Funding: This research was funded by a National Health and Medical Research Council (NHMRC) Project Grant (ID: 511125).
Prospective observational study of dementia and delirium in the acute hospital setting
Version of Record online: 27 FEB 2013
© 2012 The Authors; Internal Medicine Journal © 2012 Royal Australasian College of Physicians
Internal Medicine Journal
Volume 43, Issue 3, pages 262–269, March 2013
How to Cite
Travers, C., Byrne, G., Pachana, N., Klein, K. and Gray, L. (2013), Prospective observational study of dementia and delirium in the acute hospital setting. Internal Medicine Journal, 43: 262–269. doi: 10.1111/j.1445-5994.2012.02962.x
Conflict of interest: None.
- Issue online: 27 FEB 2013
- Version of Record online: 27 FEB 2013
- Accepted manuscript online: 24 SEP 2012 02:15AM EST
- Manuscript Accepted: 18 SEP 2012
- Manuscript Received: 7 AUG 2012
- National Health and Medical Research Council (NHMRC). Grant Number: 511125
Dementia and delirium appear to be common among older patients admitted to acute hospitals, although there are few Australian data regarding these important conditions.
The aim of this study was to determine the prevalence and incidence of dementia and delirium among older patients admitted to acute hospitals in Queensland and to profile these patients.
Prospective observational cohort study (n = 493) of patients aged 70 years and older admitted to general medical, general surgical and orthopaedic wards of four acute hospitals in Queensland between 2008 and 2010. Trained research nurses completed comprehensive geriatric assessments and obtained detailed information about each patient's physical, cognitive and psychosocial functioning using the interRAI Acute Care and other standardised instruments. Nurses also visited patients daily to identify incident delirium. Two physicians independently reviewed patients' medical records and assessments to establish the diagnosis of dementia and/or delirium.
Overall, 29.4% of patients (n = 145) were considered to have cognitive impairment, including 102 (20.7% of the total) who were considered to have dementia. This rate increased to 47.4% in the oldest patients (aged ≥90 years). The overall prevalence of delirium at admission was 9.7% (23.5% in patients with dementia), and the rate of incident delirium was 7.6% (14.7% in patients with dementia).
The prevalence of dementia and delirium among older patients admitted to acute hospitals is high and is likely to increase with population aging. It is suggested that hospital design, staffing and processes should be attuned better to meet these patients' needs.
Dementia and delirium appear to be common among older patients admitted to acute hospitals and are associated with a range of adverse outcomes, including longer admissions, functional decline, institutionalisation and death.[1-5] Although accurate data are essential for the planning and provision of hospital services for these patients, most studies reporting prevalence rates for these conditions have been conducted in Europe,[1, 2] and only two recent Australian studies have reported these data.[3, 4]
The most recently reported Australian study was a very large-scale study (n = 253 000) of dementia in acute NSW hospitals, and results showed that 25% of patients aged ≥85 years admitted to NSW hospitals between 2005 and 2007 had a diagnosis of dementia. Because the results were based on analyses of administrative databases, however, and because dementia is only recorded as a diagnosis on hospital databases if it is deemed to contribute significantly to the cost of hospital care, the dementia prevalence is almost certainly underestimated, which the authors themselves acknowledge.
While more studies have been conducted overseas, it is unclear how accurately those findings reflect the Australian situation given important differences in health and aged care systems. In addition, widely differing prevalence rates have been reported, with dementia rates ranging from 3.9% to 63% reported,[1, 7] while reported rates for delirium range from 10% to 31% at admission and from 3% to 29% during the hospital episode. While the differing rates undoubtedly reflect variability in the hospitals and patient populations studied, the studies have also used a variety of assessment methods to classify patients' cognitive status. Importantly, many studies have used Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R) criteria to diagnose dementia and delirium,[1, 8, 9] which tends to underdiagnose both conditions in comparison with the currently accepted DSM-IV criteria.[10-12]
Hence, the aim of the present study was to obtain accurate prevalence and incidence data for dementia and delirium in older patients admitted to acute hospitals in Queensland. These results will complement other recent Australian data and provide information necessary for planning and providing dementia-related hospital services.[3, 4] Results are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines that were developed to improve the transparency and quality of reporting of observational studies.
A prospective observational study of patients aged 70 years and older admitted to general medical, general surgical and orthopaedic wards of four acute public hospitals in South-East Queensland was undertaken. Recruitment took place between October 2008 and May 2010.
Patients aged 70 years and older admitted to selected wards in the four hospitals and expected to remain in hospital for at least 48 h (to allow an adequate observation period and complete assessments) were potentially eligible to participate in the study. The hospitals included two major teaching hospitals (>700 beds) in Brisbane and two smaller hospitals (250–280 beds) near Brisbane. The wards included two general medical and one general surgical ward from all four hospitals and one orthopaedic ward from each of the two larger hospitals, reflecting the ward types most frequently encountered by older patients in acute hospitals. We aimed to recruit equal numbers of patients admitted to each ward to ensure adequate representation of patients from each ward type. Patients admitted Sunday through Thursday were eligible for recruitment, with a maximum of three patients able to be recruited each day at any one hospital.
To minimise selection bias, a computer program was used to select patients randomly when there were more than three eligible patients on the one day at each hospital. Patients were excluded if they were transferred to a study ward from another hospital or ward, and had been admitted more than 48 h previously; were immunocompromised; and hence in isolation or where death was imminent.
Ethical approval was received from the ethics committee of the University of Queensland and from each hospital's Human Research Ethics Committee prior to commencement. Informed written consent was obtained from each patient (or their legal guardian) prior to participation in the study.
Procedures and assessments
Prior to patient recruitment, research nurses with extensive geriatric experience completed a 2-day training programme in using the assessment instruments. This required the completion of five real-life practice cases (excluded from the study sample) that were discussed with the trainer to ensure that competence in using the instruments was achieved. They subsequently undertook comprehensive geriatric assessments and administered standardised assessment instruments to each patient within 48 h of admission. If a patient had surgery requiring a general anaesthetic (GA) within 36 h of admission, the assessment was completed 72 h following surgery. The assessment results were used to assist independent physicians to make a clinical diagnosis of cognitive impairment (see next section).
The interRAI Acute Care (interRAI AC) instrument was used to obtain detailed information about each patient's physical and cognitive status, and psychosocial functioning prior to the onset of the illness/condition, at admission (based on the patient's first 24 h in the ward) and at discharge.[14, 15] The interRAI AC is psychometrically sound and encompasses 12 domains, including cognition, communication, mood and behaviour, activities of daily living (ADL) and instrumental ADL (IADL), continence, nutrition, skin condition, falls, medical diagnosis, advance directives and discharge potential. It is completed using all available sources of information, including the patient, family caregivers and hospital staff. The research nurse also visited the wards daily to determine whether any adverse events, including acute change in mental status, had occurred.
The mini-mental status examination (MMSE) was administered to assess cognitive functioning, and patients were screened for delirium using the confusion assessment method (CAM). If an acute change in the patient's mental status was documented during the hospital stay, the CAM was re-administered. The vulnerable elders survey 13 (VES-13), a 13-item self-report questionnaire, was administered to identify vulnerable older people, defined as ‘older people at increased risk of death or functional decline’. The geriatric depression scale 5 was administered to all patients who were considered capable of completing it (i.e. patients with an MMSE score >14).
The 16-item informant questionnaire on cognitive decline in the elderly (IQCODE) was administered to the patient's family caregiver or someone who knew the patient sufficiently well to provide information regarding changes in the patient's cognitive functioning over the previous 10 years, if available. The patient's relative or caregiver was also asked to complete the neuropsychiatric inventory questionnaire[22, 23] to measure the frequency and severity of 12 psychological and behavioural problems that may have been present in the week prior to the onset of the patient's illness/condition.
Diagnosis of cognitive impairment
Two independent expert physicians (specialist geriatricians and psychogeriatricians) performed case reviews to determine whether the patient was likely to have had dementia prior to their current illness/condition. Cases reviewed included those where the MMSE had either not been completed (because of patient incapacity) or was ≤26, while 50% of cases with MMSE scores between 27 and 30 (selected using the randomisation function in SPSS) were selected for review. Physicians were asked to determine whether patients met the criteria common to the types of dementia listed in DSM-IV (criteria A and B, and the criterion indicating that ‘the deficits do not occur exclusively during the course of a delirium’), in which case dementia was considered likely. If these criteria were not met, dementia was considered unlikely. To make this decision, physicians had access to ALL available patient information, including the patient's chart, and ALL assessments completed by the research nurses with the exception of the interRAI AC cognition items (reserved for validity studies). They worked in pairs that were regularly rotated to avoid any potential decision-making bias, and each case was reviewed independently by each physician in each pair and then jointly discussed to arrive at a decision by consensus. Physicians were not asked to specify the dementia type.
Physicians were also asked to decide whether patients met the DSM-IV criteria for delirium, at admission or subsequently (incident delirium). All cases reviewed for dementia were also reviewed for delirium, with the addition of cases with a positive score on any CAM item or where the CAM was repeated. If the patient was considered to meet ALL three DSM-IV core criteria (A, B, C), delirium was considered likely; otherwise, delirium was considered ‘unlikely’. Physicians were not asked to specify the aetiology.
Differences between groups were assessed using the chi-squared statistic for categorical data and Student's t-test for continuous variables using SPSS for Windows version 18.0 (SPSS Inc., Chicago, IL, USA). To control for potentially inflating the type I error rate secondary to multiple comparisons, the P-values (reported in Tables 1-3) were adjusted according to the method recommended by Benjamini and Yekutieli. Multivariate logistic regression was performed to assess the importance of hospital site as a predictor of dementia.
|Condition/ward type||Age group||Total (n = 493), n (%)||χ2 (df)||P-value|
|70–79 years (n = 246), n (%)||80–89 years (n = 209), n (%)||90+ (n = 38), n (%)|
|General medical||15 (6.1)||48 (22.9)||13 (34.2)||—||—||—|
|General surgical||6 (2.4)||11 (5.3)||2 (5.3)||—||—||—|
|Orthopaedic||0||4 (1.9)||3 (7.9)||—||—||—|
|Total||21 (8.5)||63 (30.1)||18 (47.4)||102 (20.7)||50.0 (2)||<0.001**|
|Delirium at admission†|
|General medical||8 (3.3)||24 (11.5)||5||—||—||—|
|General surgical||0||7 (3.4)||0||—||—||—|
|Total||8 (3.3)||34 (16.3)||6 (15.8)||48 (9.7)||23.5 (2)||<0.001**|
|General medical||9 (3.6)||7 (3.4)||2 (5.3)||—||—||—|
|General surgical||10 (4.1)||3 (1.4)||0||—||—||—|
|Total||19 (7.7)||13 (6.2)||2 (5.3)||34 (6.9)||0.14 (2)||0.93|
|Delirium superimposed on dementia‡|
|General medical||4 (1.6)||19 (9.1)||3 (7.9)||—||—||—|
|General surgical||5 (2.0)||4 (1.9)||0||—||—||—|
|Orthopaedic||0||3 (1.4)||1 (2.6)||—||—||—|
|Total||9 (3.6)||26 (12.4)||4 (10.5)||39 (7.9)||12.4 (2)||0.002*|
|Characteristic||Dementia (n = 102), n (%)||No dementia (n = 391), n (%)||t χ2 (df)||P-value|
|Age (mean, SD)||84.7 years (5.8)||79.3 years (6.2)||−8.02||0.01*|
|Sex||F = 68 (66.7)||F = 220 (56.3)||3.6 (1)||0.22|
|M = 34 (33.3)||M = 171 (43.7)|
|Married/significant other||40 (39.6)||176 (45.8)||10.9 (3)||0.05|
|Widowed||51 (50.5)||132 (34.4)|
|Never married||6 (5.9)||36 (9.4)|
|Separated/divorced||4 (4.0)||40 (10.4)|
|Nil formal||0||2 (0.6)||4.9 (4)||0.88|
|Primary level||14 (17.3)||35 (9.6)|
|Secondary level||53 (65.4)||243 (66.9)|
|Certificate/trade||11 (13.6)||63 (17.4)|
|Tertiary level||3 (3.7)||20 (5.5)|
|Primary language (English)||94 (93.1)||360 (92.5)||0.03 (1)||1.00|
|Private residence||74 (72.5)||357 (91.3)||48.8 (3)||0.01*|
|Residential aged care facility|
|Low-level care||7 (6.9)||15 (3.8)|
|High-level care||19 (18.6)||7 (1.8)|
|Other†||2 (2.0)||12 (3.1)|
|General medical ward||76 (74.5)||218 (55.8)||11.9 (2)||0.02*|
|General surgical ward||19 (18.6)||121 (30.9)|
|Orthopaedic ward||7 (6.9)||52 (13.3)|
|Assessment result/outcome||Dementia (n = 102)†||No dementia (n = 391)†||t||P-value|
|Mini-mental status examination score, mean (SD)||18.7 (6.7)||26.1 (3.2)||15.4||<0.01*|
|Informant questionnaire on cognitive decline in the elderly (IQCODE)‡, mean (SD)||4.3 (0.69) (n = 49)||3.2 (0.56) (n = 120)||−10.5||<0.01*|
|Vulnerable elders survey 13 (VES-13)§, mean (SD)||7.7 (2.5)||4.6 (3.2)||−8.9||<0.01*|
|Geriatric depression scale – 5 (GDS-5)¶, mean (SD)||1.3 (1.3)||0.8 (1.2)||−2.9||0.01*|
|Neuropsychiatric inventory questionnaire (NPI-Q)††, mean (SD)||7.3 (5.8) (n = 52)||2.7 (3.8) (n = 159)||−6.5||<0.01*|
|interRAI AC IADL scale‡‡ (premorbid), mean (SD)||30.6 (14.9)||14.2 (13.2)||−9.6||<0.01*|
|Length of hospital stay (days), mean (SD)||12.0 (11.8)||9.8 (10.0)||−1.9||0.15|
|Characteristic/patient outcome||Dementia, n (%)||No dementia, n (%)||χ2 (df)||P-value|
|interRAI AC ADL scale (premorbid)§§||80.6 (6)||<0.01*|
|0 Independent||40 (39.6)||317 (81.7)||—||—|
|1 Supervision required||10 (9.9)||17 (4.4)||—||—|
|2 Limited impairment||16 (15.8)||25 (6.4)||—||—|
|3 Extensive assistance required ‘1’||15 (14.9)||11 (2.8)||—||—|
|4 Extensive assistance required ‘2’||7 (6.9)||4 (1.0)||—||—|
|5 Dependence||12 (11.9)||14 (3.6)||—||—|
|6 Total dependence||1 (1.0)||0||—||—|
|Delirium at admission||24 (23.5)||24 (6.1)||27.8 (1)||<0.01*|
|Incident delirium (delirium not present at admission)||15 (14.7)||19 (4.8)||18.0 (1)||<0.01*|
Of 768 patients aged ≥70 years admitted to the study hospitals, 733 met eligibility criteria, and 499 patients (68.1%) consented to participate. Six patients (1.2%) withdrew following initial consent, leaving data for 493 patients available for analysis. Reasons for non-participation are presented in Figure 1. There were no significant differences between study participants and those who either declined to participate in the study or withdrew in terms of age (t = 1.16; P = 0.25) or length of stay (t = 0.21; P = 0.94).
The mean age of participants was 80.4 years (standard deviation (SD) = 6.5 years; range 70–99 years), and 58.4% were female (n = 288). The majority was admitted to general medical wards (n = 294; 59.6%), while 140 (28.4%) were admitted to general surgical and 59 (12.0%) to orthopaedic wards. The average MMSE score was 24.8 (SD = 4.9; range 0–30). One hundred and two (20.7%) patients had surgery requiring a GA, of whom 59 (57.8%) were assessed by the research nurse prior to surgery (i.e. within the first 48 h of admission) while 42.2% (n = 43) were assessed following the procedure.
Independent physicians reviewed 75.5% of cases (n = 372) and following review, 102 patients (20.7%) were considered likely to have dementia. Of these, 43 (42.2%) had a diagnosis of dementia documented in their medical record, and proxy consent was provided for 46 (45%) patients. Of patients with an MMSE score between 27 and 30, only four (1.9%) were considered to have dementia. While preliminary analysis (chi-square) showed the dementia prevalence to be significantly lower at one study hospital, results of the multivariate analysis showed that hospital site was not an independent predictor of dementia when adjusted for age and, hence, it was considered reasonable to present aggregated data across the four hospitals for patients with and without dementia.
The prevalence of cognitive impairment by age group and ward type are presented in Table 1. The dementia prevalence was significantly higher in older compared with younger patients with rates of 30.1% (95% confidence interval (CI) 23.9–36.3%) in patients aged 80–89 years and 47.4% (95% CI 31.5–63.3%) in patients ≥90 years compared with 8.5% (95% CI 5–12%) in patients 70–79 years. The overall prevalence of delirium at admission was 9.7% (n = 48), while 34 (7.6%) patients without delirium at admission developed delirium during their hospital stay (incident delirium). The delirium rate at admission was significantly higher in patients aged 80–89 years (16.3%; 95% CI 11.3–21.3%) and in patients ≥90 years (15.8%; 95% CI 4.2–27.4%) compared with patients 70–79 years (3.3%; 1.1–5.5%). By comparison, incident delirium did not differ significantly across age groups.
Key demographic and clinical characteristics of patients with and without dementia are shown in Table 2. Patients considered to have dementia were more frequently admitted to general medical wards (n = 76; 25.9%) than general surgical (n = 19; 13.6%) or orthopaedic wards (n = 7; 11.9%). They were also significantly older, were more likely to be admitted from a residential aged care facility and performed more poorly on all measures administered than patients without dementia (see Tables 2 and 3). They were more vulnerable as measured by the VES-13, had significantly poorer functional ability as measured by the interRAI AC ADL and IADL scales, and had significantly more neuropsychiatric symptoms than patients without dementia. Patients judged likely to have dementia were also more likely to have delirium at admission (n = 24, 23.5%) and to develop incident delirium (n = 15; 19.2%) than patients without dementia (n = 24, 6.1%; n = 19, 5.2% respectively)(see Table 3).
This is the first prospective study reporting the prevalence and incidence of dementia and delirium in older patients admitted to acute hospitals in Queensland. A substantial proportion of patients was found to have cognitive impairment (29.4%), with an estimated 20.7% of patients considered likely to have dementia prior to the onset of their current illness/condition. A further 8.7% without evidence of dementia had delirium at admission or developed delirium during their hospital stay. Patients with dementia were significantly older and had poorer functional ability and more behavioural and psychological symptoms than same-aged patients without dementia – findings also recently reported elsewhere. Their greater vulnerability and poorer functional ability means that these patients would have required considerably more supervision and assistance from hospital staff than patients without dementia, posing a considerable challenge for nursing staff who face considerable time constraints in the busy hospital environment.[27, 28] If the numbers of patients with dementia admitted to acute hospitals increase in the near future as expected, current nursing deployments and training are unlikely to be adequate to care optimally for these patients, perhaps further increasing their risk of adverse outcomes.
The overall dementia prevalence rate of 20.7% is consistent with studies conducted overseas, and in their review of dementia studies, Mukadam and Sampson reported rates of 12.9–26.8% in studies that had screened for delirium. While far higher rates have been reported (e.g. 43.3%, 63%),[29, 30] delirium was not screened for in those studies, with the result that some patients with delirium may have been misclassified as having dementia and hence overinflating the rate. Like others,[1, 3] we also found the dementia prevalence increased with increasing age, with almost half of the oldest patients (≥90 years) found to have dementia in our study – a figure considerably higher than the 25% reported by Draper's group, which they acknowledge as an underestimate. Because patients in the current study were comprehensively assessed for cognitive impairment, the current figures are likely to reflect more accurately the true dementia prevalence in acute Australian hospitals.
Although the delirium rates observed in patients with dementia in our study appear to be comparable with the rate reported by Draper and colleagues (17% in patients aged ≥75 years with dementia), the rates are less readily comparable, as rates for prevalent delirium and incident delirium were not reported separately in that study. By comparison, the Melbourne delirium study found an overall delirium rate of 18% in patients aged ≥65 years at admission that increased to 89% in patients with pre-existing cognitive impairment. The authors acknowledged that the high prevalence may be due to misclassifying some cases of incident delirium as prevalent delirium, and while 89% is likely to be an overestimate, we may have underestimated the rate partly as a consequence of excluding dying patients. Together, however, these findings show that the delirium rate in patients with dementia is substantially higher than in patients without dementia and that delirium rates for these groups should be presented separately. Failure to distinguish between patients with and without pre-existing dementia or cognitive impairment when reporting delirium rates very likely explains some of the marked variation in delirium rates previously reported.[4, 8]
Strengths of the study
Strengths of this study include its multicentre prospective design, relatively large sample size and robust diagnostic approach. The collection of comprehensive patient data within 48 h of admission from multiple sources means that the data regarding the patient's cognitive state at admission are likely to be accurate. Careful monitoring of patients throughout the admission also increases the accuracy of data regarding incident delirium. While chart review by independent physicians allows for a high level of confidence regarding the accuracy of patient diagnoses of dementia and delirium, it falls short of the ‘gold standard’, that is, face-to-face patient assessment by independent physicians. While this approach would have strengthened the study quality, it was not, however, a feasible option in this instance.
Limitations of the study
Limitations of the study include the possibility of selection bias from several sources, including the requirement that patients have an expected hospital stay of at least 48 h (perhaps biasing the sample towards patients with more serious conditions) and the recruitment of participants between Mondays and Fridays only. Another potential source of bias derives from the sizeable number of patients who declined to participate in the study. However, there was no difference in the length of admission between participants and non-participants, suggesting that the two groups did not differ in terms of illness severity.
Although we aimed to accurately identify dementia and delirium because not all cases were independently reviewed, it is possible that some patients may have been misclassified. However, because only a minority of cases with the highest MMSE scores was not reviewed, we consider any such underestimation to be slight. In addition, some cases were assessed several days after their admission, and consequently, delirium at admission may not have been accurately detected in those cases. However, because physicians had access to all documentation including the patient's medical record, any underestimation is again considered to be slight. Further, while IQCODE scores were available for 48% of patients with dementia, they were only available for around 30% of patients without dementia, which may also be a potential source of dementia underestimation. When available to physicians, however, IQCODE data were helpful but no decision rested on these data alone. Hence, dementia underestimation because of these missing data is considered to be low, and the greater availability of caregivers to complete the IQCODE for patients with dementia is consistent with greater levels of dependency. Finally, although research nurses visited the ward daily during the patient's admission, the possibility remains that not all cases of incident delirium were detected.
Cognitive impairment, including dementia and delirium, is common among older patients admitted to acute hospitals in Queensland and is likely to increase over the coming decades. Given these patients are more impaired and vulnerable than patients without dementia and as previous studies have shown that hospitals generally, and acute wards in particular, are ill-equipped to cater for cognitively impaired patients,[27, 28] it is suggested that hospital design, staffing and processes should be increasingly attuned to meet these patients' needs better.
- 1A systematic review of the prevalence, associations and outcomes of dementia in older general hospital inpatients. Int Psychogeriatr 2010; 18: 1–12., .
- 5Australian Institute of Health and Welfare (AIHW). Dementia in Australia: National Data Analysis and Development. Canberra: AIHW; 2007.
- 6Australian Institute of Health and Welfare (AIHW). National Health Data Dictionary, Version 15. Canberra: AIHW; 2010.
- 9American Psychiatric Association (APA). Diagnostic and Statistical Manual of Mental Disorders, 3rd edn Rev. Washington, DC: APA; 1987.
- 10American Psychiatric Association (APA). Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington, DC: APA; 1994.
- 17The Confusion Assessment Method (CAM): Training Manual and Coding Guide. New Haven: Yale University School of Medicine; 2003..
- 20Royal Australian College of General Practitioners (RACGP). Medical Care of Older Persons in Residential Aged Care Facilities. Melbourne: RACGP; 2005.
- 24SPSS Inc. SPSS Advanced Statistics 18.0. Chicago, IL: SPSS Inc.; 2010.
- 25The control of the false discovery rate in multiple testing under dependency. Ann Stat 2001; 29: 1165–1188., .