Identification and characterization of errors and incidents in a medical intensive care unit

To assess the frequency, type, consequences, and associations of errors and incidents in a medical intensive care unit (ICU).

Two-hundred and sixteen consecutive patients with predominantly cardiovascular and pulmonary disorders admitted between December 2002 and February 2003 were enrolled. Demographic data, SAPS II, and TISS-28 were obtained for all patients. Prior to patient enrolment all staff members (physicians, nurses, physiotherapists) were repeatedly encouraged to make use of the Incident Report Form (IRF) and detailed descriptions on how, why and when to use the IRF were provided.

During the observation period of 64 days, 50 errors involving 32 patients (15%) were reported. Patients subjected to errors were more severely ill (SAPS II 42 +/- 25 vs. 32 +/- 18, P < 0.05), had a higher hospital mortality (38% vs. 9%), and a longer ICU stay (11 +/- 18 vs. 3 +/- 5 days, P < 0.05). Gender, age and TISS-28 were equally distributed. Each day of ICU stay increased the risk by 8% (odds ratio 1.078, 95% confidence interval 1.034-1.125, P < 0.001), and by 2.3% per SAPS II point (odds ratio 1.023, 95% confidence interval 1.006-1.040, P < 0.001). The majority of errors and incidents were judged as 'human failures' (73%), and 46 errors and incidents (92%) as 'avoidable'.

The identification and characterization of errors and incidents combined with contextual information is feasible and may provide sufficient background information for areas of quality improvement. Areas with a high frequency of errors and incidents need to undergo process evaluation to avoid future occurrence.