Passive Surveillance of Transient Ischemic Attacks in the Emergency Department: Validity of Administrative Data and Determinants of Inaccuracies in Data Coding
Abstract
Stroke is a leading cause of morbidity and transient ischemic attacks (TIA) are an ideal target for stroke prevention strategies. Administrative data are an important source of information for TIA research, but they have not been validated in the emergency department (ED). We aimed to determine the validity of TIA codes in the Canadian ED administrative database and the predictors of accurate TIA coding. We studied patients presenting to the ED with acute neurological symptoms. The National Ambulatory Care Reporting System (NACRS) database diagnosis codes were compared to the ED chart re-abstraction and 90-day final clinical diagnoses to obtain the sensitivity, specificity, and predictive values. The sensitivity of TIA codes was low to moderate (37-64%), but the specificity was high (82-93%). Quality of physician documentation was an important predictor of data accuracy. Our findings inform TIA research and surveillance methods and we identify an opportunity for improving administrative data quality.
Description
Keywords
Epidemiology
Citation
Yu, A. Y. (2016). Passive Surveillance of Transient Ischemic Attacks in the Emergency Department: Validity of Administrative Data and Determinants of Inaccuracies in Data Coding (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/27268