Analysis of outcomes following emergency admission in patients with dementia and/or cognitive impairment and/or delirium

Abstract

Background: In general hospital settings, acute admission patients with confusion, whether due to dementia, delirium or other cause often get a worse outcome compared with other patients. In addition, unidentified dementia or cognitive impairment is frequent in general hospital, which contributes to some healthcare outcomes such as length of stay, mortality and re-admission for the dementia population.A systematic review on prevalence, associations and outcomes of dementia in older people admitted to general hospital, which has been the only review that covers a comprehensive comparison of heterogeneous studies so far.
Methods: Systematic searches were conducted using MEDLINE, EMBASE, PsyInfo, CINAHL and the Cochrane library. All types of studies were included in the reviews. Only those studies published in English and targeted in people aged 65 years and above were involved. Meta-analysis was performed with Cochrane’s Review Manager5.3. Heterogeneity of included studies would be measured by I2 which can explain the percentage of the variability in effect due to heterogeneity.The data come from a routine clinical identification programme called Cognitive Geriatric Assessment research (CGA), which was a retrospective cohort study conducted with patients aged above 65 years’ old who have an admission to acute hospital. Patients with admission between 01 January 2012 and 31 December 2012 were all involved in the study. The data analyse were conducted using SAS version9.3 provided by the server of the Safe Haven of Health Informatics Centre (HIC) in Dundee University. Characteristic differences across the people were checked for significance using tests and ANOVA tests. Kaplan-Meier procedures and Log-Rank Tests were conducted to describe the median survival time and survival difference. Cox’s proportional hazards regression model was employed to investigate the association between death and survival time with multiple predictors. To check the validity of the survival model, Kolmogorov-Type Supremum Tests (49) with the ‘assess’ statements in SAS were also added in the final model
Results: There were total of 14 papers included in the systematic review. Meta- analysis of 6 papers with 30 days’ mortality showed that people with dementia had significantly greater mortality by 11% (95% CI: 6%-16%, p<0.001) compared with people without dementia, though with significant heterogeneity (I2 = 68%, p = 0.01). The longest stay was 26.1 days; the minimum length was 4.6 days.In unadjusted Cox’s model, the hazard of death was associated with dementia, clinical delirium, FSD and CI as well as some demographic factors. When adjusting for the four conditions, age, gender CCI and SIMD, the hazard of death for patients with dementia was estimated to decrease slightly from 1.42 unadjusted to 1.17 with a 95% confidence interval of 1.00 - 1.38 . Similarly, those with clinical delirium had 1.23 times greater hazard ratio (HR) to death (95% CI: 1.10-1.37) which also declined from 1.39 when no factors were adjusted for. Although FSD and CI were shown to significantly increase the risk of death with hazard ratios of 1.53 and 1.35 by themselves, they were not significant predictors anymore in the adjusted model. Age, male gender and CCI always significantly contributed to predicting the hazard of death no matter what was adjusted for.
Conclusion: During admissions for the elderly, the four confusion conditions are prevalent worldwide. Each confusion condition is related to worse outcomes in general hospital settings. People with dementia, delirium, cognitive impairment and FSD always do badly in terms of survival time. Dementia and delirium indeed have independent significant influence on mortality when other factors are taken account of in an adjusted Cox regression model. It is crucial to identify CI in a timely way, which potentially could decrease mortality. For patients who have already been diagnosed with some chronic diseases, it will also benefit them if their CI can be detected early and possible treatment earlier.

Date of Award	2015
Original language	English
Awarding Institution	University of Dundee
Supervisor	Peter Donnan (Supervisor)