JMIR Publications

ABSTRACT

Background:

The effective utilization of Electronic Health Record (EHR) data for clinical or research purposes heavily depends on the fitness of those data for the intended use. However, there is lack of operationalized and standardized framework implementations to evaluate the suitability of EHR data for secondary use, which might lead to inconsistent quality in the outcomes of data use projects (DUPs). This research focuses on the Medical Informatics for Research and Care in University Medicine (MIRACUM) Data Integration Centers (DICs), and delves into empirical insights on addressing and automating the Fitness-for-Purpose of Clinical Data in German DIC-settings.

Objective:

1. To capture and discuss how MIRACUM DICs are evaluating and enhancing the fitness-for-purpose of observational healthcare data, and to examine there the extent to which the observed practices do align with existing recommendation 2. To identify and analyze the requirements for designing and implementing a solution to support a computer-assisted evaluation of the Fitness-for-Purpose of EHR data within the MIRACUM DICs

Methods:

In this investigation, we followed a qualitative study including the approach of Braun et al. We conducted an open-ended survey with 17 participants across the DICs of ten German University hospitals affiliated with the MIRACUM Consortium. To analyze the data, we utilized the thematic analysis framework along with data visualization techniques. Our qualitative method followed an inductive approach.

Results:

All ten MIRACUM DICs participated, revealing various approaches to assessing data’s fitness-for-purpose; including applying the 4-eyes principle and data consistency checks through e.g. cross-system data values comparison. However, the observance of DUP-related feedback loop about the fitness of intended data, and the usage of dynamic self-designed dashboards were identified as common way for mentoring the data fitness-for-purpose. Nine key requirements for a computer-assisting solution were identified, emphasizing for instance flexibility, understandability, extendibility and practicability. Furthermore, the participants were found to employ heterogeneous types of data repositories for evaluating DQ-criteria for DUPs, and practical strategies to communicate on them through research and clinical teams.

Conclusions:

This study identifies and discusses relevant divides between current practices in MIRACUM DICs and existing recommendations, offering insights into complexities involved in assessing and reporting on the clinical data’s fitness-for-purpose. It also provides valuable input for the development and the cross-location integration of an automated solution tailored for this purpose. This may encompass simple statistical comparisons up to advanced machine-learning based algorithms for operationalizing the existing framework like the 3x3 DQA Framework for real world applications. The findings also address a roadmap framework that should serve as foundational evidence for future design and implementation studies towards enhancing data quality assessments for specific DUPs within observational healthcare settings.