Unique Device Identifiers for Cardiac Implantable Electronic Devices – A Data Quality Assessment of Administrative Hospital Data in Portugal

OBJECTIVES: Medical devices utilization may be traceable in administrative hospital data in Portugal through an individual unique device identifier (UDI) that can be linked to information about the device brand, model and purchase price. This study aimed to assess the quality and usefulness of UDIs to identify and study cardiac implantable electronic devices (CIEDs).

METHODS: Retrospective study of hospital discharges occurring during 2015-2016 with documented insertion/replacement of CIEDs, identified through the presence of a UDI or a procedure code according to the ICD-9-CM or ICD-10-PCS. Discharges with a transfer in or out status were excluded.

The number of CIED discharges was summarized by type (pacemaker (PM), cardiac resynchronization therapy (CRT), implantable cardioverter-defibrillator (ICD), CIED leads) and by identification approach (UDI vs. ICD-9/10-PCS). The sensitivity and positive predictive value (PPV) of UDI were calculated using the ICD-9/10-PCS as the gold standard. False positives and false negatives were compared across hospitals. The top implanted CIED models were identified and the distribution of purchase prices across hospitals was characterized.

RESULTS: We identified 20 590 discharges occurring in 41 hospitals (84% inpatient). Of these, 2810 (14%) were identified through UDIs and 20569 (99.898%) through ICD-9/10-PCS codes. The sensitivity of UDIs was lower than 0.21 regardless of the type of CIED considered. The PPVs were generally high (>0.8) except when considering CRT-P and Leads separately. There were 169 distinct CIED models (57 PM, 10 CRT-P, 18 CRT-D, 37 CDI, 47 Leads). CRT-Ds were the most expensive CIED (median 14525€) followed by CDI, CRT-P and PM. The price of CIEDs varied across hospitals, and variability was larger in more costly devices. However, price information disclosed by hospitals documenting UDIs presented issues (missing/outliers).

CONCLUSIONS: Compared to ICD-9/10-PCS, UDIs are not reliable to identify CIED discharges. However, an effective implementation of UDIs holds potential benefits for health technology assessment and ICD coding.