Etiquettes of Information Sharing in Healthcare and -Research M. Hartswood, K. Ho, R. Procter, R. Slack, J. Ure and A. Voss Social Informatics Cluster, School of Informatics, University of Edinburgh avoss@inf.ed.ac.uk Secondary Use of Clinical Data • eScience projects making routine, secondary use of clinical data: – National Translational Cancer Research Centre in Edinburgh – Scottish Family Health Study – NeuroGrid, etc. • To replace and expand on existing research through new IT infrastructures: – Linking new kinds of data (genetic and proteomic volume data) – Involving more people in trials and epidemiological studies. – Translational research: linking practice, clinical research and bench sciences • Issue of Information Governance comes to the fore. Information Governance • What is new/specific about this kind of eScience research? – These projects will deal with personal, sensitive information – Needed for record linkage, long-term followup and study administration – Consenting is crucial but does not solve confidentiality issues The Grid Works Fine… • In ‘research land’ where data is usually: – Anonymised – In ‘the common’ – Kept and managed in designed-for-purpose, state-of-the-art systems – In an environment where a lot is possible (even though not everything goes) …but… • As we enter the world of the NHS, we find: – Personal, sensitive data (subject to DPA) – Held in well protected islands of ownership – Kept and managed in systems designed for purposes other than research – In a very restrictive environment where ‘very little goes’ – NHS IM&T staff are not normally familiar with Grid technologies So… • Need to establish recognised practices of data sharing and to establish trust (in systems used and between parties) • Ways to relieve the tension between requirements of information governance and the aims of eScience Etiquettes of Information Sharing • Involve/are related to practical ethical action (Jirotka et al. in press) as the two mutually elaborate each other. • By etiquettes we mean members’ understanding of ‘how things are properly done’, providing a reference for any particular action taken. • Referring to the formal rules is a notable exception (whereas in the IT world it is more likely to be the unseen routine) • Taking practical ethical action is often a collaborative endeavour • These practical concerns are not replaced by Grid technologies. Examples • From ethnographic studies of work in a deliberate self harm ward and in a community mental health team (as well as an NHS guidance document). • Illustrate the notion of etiquettes of information sharing • Relevance to eScience • Specific issues in Clinical eScience Research Example I: Resources • Use of space as a resource in realising confidentiality: “It was noteworthy that the doctors’ room was treated as an area ‘out of bounds’ to patients and patients’ relatives – patients seeking to gain the attention of the team by knocking on the door were invariably refused entry. Particular attention was given to ensuring that the door of the doctors’ room was closed during discussions concerning patients and staff members neglectful of closing the door were reminded of the importance of this.” Restrictions and Resources • Closer coupling of policy with resources when things get computerised • Often through centralised control by IT and HR departments • Often results in ad-hoced practices that are less secure or compromise confidentiality • Alternative: providing resources for the work of protecting confidentiality. Example II: Identity • Identify enquirers, so that information is only shared with the right people Staff should check that any callers, by telephone or in person, are who they say they are. There can be a significant risk of harm to a patient through impersonation by those seeking information improperly. Seek official identification or check identity by calling them back (using an independent source for the phone number). Check also that they have a legitimate right to have access to that information. Confidentiality: NHS Code of Practice, Department of Health, Version 3.0, 28.7.2003 Identity • Is the presentation of a valid Grid certificate enough to establish the identity of a requester? • Will it result in immediate release of data? • Grid seems to assume ‘always on’ operation vs. requestresponse • Data managers are not going to be experts in Grid security and will not be able to spot a forged certificate. • They will, however, have other competencies such as knowing who usually requests what etc. • Diverse ways of establishing identity and membership (a nurse, a consultant, an epidemiologist) • Even if not in routine use, these are necessary to seek assurance in case of uncertainty Example (III) • The social worker explained that the he maintained different two sets of records, paperbased ones, and brief details of his visits on the Social Work system. He said that because any social worker has the ability to look at the electronic records he would record more sensitive information in the paper record, giving the example of a diagnosis of schizophrenia as one such example. Understanding Implications • Based on what people know about a system and how data committed to it may be used, people make decisions about what to commit to the system. • Use of the system and its affordances as a resource for ensuring confidentiality. • Implications for the use of the data (especially the secondary use) • Grid systems make the user’s task more complicated and there will need to be support to remedy this. Training Data Governance in Research • Caldicott guardians in universities? • Working out practical guidance, e.g., – – – – Budgeting for security: e.g., to create separate space Make arrangements and implications visible Locking computers, configuring them to auto-lock Portable machines and work from home, e.g., “carry in a bag or rucksack” – Backups: keeping sensitive data separate – Ensuring no personal sensitive data is included in presentations Confidentiality Issues in Clinical eScience Research • Some examples we have come across: – Feeding results back to persons and to healthcare services – Familial links: cases as gatekeepers to their relatives and their data – GPs as gatekeepers to potential population controls – Informed consent: do people understand how their data may be used? – Food Frequency Questionaires – importance of the return address – Recruitment Conclusions: Implications for Design • Some of the issues cannot be resolved once and for all. • Practical ethical action is an unavoidable part of work. • There are no non-functional requirements. • Embedding, controlling rules vs. providing resources to work in accordance with the rules (in letter or spirit).