A2 The potential and pitfalls of anonymised data linkage for follow-up of patient outcomes in prehospital emergency care research. Issue 11 (14th October 2011)

• Record Type: Journal Article
• Authors: Sánchez, A; Snooks, H; Cheung, I; Williams, C; Jones, K
• Is Part Of: Emergency medicine journal. Volume 28:Issue 11(2011)
• Page Start: e2
• Page End: e2
• Publication Date: 2011-10-14
• Subjects: Emergency medicine -- Periodicals; 616.02505
• DOI: 10.1136/emermed-2011-200645.2
• Languages: English
• ISSNs: 1472-0205

Abstract

Abstract : Background: Due to the nature of collection, storage and retrieval of data in emergency prehospital care, following up patient outcomes can be challenging. Data are often of poor quality, data structures within which they are housed complex and difficult to interrogate, and outcome data can be owned by various providers of care. Issues of ethics, consent and research/information governance further complicate matters. The Secure Anonymised Information Linkage (SAIL) databank is currently pioneering a system that collects routine social and heath records from numerous providers and assembles them within one structure that allows patient data to be stored, linked, accessed and analysed for research purposes without the need for individual (identifiable) personal data. Methods: This system was tested within a cluster Randomised Controlled Trial (RCT) in prehospital emergency care. The Support and Assessment for Fall Emergency Referrals 1 (SAFER 1) trial, is designed to evaluate the use of computerised clinical decision support software to assist paramedics to assess and decide care for older people who have fallen, for whom an ambulance has been called. Primary and secondary outcomes utilise follow-up measures from the ambulance service; hospital trusts (A&E and inpatients) and from the ONS (deaths). Eligible patients are identified from ambulance dispatch and on-scene clinical records and are then tracked anonymously via SAIL to follow-up these outcomes. Results: The initial matching rate was high at 89% (66/74). Details of unmatched patients were returned for further checking and, where possible, supplementation with identifiers. Results of second stage matching are currently pending. Conclusions: The unmatched cases are a result of poor quality of dispatch/on-scene data. Lengthy information governance processes within SAIL and a time lag until data are available have meant that follow-up data have not yet been provided. Risk of non-provision within the study period is now being assessed by the study and SAIL teams. Anonymised data linkage offers an attractive alternative to lengthy processes of consent and follow-up of identifiable outcomes across NHS providers. On the question of whether it works within the constraints of an externally funded RCT—the jury is still out.