Comparison of a Trigger Tool and voluntary reporting to identify adverse events in a paediatric intensive care unit.

Reduction of adverse events depends on accurate detection. The utility of a Trigger Tool to detect and classify severity of adverse events in an intensive care unit of a paediatric university hospital was compared to voluntary reporting. Sixty patient records were randomly selected from 314 admissions over three months. Events detected by the Trigger Tool were classified by two independent investigators as insignificant, minor, moderate, major or catastrophic. Examination of each record required, on average, 40 minutes. Ninety-eight adverse events (1.66/patient) were detected in 59 available records. Mean admission was 2.77 days. The incidence of adverse events was 59.9 per 100 patient days or 0.60 events per patient per day. The number of events detected by the Trigger Tool was related to duration of admission (r=0.70, P <0.0001) and risk of mortality on admission (r=0.50, P=0.0001) but not to age. The inter-rater agreement on detection of adverse events was good. Investigator One detected 66 adverse events while Investigator Two detected 93 events (kappa 0.63). Of the 61 events detected by both investigators, the agreement of classification of severity was very good (kappa 0.89). Of the 56 events rated similarly by both investigators, 13 (23%) were insignificant, 19 (34%) were minor, 17 (30%) were moderate, 4 (7%) were major and 3 (6%) were catastrophic. Only four adverse events had been reported voluntarily, of which two were detected using the Trigger Tool. Whereas the Trigger Tool is a simple, efficient and robust method, voluntary reporting is inadequate and captures very few adverse events in the intensive care unit environment.