Motivating Subjects: Data Sharing in Cancer Research

Jennifer Tucker, Ph.D. January 2010

Presentation Overview

• Setting the Stage • Case Study Questions and Methods • Key Findings• Recommendations

Setting the Stage

• This presentation reports key findings from my dissertation research, conducted in 2008-2009.

• Degree Program: Science and Technology Studies (STS), a field that integrates sociology, history and philosophy to explore how science and technology gets done in our society, and to what ends.

• I used a real life case study – cancer research - to explore overarching research questions about the role of motivation and emotions in science and technology.

Social NormsPublic Discourse

Institutional-Level Arguments

Individual MotivesIndividual Beliefs & Sensemaking

Experiences Seen & Heard

The Case Study: Data Sharing in Cancer Research

• Today’s cancer research enterprise increasingly relies on molecular medicine – the analysis of large data sets that can reveal disease patterns.

• It is argued that this process is best supported when researchers share data with others through “team science” – more data sharing leads to more knowledge.

• The current evolution in scientific approach is occurring in a highly decentralized research landscape, where individual achievement is the emphasis.

Case Study Questions

• What motives and emotions are linked to a researcher’s decision to share cancer research data?

• How do these motives and emotions align, or not align, with social norms and discourse related to the benefits or risks of data sharing?

Research Method

• Qualitative research, centered on a large biomedical informatics program housed in the U.S. National Institutes of Health, National Cancer Institute

• Interviews with 42 people from September 2008 – April 2009.

• Review of published materials related to data sharing and the NCI program: articles, websites, editorials, outreach materials

Technology Economic

Legal Personal

Four Perspectives

Technology

Technology: Social Messaging

• Establishing Technological Inevitability of Data Sharing• Emphasis on Interoperability: Shared Standards • Increasing Visibility of Scientific Process and Access

Technology: Personal Experience

• Grid Sharing and Standards Considered Unrealistic• Point to Point Sharing Is the Norm • Questions about the Desirability and Timing of Sharing

Most support the ideal that data sharing helps the advancement of science conceptually. The objections are pragmatic: when and whether to share their own data.

Economics: Driving the Value of Data

• Scarcity (Data Availability)• Perceived Investment (Labor)• Future Potential (Knowledge & Financial)• Endurance (Longevity)

Economic Disincentives to Data Sharing

• NIH Data Sharing Policy: Not Compelling • Data Sharing Labor Not Recognized or Rewarded• Authorship: Only First or Last Name Counts• Tenure: Metrics Reward Individual

Data Sharing is a process, but reward structures are artifacts and products based.

Rewarding data sharing would require a fundamental shift to recognizing “Science as a Service.”

Legal-Regulatory

Protecting Human Subjects in Research & Patient Privacy

Intellectual Property Protection

Clear Disincentives to Data Sharing

• Institutional Review Boards (IRB): Power and Interests• HIPAA: Scary and Significant Penalties • Deidentification: Truly Possible? • Patents: Reach Through Potential

Personal

• Data Sharing as an Extension of Identity• Data Sharing as Relationship Building • The Grid as Surveillance: Decreasing Autonomy

At the social level, scientists are chided for not sharing. What actually inspires sharing? “My friends are on the Grid – I should be on the Grid too.”

What About the Patients?

• For researchers, patients not a central motivator on a daily basis – Incentive: Increasing sample sizes– Disincentive: Legal concerns, Value of specimens

The Role of Metaphor

• Metaphor analysis emerged as a useful analytical tool for comparing social level discourse with individual stories related to data sharing.

• Impacts how we consider communications: Are we mastering the chaos of the tsunami… or….

If Science Is the Wave…..

“We’re all little waves bouncing around in the inside of a kiddie pool.”

“A lung image database is breathing life into ‘medical grid’ vision”

Advocacy Encouraging Data Sharing

(Motivators)

Advocacy forData Sharing Concerns

(Demotivators)

Technology As Central Agent

People As Central Agent

Data sets are something to be “squeezed” (or “wring it out”) until there is no value left, leaving it as “freeware” to then share.

The caBIG technologies will also allow researchers to tap into an ocean of raw published data

Applications that talk to each other; Let Data Speak to Data.

Data sharing is like dating; data sharing is a scientific marriage; Data sharing occurs most often in data clubs. we are all in the kiddie pool.

Only then will be it be possible to fully exploit the mountains of samples, and reams of data. We are in the midst of an explosion of knowledge about cancer as a disease process….. Within cancer research there exists a “Tower of Babel” problem.

Researchers are afraid of getting scooped if they share their data.

Embracing the individual diversity of members and connecting them

caBIG provides libraries of resources or banks for cancer research

Think of it as an organic bank account. You put your biomaterial in and earn medical interest in the form of knowledge and therapies that grow out of that deposit

“Raw” data as noise, crap, or as additional stuff you have to wade through.

Large Scale Metaphor

(Fear Based)

Small Scale Metaphor

(Relationship Based)

Small Scale Metaphor

(Control Based)

Scale-Less Metaphor

(Form-Less)

Social Level Arguments… You Should Share

• Data sharing encouraged as a means to respond to vast amounts of data being generated by science today.

• Data sharing positioned as good for scientific discovery and progress, and as a vital step towards patient care.

• Emphasis on standards-based sharing, allowing data to be posted and exchanged impersonally on technology grids.

• Institutional rewards are still driven by individual achievement: measured by grants and publications.

• There are some data sharing “strings” attached to large grants – but few measures in place to ensure it happens.

• Regulations and legal requirements complicate ability to share data in the ways advocated.

Individual Motives... Why People Actually Share

• People share data when it is advantageous to do so: increased sample size, shared publication/grant, or getting something back later.

• They will also generally share if doing so has little cost (data is of perceived low value) and takes little effort.

• Point-to-point data sharing within a trusted collaborative relationship is the norm – “you cannot underestimate the personal element of sharing.”

• Standards-based sharing on technology grids seen as overly burdensome, and by some, even insulting.

• Data are perceived as extensions of oneself and one’s work – to be shared with trusted colleagues and when the terms are clear up-front.

• The greatest fear is being “scooped” – being beat to publication by your own data.

Recommendations: Points of Focus and Communication

• Lower the technological bar to entry, create a community-based marketplace for data

• Better document the invisible labor of data sharing, and its necessity in meeting data sharing mandates

• Influence institutional structures: Deans (Tenure Systems), Citation Indices (Data Sets = Publication Reward), Sharing Mandates (With Accountability)

• Demonstrate low risk, high impact opportunities (pre-competitive problems helped through data sharing)

• Reorient language from communalism to communities – the PEOPLE bridges across which data are shared.

Links and Contact

Jennifer Tucker, PhD. www.tuckertalk.net

jtucker@tuckertalk.net

Link to Abstract and Full Dissertation: http://scholar.lib.vt.edu/theses/available/etd-09182009-161937/