Frequency and determinants of drug administration errors in the intensive care unit*

OBJECTIVE: The study aimed to identify both the frequency and the determinants of drug administration errors in the intensive care unit. DESIGN: Administration errors were detected by using the disguised-observation technique (observation of medication administrations by nurses, without revealing the aim of this observation to the nurses). SETTING: Two Dutch hospitals. PATIENTS: The drug administrations to patients in the intensive care units of two Dutch hospitals were observed during five consecutive days. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: A total of 233 medications for 24 patients were observed to be administered (whether ordered or not) or were observed to be omitted. When wrong time errors were included, 104 administrations with at least one error were observed (frequency, 44.6%), and when they were excluded, 77 administrations with at least one error were observed (frequency, 33.0%). When we included wrong time errors, day of the week (Monday, odds ratio OR 2.69, confidence interval CI 1.42-5.10), time of day (6-10 pm, OR 0.28, CI 0.10-0.78), and drug class (gastrointestinal, OR 2.94, CI 1.48-5.85; blood, OR 0.12, CI 0.03-0.54; and cardiovascular, OR 0.38, CI,0.16-0.90) were associated with the occurrence of errors. When we excluded wrong time errors, day of the week (Monday, OR 3.14, CI 1.66-5.94), drug class (gastrointestinal, OR 3.47, CI 1.76-6.82; blood, OR 0.21, CI 0.05-0.91; and respiratory, OR 0.22, CI 0.08-0.60), and route of administration (oral by gastric tube, OR 5.60, CI 1.70-18.49) were associated with the occurrence of errors. In the hospital without full-time specialized intensive care physicians (which also lacks pharmacy-provided protocols for the preparation of parenteral drugs), more administration errors occurred, both when we included (OR 5.45, CI 3.04-9.78) and excluded wrong time errors (OR 4.22, CI 2.36-7.54). CONCLUSIONS: Efforts to reduce drug administration errors in the intensive care unit should be aimed at the risk factors we identified in this study. Especially, focusing on system differences between the two intensive care units (e.g., presence or absence of full-time specialized intensive care physicians, presence or absence of protocols for the preparation of all parenteral drugs) may help reduce suboptimal drug administration.