What is Clinical Data Linkage?
‘Clinical Data Linkage (CDL)’ or ‘Record linkage’ are terms used interchangeably to describe the process of bringing together two or more records relating to the same individual or entity (e.g. family). A good example occurs in the health field where cross-referencing of different health information sources occurs. The art of record linkage can be quite challenging if there are multiple data custodians involved and if the infrastructure of the health system is heterogeneous. A theory of record linkage goes back to 1969 when Fellegi and Sunter introduced mathematical algorithms to link two or more sets of data that belonged to the same entities (Fellegi & Sunter, 1969). There have been other methods of record linkage based on vector methods and decision trees but no method has surpassed the Fellegi and Sunter model (Christen, 2013).
CDL is needed because individual identifiers (e.g. an individual driver’s license number, health identifier number, hospital patient number) are unique in different settings and may not be able to connect different services (Christen, 2012; Christen & Churches, 2006). CDL allows information from multiple sources to be joined together to produce richer data sets for research purposes and has wide applicability in public health and epidemiological research.
In SCI, linkage has been used by researchers in predicting mortality after traumatic SCI (Hagen, Lie, Rekand, Gilhus, & Gronning, 2010), survival after the injury (O’Connor, 2005), looking for patterns of morbidity and rehospitalisation after SCI and incidences and patterns (Middleton, Lim, Taylor, Soden, & Rutkowski, 2004),
The following section will explore how data linkage is applied internationally and locally with respect to co-ordination of disparate datasets, and in particular its practical application to making available health information for patient conditions with high medical intervention, e.g. for chronic diseases.