Annual Meeting of the NCI Cohort Consortium (Abstract Submission): Submission #1

Lightning Talks Abstract
------------------------
Presenter's First Name: : Konrad
Presenter's Last Name:: Stopsack
Title (eg: professor, assistant professor, chair, etc):: Professor
Degree(s): MD MPH
Contact Email:: stopsack@leibniz-bips.de
Organization:: Leibniz Institute for Prevention Research and Epidemiology – BIPS
Project Title:: An R Package for the Health Professionals Follow-up Study to Increase Accessibility and Reusability of Legacy Data Structures
Additional Authors:
1. First Name: Konrad
   Middle Initial: H.
   Last Name: Stopsack
   Degree(s): M.D. M.P.H.
   Organization: Leibniz Institute for Prevention Research and Epidemiology – BIPS
2. First Name: Hannah
   Middle Initial: E.
   Last Name: Guard
   Degree(s): M.S.
   Organization: Harvard T.H. Chan School of Public Health
3. First Name: Caroline
   Last Name: Himbert
   Degree(s): Ph.D.
   Organization: Huntsman Cancer Institute
4. First Name: Yuliya
   Last Name: Leontyeva
   Degree(s): Ph.D.
   Organization: Harvard T.H. Chan School of Public Health
5. First Name: LeeAnn
   Last Name: Lucas
   Degree(s): M.S.
   Organization: Harvard T.H. Chan School of Public Health
6. First Name: Colleen 
   Middle Initial: B.
   Last Name: McGrath
   Degree(s): M.S.
   Organization: Harvard T.H. Chan School of Public Health
7. First Name: Jane
   Middle Initial: B.
   Last Name: Vaselkiv
   Degree(s): M.P.H.
   Organization: Harvard T.H. Chan School of Public Health

Abstract::
Background: With long-term follow-up of prospective cohorts, not just participants but also data structures and data pipelines age. Accessibility and reusability of such legacy data can be major barriers to analysis projects, compounded by analytic software with vendor lock-in. 

Methods and Results: One approach to improving accessibility and reusability is the creation of open-source software that leaves the original data storage intact and provides well-documented, transparent translations of legacy data structures into modern data objects, which are then directly usable in reproducible analytic pipelines. The test case, presented here, is the R package {hpfs} for loading and reshaping data from the Health Professionals Follow-up Study. This software has been developed with a team-based approach, fully online and with collaboration and quality control supported by a version control system. This effort has required a deep understanding of the cohort data as well as emphasis on consistent software design. Initial application experience suggests that this approach eliminates some common pitfalls in analyses and provides major efficiency gains.

Conclusion: Transparent translations of legacy data by open-source software are one approach to substantially improve accessibility and reusability of valuable cohort data.