Background: Delirium occurs commonly among older hospitalized patients
and is frequently not recognized. In an effort to identify tools usef
ul to clinicians in the diagnosis of delirium, test characteristics of
four screening instruments were compared. Methods: Patients 65 years
of age or older who were admitted to one of four medical and surgical
wards of a university teaching hospital were followed up prospectively
. Potential subjects were excluded if unavailable for interviews or di
scharged within 48 hours of admission, or if judged too impaired to pa
rticipate in the daily interviews. Research assistants administered fo
ur instruments used to detect delirium: Digit Span Test, Vigilance 'A'
Test, Clinical Assessment of Confusion, and Confusion Assessment Meth
od. Abnormal scores on these tests or suspicion of acute confusion pro
mpted a referral to the clinician-investigators who then assessed the
patient daily for delirium based on the Diagnostic and Statistical Man
ual of Mental Disorders, Revised Third Edition criteria. Results: Deli
rium occurred in 64 (14.8%) of 432 subjects. The positive likelihood r
atios for all of the instruments were significantly more than 1. The i
nstruments remained useful when applied to selected subgroups: subject
s in whom acute mental status changes were documented, subjects on sur
gical services, and subjects with impaired cognitive status on admissi
on. Combinations of any two instruments did not perform substantially
better than the instrument with the best test characteristics: the Cli
nical Assessment of Confusion. All instruments were more useful at con
firming delirium than in excluding it. Conclusion: The four instrument
s studied, which are suitable for use at the bedside, can aid the clin
ician in identifying patients likely to be suffering from delirium.