benefit the
public and protecting people’s privacy therefore When the purpose of a proposed use or sharing
remains a key challenge for policy makers, of data is clearly defined, and presented
frontline staff, advocacy groups and the public transparently, participants more readily accepted
at large if the ambitions for data to be used as a it. Across all of the workshops, there was a
tool for delivering public benefit are to be realised. general resistance to data being shared and/
or linked for speculative purposes or ‘fishing
expeditions’, as this was felt to be one of the
Principles for assessing public benefit quickest ways to lose public support for the use
of data held about them. It was also argued
Discussions during the workshops raised a range that clarity and transparency of purpose helped
of issues related to appropriate purposes, groups provide protection (for both the data subject and
of beneficiaries, types of data being shared, and the organisations involved) against data being
the privacy incursions and associated risks that used for purposes beyond that for which it was
need to be identified, acknowledged and weighed initially shared.
up before the relative acceptability of different
forms of data sharing to deliver public benefits can Additionally, for a use of data to be accurately
be determined. Three clear principles, however, described as delivering ‘public benefit’, workshop
emerged throughout the workshops as being participants tended to agree that it needed to be
necessary conditions for public service providers geared towards a publicly acceptable purpose.
to gain the social licence to share and use data There were, however, a wide range of opinions
more widely based on the promise of delivering expressed as to where the boundaries of public
public benefits: purpose, proportionality and support would lie, ranging from undisputed
responsibility. ‘social goods’ like improved community safety
and public health at one end of the spectrum,
The remainder of this chapter looks at each of to preventing fraud and maximising public
these principles in turn and presents the various sector revenue by enabling punitive action to
factors participants weighted up, as well as the be taken against individuals at the other. This
types of questions they debated, when assessing suggests that there is a need for further direct
the potential value of an initiative. Presented in engagement with the public to better understand
order of the frequency and intensity in which where their boundaries of acceptability lie within
they were emphasised across the workshops, this the context of data sharing.
provides a cumulative map of the factors that
were considered most important in determining A number of additional factors impacted on
acceptable use of data to deliver public benefits. participants’ assessments of the acceptability
of purposes geared towards delivering public
benefit:
 Data for Public Benefit 33

• That the use of data provides direct and • That the use of data impacts on multiple
tangible benefits to individuals. Delivering beneficiaries such as individuals, services
demonstrable benefits to individuals’ and the wider public. The acceptability of
wellbeing was identified as an important data sharing activities was seen to rise if the
criterion for assessing an initiative as delivering use of data was able to advantage multiple
‘public benefit’. It was also acknowledged beneficiaries. The likelihood of ‘knock-on’ or
however, that in many cases, the uses of data ‘ripple effects’ increased acceptability, as did
that have the most potential to directly benefit the ability to maximise these benefits in long-
individuals were those that also involved term, sustainable ways.
the greatest incursions into individuals’
privacy (and thus also presented the greatest The key questions that participants asked
reputational risk to service providers). While themselves when seeking to determine
gaining the consent of individuals for data whether a use of data could be justified on the
about them to be shared was universally seen basis of providing multiple benefits were:
as the best way of overcoming these concerns, Who are the beneficiaries of this
it was also recognised that this was not always exercise and how much,
feasible. proportionally, do they benefit? How ready
are the public to recognise uses of personal
When considering the acceptability of data data that appear, particularly in the short
shared without consent, in order to directly term, to primarily benefit service providers
contribute to an individual’s wellbeing, key as delivering wider public benefits?
questions participants asked themselves were:
In what circumstances does the • That the purpose aligns with what the
benefit offered outweigh the public would expect. This was considered
incursion on privacy? Are there to be a significant factor in determining the
circumstances in which the use of personal acceptability of data use in the majority of the
data without consent presents such a workshops. Participants recognised that this
minimal harm to an individual’s privacy is challenging, given that the public may not
that the incursion is justified? have a clear understanding of the opportunities
that the data held about them might offer
• That the use of data delivers positive for informing service improvements, or have
social outcomes. When data sharing a defined expectation of how data may be
initiatives were seen as capable of exerting a productively used by service providers. However,
measurable, beneficial impact on the lives of it was seen as important to maintaining public
people and communities (eg through reduced trust that service providers did not overstep the
social isolation, reduced inequalities and local boundaries of what the public could reasonably
area regeneration), then these uses were expect data they had provided to be used for.
generally seen as being more acceptable than
when data was used for punitive purposes. The key questions that participants asked
themselves when assessing whether a
Key questions participants asked themselves potential use of data would be acceptable
when considering this were: included:
Is the way data is shared and used How aware are the public that data
making a direct contribution to these held about them could be used for
outcomes? Would the public recognise secondary purposes? Is it reasonable to
these outcomes as acceptable uses of data expect that data collected by public
held about them? At what point do service organisations may be used for this
initiatives that may benefit the majority, purpose? Was the information initially
but have the potential to have a negative provided in a context in which the public
impact on other sectors of society, become would expect it to be kept confidential?
unacceptable?
34 Data for Public Benefit

• The nature of the problem being social issues) perceived acceptability tended
addressed. As noted earlier in this report, to increase, regardless of the potential impact
some types of social problems were seen on individual privacy. This was in part due to
as more important, and more difficult a belief among participants that unintended
and complex, to address than others eg negative consequences were more likely
homelessness, violence in communities and to emerge when initiatives were primarily
social isolation among the elderly. Some reactive or ‘plastered over’ symptoms of a
participants considered that if a data sharing bigger problem.
initiative had the potential to address a
significant social problem or provide life- When attributing value to a potential use of
changing benefits to vulnerable individuals or data using this as a criterion the key questions
those suffering from multiple disadvantages, that participants asked themselves were:
then even in cases where this might entail Is this use of data getting to the root
significant incursions into an individual’s cause of the problem or addressing
privacy and/or result in the loss of service users its symptoms? Can it contribute to long-
trust, this could be considered an acceptable term social change? Is the short-term risk
risk and therefore an acceptable use of data. to services from using data in this way
worth the long-term gain?
Key questions participants asked themselves
in this context were: • That the use of data minimises negative
Are the potential benefits to effects. We noted earlier that, alongside
individuals worth breaching any benefits they may offer, many potential
individual privacy? Can this problem be uses of data by public service providers may
addressed better without sharing personal result in negative consequences (intended
data? Will wider society, if not the or unintended) for individuals and groups,
individuals involved, accept this type of for instance through excessive incursions
activity as one delivering public benefits? into privacy; punitive action; the risk of
stigmatisation/discrimination; or the diversion
of resources from one area/sector of the
‘We can mitigate the risks but population to allow focus to be given to
we can’t change the benefits.’ another. At some of the workshops, this was
WORKSHOP PARTICIPANT, MANCHESTER a central consideration for participants when
assessing the acceptability of opportunities for
data sharing.
• That the use of data achieves long-term
impacts. While recognising that the better Key questions participants asked themselves
use of data can deliver immediate benefits in these cases were:
to individuals, in some of the workshops, Do any potential negative
participants tended to attribute more value consequences outweigh potential
to examples that were able to deliver long- benefits? Are there ways to minimise
term, systematic or strategic benefits. This negative effects? If not, are the expected
included benefits that were likely to apply benefits to individuals, groups within
mainly to future, rather than current, service society or services themselves worth
users. When data sharing initiatives were accepting the negative public response
assessed as addressing the root-causes of that may result?
problems (for individuals or in relation to wider
 Data for Public Benefit 35

That data use should be of data being used for purposes other than
proportional relative to its that for which it was provided or shared.
intended outcome
Key questions participants asked therefore
There were a number of factors that contributed when assessing whether proposed uses of
to assessments of proportionality across the data seemed proportional were:
workshops: Are there clear restrictions on what
this data can be used for? Are the
• That the data use minimises the amount people who have access to the data aware
of data shared. Concern was expressed that of these limitations? Is access, particularly
when a large number of data sets are linked, to personally identifiable data, restricted
this increases the risk that data subjects could to a ‘need-to-know’ basis?
be re-identified. In other cases discussed, the
sheer number of data sets involved began • The likelihood of risks being realised.
to feel to some participants like an instance While a number of potentially significant risks
of doing this ‘because we can’ rather than to individual privacy and to the reputation
‘because it is needed’. of service providers were identified during
the workshops, many participants felt it was
The key question participants asked important to balance the severity of the
themselves when assessing whether proposed impact with the likelihood of it happening.
uses of data were proportionate was: There was a general sense across most
Is all the data shared/linked strictly workshops that well conceived, well designed
necessary for the intended purpose? and well governed data sharing initiatives
should be able to mitigate against unintended
• Whether there is a need for personally consequences, thus minimising the likelihood
identifiable data to be shared in order to of potential risks being realised. It was also
achieve the intended outcome. Whether the proposed that the existence of risk should not
data used needed to be personally identifiable automatically discount an opportunity being
was highlighted as a central consideration in pursued if there was significant potential to
determining proportionality and ultimately, deliver pubic benefits.
acceptability. It was widely agreed across all of
the workshops that data should be anonymised On this basis the key questions participants
by default, unless there was a clear and valid asked themselves when considering the
reason for it to be personally identifiable. proportionality of risk were:
How likely were potential negative
Key questions workshop participants asked impacts? How confident were they
themselves when determining whether a that the risks associated with a particular
proposed use of data was proportionate were: use of data could be effectively mitigated?
Is the data being shared personally On balance, was the proposed use still
identifiable? Does it need to be? desirable?
Would there be another way of achieving
the desired outcomes without using • The sensitivity of the data being used.
personally identifiable information? Although the type of data shared was
discussed as a factor in assessing risk, when
• Clear parameters for data use. A central it came to concluding overall acceptability
factor in assessing the acceptability of the through balancing up risks and benefits, this
sharing of personal data was that there should was not a principal consideration for most
be clear boundaries in place regarding who is participants. Instead, it was widely held that
able to access it and for what purposes. This the purpose of the data sharing should to be
was generally considered to be a valuable allowed to define the type of data that was
protection against ‘mission creep’ and the risk needed.
36 Data for Public Benefit

Key questions that participants asked available to service providers. However, when
themselves when considering the use of using personal data was assessed as being
personally sensitive data were: able to deliver benefits in a more efficient
Is this specific type of data needed and effective way than other approaches,
to achieve the identified purpose? its acceptability increased as an approach
Are there any particular reasons to deserving of public investment.
consider this type of data more sensitive
than other forms of data in this context? Key questions that participants asked
Are there sufficient protections in place to themselves were:
minimise the risk of harm to individuals Is the data service providers have
from the use of this type of data? access to reliable, complete and
timely? Is the use of personal data about
service users or the wider public the ‘best’
way to achieve the desired ends?
The responsibility of data uses
• That data can be used and shared securely.
To be a justifiable use of public resources, Alongside purpose and proportionality,
participants noted that the benefits likely to be the processes involved in data sharing and
achieved by a particular use of data have to be use are also vitally important for assessing
balanced against not just the risks, but the overall acceptability. Workshop participants believed
‘costs’ involved in delivery. there needed to be consideration of ‘the
how, not just the why’ in any assessment of
potential data uses. The ability to give an
‘The benefits achieved by the unequivocal response to public questions
use of data have to outweigh relating to how data was being used, including
not just the risks, but also the effort, a clear statement of the storage and processing
time, financial outlay, resources and procedures, access restrictions and the controls
opportunity costs involved.’ in place to protect individual privacy, was seen
WORKSHOP PARTICIPANT, SHEFFIELD as vital to extending service providers social
licence to use data in new and different ways to
improve service planning and delivery.
A variety of factors were identified within the
workshops that contributed to assessments of
whether a particular opportunity for data sharing ‘Fears about security and risk
could be considered an efficacious use of public shouldn’t be a barrier for good
resources: projects that have potential; we just
need to be clear about the safeguards we
• That it is a ‘good’ use of data. When this put in place to mitigate risks.’
was adopted as a criterion for assessment, WORKSHOP PARTICIPANT, LEEDS
the discussion focused both on the quality of
the data (in terms of accuracy and timeliness)
and whether the use of personal data (given Key questions that participants asked
known public concerns around how data themselves when considering risks to data
about them was used) added significant value security were:
to decision making or policy implementation. Are the proposals fully compliant
Participants also noted that there was a need with current data protection
for caution in relation to the increased reliance regulations? Are there sufficient data
on quantitative data as a ‘catch-all’ solution governance procedures in place to ensure
for improving public service provision, stressing that security risks are minimised? What
instead that access to data about service more could be done to mitigate security
users was only one source of intelligence and/or privacy concerns?
 Data for Public Benefit 37

• That the use of data will deliver the ultimately determined that the benefits
intended outcomes. Here discussions tended – to service providers, individuals and/or
to focus on whether there was confidence that the wider public – outweighed the
the proposed use of data sharing would be potential harm the initiative could cause?
able to deliver the benefits intended.
• That the use of data is defensible. Are
Key questions that participants ask themselves public service bodies able and willing to make
when considering this were: a publicly acceptable defence of their decision
Does the hypothesis underpinning to use data in a particular way if challenged.
the rationale for data sharing ‘ring These discussions allowed for the fact that
true’? Does using data in this way expose while something may be able to be classed as
an organisation to additional risks if the legally, or procedurally, correct there were also
expected benefits from the use of data are ethical, political and normative considerations
not realised? that impacted upon acceptability. Overall,
workshop participants considered that
• That the use of data is justifiable. In the these reservations, while valid, should not be
workshop discussions, it was recognised that allowed to obstruct opportunities for data
the public and service providers (and even sharing that have the potential to deliver
different branches of the public sector) are substantial benefits to individuals, services and
likely to evaluate the risk/benefit trade-off communities. Fundamentally, it was asserted
differently depending on their own priorities. that that public agencies have to ‘make tough
It was felt that this was particularly likely to be decisions all the time’ regarding how they
the case when the beneficiaries may not be deploy their resources, and that the decision of
the ones shouldering the risks: for example, if when, how and why to share data was not so
the risk is to individual privacy, but the benefit very different.
is to service provider efficiency, or alternatively,
if the benefit is to vulnerable service users, The key questions that participants asked
but the risk relates to wider public trust in how themselves when determining whether a
services use the information they hold about controversial or potentially unpopular use of
individuals. data was acceptable were:
Will this deliver substantial benefits
The key question that participants asked to the public? Are there ways for
themselves when considering justifiability was: public service providers to minimise the
Can the service demonstrate that risks it poses? Are we, as public service
they had considered the potential providers able, and willing, defend our
impacts, done whatever they could to decision to use data this way in the face of
mitigate negative consequences, and challenges?
38 Data for Public Benefit

Conclusions following the workshop process

While fully informed consent may be the ‘gold standard’ for using personal data about individuals, public
services already have a wide range of permissions to use data to inform service delivery and planning in
ways that do not require this level of public awareness. Making the decision about which of these powers
they use, and for what purposes, requires service providers to balance and assess the relative benefits and
risks to individuals, and services themselves, in the choices they make about data use.

There are a variety of privacy impact assessment tools and planning frameworks that have been
produced to help organisations assess the acceptability of opportunities for data sharing within
and between one another. Most of these, however, tend to focus on aspects of the data itself: the
identifiability of the data used; the additional precautions necessary when using personally sensitive
data; considerations of consent; the reliability of the data; and precautions needed to ensure data
security and mitigate risks of misuse or accidental disclosure causing harm. To date, very little focus
has been given to supporting organisations to evaluate dimensions of benefit and risk, the types of
purposes that they, and the wider public, value most, and the issues that they and the public are most
concerned about.

Across all of the workshops it was also stressed that if public confidence and support for increased data
sharing for a greater array of purposes is to be translated into the social licence for organisations to use
data more widely, then the public also needs to have the opportunity to contribute to the discussions.
In order to facilitate this, service providers need to be in a position to more clearly articulate to the
public the potential benefits that data use can bring. Without a shared language to talk about public
benefits, service providers can find it difficult to articulate any rationale behind the choices they make
regarding when, and when not, to make use of the data they hold.

This report, by analysing the criteria workshop participants drew upon to rationalise their choices,
attempts to address this gap. By identifying points of common ground among the range of service
providers that participated in the workshops, we have been able to identify a number of elements that
are seen as necessary for a data sharing initiative to be described as producing ‘public benefit’:

1. That it enables high quality service delivery which produces better outcomes for people, enhancing
their wellbeing;

2. That it delivers positive outcomes for the wider public, not just individuals;

3. That it uses data in ways that respect the individual, and their privacy, not just in the method of
sharing, but also in principle;

4. That it both represents and supports the effective use of public resources (money, time, staff) to
enable the delivery of what people need/want from public services;

5. That the benefits are tangible, recognised and valued by service providers and the wider public.
 Data for Public Benefit 39

6. Moving Forward.... balancing

the risks of using data against
ambitions to deliver ‘Public Benefit’
This project aimed to establish a greater understanding of how different groups
(public sector service providers, the voluntary sector and advocacy groups)
make sense of and balance the trade-offs inherent in using and sharing data
to improve the provision of public services.

The previous chapters describe how a diverse different types of public benefits in ways that
range of service providers approached this, acknowledge public concerns regarding how
drawing out common approaches and recurring personal data is used alongside service providers’
considerations to provide an overview of how ambitions to use data more effectively.
stakeholders from across the public and voluntary
sectors tend to assess the potential risks and These questions are not exhaustive, nor mutually
rewards associated with data sharing. exclusive. Our research, however, suggests that,
when addressed collectively, the more answers
Our goal, however, is not to simply cast light on that fall towards the right-hand side of the scale,
the tensions inherent in using personal data in the more likely an initiative is to be assessed as
the context of public service delivery. We also acceptable to stakeholders and the wider public,
want to suggest a way forward by proposing a and valued as delivering public benefits.
framework for talking about, and evaluating,

Framework for Evaluating Public Benefit

The workshop process identified five criteria that need to be considered when determining whether a
potential data sharing activity can be described as delivering ‘public benefit’:

1. That it enables high quality service delivery which produces better outcomes for people, enhancing
their wellbeing;
2. That it delivers positive outcomes for the wider public, not just individuals;
3. That it uses data in ways that respect the individual, and their privacy, not just in the method of
sharing but also in principle;
4. That it both represents and supports the effective use of public resources (money, time, staff) to
enable the delivery of what people need/want from public services;
5. That the benefits are tangible, recognised and valued by service providers and the wider public.

The framework on the following pages presents each of the elements alongside a set of evaluation
questions that can be applied to a proposed data sharing initiative. The questions have been selected
in order to initiate discussions about the various dimensions that data sharing proposals can be measured
against, in order to evaluate its potential to deliver ‘public benefit’.39

39
39 The format of this framework tool has been inspired by the scale presented in the Cabinet Office (2016) Data Science Ethical Framework https://assets.publishing.
service.gov.uk/government/uploads/system/uploads/attachment_data/file/524298/Data_science_ethics_framework_v1.0_for_publication__1_.pdf.
 40
Framework for assessing the merit of a data sharing activity to deliver ‘public benefit’

Data for Public Benefit

Checklist covering key Where does your data sharing proposal lie on the scale?
elements of ‘public
benefit’

1. That the use of Does the individual’s whose data is being used
data enables high directly benefit from improved service provision? No direct benefit Clear and direct
quality service provided to the benefits provided to
delivery which individual the individual
produces better
outcomes for Is the use of data able to deliver long term, life
people, enhancing changing benefits to individuals? Unlikley to deliver Very likely to deliver
their wellbeing. long-term benefits long-term benefits

Will the use of data have punitive impacts on

individuals? High risk of Very low risk of
punitive impacts punitive impacts
for individuals

What is the risk that individuals or communities

will suffer unintended negative consequences as a High risk of Very low risk
result of the data sharing i.e. as a result of stigma, unintended of unintended
discrimination or inappropriate targeting? negative negative
consequences consequences

2. That the use of Does the use of data deliver wider social benefits?
data delivers The benefits are There are
positive outcomes just for individuals multiple types of
for the wider beneficiaries
public, not just
individuals; Does the initiative address the root causes of social
problems/issues? The approach It addresses the
addresses the root cause
symptoms

Can the use of data inform strategic service changes

in ways that could have significant impacts on both Limited strategic High strategic value
public service providers and community wellbeing in value to inform to inform service
the future? service changes changes
Framework for assessing the merit of a data sharing activity to deliver ‘public benefit’

Checklist covering key Where does your data sharing proposal lie on the scale?
elements of ‘public
benefit’

3. That the initiative Is only data that is necessary to achieve the purpose
uses data in ways being shared? Wide data sets Sharing strictly
that respects the shared without limited to necessary
individual, and filtering data
their privacy, not
just in the method Does the use of data respect the privacy of the
of sharing but also individual data subject? Data routinely Data anonymised
in principle. shared in where possible and/
personally or proactive steps
identifiable forms taken to reduce
privacy intrusions

If data is being shared without consent, is it practical

to gain informed consent without compromising the No efforts have Gaining informed
value of the data? been made to consent is not
seek consent viable

Could the public reasonably expect data about them

to be used for this purpose? Very unlikely to Very likely to expect
expect data to be data to be used this
used this way way

How compatible is this use with the reason the data

was originally collected? Very incompatible Very compatible

4. That the use Does the use of data enable the delivery of ‘better’
of data both public services, able to respond to what the majority Unlikely to lead to Likely to deliver
represents, and of people need/want? more responsive better, more

Data for Public Benefit

supports, the service provision responsive public
effective use of services
public resources
(money, time, Does the use of data enable strategic planning that
staff). will deliver long-term efficiencies? Unlikely to deliver Likely to deliver
efficiency savings efficiency savings

41
 42
Framework for assessing the merit of a data sharing activity to deliver ‘public benefit’

Data for Public Benefit

Checklist covering key Where does your data sharing proposal lie on the scale?
elements of ‘public
benefit’

5. That the use Are the benefits tangible and measurable?

of data creates Benefits are Benefits are
benefits that are difficult to define tangible and
visible, recognised measurable
and valued by
service providers Are the benefits delivered able to be clearly attributed
and the wider to the use of data? Benefits Benefits clearly
public. not directly attributable
attributable.

Are the benefits able to be clearly communicated to

the public? Will need Benefits will be
considerable immediately
explanation to recognisable to the
demonstrate wider public
value to
the public

Is the organisation able, and willing, to defend their

use of data in this way if objections were raised? Maybe not. Absolutely

The proposed use of data Some issues – The proposed use

may not be recognised as think about whether of data is likely
delivering public benefits there are ways to to be assessed as
(but may still be justified move answers delivering ‘public
on different grounds) further to the right benefit’
 Data for Public Benefit 43

As already highlighted throughout this report, in relation to uses that the public may not
deciding whether or not a particular use of data ostensibly support.
is acceptable in the context of public service
provision relies on a variety of technical and The current disparity in approaches to assessing
operational conditions as much as ethical and risk and benefit presents significant challenges,
political factors considered in this framework. This as it can result in a ‘patchwork’ approach across
framework is therefore only intended to be used services and geographies, with little shared
to assess a specific data sharing proposal once an knowledge and understanding of best (and
organisation has already established the overall worst) practice. Therefore, a more strategic
veracity of an initiative: for example, that there is discussion is required, at both local and national
a clear purpose to the data sharing; that there is a levels, and we hope our framework can provide a
legally compliant route for the data to be shared; useful contribution to this.
that the data that is proposed to be used is robust
and reliable; the data can be securely stored, used
and transmitted; that there are effective oversight Involving the public in the discussion
and management arrangements in place to
protect how the data will be used, and mitigate If the social licence for greater data sharing is to
the risks associated with accidental disclosure be realised to deliver public benefit, then the public
or misuse to individuals and service providers; also need to have the opportunity to be involved
and that the use of data is an efficacious way of in the discussions. This should not be done simply
achieving the desired outcome (that it is the most to generate greater public approval for the use
efficient and effective use of public resources to of data. Rather, it should be undertaken in ways
address the problem). that enable the public to help shape the future of
data use by engaging in informed and meaningful
dialogue with service providers regarding their
How this framework can be used aspirations for how public services should be
within (and across) organisations provided and their concerns about how data about
them should be used.
As highlighted previously in this report, service
providers do not currently have a clear and For those organisations and partnerships that
consistent set of criteria to use to identify and want to be at the forefront of data sharing good
evaluate the relative public benefits achievable practice, and who are willing to have the types
through the better use of data. This framework of difficult conversations that will be required to
is designed to help service providers and data co-produce acceptable conditions for data use,
controllers (in the first instance) clarify the then a starting point of negotiating a shared
potential public benefits of data sharing in public understanding of the public benefits people want
service provision, in order to be able to better and expect from data use seems vital.
articulate, and/or justify, the decisions they make
about uses of the data they hold. To enable this, public service providers need to get
better at informing the public about how data is
Internally working through these questions will currently collected by service providers and the
help service providers clarify and articulate some different ways that this information could be
of the difficult decisions they make regarding how used, linked and shared to enhance the delivery of
they use data: defining and justifying the trade- public services. Alongside this, they have a parallel
offs that they are willing to make regarding the responsibility to raise public awareness about the
type and sensitivity of data used; the different implications of not using the data that is available
organisations they include in data sharing to its full capacity, and the impacts this could have
arrangements; the incursions into individual on how services can be delivered for individuals
privacy and the balance of consent and public and for the wider public. Achieving this will involve
awareness they are willing to accept, particularly being able to articulate the underlying tensions
44 Data for Public Benefit

and trade-offs required between using data to While in the short term, the efforts required to
deliver benefit and protecting privacy clearly to begin this process are not insignificant, in the long
the public, and the framework provided above can term, the costs to service providers of not taking
help to do that. the public with them on this journey are likely to
be much higher. To not begin these conversations
For those organisations that are prepared to now may undermine the reputation of public
invest the time and work with the public, then service agencies, hamper their ability to resolve
this framework will also provide a useful tool disputes, and ultimately constrain their ability to
for initiating conversations with service users, use data in modern, beneficial and potentially
community representatives, and ultimately the transformative ways.
wider public, in ways that will increase their
understanding of the complexities involved.
 Data for Public Benefit 45

Annex A: Workshop Methodology

In June-July, 2017, staff from Involve designed and illustrated how data is being shared and used
led a series of six workshops in a diverse range of by public service providers across the country for
local authority areas across England to explore a variety of purposes, that are all described as
how professionals from the public and voluntary delivering ‘public benefit’. Each set of examples
sectors understand, define and value the public was constructed to stimulate debate and
benefits that may be delivered by the better discussion (rather than to illustrate good practice)
use of data. The purpose was to begin to make and cut across a range of sectors (housing,
sense of where an acceptable balance between health, welfare, social care and community
risks and benefits may lie for data sharing in the safety). Each involved the use of different types
context of public service delivery and to establish of personal data, shared with different types of
a framework for continuing these discussions with organisations, and featured a variety of intended
both stakeholders and the wider public. outcomes. As such, the examples were chosen to
encourage participants to take a deep-dive into
To help focus discussions during the workshops, their own understanding of the potential benefits
a Background Briefing Paper was distributed to and risks of data sharing. Building in complexity
attendees in advance. This paper provided: throughout the workshop, the examples
challenged participants to develop shared
• Information about the purpose of the criteria to assess the proportionality, and relative
workshop and the rationale for the project; acceptability, of using different types of data for
• An outline of the policy drivers for the better different purposes.
use of data to support public service delivery;
• A summary of the legal context for data Throughout the workshop series participants also
capture and sharing; benefited from expert input from:
• Definitions of ‘personal data’ and ‘sensitive
personal data’ and the protections afforded to • Understanding Patient Data’s research into
each; the best language to use when discussing the
• Information about the legal basis for a ‘right different forms in which personal data can be
to privacy’ and the privacy implications of shared, and explaining anonymisation and the
data sharing; likelihood of re-identification;
• An overview of the key areas of tension
between data sharing, public benefits and • Members of the National Data Guardian’s
individual privacy. Panel, highlighting how challenges in relation
to data sharing, privacy and the public’s
Workshop Design expectations of how data about them is used,
are being navigated in a health and social care
The workshops were designed to involve a context;
mixed group of participants, primarily working
in small groups (four to six people) in defining, • Staff from the Open Rights Group, discussing
interrogating and evaluating the ‘public benefits’ how the legislative context for data sharing
that could (or could potentially) be achieved is changing and the risks and opportunities
by greater data sharing between organisations these changes bring.
involved in delivering public services.
These contributors (alongside observers from
In the workshops, a series of examples and case Carnegie UK Trust and the Centre for Information
studies were used to prompt discussions40. These Sharing Excellence) ‘sat in’ on the discussions
at various workshops to provide information
40 The examples prepared by Involve for use during the workshops can be
viewed at https://www.involve.org.uk/wp-content/uploads/2017/07/
and respond to questions, but did not actively
workshop-examples.pdf and the case studies at https://www.involve.org. participate in the deliberations.
uk/wp-content/uploads/2017/07/3-case-studies-data-sharing.pdf
46 Data for Public Benefit

Various exercises were developed for the series 5. Expert inputs to help frame the discussions;
of workshops, and not every exercise was used in
each location. Instead, the facilitators selected 6. Working in small groups to ‘define’ public
exercises in response to the size of the group, the benefit;
interests of participants, the local context and the
degree to which questions around data sharing 7. Identifying criteria to rank the relative
were already familiar to participants. The goal acceptability of different contexts, forms
was to prompt increasingly in-depth deliberations and purposes of data sharing;
about the public benefits associated with data
sharing. Activities used across the workshop series 8. Using a case study to explore the
included: acceptability of a systematised use of data
linking within a local authority area;
1. Introducing and framing the conversation
9. Using case studies to explore whether the
2. Exploring participants initial impressions of type of personal information shared, the
the opportunities and challenges associated purpose or who it is shared with has the
with data sharing most impact on assessments regarding
acceptability; and
3. Using examples of data sharing practice
to identify dimensions of public benefit: 10. Plenary discussion focussing on insights,
ranking the examples in order of their reflections and ‘take-aways’ from the
ability / potential ability to deliver public workshop.
benefits (negotiating criteria for ranking the
examples at each table); At the conclusion of each workshop a local area
report was produced and distributed to participants.
4. Using examples of data sharing practice to These were designed to give local stakeholders a
identify areas of risk and concern, and then record of the debates and deliberations that took
ranking them in order of risk (negotiating place in their area in order to support participants
criteria for ranking the examples at each to continue the discussions they had started with
table); colleagues, partners and local policymakers.
 Data for Public Benefit 47

Annex B: Spectrum of Identifiability

These terms were developed by Understanding Anonymised data
Patient Data through an extensive engagement
process with healthcare workers and the public to The Information Commissioner’s Office gives
determine best language to use when discussing guidance about what details must be removed
identifiability and anonymisation with a non- or masked, and the safeguards that must be
expert audience.41 They found that using pictures followed to anonymise data effectively. There are
is the most helpful way to explain these concepts. two different types of anonymised information:
one individual-level, one grouped.
At one end of the spectrum, a person is
fully identifiable. As you remove or encrypt • De-personalised data – This is information
information, you blur the image more and more, that does not identify an individual, because
and it becomes more difficult to identify who that identifiers have been removed or encrypted.
person is. At the other end of the spectrum, it is However, it would, in theory, be possible to
not possible to identify who someone is — they reverse that process and re-identify someone,
are effectively anonymous. so safeguards are still important. It is just like a
blurred photo of someone. We can’t immediately
Personally identifiable data see who the person is, but we know it is a specific
person. If we had the right computer power, and
This is information that identifies a specific really needed to know who the person was, it
person. Identifiers include: name, address, might be possible to work it out. There are strict
full postcode, date of birth or NHS number. safeguards on how de-personalised information
Personally identifiable information will always can be used, because there is the potential that
be stored in a highly secure way. There are strict it might be possible to re-identify someone.
laws that safeguard how personally identifiable
information can be used if you are not asked for • Anonymous data – This is information from
consent. There are also sanctions under the Data many people combined together, so that it
Protection Act if personally identifiable data is would not be possible to identify an individual
misused. from the data. It may be presented as general
trends or statistics. Because it would not be
41 Understanding Patient Data (2017) What are the best words to use when
talking about data https://understandingpatientdata.org.uk/what-are- possible to identify someone, this information
best-words-use-when-talking-about-data does not need special protection and can be
published openly.
48 Data for Public Benefit

Annex C: Legal Context for Data

Capture and Sharing
The text below replicates the background • section 25 A, B and C of the Health and Social
information provided to workshop participants Care (Safety and Quality) Act 2015 which
about the current legal context for data capture places a legal duty on health and adult social
and sharing. It is included here to provide care organisations to share information when
additional background and context for readers it will facilitate care for an individual.
less familiar with this field.
In May, 2017, the Digital Economy Act was
A public body may only share data if it has passed by Parliament to enable greater data
legal authority to do so. The first question that access for defined public interest purposes by
agencies wishing to share data need to ask public authorities. Broadly defined, clause 30 of
therefore is whether they have the expressed the Digital Economy Act contains provisions for
or implied legal powers to perform a function a ‘single gateway to enable public authorities,
necessitating data sharing. The power may be set specified by regulation, to share personal
out expressly in statute, or it may be implied from information for tightly constrained reasons
the body’s other statutory powers and functions. agreed by parliament, where its purpose is to
improve the welfare of the individual in question.
Until very recently, the legal power to share data To use the gateway, the proposed sharing of
has come from a variety of specific legislative information must be for the purpose of one of
‘gateways’ by which information can be disclosed the specified objectives, which will be set out in
or received for particular purposes. Examples of regulations.’42
such permissive statutory gateways include:
The Digital Economy Act therefore provides new
• section 115 of the Crime and Disorder Act legal mechanisms to allow data sharing between
1998, allowing anyone to pass information specified public-sector bodies to support the
to certain authorities if it is necessary or better use of data for targeted interventions;
expedient for the purposes of any provision of improving the welfare of citizens; reducing debt
the Act; owed to the public sector; fraud prevention; the
sharing of civil registration information; and
• section 14 of the Offender Management producing better statistics and research. These
Act 2007, allowing data sharing between powers are to be regulated by codes of practice
specified bodies for various purposes relating that have yet to be published.
to offenders;
Health data, however, is considered particularly
• section 111(1) of the Local Government Act sensitive and there are additional restrictions
1972, providing that they ‘shall have power and conditions on its sharing, including the
to do anything . . . which is calculated to common law duty of confidentiality. The Digital
facilitate, or is conducive or incidental to, the Economy Act, for example, explicitly excludes
discharge of any of their statutory functions’; the use of health data from its permissions for
research purposes and health services are not
• section 6 of the Crime and Disorder Act 1998 currently included in the list of specified public
which gave police and local authorities the bodies. The 2013 Caldicott Review of Information
implied power to share data to formulate and Governance established four legal bases for
implement strategies for reduction of crime in processing personal confidential health and social
their area; and
42 UK Government (2017) Digital Economy Act http://www.legislation.gov.
uk/ukpga/2017/30/part/5/enacted (accessed 28/5/2017)
 Data for Public Benefit 49

care information which meet the common law has the right to respect for his private and family
duty of confidentiality. These are: with consent, life, his home and his correspondence44. It also
through statute, through a court order and ‘when must comply with the requirements of the Data
the processing can be shown to meet the “public Protection Act 1998 (DPA).45
interest test”, meaning the benefit to the public of
processing the information outweighs the public From May, 2018, the General Data Protection
good of maintaining trust in the confidentiality Regulation (GDPR)46 will apply in the UK, and the
of services and the rights to privacy for the Government has confirmed that leaving the EU
individual concerned.’43 will not affect the commencement of the GDPR.47
The GDPR will replace the Data Protection
Once it has been established that the parties Directive 95/46/ec as the primary law regulating
have the necessary powers to share data, the how personal data is protected and is intended
next step is to consider whether the proposed to harmonise data privacy laws across Europe, to
sharing is compatible with other legal provisions protect and empower all EU citizens’ data privacy
regulating the use of personal data. For example, and to reshape the way organisations across the
data sharing by public authorities must also region approach data privacy.
comply with the European Convention of Human
Rights (now part of the UK domestic law as a 44 UK Government, (1998) Human Rights Act https://www.legislation.gov.
result of the Human Rights Act 1998), and in uk/ukpga/1998/42/pdfs/ukpga_19980042_en.pdf

particular Article 8, which provides: Everyone 45 UK Government (1998) Data Protection Act https://www.legislation.gov.
uk/ukpga/1998/29/contents (accessed 28/5/2017)
46 European Union (2016) General Data Protection Regulation https://gdpr-
43 National Data Guardian (2013) Information: To Share Or Not To Share? info.eu/ (accessed 4/12/2017)
The Information Governance Review https://www.gov.uk/government/ 47 Information Commissioner’s Office (2017) Overview of the General Data
uploads/system/uploads/attachment_data/file/192572/2900774_ Protection Regulation (GDPR) https://ico.org.uk/for-organisations/data-
InfoGovernance_accv2.pdf protection-reform/overview-of-the-gdpr/
50 Data for Public Benefit

Annex D: Defining Personal Data

The Data Protection Act (1998) defines personal ethnic origin, political opinions, religious beliefs,
data as being ‘data which relate to a living trade union membership, physical or mental
individual who can be identified from those data, health, sexuality and criminal justice history,
or from those data and other information which and needs to be treated with greater care than
are in the possession of, or are likely to come other personal data. There are also additional
into the possession of, the data controller, and protections included within the Act to ensure that
includes any expression of opinion about the sensitive personal data is processed and stored
individual and any indication of the intentions of securely and its use is controlled.
the data controller or any other person in respect
of the individual.’48 The General Data Protection Regulation, which
comes into force in May 2018, provides a more
The Act further notes that there are some detailed and expansive definition of personal
forms of personal data that are likely to be of a data and makes it clear that information such as
private nature are additionally sensitive because online identifiers – such as an IP address – can be
information about these matters could be used personal data49, reflecting changes in technology
in a discriminatory way. Sensitive personal data and the way organisations collect information
is taken to include information related to racial or about people.

48 UK Government (1998) Data Protection Act https://www.legislation.gov. 49 European Union (2016) General Data Protection Regulation https://gdpr-
uk/ukpga/1998/29/contents (accessed 28/5/2017) info.eu/ (accessed 4/12/2017)
 Data for Public Benefit 51

Annex E: The Right to Privacy

Privacy is a qualified, fundamental human Information dissemination
right. The right to privacy is articulated in all of
the major international and regional human • Breach of confidentiality – Breaking a promise
rights instruments, including the United Nations to keep a person’s information confidential
Declaration of Human Rights (UDHR) 1948, Article • Disclosure – Revelation of information about a
12 which states: person that impacts the way others judge her
character
‘No-one shall be subjected to arbitrary interference • Increased accessibility – Amplifying the
with his privacy, family, home or correspondence, accessibility of information
nor to attacks upon his honour and reputation. • Distortion – Dissemination of false or
Everyone has the right to the protection of the law misleading information about individuals
against such interference or attacks.’50
Invasion
Individual privacy is, in the context of data sharing
debates, fundamentally the ability of individuals • Intrusion – Invasive acts that disturb one’s
to choose when they wish to disclose personal tranquillity or solitude
information about themselves, and who they want • Decisional interference – Incursion into the data
to disclose this information to. Individual privacy, subject’s decisions regarding her private affairs
therefore, can be threatened or breached through
a number of practices associated with data Further, unlike most other rights, an individual’s
sharing, each of which has the ability to produce privacy can be compromised without them
a different form of harm. The list below, adapted necessarily being aware that is it taking place:
from Solove’s A Taxonomy of Privacy51 itemises ‘With other rights, you are aware of the
the types of harm that different aspects of the interference – being detained, censored, or
process and outcomes of data sharing can have on restrained. With other rights, you are also aware
individual privacy: of the transgressor – the detaining official, the
censor, or the police.’52
Information processing
The Data Protection Act gives individuals certain
• Aggregation – The combination of various specific rights over their personal data. These
pieces of data about a person include:
• Identification – Linking information to
particular individuals • the right to access personal data held about
• Insecurity – Carelessness in protecting stored them;
information from leaks and improper access • the right to know how their data is being used;
• Secondary use – Use of information collected and
for one purpose for a different purpose • the right to object to the way their data is
without the data subject’s consent being used.
• Exclusion – Failure to allow the data subject to
know about the data that others have about
her and participate in its handling and use,
including being barred from being able to
access and correct errors in that data

50 United Nations (1948) Declaration of Human Rights http://www.un.org/

en/universal-declaration-human-rights/ (accessed 28/5/2017)
51 Solove, D.J. (2006) A Taxonomy of Privacy. University of Pennsylvania 52 Privacy International (2017) What is Privacy?
Law Review 154(3): 477-560. https://www.privacyinternational.org/node/54 (accessed 28/5/2017)
52 Data for Public Benefit

Individuals can object when the use of their data is being used. ‘However, they do not need
personal data is causing them ‘substantial, to comply with the request unless there is
unwarranted damage or substantial, unwarranted damage or distress and this is substantial and
distress’. The objection can be to a particular use unwarranted.’ 53
of information or to the fact an organisation is
holding their personal data at all. Organisations
are required by law to respond to individuals 53 Information Commissioner’s Office (2011) Data Sharing Code of
who object in writing to the way their personal Practice. https://ico.org.uk/media/for-organisations/documents/1068/
data_sharing_code_of_practice.pdf
Data for Public Benefit 53
For more information
Carnegie UK Trust
Andrew Carnegie House Tel: +44 (0)1383 721445
Pittencrieff Street Email: [email protected]
Dunfermline www.carnegieuktrust.org.uk
KY12 8AW

Involve
18 Victoria Park Square Tel: +44 (0) 20 3745 4334
London Email: [email protected]
E2 9PF www.Involve.org.uk

Understanding Patient Data

215 Euston Road Tel: +44 (0)20 7611 5765
London Email: [email protected]
NW1 2BE www.understandingpatientdata.org.uk

Lead Author: Kaela Scott

Contributors: Simon Burall, Nicola Perrin, Philippa Shelton, Douglas White, Gail Irvine and Anna Grant.

April 2018

Carnegie United Kingdom Trust

Registered charity SC 012799 operating in the UK
Registered charity 20142957 operating in Ireland
Incorporated by Royal Charter 1917