Ongoing evaluation of sedation programs and individual physician performance are important elements in a successful sedation program. In our institution, quality improvement (QI) forms are used to track adverse events occurring during procedural sedation. Despite widespread use of similar forms, there are little available data to support the accuracy of such systems in reflecting actual events that occur during sedations. Our objective was to assess the accuracy of QI forms in reporting adverse events associated with propofol sedations.
Since January 2007, sedation providers in our hospital have filled out a paper QI form after the completion of sedation, and the data are subsequently entered into a computerized database. We had previously reviewed 1649 charts of patients sedated with propofol from 2005 to 2009. Based on this chart review, we identified 67 patients who experienced adverse events during propofol sedation between January 2007 and September 2009. The QI database was queried about these 67 patients to determine the concordance between the QI forms and medical charts.
Only 40 of the 67 patients (60%) were identified as having adverse events in the QI database. We were able to recover information for 56 of the 67 patients from the QI database. The information in 29 of the 56 QI sheets (50%) matched the events as recorded in the chart. In the 27 QI sheets that did not match the medical charts, the majority were cases where events such as insertion of oral airway (11), administration of continuous positive airway pressure (3), or increased oxygen requirement (3) were not recorded in the QI form. There was 1 emergent anesthesia consult and 3 procedures that were aborted because of complications that were not recorded in the QI sheet. For the remaining 7, an adverse event was recorded in the QI database but not in the medical chart. We also found that premedication with midazolam during propofol sedations increase the likelihood of adverse events in our initial propofol study. Of the 67 charts identified with events, 18 patients were premedicated with Versed, but this was recorded in the QI form in only 9 cases (50%).
Reporting adverse events on paper‐based QI forms appears to be inaccurate in our institution. There are likely multiple reasons for underreporting adverse events. Sedation providers enter the information at the conclusion of the sedation rather than concurrently. At the end of a successful sedation, minor interventions are often not believed to be significant and are not recorded. The structure of the QI forms may also have posed a barrier for capturing all events. As our institution moves forward to electronic sedation forms in the near future, there may be opportunities to create a program in which the QI form is automatically generated from the actual sedation forms.
M. Srinivasan ‐ none; K. Hamlin ‐ none; Y. Daud ‐ none; D. Carlson ‐ none