Diagnostic Performance of Delirium Assessment Tools in Critically Ill Patients: A Systematic Review and Meta-Analysis

Abstract

Background: Critical care nurses are in the best position to detect and monitor delirium in critically ill patients. Therefore, an optimum delirium assessment tool with strong evidence should be identified with critical care nurses to perform in the daily assessment. Aim: To evaluate and compare the diagnostic performance of delirium assessment tools in diagnosing delirium in critically ill patients. Methods: We searched five electronic databases including the Cochrane Library, PubMed, Embase, CINAHL, and a Chinese database for eligible diagnostic studies published in English or Mandarin up to December 2018. This diagnostic test accuracy meta-analysis was limited to studies in intensive care unit (ICU) settings, using the Diagnostic and Statistical Manual of Mental Disorders (DSM) as a standard reference to test the accuracy of delirium assessment tools. Eligible studies were critically appraised by two investigators independently. The summary of evidence was conducted for pooling and comparing diagnostic accuracy by a bivariate random effects meta-analysis model. The pooled sensitivities and specificities, summary receiver operating characteristic curve (sROC), the area under the curve (AUC), and diagnostic odds ratio (DOR) were calculated and plotted. The possibility of publication bias was assessed by Deeks’ funnel plot. Data Synthesis: We identified and evaluated 23 and 8 articles focused on CAM-ICU and ICDSC, respectively. The summary sensitivities of 0.85 and 0.87, and summary specificities of 0.95 and 0.91 were found for CAM-ICU and ICDSC, respectively. The AUC of the CAM-ICU was 0.96 (95% CI, 0.94–0.98), with DOR at 99 (95% CI, 55–177). The AUC of the ICDSC was 0.95 (95% CI, 0.92–0.96), and the DOR was 65 (95% CI, 27–153). Linking Evidence to Action: CAM-ICU demonstrated higher diagnostic test accuracy and is recommended as the optimal delirium assessment tool. However, the results should be interpreted with caution due to the between-study heterogeneity of this diagnostic test accuracy meta-analysis.