Data Reliability in Large Data Bases
This study of data reliability is based on the creation of a 2.1 million records data file at the National Mass Radiography Service of Norway. Redundant information about 106.228 persons in the file gave an opportunity to study error situations and error rates in a medical information system.
Assigning data to wrong persons was estimated to occur in less than 0.05% of the cases. Errors in data elements that did not serve as identificators were also studied. Height and weight errors in the file were mainly due to misreading of handwrittten figures in the punching operation.
The paper ends with a discussion of error liability and its consequencies for medical systems.
KeywordsHeight Difference Identification Element Large Data Base Wrong Identification Medical Information System
Unable to display preview. Download preview PDF.