Trauma registry record linkage: methodological approach to benefit from complementary data using the example of the German Pelvic Injury Register and the TraumaRegister DGU®
1 Department of Trauma, Hand and Reconstructive Surgery, University of Saarland, Kirrbergerstraße 100, 66421, Homburg/Saar, Germany
2 AUC - Academy of Trauma Surgery, Landwehrstraße 34, 80336, Munich, Germany
3 Department of Trauma and Orthopaedic Surgery, University of Witten/Herdecke, Cologne-Merheim Medical Center (CMMC), Ostmerheimerstraße 200, 51109, Cologne, Germany
4 Institute for Evaluative Research in Orthopedic Surgery, University of Bern, 3014, Bern, Switzerland
5 Institute for Research in Operative Medicine (IFOM), Faculty of Medicine, University of Witten/Herdecke, 51109, Cologne, Germany
BMC Medical Research Methodology 2013, 13:30 doi:10.1186/1471-2288-13-30Published: 5 March 2013
In Germany, hospitals can deliver data from patients with pelvic fractures selectively or twofold to two different trauma registries, i.e. the German Pelvic Injury Register (PIR) and the TraumaRegister DGU® (TR). Both registers are anonymous and differ in composition and content. We describe the methodological approach of linking these registries and reidentifying twofold documented patients. The aim of the approach is to create an intersection set that benefit from complementary data of each registry, respectively. Furthermore, the concordance of data entry of some clinical variables entered in both registries was evaluated.
PIR (4,323 patients) and TR (34,134 patients) data from 2004-2009 were linked together by using a specific match code including code of the trauma department, dates of admission and discharge, patient’s age, and sex. Data entry concordance was evaluated using haemoglobin and blood pressure levels at emergency department arrival, Injury Severity Score (ISS), and mortality.
Altogether, 420 patients were identified as documented in both data sets. Linkage rates for the intersection set were 15.7% for PIR and 44.4% for TR. Initial fluid management for different Tile/OTA types of pelvic ring fractures and the patient’s posttraumatic course, including intensive care unit data, were now available for the PIR population. TR is benefiting from clinical use of the Tile/OTA classification and from correlation with the distinct entity “complex pelvic injury.” Data entry verification showed high concordance for the ISS and mortality, whereas initial haemoglobin and blood pressure data showed significant differences, reflecting inconsistency at the data entry level.
Individually, the PIR and the TR reflect a valid source for documenting injured patients, although the data reflect the emphasis of the particular registry. Linking the two registries enabled new insights into care of multiple-trauma patients with pelvic fractures even when linkage rates were poor. Future considerations and development of the registries should be done in close bilateral consultation with the aim of benefiting from complementary data and improving data concordance. It is also conceivable to integrate individual modules, e.g. a pelvic fracture module, into the TR likewise a modular system in the future.