Phaik Yeong Cheah, B.Pharm, MSc (Bioethics), PhD Data ethics Phaik Yeong Cheah, B.Pharm, MSc (Bioethics), PhD Associate Professor, Oxford University Head, Bioethics & Engagement, Mahidol Oxford Tropical Medicine Research Unit (MORU) New York City, 2 Aug 2018

From 1st July 2018

What do we mean data sharing in research Individual level health research data De-identified Sharing beyond primary research team

Quinine vs Artesunate for severe malaria in African children Illustration Quinine vs Artesunate for severe malaria in African children Randomised controlled trial in 11 sites in 9 countries Quinine vs artesunate 1:1, death is the primary outcome 5425 children with severe malaria enrolled Artesunate was proven better than quinine Dondorp et al, Lancet, 2010

What are the arguments for sharing data What are the arguments for sharing data? What are the arguments against sharing data?

Why data sharing? Transparency Cost effective New uses for data already collected Bigger more powerful datasets Reanalysis – new findings Different statistical methods Better data quality overall Respecting participants Improve science Improve health

Why not? Confidentiality Misuse of data Rogue analysis, cherry picking Extra resources Data quality Data curation is expensive Primary researchers Exacerbates existing inequality Ownership issues Group harms Any moral objections Consent issues

Types of consent Which do you think works best? Consent type Description No consent Do not obtain donor consent Blanket consent Consent to future research with no limitations Broad consent Consent to future research with limitations (or governance) Checklist Particioant chooses which types of studies/projects allowed Reconsent Consent for each specific future study Which do you think works best?

Consent When data are collected and stored for research purposes, either specific informed consent for a particular use or broad informed consent for unspecified future use must be obtained from the person from whom the data were originally obtained. The ethical acceptability of broad informed consent relies on proper governance. CIOMS 2016 Guideline 12 - Data Sharing

Challenges around broad consent Consent for the primary study is challenging enough Analogy – “I am late for a date….” Is broad consent considered “informed” consent? (2 opposing views) What about children? What aspects of data sharing are considered most important to inform participants? How to explain data sharing to potential participants? Little empirical data

Thailand study – broad consent Cheah et al, BMC Med Ethics 2018. Submitted

Location – Bangkok and Tak Province Tak Province is an area of convergence Karen and Burmese migrants Multi resistant malaria, material and child health problems A typical T-CAB meeting Lwin KM et al. BMC Med Ethics 2015 Cheah et al, Int Health 2010

What information should be provided to participants? Some findings What information should be provided to participants?

Healthy volunteer study vs very ill e.g. severe malaria

Broad consent Broad consent is imperfect, so governance is important Conclusions from our study Broad consent Broad consent is imperfect, so governance is important Information will have to be tailored per context – or we crowd out important information Empirical research and *community engagement is important *More on community engagement if we have time….

What does governance mean? Having a review process eg. data access committee Who is applying? What will the data be used for? Social value - improvement health and science? How would the primary investigators be credited? Any potential for group harms Any moral objections to certain research Any potential for breach of confidentiality Data access agreement

Many research questions What are the REAL harms and REAL benefits of data sharing? Who should be on a data access committee? What should their roles be? How do we make it equitable?

Quinine vs Artesunate for severe malaria in African children Illustration Quinine vs Artesunate for severe malaria in African children Randomised controlled trial in 11 sites in 9 countries Quinine vs artesunate 1:1, death is the primary outcome 5425 children with severe malaria enrolled Artesunate was proven better than quinine Dondorp et al, Lancet, 2010

The study was funded by a charitable UK funder (except for drugs) The drugs used were donated by a pharmaceutical company No prior consent for data to be shared with third parties The trial resulted in change in WHO policy for treatment of malaria Artesunate registered in many countries Study can/will never be repeated

Time to vote A pharmaceutical company requests for data for registration of IV artesunate in USA Researchers in Europe want to merge datasets of severe malaria to find out causes of mortality in severe malaria Vote: Should the data be shared with the pharmaceutical company? Should the data be shared with the European researchers?

And if time permits…

Ethics guidelines are not enough Some actions required prohibited encouraged permissible Some areas covered by guidance are relatively uncontroversial, others make a judgement in areas of ongoing controversy amongst competing ethical obligations eg new areas like data sharing What can we do? Community engagement can help address some of these issues Examples of community engagement…

Young Persons Advisory Board – Angkor Hospital for Children, Cambodia Sremom, Coordinator of YPAG

Community Advisory Board (CAB)– on the Thai-Myanmar border A typical T-CAB meeting

Play video – 7 mins https://youtu.be/HFYJWfGoVeQ

Questions? phaikyeong@tropmedres.ac