Data Consistency in a Voluntary Medical Incident Reporting System

Voluntary medical incident reporting systems are a valuable source for studying adverse events and near misses. Unfortunately, such systems usually contain a large amount of incomplete and inaccurate reports which negatively affect their utility for medical error research. To investigate the reporting quality and propose solutions towards quality voluntary reports, we employed a content analysis method to examine one-year voluntary medical incident reports of a University Hospital. Results indicate that there is a large amount of inconsistent records within the reports. About 25% of the reports were labeled as "miscellaneous" and "other". Through an in-depth analysis, those "miscellaneous" and "other" were substituted by their real incident types or error descriptions. Analysis shows that the pre-defined reporting categories serve well in general for the voluntary reporting need. In some cases, human factors play a key role in selecting accurate categories since reporters lack time or information to complete the report. We suggest that a human-centered, ontology based system design for voluntary reporting is feasible. Such a design could help improve the completeness and accuracy, and interoperability among national and international standards.