Lessons Learned (so far) about PCORI Grantee Data Sharing Amy - - PowerPoint PPT Presentation

ANNUAL MEETING

• OCT. 31-NOV. 2, 2017

ARLINGTON, VA

#PCORI2017

Lessons Learned (so far) about PCORI Grantee Data Sharing

Amy Pienta

Associate Research Scientist, ICPSR @apienta 11/1/2017

ANNUAL MEETING | #PCORI2017

Amy Pienta, Ph.D.

Disclosures

Rela latio ionship ip Company ny(ies es) Speakers Bureau Advisory Committee Qualitative Data Repository, Syracuse University Consultancy Review Panel Ad Hoc NIH, NSF, NIDILRR Board Membership Honorarium Ownership Interests Ingenex Digital Marketing

2 November 5, 2017

ANNUAL MEETING | #PCORI2017

3

Objectives

At the conclusion of this activity, the participant should be able to:

• Understand the Goals of the Pilot Repository Project at

ICPSR at the University of Michigan

• Understand Select Lessons Learned when PCORI

Grantees Share Data with a Repository

November 5, 2017

ANNUAL MEETING | #PCORI2017

4

Pilot Project Goals

• The ICPSR repository is working with a variety of PCORI

data collection projects (completed and in progress) to archive data (of varying types) with ICPSR

• We will document and record how it was to work with

the various research projects in hopes of enabling PCORI to plan for broader data sharing activities amongst its grantees

• Create a demonstration repository for PCORI, initially

for internal review and use.

November 5, 2017

ANNUAL MEETING | #PCORI2017

5

Pilot Project Progress

Working with 4 data collection projects (completed and in progress) Pre-Deposit Stage -- 1 project depositing soon; Consulting with 2 projects on what to send Deposit Stage -- None currently, but 1 project completed this stage Curation Stage -- 1 project in preparation for release Data Release Stage -- None

November 5, 2017

ANNUAL MEETING | #PCORI2017

November 5, 2017 6

Pre-Deposit Stage: Lessons Learned

• Data Sharing Materials and Resources sent to grantees (Tips,

Checklist, IRB Planning Guidance)

• Working with a wide Variety of Data Types
• Described as challenging – EHR, CMS hospital profiles, admin data
• IRB Issues –Some IRBs want to approve deposit and

dissemination agreement

• Need for Understandable Terminology – limited-use refers to

data requiring safe harbor method of de-identification vs. restricted use.

• Grantees require Guidance on Using Multiple Repositories for

the same data

• Authoritative source of data, version control, interoperability

ANNUAL MEETING | #PCORI2017

November 5, 2017 7

Pre-Deposit Data Review (one study)

• Where We Need More Information/Recommendations:
• Raw data has higher disclosure risk compared to summary

variables

• Clarify some variable and value labels
• Identify missing data codes (and standardize where possible)
• Recommend random generation of CaseIDs
• Categorize some continuous/ordinal variables
• Alter some of the date variables

ANNUAL MEETING | #PCORI2017

November 5, 2017 8

Challenge: Deciding Who Does What

• Data archiving best practice requires more attention to

detail and to careful documentation than is typically required for analysis.

• Research Team Data Preparation – Team knows the data well

and the methodology, but is not at arm’s length to know what a secondary user would need to know to use the data well into the future.

• Data Repository Curation Services – Repository understands

required documentation and how to improve usability; also has tools available to reduce much of the routine work. Repository staff may require some input from the researcher.

ANNUAL MEETING | #PCORI2017

November 5, 2017 9

Additional Guidance for Grantees

• Clarification about what data to deposit in an archive
• e.g., raw data, analysis files of summary variables, derived

and scale variables, open-text responses; all time points, only from subjects that consented at some point-not always at beginning of data collection

• Code instead of data
• Guidance on Data File Structure – 1 row/case vs.

hierarchical vs relational database

ANNUAL MEETING | #PCORI2017

November 5, 2017 10

Learning About Related Publications

• Starting point we used the PCORI page to identify related

publications (also will collect info from PIs)

• PCORI Page links to HRSP (Health Services Research

Projects) and NCT (ClinicalTrials.gov) IDs which give MeSH terms, keywords, and grant numbers

• NIH Grant Number (via PMCID) often associated with a

much wider set of publications (how does this field associate grants and pubs with one another)

• Primary use of data vs. secondary use (harder to capture)
• Large Number of PIs – difficult to track

ANNUAL MEETING | #PCORI2017

November 5, 2017 11

Relevant Metadata Standards

• MeSH is the National Library of Medicine's controlled

vocabulary thesaurus.

• Consists of sets of terms naming descriptors in a hierarchical

structure that permits searching at various levels of specificity

• MeSH thesaurus is widely accepted and used
• Useful for describing the topical aspects of the data
• A preliminary search of terms for the one study – able to identify

several terms appropriate and sufficient for discovery

• Also, SNOWMED-CT; PICO; Consumer Health Vocabularies

ANNUAL MEETING | #PCORI2017

November 5, 2017 12

Looking Forward

• Continuing to work with the 4 PCORI-funded projects
• Populating the internal data repository in order to offer

a full demonstration of a PCORI repository

• Talking to a broader set of PCORI-funded projects
• Making recommendations on data governance,

technology, and promotion

ANNUAL MEETING | #PCORI2017

13

Learn More

• www.pcori.org
• info@pcori.org
• #PCORI2017
• ICPSR: www.icpsr.umich.edu

November 5, 2017

ANNUAL MEETING | #PCORI2017

14

Questions?

November 5, 2017

ANNUAL MEETING | #PCORI2017

November 5, 2017 15

Thank You!

Amy Pienta

Associate Research Scientist @apienta 11/1/2017