Abstract
Background: Delirium is a common, underdetected problem that has short- and long-term negative sequelae for critically ill patients. Prompt and accurate delirium identification by nurses can ensure early intervention and treatment to help minimize adverse outcomes.
Objectives: To evaluate the relationship between an educational program and the accuracy of registered nurses' (RNs') documentation of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU), a delirium screening tool.
Methods: In a medical ICU at a tertiary academic medical center from September 2015 to March 2016, RNs were reinstructed on use of the CAM-ICU. Registered nurse assessment data were collected retrospectively for 12 months before and after intervention and were compared against the CAM-ICU algorithm using [chi]2 analysis.
Results: A total of 10 736 RN assessments in 1020 patients preintervention and 11 068 in 951 patients postintervention were evaluated. Overall RN accuracy improved from 78% to 80% (P = .054). The algorithm determined delirium to be present in 32% versus 30% of all patients preintervention and postintervention, respectively; there was no difference in rate of nurse detection of delirium preintervention and postintervention (54% vs 55%, not statistically significant). The percentage of "inappropriate unable to assess" ratings by nurses decreased from 42% to 37% postintervention (P < .05).
Conclusions: After a comprehensive training initiative, there was no significant improvement in CAM-ICU documentation and no improvements in patient delirium identification. Future quality improvement efforts should target reducing the number of assessments that RNs judge to be "unable to assess." Clinical practice must evolve to routinely incorporate RN delirium assessments into the patient's plan of care.