Determining the test characteristics of claims-based diagnostic codes for the diagnosis of venous thromboembolism in a medical service claims database.

PURPOSE: To determine the test characteristics of diagnostic codes within a medical service claims database for deep vein thrombosis (DVT) and pulmonary embolism (PE).
METHODS: The Regie de l' Assurance Maladie du Québec (RAMQ) administers the health insurance program in Québec, Canada. RAMQ claims data were obtained for subjects with objectively diagnosed DVT with or without PE who were participants in the Venous Thrombosis Outcomes (VETO) Study from April 2001 to July 2002. Using the date of DVT and PE diagnosis in the VETO record as the reference standard, the proportion of subjects correctly classified by RAMQ diagnostic codes was determined for the exact date of DVT and PE occurrence and for four expanded time windows around this date.
RESULTS: In all, 355 VETO patients were included, 301 with DVT alone and 54 with DVT and PE. Overall, 97% of VETO cases had a RAMQ diagnostic code for DVT and 82% of VETO cases with PE had a RAMQ diagnostic code for PE. Sensitivity for DVT and PE was 52% (95% confidence interval (CI), 47-57) and 35% (95% CI, 23-49), respectively for the exact date of diagnosis, and 87% (95% CI, 83-90) and 78% (95% CI, 64-88), respectively for a 60-day window around this date. As all VETO participants had DVT, specificity for the diagnosis of DVT could not be determined.
CONCLUSION: Diagnostic codes within a medical service claims database are relatively sensitive indicators for DVT and PE, and use of claims data for VTE research purposes can be considered.