Cross-organizational workflow in radiology: an empirical study of the quality of shared metadata elements in Region Vastra Gotaland, Sweden.

Acta radiologica (Stockholm, Sweden : 1987)

PubMedID: 23528568

Lindsköld L, Wintell M, Edgren L, Aspelin P, Lundberg N. Cross-organizational workflow in radiology: an empirical study of the quality of shared metadata elements in Region Vastra Gotaland, Sweden. Acta Radiol. 2013;54(6):676-83.
BACKGROUND
Challenges related to the cross-organizational access of accurate and timely information about a patient's condition has become a critical issue in healthcare. Interoperability of different local sources is necessary.

PURPOSE
To identify and present missing and semantically incorrect data elements of metadata in the radiology enterprise service that supports cross-organizational sharing of dynamic information about patients' visits, in the Region Västra Götaland, Sweden.

MATERIAL AND METHODS
Quantitative data elements of metadata were collected yearly from the first Wednesday in March from 2006 to 2011 from the 24 in-house radiology departments in Region Västra Götaland. These radiology departments were organized into four hospital groups and three stand-alone hospitals. Included data elements of metadata were the patient name, patient ID, institutional department name, referring physician's name, and examination description.

RESULTS
The majority of missing data elements of metadata was related to the institutional department name for Hospital 2, from 87% in 2007 to 25% in 2011. All data elements of metadata except the patient ID contained semantic errors. For example, for the data element "patient name", only three names out of 3537 were semantically correct.

CONCLUSION
This study shows that the semantics of metadata elements are poorly structured and inconsistently used. Although a cross-organizational solution may technically be fully functional, semantic errors may prevent it from serving as an information infrastructure for collaboration between all departments and hospitals in the region. For interoperability, it is important that the agreed semantic models are implemented in vendor systems using the information infrastructure.