Identifying venous thromboembolism and major bleeding in emergency room discharges using administrative data

Area of Science:

• Medical Informatics
• Health Services Research
• Clinical Epidemiology

Background:

• Administrative data, including International Classification of Diseases, 10th revision (ICD-10CM) codes, are increasingly used to identify clinical events.
• The accuracy of these codes for capturing venous thromboembolism (VTE) and major bleeding (MB) in Canadian emergency department discharge records has not been established.

Purpose of the Study:

• To evaluate the validity of ICD-10CM codes for identifying VTE and MB events in emergency department discharge data.
• To determine the positive predictive values (PPVs) for VTE and MB codes.

Main Methods:

• A retrospective chart re-abstraction study was conducted at a single institution in London, Canada.
• Adult patients with VTE or MB codes in the Canadian Institute for Health Information National Ambulatory Care Reporting System between July 2002 and March 2014 were identified.
• A random sample of 50 patients for each condition underwent blinded diagnostic adjudication by two independent abstractors to calculate agreement and PPVs.

Main Results:

• ICD-10CM codes demonstrated very good ability to identify major bleeding events, with an overall PPV of 88%, particularly for intracranial and lower gastrointestinal bleeds.
• In contrast, ICD-10CM codes for VTE showed moderate ability, with an overall PPV of 49%.
• Codes for pulmonary embolism (PE) performed better than those for deep venous thrombosis (DVT).

Conclusions:

• ICD-10CM codes for VTE have moderate predictive value for deep venous thrombosis and pulmonary embolism in emergency department discharges.
• Codes for major bleeding events demonstrate very good ability and are adequate for research purposes.
• Further validation may be needed for VTE coding in administrative datasets.