The design and implementation of a new surveillance system for venous thromboembolism using combined active and passive methods

Estimates of venous thromboembolism (VTE) incidence in the United States are limited by lack of a national surveillance system. We implemented a population-based surveillance system in Oklahoma County, OK, for April 1, 2012 to March 31, 2014, to estimate the incidences of first-time and recurrent VTE events, VTE-related mortality, and the proportion of case patients with provoked versus unprovoked VTE. The Commissioner of Health made VTE a reportable condition and delegated surveillance-related responsibilities to the University of Oklahoma, College of Public Health. The surveillance system included active and passive methods. Active surveillance involved reviewing imaging studies (such as chest computed tomography and compression ultrasounds) from all inpatient and outpatient facilities. Interrater agreement between surveillance officers collecting data was assessed using κ. Passive surveillance used International Classification of Disease, Ninth Revision (ICD-9) codes from hospital discharge data to identify cases. The sensitivity and specificity of various ICD-9-based case definitions will be assessed by comparison with cases identified through active surveillance. As of February 1, 2015, we screened 54,494 (99.5%) of the imaging studies and identified 2,725 case patients, of which 91.6% were from inpatient facilities, and 8.4% were from outpatient facilities. Agreement between surveillance officers was high (κ ≥0.61 for 93.2% of variables). Agreement for the diagnosis of pulmonary embolism and diagnosis of deep vein thrombosis was κ = 0.92 (95% CI 0.74-1.00) and κ = 0.89 (95% CI 0.71-1.00), respectively. This surveillance system will provide data on the accuracy of ICD-9-based case definitions for surveillance of VTE events and help the Centers for Disease Control and Prevention develop a national VTE surveillance system.