Emergency department (ED) visits for syncope are common. Validation of ED administrative diagnostic coding for syncope is required before these codes can be used for health services research. We performed a retrospective multicentre chart review using a regional ED database in British Columbia. We identified adults who visited 1 of 3 high-volume urban EDs between 2010 and 2015. Cohort 1 included 350 ED visits for patients with a presenting complaint (PC) of syncope/presyncope, a discharge diagnosis (DD) of syncope and collapse, or both. Cohort 2 included 100 patients with ED visits with neither a PC of syncope/presyncope nor a DD of syncope and collapse. The reference standard was abstractor conclusion regarding syncope and presyncope ("definite/very likely" vs "possible" vs "unlikely" vs "absent") after structured review of ED medical records. We found that in cohort 1, syncope or presyncope were definite/very likely or possible in 96% of visits with a PC of syncope/presyncope and a DD of syncope and collapse. Syncope alone was definite/very likely in only 56% of visits. In cohort 2, syncope was definitely absent for 94% of patients. The reference standard showed excellent face validity and abstractor inter-rater agreement (Cohen κ > 0.80). Vital signs and orthostatic vital signs were not documented for 8% and 84% of visits, respectively. Our results suggest that a PC of syncope/presyncope combined with a DD of syncope and collapse is highly predictive of syncope or presyncope. These findings will inform design and interpretation of syncope health services research.
Copyright © 2017 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved.