Abstract
Aims: To synthesise the evidence on and to compare the diagnostic accuracy of the Nu-DESC and CAM in detecting postoperative delirium among hospitalised patients.
Design: Systematic review and diagnostic meta-analysis.
Data sources: The PubMed, Embase, Cumulative Index to Nursing and Allied Health Literature, ProQuest Dissertations and Theses A&I, and PsycINFO databases were systematically searched from their inception to February 10, 2023.
Results: In total, 10 (n = 1950) and seven (n = 830) reports were included for the Nu-DESC and CAM, respectively. For Nu-DESC and CAM, the pooled sensitivities were 0.69 and 0.65, respectively, while the summary specificities were 0.99 for Nu-DESC and 0.92 for CAM. The pooled specificity differed significantly between the two tools (p < 0.001), despite comparable pooled sensitivities. The duration of stay in the intensive care unit significantly moderated the summary specificity of Nu-DESC (B = -0.0003, p = 0.009). Regarding CAM, the percentage of female participants showed a positive correlation with its pooled sensitivity (B = 0.005, p = 0.02). Furthermore, studies where clinical specialists served as assessors demonstrated a higher summary sensitivity than those assessed by nurses (0.87 vs. 0.25, p = 0.01).
Conclusion: The sensitivities of the Nu-DESC and CAM for detecting postoperative delirium did not achieve optimal levels. Therefore, developing more accurate tools to detect postoperative delirium by integrating features from related risk factors or incorporating technology-based algorithms to enhance the screening capability is warranted.
