RESUMEN
OBJECTIVE: There is no valid instrument to aid delirium detection in Hong Kong. The objective of this study was to investigate the effectiveness of the Confusion Assessment Method Diagnostic Algorithm (CAM algorithm) and the bilingual version of Nursing Delirium Screening Scale (Nu-DESC) among geriatric inpatients in a Chinese population. METHODS: Between January and March 2007, 100 newly admitted geriatric patients were assessed by physician and bedside nurses, using the CAM algorithm and bilingual version of Nu-DESC, respectively. The two instruments were compared with a gold standard, the psychiatrist's DSM-IV-based diagnosis. Receiver operating characteristic curve (ROC) was used in conjunction with sensitivity and specificity measures to assess the performance of the tools. RESULTS: The prevalence of delirium was 25%. The ROC curve of Nu-DESC showed at the optimal cutoff of >0 a sensitivity of 0.96 and specificity of 0.79. CAM had a sensitivity of 0.76 and specificity of 1. Underlying dementia did not affect the validity of both instruments. Average time of Nu-DESC administration was 1 min/shift and CAM was 10 min. CONCLUSIONS: The bilingual version of Nu-DESC is a sensitive screening tool and the CAM algorithm is an accurate diagnostic instrument for detection of delirium in geriatric inpatient population.