A Win:Win for Data Access: Balancing Public Good with Privacy Concerns Legal Framework for...

A Win:Win for Data Access:Balancing Public Good with

Privacy Concerns

Legal Framework for e-Research Conference

12 July 2007

Gold Coast

Professor Fiona Stanley ACDirector: Telethon Institute for Child Health Research

Executive Director: Australian Research Alliance for Children and Youth

Outline

1. Rationale for data sharing and e-research

2. Population data and record linkage

3. How seeking consent can lead to bias

4. Harmonising privacy and access. Can we have a win:win?

1. Rationale for data sharing and e-research

4

Wor

king

Gro

up o

n D

ata

for

Sci

ence

Working Group on Data for Science

Wisdom

Knowledge

Information

Data

Wisdom

Knowledge

Information

Data

From Data to Wisdom:Pathways to Successful Data Management for Australian Science

Report of the Working Group on Data for Science

Presenters:Professor Robin Batterham AOProfessor Fiona Stanley AC

Prime Minister’s Science, Engineering and Innovation Council

5

Wor

king

Gro

up o

n D

ata

for

Sci

ence


Our Vision

Australia is managing increasing volumes and complexity

of data to enhance our country’s scientific, economic and

social prosperity and to protect it from threats

Wisdom

Knowledge

Information

Data

Wisdom

Knowledge

Information

Data

Wisdom

Knowledge

Information

Data

Wisdom

Knowledge

Information

Data

6

Wor

king

Gro

up o

n D

ata

for

Sci

ence


Science Data Challenges

• Exponential increase in data assets

• Lack of data in some vital areas

• Increasing diversity of data

• Vulnerability of data

• Lack of capacity in data management

• Missed opportunities to collaborate

• Impediments to discover, preserve, share and re-use data

• Lack of relevant skills

• Lack of global engagement

7

Wor

king

Gro

up o

n D

ata

for

Sci

ence


Vulnerability of Data Assets

Source: The New Yorker

8

Wor

king

Gro

up o

n D

ata

for

Sci

ence


Key Issues

• Digitisation

• Capture

• Preservation

• Storage

• Accessibility

• Security

• Privacy

• Discoverability

• Integration

• Interoperability

• Sharing

• Re-use

9

Wor

king

Gro

up o

n D

ata

for

Sci

ence


OECD

“fostering broader, open access to and wide use of

research data will enhance the quality and productivity

of science systems worldwide.”

ICSU

“Scientific data and information should be as widely

available and affordable as possible…”.

Importance of Open Access and Data Sharing:International Perspective

10

Wor

king

Gro

up o

n D

ata

for

Sci

ence


Bureau of Meteorology

“… foregoing proprietary rights to data and making

them freely available actually benefits the individual as

well as the community at large …”

Importance of Open Access and Data Sharing:Local Perspective

A National Strategic Framework for Scientific Data

• Recommendation 1– That Australia’s government, science, research and

business communities establish a nationally supported long-term strategic framework for scientific data management including guiding principles. policies, best practices and infrastructure

• Recommendation 2– That a high-level expert committee be established to provide

the leadership role in progressing the formation of the long-term strategic framework for scientific data management

The National Network of Digital Repositories

• Recommendation 3– That the necessary policy and programmes be

implemented with a view to establishing a sustainable publicly funded network of federated digital repositories

• Recommendation 4– That the expert committee consider the

development of a strategic roadmap for the implementation and evolution of the national network of federated digital repositories

Data Management, Access, Sharing and Collaboration - Changing the

Culture• Recommendation 5

– That standards and standards-based technologies be adopted and that their use be widely promoted to ensure interoperability between data, metadata, and data management systems, providing authentic users of the data with appropriate processes and safeguards.



– That the principle of open equitable access to publicly-funded scientific data be adopted wherever possible and that this principle be taken into consideration in the development of data for science and programmes.

– As part of this strategy, and to enable current and future data and information resources to be shared, mechanisms to enable the discovery of, and access to, data and information resources must be encouraged



– That funding agencies offer incentives to encourage researchers and institutions to:

• Develop data management plans for each research grant application involving data collection and generation, and that standards be made freely available and widely disseminated so as to encourage best practice in data management

• Introduce policies and practices to encourage collaboration and sharing of data across Australia’s scientific research institutions and across agencies

• Analyse and re-use existing data

Ensuring there are no Regulatory Impediments

• Recommendation 8– That funding agencies such as the NHMRC and

ARC ensure that best practices and policies are developed and followed that allow bonafide researchers to access individual population data, including and linking of data from multiple sources, whilst protecting privacy, and ensuring that ethics committees fully understand these policies and their rationale.

Ensuring there are no Regulatory Impediments

• Recommendation 9– That in the context of developing the strategic

framework for scientific data management, Australia’s intellectual property approaches be checked to ensure they do not impede the sharing of data.

– In particular it should take into account the OECD Committee for Scientific and Technological Policy guidelines on access to research data and the International Council for Science statements about the benefits of sharing data.

Skills for Data Management

• Recommendation 10– That data management expertise becomes a core skill for

researchers, including graduate and postgraduate science students across all disciplines, and that they receive data management training as part of their education.

• Recommendation 11– That the Australian Government give early consideration to

the findings of the e-Research Coordinating Committee regarding changing research behaviour, practices and skills.

CHILD

Community School

Family

Cultural

Economic Political

Social

CIVIL SOCIETY UNCIVIL SOCIETY

Focus on:

Equality/diversity

Trust, care

Collective good

Valuing parents

Valuingchildhoods

Prevention more than cures

Protected environments

Safe places for all

Effective use of helpful technologies

Child needs as well as adults

Workplace

Accepting of:

Inequalities

Fear, violence

Priority for material wealth

Parents not valued

Fast tracking childhoods

Cures more than prevention

Environmental degradation

Safe places for the few

Excessive use of damaging technologies

Adults needs more than children’s

Ecological contexts shaping child development

Trends in Problems Affecting Children & Youth in Today’s World

• Many are increasing in incidence

• Complex problems (eg mental health, obesity)

• Demand complex information to monitor, study & prevent them

• Costly to treat & manage

• Crisis in child & youth services (health, mental health, education & crime)

• Research in silos

• Services in silos

“Law and ethics in population health are having a renaissance. Once fashionable during the Industrial and Progressive eras, the ideals of population health began to wither with the rise of liberalism in the late

20th century. In its place came a sharpened focus on personal and economic freedom. Political attention

shifted from population health to individual health and from public health to private medicine.”

Prof Lawrence O Gostin 2004University of Georgetown

Law and Population Health

New Yorker March 2005

2. Population data and record linkage

What is Record Linkage?

• Brings together records from different sources, relating to the same individual

• Used for: – administration or case management – population based research and policy

• Focus today: on public good i.e. monitoring, research & evaluation to improve the health & wellbeing of the population

WA Maternal and Child Health Research Data Base 1977-2004

1970’s • Public concerns re thalidomide and adverse effects of perinatal care1980/81• Establish 1st Australian linkage of births, deaths and midwives (perinatal)

records (total population)• Establish registers of cerebral palsies and birth defects to link to data base1982/83• Link computerised hospitalisations1990 onwards• Ongoing MCHRDB2004• WA data linkage system

WA Data Linkage Unit

MCHRDB 2004 Onwards

CORE(1980-current)

Midwives notifications*

Birth registrations Hospital Morbidity*Death registrations*

National registers

National Death IndexNational Cancer

Registry

Family links

Marriage registrationsElectoral roll

Birth/death registrations

Commonwealth Data

PBS (Prescription Drug Use)

MBS (Medicare)Australian Childhood

Immunisation Register

Registries

Cerebral PalsyBirth DefectsIntellectual DisabilityCancer

Mental Health

ICHR Studies

Raine StudyWATCH

RASCALSChild Health Survey

Aboriginal Child Health Survey

* geocoded

Advantages of WA Population Data & Record Linkage

eg MCHRDB• 100% sample: unbiased, no one excluded• Cheap cf. seeking consent/ surveys• Valid & reliable data on sensitive issues• Reduces survey burden on populations• Fast, effective linkage technology• Privacy protected• Better data for policy, planning, evaluation• Improve administrative data

Limitations of WA Population Data & Record Linkage

• Information only available on items and outcomes recorded in data bases (breadth > depth)

• Privacy issues still need to be addressed eg ethics committees, understanding of public good by the community

• Need better, complete denominators• Changes in diagnostic classifications present challenges for

temporal analyses• (In)accuracy of recorded information• Incomplete ascertainment• Sample size for rare disorders (APSU)

Antenatal factors in later disease/ disability

• Trends in diseases & disability (complete) • Environmental exposures & later diseases • Birth outcomes in psychiatric patients• Intra-uterine growth & teenage mental illness• Pregnancy problems & later childhood

diseases

Evaluation of medical care

• Increased very preterm survival - problems in survivors

• IVF & cerebral palsy• IVF & birth defects• Reasons for and impact of, increasing

caesarean sections• Effects of increased obstetric intervention• Adverse drug effects

Evaluation of health promotion

• Prevention of cot deaths

• Folate campaign for spina bifida

• Childhood vaccination coverage

• Anti-smoking programs

3. How seeking consent can lead to bias

Consent

• Consent is essential for all research involving participation of individuals– Questionnaires, interviews– Donate blood, tissues– Drug trials etc

• Not all research requires consent

Bias

• Bias is the distortion of the true relationship between exposure and outcome due to flaws in either study design or analysis

• Can give wrong answers

Bias from Non-Participation• Inability to trace/ contact (most common)• Refusal (rare)

Both of these groups very different from participants

• Magnitude and direction not predictable• Not quantifiable• Explain differences in risks between studies• Poor information for health services and epidemiological

research

Source:Hansen, Kurinczuk, Bower & Webb 2002

Cumulative Prevalence of Birth Defects after ICSI & IVF

Does TOP Increase the Risk of Later Breast Cancer?

• Recent meta analysis (53 studies)• Retrospective studies with variable reporting &

response rates RR=1.14(1.09-1.19)• Record linking of abortion data to cancer register data

RR=0.93(0.89-0.96)• Comparisons of abortion registry & self report data

– 24% women with breast cancer and– 27% women without breast cancer reported incorrectly that they had

never had an induced abortion– 27% of women reporting a spontaneous abortion did not report it 20

years later

Record Linkage Impact of Informed Consent

• Tu et al (2004) analysed the impact of informed consent on characteristics of participation in the Canadian Stroke Registry

• Overall participation rate of eligible patients was 39% in Phase 1 & 51% in Phase 2

• Selection Bias - lower in-hospital mortality rate among participants

• Expensive ($500,000 over 2 years for consent alone)

7

Consent Rate for Health Research

0%

20%

40%

60%

80%

100%

Population

Pro

po

rtio

n Rochester Clinic

Mayo Clinic

Australia In Person

Australia GP Clinic

USA mailed survey

UK mailed survey

Finland mailed survey

Australia mailed survey

Arguments for not Seeking Consent in Population Based Record Linkage

(for all other studies wherever practical, informed consent should be sought)

• Research is of great benefit to society• Evidence of bias when consent is sought - serious

misinterpretation with implications for health, health services, other

• Very small risks to individuals involved (and may be significant benefits)

• Impractical to obtain consent• Too costly to obtain consent• May actually protect privacy more than in studies

seeking consent (DLU)

Proportion of Ethics Approved Research Projects using Name Identified & Data Linked

Administrative Health Information WA 1990-2003

0

10

20

30

40

50

60

70

80

90

100

Year of Approval

Pro

po

rtio

n o

f R

esearc

h P

roje

cts

(%

)

% data-linked

% name-identified

WA DataLinkage System

Figure Proportions of ethics-approved research projects (N=408) using name-identifiedand data-linked administrative health information in Western Australia 1990-2003

Source: Trutwein, Holman & Rosman (2006)

4. Harmonising privacy and access.

Can we have a win:win?

Record linkage without consent is allowed:

• We are guided by the National Health & Medical Research Council (NHMRC) Australian Health Ethics Committee– Guidelines for researchers and for Human

Research Ethics Committees ( HREC)– Balance public right to privacy against public

right/interest in proposed research/activity– Influenced Privacy Act of 2001

NHMRC Guidelines for Epidemiological Research

• Role of Ethics Committees (composition, information, guidelines)

• Reasons for data collection• Reasons why impracticable to seek consent• Protection of identifiable data- security standards etc• Nominated custodians, etc• De-identified & anonymous analysis• Conform to National Privacy Principles

NHMRC 2001

WA Data Linkage Unit Best Practice for Cross Jurisdictional Linkage(harmonising privacy & encouraging access)

1. Obtain files of identified individual’s records from Custodians (eg. Births, prescriptions, birth defects registries)

2. Link identifiers, but no access to clinical/sensitive data3. Strip off identifying information4. Return to each Data Custodian with a project ID5. Researchers apply to:

1. Institutional NHMRC Ethics Committee2. DLU Confidentiality Committee3. DLU Advisory Committee

6. Researchers go to Data Custodians to obtain de identified linked data set. No individual data available or used(eg. analyse drugs in pregnancy - effects on birth outcomes & birth defects)

Privacy Concerns

• Can’t address if not known

• Public understanding of– Research– Trade offs (protecting privacy/ allowing

access)– Contexts and current legislation and

processes

Canadian privacy activities2004-2006

• 4 workshops on harmonising privacy

• Privacy toolkit

• Privacy audits

• Privacy officers

Source: Slaughter et al, 2006

UK Privacy & Medical Research

Personal data for public good: using health information in medical research

• Increased complex laws/regulations• Variable interpretation• Many projects blocked/delayed• Increase in costs• Poor public awareness of value and methods

of researchCommentary Lancet, 2006

Source: Academy of Medical Sciences, 2006

Our health system needs effective evaluation

• Spiraling costs of care• Increase complex diseases• New technologies, drugs - harmful side

effects• Patient expectations• Concerns over safety• Poor data for service planning

New Yorker May 2003

Are patients morally obliged to participate in research projects as a “mandatory

contribution to public good”, particularly for those aimed at preventing serious harms and

providing important benefits?

John HarrisProf of Bioethics

University of Manchester

“We are optimistic that a win:win is possible: where privacy is protected, where important

health research can proceed. There is evidence that a shift in thinking is possible on

behalf of ethics review committees, regulators, researchers and the public.”

Dr E Meslin 2006

Director Indiana University

Centre for Bioethics

Towards a win:win

1. HRECs understand and accept current guidelines which allow identifiable data for research without consent if privacy issues addressed and rationale acceptable

2. Develop Australian privacy and research best practice for the researchers using personal data

3. Public aware of how personal records are used and how research is done

4. e.Health: ensure records can be used to evaluate health system

“ In the late 20th century, scholars and politicians posed a key question ‘What desires and needs do you have as an autonomous rights bearing person

to privacy, liberty and free enterprise?’ Now it is important to ask another kind of question ‘What

kind of community do you want and deserve to live in, and what personal interests are you willing to forgo to achieve a good and healthy society?’ “

Prof Lawrence O Gostin 2004

University of Georgetown

A Win:Win for Data Access: Balancing Public Good with Privacy Concerns Legal Framework for...

Documents

Transcript of A Win:Win for Data Access: Balancing Public Good with Privacy Concerns Legal Framework for...