Administrative codes inaccurately identify recurrent venous thromboembolism: The CVRN VTE study

Area of Science:

• Medical Informatics
• Epidemiology
• Health Services Research

Background:

• Administrative data, often utilizing diagnosis codes, are commonly used to identify venous thromboembolism (VTE) events.
• Accurate identification of recurrent VTE is crucial for patient management and research.
• The validity of using International Classification of Disease, 9th Revision (ICD-9) codes for this purpose requires careful assessment.

Purpose of the Study:

• To evaluate the accuracy of ICD-9 codes in identifying recurrent VTE events.
• To determine the positive predictive value (PPV) of ICD-9 codes for recurrent VTE.
• To compare the PPV of VTE codes across different clinical settings (inpatient, ED, ambulatory).

Main Methods:

• A cohort of 5497 adults with confirmed incident VTE was analyzed.
• Subsequent clinical encounters with ICD-9 VTE codes were identified across four healthcare systems.
• Medical records were reviewed to validate new VTE events, and PPV was calculated.

Main Results:

• Out of 2397 potential recurrent VTE encounters identified by ICD-9 codes, only 31.1% were confirmed.
• Encounters with VTE codes in the primary position had a higher PPV (61.3%) compared to secondary (35.4%) or outpatient codes (20.3%).
• PPV was low for specific VTE types, including pulmonary embolism (29.9%) and deep vein thrombosis (38.3% lower extremity, 37.7% upper extremity).

Conclusions:

• ICD-9 codes demonstrate low accuracy in identifying new VTE events in patients with a prior VTE history.
• The position of the VTE code within the encounter record influences its reliability.
• Relying solely on ICD-9 codes for recurrent VTE identification in administrative data may lead to significant misclassification.