1.The Epidemiological Study of Venous Thromboembolism and Bleeding Events Using a Japanese Healthcare Database
Takuhiro Yamaguchi ; Takeshi Fuji ; Masao Akagi ; Yasuyuki Abe ; Mashio Nakamura ; Norikazu Yamada ; Eisei Oda ; Daisuke Matsubayashi ; Kaori Ota ; Masafumi Kobayashi ; Daiju Matsui ; Jumpei Kaburagi ; Yasuyuki Matsushita ; Atsushi Harada
Japanese Journal of Drug Informatics 2015;17(2):87-93
Objective: The aim of this study was to establish the proper definitions of venous thromboembolism (VTE) and bleeding events for a healthcare database in Japan.
Study Design: Validation study.
Methods: The study comprised patients with VTE or who had undergone orthopedic surgery of the lower extremities and whose outpatient or inpatient medical information from April 1, 2008 to September 30, 2013 was available. The source population of the database was derived from 100 acute-care hospitals. The endpoints were VTE events (deep venous thrombosis [DVT], pulmonary thromboembolism [PE]) and bleeding events (bleeding requiring blood transfusion, intracranial hemorrhage, intraocular hemorrhage, upper gastrointestinal [GI] bleeding, and lower GI bleeding). The frequent events with laboratory data were randomly extracted and evaluated, while all the infrequent events with laboratory data were extracted and evaluated. Positive predictive value (PPV) was defined as the proportion of events judged to be clinical by medical experts of all the extracted events. First, we conducted a test with a small number of cases and then revised the definitions of events. Second, we extracted and evaluated data in 50 patients for VTE and bleeding events patients, based on which we defined the target PPV level between 60 and 70%.
Results: Of the 5,044,743 patients in the database, 36,947 patients underwent orthopedic surgeries of the lower extremities and 3,578 patients experienced a VTE event. The PPV at the first evaluation was 80.0% (8/10) for DVT, 57.1% (4/7) for PE, and 27.3% (6/22) for bleeding events. At the second evaluation using the revised definitions, the PPV were 75.0% (42/56) for VTE and 73.3% (33/45) for bleeding events. Overall, the PPVs for VTE and bleeding events were over 70%. The PPV of the VTE events were 76.9% (30/39) for DVT and 70.6% (12/17) for PE. The PPVs of each type of bleeding event were over 70% except for intracranial hemorrhage (44.4%, 4/9).
Conclusion: The PPV was high for VTE events (75.0%) and bleeding events (73.3%). The definitions used in this study are rational for the identification of VTE, DVT, PE, and bleeding events in the healthcare database in Japan. The definition for each type of bleeding event should be investigated in further studies.