gms | German Medical Science

Objective: Morbidity and mortality conferences are a well known and accepted tool for measuring quality in medicine. The data collecting process is an important but also unpopular issue for further data analyzing processes. We analyzed the data processing during a complication conference, which is hold in the daily morning conference session once a week. The data generating process – the data registering process but also the data quality for further analyzing processes was evaluated.

Method: The analyzed data-process includes the paper-based data source (performed operation-procedures), the paper-based standardized adverse event registration sheet, the electronical database & the "human factor" during in the data handling process.

Results: The results showed that in approximately 12% out of our performed operations-procedures the registered diagnosis or therapy, in the data-sheet for the conference, was wrong. The registration on the standardized paper-sheet showed that only 1% of the sheets where fully filled out correctly (human-factor). This error-source was transferred directly in to the database, especially because this process was performed by a medical lay person (secretary). In total, only 94% of all registered adverse events were correctly transferred in to the database. Another black box for neurosurgeons is the registration of complications seen during the patients' course on the ICU or internal medicine related adverse events. These kinds of complications have been registered only in 2% of all cases. In comparison the registration quote was significantly higher due to surgical related complications (12%).

Conclusions: Morbidity & mortality conferences are one important step in patient quality management. The detection of patients with adverse events is the first step in data processing. At the end the data reliability is important for further analyzing processes. The conclusion is, the data collecting process has to be supervised by an expert during the complete collecting and molding process. Otherwise the data generating process leads at the end to unfeasible data.