CHALLENGES OF HEALTH DATA‐LINKAGE IN AUSTRALIA

Abstract

Listen

Translate article

Background: Data linkage is a valuable research technique reducing data collection costs and permitting more detailed and extensive analyses. We sought to undertake a national, multijurisdictional data‐linkage project to examine the perioperative outcomes of patients on renal replacement therapy, an important but under‐researched area. Despite its value as an important tool for improving patient‐orientated research and health service policy, establishing such linkages remains challenging. Access to multijurisdictional linked data involves parallel communication between multiple bodies which can be unwieldy and is further hampered by a lack of consistency of procedures between different states and territories. Aim: We highlight some of the difficulties that we experienced. Results: Data Linkage Units: (a) lack of understanding of medical terminology and methodological expertise, (b) variable responsiveness and timeliness of reviews; Data Custodians: (a) response times can be slow, regular reminders are necessary, with data‐linkage projects viewed as excess responsibility placed on custodians who thus only deal with them opportunistically, (b) correct application of privacy guidelines when working with non‐identifiable data. This can also help streamline processing of data‐linkage ethics requests; Human Research Ethics Committee: a) Inconsistent interpretation of National Mutual Acceptance Memorandum of Understanding across jurisdictions leading to duplication of effort, b) lack of universal understanding of data‐linkage process itself; Data‐Registries: Lack off transparent and streamlined application process to guide prospective researchers; Researchers: cost, time, staffing and appreciating limitations of datasets. Conclusions: Existing policies and practices are stifling research enterprise and opportunity. Widespread changes in culture and processes will take time and till then researchers will need to be persistent, exercise patience and unfortunately experience long delays and compromised analyses.