Overcoming barriers to the use of patient data for Research and Development between the National Health Service and Higher Education

8/9/2019 Overcoming barriers to the use of patient data for Research and Development between the National Health Servic

1/18

Charles Thackrah Building

101 Clarendon Road

Leeds, United Kingdom

LS2 9LJ

Tel. +44 (0) 113 343 4961

www.ychi.leeds.ac.uk

[email protected]

Yorkshire Centre forHealth InformaticsLEEDS INSTITUTE OF HEALTH SCIENCES

Overcoming barriers to the use of patient data for Research

and Development between the National Health Service and

Higher Education

A Briefing Paper for the UK Faculty of Health Informatics

March 2009


2/18

1

Overcoming barriers to the use of patient data for Research and

Development between the National Health Service and Higher Education

In this briefing paper and accompanying presentation we discuss the key issues, risks andopportunities associated with the sharing of patient data for Research and Development (R&D)

between Higher Education and the National Health Service (NHS), identify lessons that could be

learned from countries outside of the UK, present a series of practical ideas and recommendations on

how to better facilitate the use of information and data for health research between Universities

involved in clinical education and research and the NHS, and discuss throughout how the UK Faculty

of Health Informatics could best support the implementation of these recommendations in the future.

1. Introduction

In 2005, the UK government committed to develop the research capability of NHS Information

Technology (IT) systems in order to facilitate the recruitment of patients to clinical trials and to gather

data to support work on the populations health, and effectiveness of health interventions. This

included promoting the Secondary Uses Service (SUS) which is an example of a source of

comprehensive data enabling a range of reporting and analysis. In support of this development, the

National Institute for Health Research Clinical Research Network Coordinating Centre (NIHR CRN

CC) (formerly the UK Clinical Research Collaboration [UKCRC]) and Wellcome Trust carried out

paper-based research simulations and made recommendations1 on the technical, regulatory and

governance issues surrounding the use of electronic patient records for research and health benefit.

The NHS Connecting for Health Research Capability Programme (RCP)2 was established to carry

these recommendations forward alongside the Office for Strategic Coordination of Health Research

(OSCHR) E-Health Records Research Board3 who act primarily as an external reference group for

RCP and in addition provide strategic oversight of e-health records research in the UK. E-health

1UK Clinical Research Collaboration and Wellcome Trust (2007) Use of Electronic Patient Records for

Research and Health Benefit available at: http://www.wellcome.ac.uk/About-

us/Publications/Reports/Biomedical-science/WTX039411.htm

2 NHS Connecting for Health Research Capability Programme available at:

http://www.connectingforhealth.nhs.uk/systemsandservices/research

3 OSCHR E-Health Records Research Board available at: http://www.nihr.ac.uk/about/Pages/about_oschr.aspx


3/18

2

records research is being stimulated by the Wellcome Trust, in partnership with the Economic and

Social Research Council (ESRC), the Engineering and Physical Sciences Research Council (EPSRC)

and the Medical Research Council (MRC) with for instance their current research call4.

2. Organisation of Clinical Research in the UK

The UK has a strong history in clinical research, however, there is a common understanding that the

costs of research are high and as a society we are duty-bound to maximise the return on research

investment Hence efforts are being made to streamline the research process. These include the

emergence of an NHS research strategy5, the pulling together of all government funded applied NHS

and social service research under the National Institute for Health Research (NIHR) and the attempted

alignment of the academic, NHS and industry agendas. Government-funded basic research is

delivered through research councils such as the Medical Research Council (MRC), Economic and

Social Research Council (ESRC), the Engineering and Physical Sciences Research Council (EPSRC).

In addition charities such as the Wellcome Trust and Nuffield Foundation support both basic and

applied health research.

The UK Clinical Research Network comprises of a number of Clinical Research Networks (CRNs).

Clinical Research Networks in England have been established to support work in the areas of cancer,

dementias and neurodegenerative diseases, diabetes, medicines for children, mental health, primary

care and stroke. Each CRN consists of a number of Local Research Networks that coordinate and

facilitate the conduct of clinical research within a specific geographical area of England. Each

Network, with the exception of the Primary Care Research Network, is overseen by a national

Coordinating Centre. A major function of each of the CRNs is to increase involvement and

recruitment into clinical trials. For example the National Cancer Research Network (NCRN) in

England has 33 cancer research networks across England, closely aligned to cancer service networks,

which are currently conducting around 400 clinical trails. Alongside these networks there are disease

4 Wellcome Trust (2007/8) Electronic patient records and databases in research available at:

http://www.wellcome.ac.uk/Funding/Biomedical-science/Grants/Other-initiatives/WTD028245.htm

5 Department of Health (2006) Best Research for Best Health: a new national health research strategy

available at:

http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/DH_4127127


4/18

3

registries which identify all cases of a particular disease in a defined population. A report 6

commissioned by the Department of Health Policy Research Programme identified 250 registries in

use in England, and with an estimation of around 400 registries in total. Studies based on registers can

supplement data available from randomised studies and should be seen as an important adjunct to

rather than a replacement for them.

3. Types of Clinical Trials Research

Clinical research studies can take many forms and their design has become something of a science in

its own right. All have in common the need to answer a research question and to do so by gathering,

analysing and interpreting data. Clinical research mainly falls into two general categories:

experimental and observational. Experimental trials can also be sub-divided into two: randomised and

non-randomised. Observational studies can be either analytical or descriptive. Whether they are

observational, interventional or whether they employ quantitative or qualitative outcomes it is

necessary to guard against bias and to provide evidence of sufficient power to avoid false positive or

false negative conclusions. Issues of scale, cost, feasibility, ethics and utility have also to be factored

into the designs. In some fields, especially cancer, where the number of competing new interventions

is large recruiting sufficient subjects is becoming rate limiting and great efforts are being made to

invite all possible patients to become participants.

4. Current Uses of Data

In supporting research, patient data can be used in a number of ways. This includes being utilised in

clinical trials designed to test the safety and/or effectiveness and/or cost-effectiveness of healthcare

interventions, identifying potential participants in specific research trials to seek their consent,

providing data from routine care for analysis according to epidemiological principles, to identify

trends and unusual patters indicative of more detailed research, and providing specific datasets for

defined approved research projects. Indeed, the traditional model of clinical research has been based

around the collection of data in support of a specific research objective, which may include

developing a specific data collection process or retrospectively seeking to collect data from existing

sources. It is against this background that new approaches to harvesting routinely captured clinical

data have a potentially significant role. Each of these approaches will present specific issues but one

challenge for the Health Informatics (HI) community is to identify the generic questions and to ensure

6John Newton & Sarah Garner (2002) Disease Registers in England available at:

http://www.sepho.org.uk/Download/Public/5445/1/disease_registers_in_england.pdf


5/18

4

that developments in this field remain coherent. For example, promoting standards as regards to the

way in which data is captured and represented in the databases to ensure interoperability, reusability

and greater reliability of the data in a research context.

With the growing number, scale and richness of the available clinical data sources the opportunities to

exploit these sources have increased. There are many examples of where such approaches have been

taken already, for example The General Practice Research Database (GPRD)7 is a clinical dataset

which has been running since 1987 and includes 39 million person years of research quality data from

over 460 practices. However the degree of sophistication of databases varies with many still

disconnected across disparate systems. Many are unstructured and not coded, and require manual data

entry between systems. The current question is whether, and how, the potential to support research

can be enhanced and exploited.

5. Future Uses of Data

With the development of strategic clinical databases in the NHS there is an opportunity to support

real-time collection of clinically relevant outcomes. For example, the SystmOne8 database used by

GPs in Yorkshire and the Humber holds information approaching 50% of the population (4.5 million)

in a single database instance. This database collects, in real-time, Read/SNOMED-CT coded patient

events of the type of interest to clinical researchers. These outcomes could potentially be made

available to researchers using either push (XML messages sent to research systems on suitably tagged

patients) or pull approaches (authorised database searches) which considerably improve on the current

situation. Extending this model to facilitate targeted data capture would provide other generic

benefits. For example, a suitably designed architecture could provide messaging interfaces for

accessing current patient records and relevant clinical databases which would enable the identification

of family members of index cases who are marked as carrying a high risk cancer gene, such as

BRCA1. This currently is a difficult and expensive task reliant on family history questionnaires

administered manually. A number of models of data use could be envisaged:

Providing access for researchers to clinical systems to conduct work on the data in situ. Forexample identifying patients who have consented to follow-up and allowing controlled access

to their clinical records to collect health activity data.

7 General Practice Research Database available at: http://www.gprd.com/

8 TPP SystmOne available at: http://www.tpp-uk.com/


6/18

5

Transferring data sets from NHS systems to research systems in bulk with or withoutanonymisation. For example exporting large collections of data from specific sub-populations

to allow epidemiology analysis of interventions and outcomes, such as trends in epileptic

control with different drug regimens.

Harvesting individual data items off clinical systems in support of research, e.g. bloodpressure readings or drug prescriptions for a particular research cohort, which could be

retrospective.

Providing access to data on patients whose samples have been stored in bio-banks againstwhich new diagnostics are being tested. This has the potential to shorten the time to

implementation of new tests since the discriminating power can potentially be identified on

retrospectively collected clinical data rather than waiting to amass prospective information.

This model has worked successfully in many areas such as antenatal screening whereoutcomes are known at short intervals. It is much more problematic for conditions with longer

prodromal lead times.

6. What Are The Issues?

The focus of routine data collection in the NHS is to support clinical care and operational

management of patients. Unless specifically engaged, providers do not have time to collect additional

data. From a research perspective this can mean that such data collected will be of lower quality than

that collected by research staff because of lack of incentives and motivation, lack of time, andinterruptions. However, paradoxically, because the data for patient care is likely to be used to provide

safe and effective care what is collected maybe more accurate and, from some points of view, of

higher quality. Thus if one wished to measure some health impacts across a cohort the clinical

database might hold a better validated set than a specifically designed study set collected off-line. It

might also be available at lower cost. Clinical databases however will not answer all questions,

particularly prospective ones. Unless data collection procedures can be injected into the clinical

pathway to ensure that data is routinely collected on a regular basis it is unlikely that specific

questions can be prospectively researched. However, the examples of data collection for Quality

Outcomes Framework (QOF) where General Practitioners (GPs) are incentivised to collect items

about their patients do show that practise can be influenced at a macro scale. If a sufficiently serious

question needed to be addressed one could envisage the injection of a data collection process on a

widespread basis to generated large datasets very rapidly. The NIHR CRNCC report of research


7/18

6

simulations 9 highlighted four models of research similar to some of the examples above where it was

felt benefit could be identified (See Appendix A). One challenge for the HI community is to study

those simulations and to consider the implications they raise and what measures would be needed to

bring them into realistic and routine use.

Clinical researchers struggle to obtain relevant clinical outcomes for prospective cohort studies,

clinical trials and pharmaco-surveillance following the introduction of new therapies and therapeutic

strategies. The paradigm adopted is to create independent research databases and for these to be

administered in isolation from real-time clinical systems by a pool of research administrators.

Longitudinal clinical data is traditionally collected in the setting of a research clinic or postal

questionnaire, which is both labour-intensive and expensive. For the single outcome of death it is

possible to register research subjects with the Office of National Statistics (ONS) and to receive death

notifications automatically, on the same basis as such notifications are now made to NPfIT Patient

Administration Systems (PAS). For intermediate outcomes other than death (e.g. heart attack in

patients in secondary prevention trials) such notifications are not possible at present even though in

many ways such outcomes have high relevance.

Similarly, researchers find it difficult to assess baseline health needs against which to design

intervention studies especially where multiple agencies are involved in care exemplified by work in

assistive technology and disability. Here routine clinical outcome data (e.g. Hospital Episode

Statistics [HES]) is too coarse to enable fine-grained analysis or management. Stroke, a commonly

collected major outcome, in itself does not predict disability and may point to a fully recovered

episode. Matching electronic assistive technology to the independence needs of a client requires

detailed assessments of their disease progression, as well as current physical, sensory and cognitive

impairment. Using intelligent algorithms which integrate information from the patient records has the

potential to support a whole-system approach to address posture, mobility, well-being, lifestyle

choices and emotional and mental needs of each individual. For record linkage to work effectively

robust data identification methods will be needed. One is aware that the use of NHS number is far

from universal in NHS systems and most likely used to a minimal degree in research databases.

Promoting its use by researchers must be an early step.

The predominant coding standard in primary care throughout the NHS is Read, though this is

strategically moving to SNOMED-CT. However this is only one relevant code standard and many

9UK Clinical Research Collaboration (2007) UKCRC Advisory Group for Connecting for Health: The Report

of the Research Simulations available at: http://www.ukcrc.org/publications/reports.aspx


8/18

7

other codes and data recording conventions (e.g. units of measure, drugs, etc) are used within the

clinical record. Though some research data is coded in the majority of instances such data will be

recorded using idiosyncratic codes. Achieving alignment between the NHS and research

representations of data will be a challenge. However, much data in the clinical record is likely to be

recorded as free text. In part this is because it provides a richer narrative of a patients clinical

problem and partly because that systems in current use lack sophistication. Whatever the case,

technologies to extract codified information from free text should be of great benefit.

Although the benefits seem apparent, and are well documented, the power of modern information

technology to gain insights from large databases of population data has yet to be fully realised. In part

this is due to public concerns, and media sensationalism, over the confidentiality and security of data.

Further analyses are documented in a Wellcome Trust funded report10 on public perspectives on

research governance, a working group report11 on secondary uses of patient information, and a

consultation on the wider use of patient information12 supported by the Data Sharing Review13. Whilst

the public are aware of the principle and potential values of research, many conceded that their

awareness of issues surrounding such research was low. Implications for research included

researchers needing to pro-actively engage with participants desire for more transparency about, and

active involvement in, the research process including data access and scope, models of consent and re-

enforcement of database security and their potential future or other uses. However, we intend to

venture from such concerns, as they have already been investigated in great depth, and adopt a more

positive approach illustrated with working examples from other countries and the UK of the

successful uses of patient data for research and development.

10 Wellcome Trust (2007) Public Attitudes to Research Governance: A qualitative study in a deliberative

context available at: http://www.wellcome.ac.uk/About-us/Publications/Reports/Public-

engagement/WTX038446.htm

11Connecting for Health (2007) Report of the Care Record Development Board Working Group on the

Secondary Uses of Patient Information available at: http://www.connectingforhealth.nhs.uk/crdb/workstreams/

12 Connecting for Health (2008) Consultation on the wider use of patient information available at:

http://www.connectingforhealth.nhs.uk/systemsandservices/research/consultation

13Ministry of Justice (2008) Data Sharing Review available at:

http://www.justice.gov.uk/reviews/datasharing-intro.htm


9/18

8

7. What Are The Benefits And Opportunities?

Although most healthcare practitioners and institutions in the United States are not yet ready to

implement electronic medical record (EMR) systems there are many pioneers and innovators who

have overcome enormous obstacles to informing research goals with clinical practice data14

. TheMayo Clinic, that conducts more than 4,000 clinical trials per year, is working alongside IBM

applying super-computing technology to uncover correlations in clinical data. Kaiser Permanente has

automated records for its 8.4 million members and created both a transactional system for patient care

and a data warehouse with extracted data for use by researchers. The Indiana Health Service (IHS)

and Veterans Health Administration (VA) also provide data repositories for supporting research and

population analyses, improving data quality and security, and facilitating patient access to data and

health information.

European-level collaboration, as well as global partnerships, are helping establish reference

frameworks of best practices and mistakes, organisational, ethical, and economic aspects of re-use of

electronic patient data. For example the Biobanking and Biomolecular Resources Research

Infrastructure (BBMRI) has over 200 organisations in 24 EU Member States jointly planning an EU

infrastructure to deliver secure access to biological resources required for health-related research and

development intended to improve the prevention, diagnosis and treatment of disease and to promote

the health of the citizens of Europe15. The year 2007 marked the beginning of the European

Commissions seventh framework programme with a large increase in funding (63%) and the creation

of a European Research Council (Mladovsky, Mossialos & McKee, 2008). Despite the allocation of

6bn in the health research budget, little has been done to promote equitable access to the data whose

collection it will finance. Ethical and regulatory problems combined with the technical barriers of

interoperability of computing systems, make sharing health research data more complex than other

types of research data, where individual records should ideally be available to make data meaningful.

However, there are examples of countries (including the UK) where great progress has been made.

Some believe that Denmark has gathered more data on its citizens than any other country (Frank,

2000) and is one of the most successful applications of data-driven research and development. The

14 FasterCures (2005) Think Research: Using Electronic Medical Records to Bridge Patient Care and

Research available at:

http://www.fastercures.org/index.cfm/OurPrograms/ThinkResearch/Electronic_Health_Records

15 The Biobanking and Biomolecular Resources Research Infrastructure available at: http://www.bbmri.eu/


10/18

9

Danish government manages nearly 200 public databases, some since the 1930s. Denmarks Civil

Registration System (CRS) assigns to all Danish citizens a health insurance card with a unique

personal identification number. This number is recorded at all hospital admissions and is also used to

register all births, deaths, migrations to and from foreign countries, and domestic address changes

(Olesen et al., 2009). The use of this number creates huge opportunities for conducting large-scale

cohort studies such as investigations into cellular phone use and cancer (Johansen et al. 2001) and

measles, mumps, and rubella vaccination and autism (Madsen et al. 2002). For added confidentiality,

all studies must be approved by the Danish Data Protection Agency, which requires adherence to a

number of data storage rules to avoid disclosure of personal data (Olesen et al, 2009).

In the UK collaborations between the NHS and Higher Education (HE) institutions are already

underway in early adopter sites as part of The NHS-HE Connectivity Project16. The projects

objective is to achieve good interoperability between NHS and HE networks that enable secure

anytime, anywhere access by medical, nursing and allied profession students, clinical teachers and

researchers. In England and Scotland collaboration is between NHS Connecting for Health in

England, NHS National Services Scotland Information Systems Group and JANET (UK), and in

Wales this is run by the Public Sector Broadband Aggregation (PBSA) network. N3 is the broadband

network for the NHS in England and Scotland and used to form the N3 JANET Gateway. On a local

level there are instances of linkage between NHS systems and research databases, for example,

between the St Jamess Institute of Oncology and the University of Leeds.

8. Practical Ideas and Recommendations

There is not one solution to these issues. The research field is complex and the emerging technology

is still relatively immature. What is apparent is that the awareness of the new pan-sector possibilities

of researchers and IT staff whether in academe, the NHS or industry are limited by their previous

organisationally bounded experience. A major challenge is to create a common awareness of the

potential whilst at the same time increasing insight into the technical, legal and organisational

challenges. It is important also to dispel many of the myths surrounding the supposed limitations

created by strong information governance and to promote a mature debate about the protection of the

rights of the individual and the need to harvest data for the common good.

16 JANET The NHS-HE Connectivity Project available at: http://www.nhs-he.org.uk/


11/18

10

As an early UK investment in the process, the Wellcome Trust has funded a small project to define

the training needs of practitioners in the field and to build a core training package17. The project

recognises that systems in the two environments, the NHS and in research, have been built with very

different objectives, have widely divergent data models, governance frameworks, and utilise different

technical standards. This is reflected in the employment backgrounds of many of the staff involved

and apart from levels of informatics experience and skills; their knowledge of the different

environments is limited. As a first step in defining the training needs the project has surveyed 266

such users and has confirmed the poor understanding of concepts such as SNOMED-CT and HL7 in

the research community. Furthermore, it has revealed that a large proportion of people developing and

using systems are self-taught and appear to have limited understanding of basic informatics concepts

such as normalisation of data tables. The same is true of a large number of key concepts including

anonymisation and pseudonymisation and a programme of widespread education will be essential.

It is also clear that beyond the HI community much confusion exists about the risks and benefits of

allowing access to data for research purposes. The public is rightly concerned about unauthorised

access to sensitive data and its potential misuse hence there is a need to develop very robust

governance processes for research access to data backed up by an effective public education

programme. Whether this is by the establishment of new processes or reinforcement and evolution of

current procedures is an obvious target for early research. For some research the necessity for

identifiable linkage will be essential and the HI community may have to work hard to increase

awareness of the relative benefits of such access and lend credibility to the fact that such access can

be achieved without compromise to patients in other ways.

What Needs To Be Done?

Several strands of intervention need to be initiated across all levels of seniority, and form both local

and national priorities. The Faculty could take a major role in all of these, including:

1. Educating and challenging the executive leaders to influence policy and increaserepresentation at senior-level.

Champion cross-representation on Higher Education/NHS bodies e.g. Trust IT,University IT Boards.

17University of Leeds (2008/9) Training and education for the developers of databases in research and clinical

practice available at: http://www.ychi.leeds.ac.uk/eprresearch


12/18

11

Ensure Health Informatics involvement in research ethics committees.2. Educating middle-ranking research leaders and influencing the design of grant forms.

Educate NHS IT managers, Caldicott Guardians, Trust leads for R&D, andInformation Governance leads in research needs.

Establish Health Informatics involvement in local Research Design Services.3. Promoting engagement between suppliers and designers.

Include R&D needs in the procurement cycle, for example adding additional fields toNHS datasets with research use in mind.

Establish data access requirement for research on all clinical systems. Pressurising for the use of a single identifier within the care environment.

4. Increasing the Health Informatics capacity in the NHS workforce. Influence recruitment policy to ensure potential staff have the requisite Health

Informatics awareness and skills.

Include Research in Key Skills Framework for NHS IT staff.5. Building Health Informatics into teaching and research programmes in Health and Computer

Science.

Embed Health Informatics in the undergraduate curriculum. Promote intercalated medical degrees in Health Informatics and Computing. Support health related undergraduate projects in Computer Science.

In addition there are a number of practical activities which would cut across these strands to which the

Faculty could make a very positive contribution.


13/18

12

Promote guidance resources such as the Information Governance Toolkit18 andResearch Capability Programme19, as well as publications by the Wellcome Trust and

UK Clinical Research Collaboration, which are not being fully utilised across

research environments.

Support a programme of master classes for senior opinion leaders. Support a funded CPD programme for health, research and industry professionals. Promote Health Informatics within Medical and Computer Science Teaching and

Training courses in Higher Education.

9. Conclusions

The secondary uses of clinical data are already widespread amongst Higher Education institutions and

the NHS. Many resources contain personally identifiable or potentially identifiable data amidst,

justified or otherwise, concerns of consent, privacy and security. Potential benefits gained through

exploitation of such data are recognised amongst local innovation centres but are not reproduced

across national boundaries. Apart from consensus on privacy, policy and security, transparent policies

and practices, and public awareness of trust (Barrett et al. 2006; Davies & Collins, 2006; Safran et al.

2007) there must also be campaigns to improve the capacity and capability of NHS and research staff

on the effective use of clinical data for research.

Acknowledgements

This paper was written by Mark Hawker, Dr Susan Clamp, and Dr Rick Jones. We are grateful to the

many respondents for their contributions.

The Yorkshire Centre for Health Informatics is a leading international centre for health informatics

expertise, collaboration and research. Our mission is to improve heath care practice through high

quality research and evidence based education and training. The centre brings together partners from

18 Connecting for Health (2009) Information Governance Toolkit available at:

https://www.igt.connectingforhealth.nhs.uk/

19Connecting for Health (2009) Research Capability Programme available at:

http://www.connectingforhealth.nhs.uk/systemsandservices/research


14/18

13

the University, NHS and Industry to help meet the challenges in handling health information. Our

objectives are to develop knowledge through multidisciplinary research; develop best practice and

quality assurance within health informatics processes; disseminate best practice through education

and training; facilitate knowledge transfer by bridging the gap between health informatics researchers,

healthcare providers and health IT industries.

References

Barrett, G., Cassell, J.A., Peacock, J.L. & Coleman, M.P. (2006) National survey of British publics

views on use of identifiable medical data by the National Cancer Registry, British Medical Journal,

332, 1068-72.

Davies, C. & Collins, R. (2006) Balancing potential risks and benefits of using confidential data,

British Medical Journal, 333, 349-351.

Frank, L. (2000) Epidemiology: When an Entire Country Is a Cohort, Science, 287 (5462), 2398-9.

Johansen, C., Boice, J.D., McLaughlin, J.K. & Olsen, J.H. (2001) Cellular Telephones and Cancer a

Nationwide Cohort Study in Denmark,Journal of the National Cancer Institute, 93 (3), 203-7.

Madsen, K.M., Hviid, A., Vestergaard, M., Schendel, D., Wohlfahrt, J., Thorsen, P., Olsen, J. &

Melbye, M. (2002) A Population-Based Study of Measles, Mumps, and Rubella Vaccination and

Autism, The New England Journal of Medicine, 347, 1477-82.

Mladovsky, P., Mossialos, E. & McKee, M. (2008) Improving access to research data in Europe,

British Medical Journal, 336, 287-8.

Olesen, A.V., Parner, E.T., Mortensen, P.B., Srensen, H.T. & Olsen, J. (2009) Prenatal Risk Factors

for Cutaneous Malignant Melanoma: Follow-up of 2,594,783 Danes Born from 1950 to 2002, Cancer

Epidemiology, Biomakers and Prevention, 18 (1), 155-61.

Safran, C., Bloomrosen, M., Hammond, E., Labkoff, S., Markel-Fox, S., Tang, P.C. & Detmer, D.E.

(2007) Toward a National Framework for the Secondary Use of Health Data: An American Medical

Informatics Association White Paper, Journal of the American Medical Informatics Association, 14

(1), 1-9.


15/18

14

Appendix A: Report of Research Simulations (Executive Summary)


16/186UKCRC R&D Advisory Group to Connecting for Health

The UK can significantly enhance its clinical researchcapability by using, strictly within the bounds ofpatient confidentiality, the electronic patient datathat the UKs National Programmes for IT in the NHShave the potential to allow. This will have enormousbenefits for all types of clinical, public health andhealth services research and for many aspects ofpatient care.

The UK Clinical Research Collaborations (UKCRC)

Research & Development Advisory Group toConnecting for Health therefore commissioned aseries of simulations in October 2006 to providethe Department of Health Directorate of Researchand Development, and NHS Connecting for Health(NHS CfH), with detailed specifications for a range ofpossible research applications. The objective was to:

Inform future development of the NHS CareRecords Service (NHS CRS)

Highlight technical, regulatory and governanceissues

Inform plans for any further simulations and fullpilots to test the capacity of the infrastructure,using real patient data with appropriatesafeguards when this becomes feasible.

Four simulations were commissioned, based on arange of clinical research applications. These were:interventional clinical trials; surveillance; prospectivetracking of an identified cohort; and observationalepidemiological research. Detailed reports and keyfindings were presented to the Advisory Group inFebruary 2007 and form part of this report.

The simulation leads worked as a team over thisperiod and there was strong consensus in relationto both the high level and more detailed messages

emerging from their work. They have identified anumber of key data, regulatory and governance issuesthat need to be addressed for future development:

Clinical services and research share the samemission of improving patient care and patientsafety: research is integral to patient benefit

Research makes a very important contribution toassessing the completeness and quality of dataused for clinical care and health services

Leadership is needed to create the sustainableand governance infrastructure required to exploitthe research opportunities afforded by routinepatient and other data

Solutions should be addressed from a UK-wideperspective and build on the extensive experiencewith record linkage already in place

Much of this research involves informationabout groups of patients rather than individualsand hence requires anonymised rather than

identifiable data. However there will be occasionswhere data needs to be linkable (possibly byan honest broker) and comprehensive at theindividual patient level in order to have maximumvalue and to allow quality and completeness tobe validated

Where data are required at individual patientlevel, such data access will need to be topseudonymised data. Where identifiers need toretained, appropriate consent must be gained aspart of enabling access to those data

Existing UK strengths in the use of routineand other patient data for research will besignificantly enhanced by the mandated use of aunique identifier (for example NHS number) in all

EXECUTIVE SUMMARY


17/187 UKCRC R&D Advisory Group to Connecting for Health

key NHS records and activities and by ensuringthat new and existing data sets are person-based

The data made available must cover the wholepopulation, be up-to-date, and be retrospectiveover a number of years to give a rich historicalpicture of patients health and care. They mustalso be accurate and based on high-quality input

Further work to confirm the detailed requirementsfor data, which have been spelled out in each

of the individual simulation reports, will need tobe finalised. Much of the same data required forpurposes such as safety monitoring and clinicaltrial research is of interest for public health andNHS management activity including monitoringservice delivery. So there is a high degree ofcommonality in the data needs

The breadth of data needed for the potentialresearch applications explored in the simulationssupports the concept of a data switchboard, withpotential to link NHS Care Record data widelyto other data sources. Thus future strategicdevelopments should be based on this premise,rather than that of a single data warehouse

A federated structure of data sources rather thana single data warehouse would also provide aneffective infrastructure with optimal governancesystems in place. This could be an honest brokerwith responsibility for removing identifiers,linkage of data and data quality checks

Tackling regulatory and governance issuessuccessfully will be key to ensuring appropriate

access and use of the data for research purposes.

Ensuring patient confidentiality is critical. Datagovernance must be robust and at the same timecapable of facilitating research

Although for some research applications fullyanonymised data will suffice, for many researchapplications pseudonymised data is requiredto enable linking of data sets or elimination ofduplicate records. However, for other researchpurposes it will be important for patients tobe contactable in an appropriate manner.Appropriate approaches to consent will need tobe built into access mechanisms for informationwhich might be capable of being linked to a

specific patient

In order to satisfy regulatory requirements forpurposes such as pharmacovigilance and forclinical trials research, there are specific dataquality and access requirements that need to beaddressed

The dual role of the honest broker in ensuring patientdata confidentiality and security as well as scientificintegrity of data delivered to the research communitywill be key to engendering trust amongst patient,clinical professional and research communities.

The potential benefits for research will be lost unlessthese issues can be addressed.

It is critical that the needs of research be formallyprioritised so that both individual healthcare andpublic health can reap the full benefits of this NHSresource. The recommendations are summarisedbelow.


18/18

Key recommendations of theUKCRC simulations

Quick Wins

Recommendation 1: Mandate a common patientidentifier

To enable linkage of sources of data at patient levela unique patient identifier will be required: use ofthe NHS Number should be mandated in all key NHSrecords and activities, including laboratory records.

Recommendation 2: Communicate the relevance ofresearch to healthcare

There should be formal recognition that research is acore, not secondary, component of the developmentof the NHS Care Records Service as it benefitspatients directly. Objectives, strategy and resourcesneed to be committed or endorsed at the highestlevel of NHS Connecting for Health and reflected inits literature including website content.

Short Term Deliverables

Recommendation 3: Federate existing databases

A federated structure of data sources is requiredfor research. A high-level strategy to support suchan infrastructure needs to be developed togetherwith a roadmap for its delivery. This strategy shouldensure that the data made available cover the wholepopulation, are up-to-date, person-based and of highquality, and extend back over a number of years togive a rich historical picture of a patients health andcare.

Recommendation 4: Improve data quality

Data quality is of paramount importance both inthe clinical setting and for research. Data should beaccurate (relying on high quality input) and basedon a set of standards for recording and processingdata. Ongoing processes will need to be developed toimprove data completeness and quality which couldinvolve development of incentives.

Recommendation 5: Initiate governance discussions

Tackling regulatory and governance issues

successfully will be key to ensuring appropriateaccess and use of the data for research purposes.Data governance must be robust and at the same

time capable of facilitating research.

Recommendation 6: Engage key stakeholders

It is essential to engage professional audienceswho are key to implementation, particularly forthe enhancement of data quality and improvingdata access. Patient safety is of importance toall audiences and should be at the forefrontwhen communicating the value of research. Acommunication strategy regarding the joint benefitsof using patient data for research and clinical careneeds to be developed. The responsibility for this

development and those in recommendation 5 abovewill be with the Care Record Development Boardand may subsequently transfer to the NationalInformation Governance Board upon its formation.

UK-Wide Strategy: Next Steps

In informing plans for next steps, the outcomes ofthe simulations suggest that more extensive dataare required to enable research than those currentlyavailable through the Secondary Uses Service.

We recommend that an approach that relies on a

federated system of databases should be based on aUK-wide strategy.

This will require:

Initiation of pilots to link datasets, on the basis ofexisting successful examples within the UK;

Definition of methods of access to the differentsources housing the data. This should includeaccess to detailed patient-level data from primarycare, pathology services, disease registers andkey private sector services;

Future development which learns from, and buildupon, existing skills, knowledge, databases andsystems that have been developed in the UK overmany years;

Adoption of a UK-wide approach: not onlyare there examples of good practice beyondEngland that can be built upon, but the futuredevelopment should ensure compatibility acrossthe UK;

An organisation capable of managing the

specification and delivery of the requiredinfrastructure and providing the linkage anddefinition of research support services.

Documents

Overcoming barriers to the use of patient data for Research and Development between the National Health Service and Higher Education