Nurses making a diagnosis of dementia - a potential change in practice?

Objectives To compare nurse initial diagnoses following domiciliary visits with subsequent formal multidisciplinary formulation based upon the full possession of investigations, neuropsychological tests and brain imaging. Design Retrospective case note study. Setting Home-based nurse assessments, secondary care based multidisciplinary memory clinic. Participants Four hundred and four consecutive referrals over an 18-month period. Results One hundred and seventy-five patients (43.3%) were diagnosed by the multi-disciplinary team as having a dementia. Seventy-three (41.5%) were classified as Alzheimer's disease, 81 (46%) of mixed sub-type and 20 (11.4%) of vascular origin. Overall nurse-MDT diagnosis agreement was 0.88 (Kappa =0.82). Together, two nurses were able to detect dementia with 94% accuracy (Kappa = 0.88) Sensitivity was 92% and specificity 96%. The positive and negative predictive values of their judgements were 94% and 98% respectively. Nurses were able to sub-diagnose dementia with 86% (kappa = 0.76) accuracy. Multivariate logistic regression modelling showed a MDT dementia diagnosis to be independently and significantly associated with lower scores on the KOLT and total MMSE, and with the absence of biological markers of depression. Nurse dementia diagnoses were associated with lower KOLT scores and the absence of biological depression markers, high past alcohol consumption or a history of hypercholesteraemia. Conclusions In this study, structured initial assessment by a specialist nurse was shown to be an accurate method of determining a diagnosis of cognitive impairment, when compared with formal MDT judgement. The principal benefit of this approach was that signposting to subsequent care pathways was expedited. Arguably, such distributed responsibility affords a viable option for the future detection of early dementia. Copyright (C) 2007 John Wiley & Sons, Ltd.