a1 Department of Clinical Epidemiology and Community Studies, St Mary's Hospital, Montreal, Quebec, Canada
a2 Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada
a3 Department of Psychiatry, St Mary's Hospital, Montreal, Quebec, Canada
a4 Department of Psychiatry, McGill University, Montreal, Quebec, Canada
a5 Faculty of Nursing Sciences, Laval University, Montreal, Quebec, Canada
a6 Division of Geriatric Medicine, Jewish General Hospital, Montreal, Quebec, Canada
a7 Institut Universitaire de Gériatrie de Montréal; Département de médecine familiale, Université de Montréal, Quebec, Canada
a8 Division of Geriatric Medicine, Centre Hospitalier de l'Université de Montréal and Department of Medicine, Université de Montréal, Montreal, Quebec, Canada
Background: Previous studies have reported that nurse detection of delirium has low sensitivity compared to a research diagnosis. As yet, no study has examined the use of nurse-observed delirium symptoms combined with research-observed delirium symptoms to diagnose delirium. Our specific aims were: (1) to describe the effect of using nurse-observed symptoms on the prevalence of delirium symptoms and diagnoses in long-term care (LTC) facilities, and (2) to compare the predictive validity of delirium diagnoses based on the use of research-observed symptoms alone with those based on research-observed and nurse-observed symptoms.
Methods: Residents aged 65 years and over of seven LTC facilities were recruited into a prospective study. Using the Confusion Assessment Method (CAM), research assistants (RAs) interviewed residents and nurses to assess delirium symptoms. Delirium symptoms were also abstracted independently from nursing notes. Outcomes measured at five month follow-up were: death, the Hierarchic Dementia Scale (HDS), the Barthel ADL scale, and a composite outcome measure (death, or a 10-point decline in either the HDS or the ADL score).
Results: The prevalence of delirium among 235 LTC residents increased from 14.0% (using research-observed symptoms only) to 24.7% (using research- and nurse-observed symptoms). The relative risks (and 95% confidence intervals) for prediction of the composite outcome, after adjustment for covariates, were: 1.43 (0.88, 1.96) for delirium using research-observed symptoms only; 1.77 (1.13, 2.28) for delirium using research- and nurse-observed symptoms, in comparison with no delirium.
Conclusions: The inclusion of delirium symptoms observed by nurses not only increases the detection of delirium in LTC facilities but improves the prediction of outcomes.
(Received July 20 2010)
(Revised July 28 2010)
(Revised August 26 2010)
(Accepted September 01 2010)
(Online publication September 30 2010)
c1 Correspondence should be addressed to: Dr. J. McCusker, Department of Clinical Epidemiology and Community Studies, St Mary's Hospital Center, 3830 Avenue Lacombe, Montreal, Quebec, H3T 1M5, Canada. Phone: +1 514-435-3511-5062; Fax: +1 514-734-2652. Email: firstname.lastname@example.org.