The use of a clinical database in an anesthesia unit: focus on its limits.

Anesthesia information management system (AIMS) can be used a part of quality assurance program to improve patient care, however erroneous or missing data entries may lead to misinterpretation. This study assesses the accuracy of information extracted for six consecutive years from a database linked to an automatic anesthesia record-keeping system. An observational study was conducted on a database linked AIMS system. The database was filled in real time during surgical/anesthesia procedure and in the post-anesthesia care unit. The following items: name of the anesthetist, duration of anesthesia, duration of monitoring, ventilatory status upon arrival in postoperative care unit, pain scores, nausea and vomiting scores, pain medication (morphine) and anti nausea and vomiting drug consumption (ondansetron) were extracted and analysed in order to determine exhaustivity (percentage of missing data) and accuracy of the database. The analysis covered 55,946 anaesthetic procedures. The rate of missing data was initially high upon installation but decreased over time. It was limited to 5% after 3 years for items such as start of anesthesia or name of the anesthetist. However exhaustivity/completeness of some other variable, such as nausea and vomiting started as low as 50% to reach 20% at 2008. After cross analysing pain and post-operative nausea and vomiting scores with related medication consumption, (morphine and ondansetron) we conclude that missing data was due to omission of a zero score rather than human error. The follow-up of quality assurance program may use data from AIMS provided that missing or erroneous values be mentioned and their impact on calculations accurately analysed.