Quantification of NSW Ambulance Record Linkages with Multiple External Datasets. (2nd October 2015)

• Record Type: Journal Article
• Authors: Carroll, Therese; Muecke, Sandy; Simpson, Judy; Irvine, Katie; Jenkins, André
• Is Part Of: Prehospital emergency care. Volume 19:Number 4(2015:Oct./Dec.)
• Page Start: 504
• Page End: 515
• Publication Date: 2015-10-02
• Subjects: 362.18
• DOI: 10.3109/10903127.2015.1025154
• Languages: English
• ISSNs: 1090-3127

Abstract

Abstract: This study has two aims: 1) to describe linkage rates between ambulance data and external datasets for "episodes of care" and "patient only" linkages in New South Wales (NSW), Australia; and 2) to detect and report any systematic issues with linkage that relate to patients, and operational or clinical variables that may introduce bias in subsequent studies if not adequately addressed. During 2010–11, the Centre for Health Record Linkage (CHeReL) in NSW, linked the records for patients attended by NSW Ambulance paramedics for the period July 2006 to June 2009, with four external datasets: Emergency Department Data Collection; Admitted Patient Data Collection; NSW Registry of Births, Deaths and Marriages death registration data; and the Australian Bureau of Statistics mortality data. This study reports linkage rates in terms of those "expected" to link and those who were "not expected" to link with external databases within 24 hours of paramedic attendance. Following thorough data preparation processes, 2, 041, 728 NSW Ambulance care episodes for 1, 116, 509 patients fulfilled the inclusion criteria. The overall episode-specific hospital linkage rate was 97.2%. Where a patient was not transported to hospital following paramedic care, 8.6% of these episodes resulted in an emergency department attendance within 24 hours. For all care episodes, 5.2% linked to a death record at some time within the 3-year period, with 2.4% of all death episodes occurring within 7 days of a paramedic encounter. For NSW Ambulance episodes of care that were expected to link to an external dataset but did not, nonlinkage to hospital admission records tended to decrease with age. For all other variables, issues relating to rates of linkage and nonlinkage were more indiscriminate. This quantification of the limitations of this large linked dataset will underpin the interpretation and results of ensuing studies that will inform future clinical and operational policies and practices at NSW Ambulance.Key words: emergency medical services, epidemiologic methods, medical record linkage, research design