Comparative Implementation of a Brief App-Directed Protocol for Delirium Identification by Hospitalists, Nurses, and Nursing Assistants A Cohort Study

Edward R. Marcantonio, Donna M. Fick, Yoojin Jung, Sharon K. Inouye, Marie Boltz, Douglas L. Leslie, Erica K. Husser, Priyanka Shrestha, Amber Moore, Kimberlyann Sulmonte, Jonathan Siuta, Malaz Boustani, Long H. Ngo

Research output: Contribution to journalArticlepeer-review

17 Scopus citations


Background: Systematic screening improves delirium identification among hospitalized older adults. Little data exist on how to implement such screening. Objective: To test implementation of a brief app-directed protocol for delirium identification by physicians, nurses, and certified nursing assistants (CNAs) in real-world practice relative to a research reference standard delirium assessment (RSDA). Design: Prospective cohort study. Setting: Large urban academic medical center and small rural community hospital. Participants: 527 general medicine inpatients (mean age, 80 years; 35% with preexisting dementia) and 399 clinicians (53 hospitalists, 236 nurses, and 110 CNAs). Measurements: On 2 study days, enrolled patients had an RSDA. Subsequently, CNAs performed an ultra-brief 2-item screen (UB-2) for delirium, whereas physicians and nurses performed a 2-step protocol consisting of the UB-2 followed in those with a positive screen result by the 3-Minute Diagnostic Assessment for the Confusion Assessment Method. Results: Delirium was diagnosed in 154 of 924 RSDAs (17%) and in 114 of 527 patients (22%). The completion rate for clinician protocols exceeded 97%. The CNAs administered the UB-2 in a mean of 62 seconds (SD, 51). The 2-step protocols were administered in means of 104 seconds (SD, 99) by nurses and 106 seconds (SD, 105) by physicians. The UB-2 had sensitivities of 88% (95% CI, 72% to 96%), 87% (CI, 73% to 95%), and 82% (CI, 65% to 91%) when administered by CNAs, nurses, and physicians, respectively, with specificities of 64% to 70%. The 2-step protocol had overall accuracy of 89% (CI, 83% to 93%) and 87% (CI, 81% to 91%), with sensitivities of 65% (CI, 48% to 79%) and 63% (CI, 46% to 77%) and specificities of 93% (CI, 88% to 96%) and 91% (CI, 86% to 95%), for nurses and physicians, respectively. Two-step protocol sensitivity for moderate to severe delirium was 78% (CI, 54% to 91%). Limitation: Two sites; limited diversity. Conclusion: An app-directed protocol for delirium identification was feasible, brief, and accurate, and CNAs and nurses performed as well as hospitalists.

Original languageEnglish (US)
Pages (from-to)65-73
Number of pages9
JournalAnnals of internal medicine
Issue number1
StatePublished - Jan 1 2022

All Science Journal Classification (ASJC) codes

  • Internal Medicine


Dive into the research topics of 'Comparative Implementation of a Brief App-Directed Protocol for Delirium Identification by Hospitalists, Nurses, and Nursing Assistants A Cohort Study'. Together they form a unique fingerprint.

Cite this