Exploring the Possibility of Medical Device Surveillance in Patients on Peritoneal Dialysis Using a Common Data Model

TL;DR

This study explores using a common data model to monitor medical devices in peritoneal dialysis patients, finding it can detect some complications but misses others.

Contribution

The study demonstrates the potential and limitations of using a common data model for medical device surveillance in peritoneal dialysis.

Findings

Infectious complications were detected using the CDM, but mechanical complications were missed.

Data on PD catheters and adaptors were present in EHRs but not captured by the CDM.

Amended matching improved detection of some complications in another CDM database.

Abstract

Background and Objectives: Peritoneal dialysis (PD) requires well-functioning medical devices (MDs). PD complications can result in significant adverse events, including the discontinuation of PD, hospitalization, and death. This study aimed to evaluate the feasibility of detecting various PD complications and data related to MDs. Materials and Methods: A retrospective study was conducted on patients who received PD catheter insertions between January 2001 and March 2021 to evaluate PD-related complications. PD complications were evaluated through diagnostic, procedural, and MD codes using a common data model (CDM) and were compared with those from electronic health records (EHRs). The results from one CDM database were compared with those from another CDM database. Results: A total of 342 patients were enrolled. One hundred and ninety-five patients experienced PD complications more than once. Nineteen prescription codes and twenty diagnostic codes from the EHR were identified, covering 11 procedures, three MDs, and seven complications related to PD. Infectious complications were detected using the CDM, whereas mechanical complications were missed. Although data on PD catheters and adaptors were available in the EHR, they were not detected via the CDM. Some infectious and mechanical complications were identified via the CDM in the other database. After implementing amended matching, these data were detected. Conclusions: While some PD-related medical data recorded in EHRs were misrepresented or omitted during the CDM database extraction, transformation, and loading process, the CDM shows potential to serve as a source of real-world data for active surveillance.

Figures