Skip to main content Accessibility help
×
Home

Prediction of general hospital admission in people with dementia: Cohort study

Published online by Cambridge University Press:  02 January 2018

Tom C. Russ
Affiliation:
University of Edinburgh, and Scottish Dementia Clinical Research Network, National Health Service (NHS) Scotland, Murray Royal Hospital, Perth, and Centre for Cognitive Ageing & Cognitive Epidemiology, University of Edinburgh, and Division of Psychiatry, University of Edinburgh
Mario A. Parra
Affiliation:
University of Edinburgh, and Scottish Dementia Clinical Research Network, NHS Scotland, Murray Royal Hospital, Perth, and Centre for Cognitive Ageing & Cognitive Epidemiology, University of Edinburgh, UK, and UDP-INECO Foundation Core on Neuroscience (UIFCoN), Diego Portales University, Santiago, Chile
Alison E. Lim
Affiliation:
University of Edinburgh
Emma Law
Affiliation:
Scottish Dementia Clinical Research Network, NHS Scotland, Murray Royal Hospital, Perth
Peter J. Connelly
Affiliation:
Scottish Dementia Clinical Research Network, NHS Scotland, Murray Royal Hospital, Perth
John M. Starr
Affiliation:
University of Edinburgh, and Scottish Dementia Clinical Research Network, NHS Scotland, Murray Royal Hospital, Perth, and Centre for Cognitive Ageing & Cognitive Epidemiology, University of Edinburgh, UK
Corresponding
E-mail address:
Rights & Permissions[Opens in a new window]

Abstract

Background

People with dementia are extremely vulnerable in hospital and unscheduled admissions should be avoided if possible.

Aims

To identify any predictors of general hospital admission in people with dementia in a well-characterised national prospective cohort study.

Method

A cohort of 730 persons with dementia was drawn from the Scottish Dementia Research Interest Register (47.8% female; mean age 76.3 years, s.d. = 8.2, range 50–94), with a mean follow-up period of 1.2 years.

Results

In the age- and gender-adjusted multivariable model (n = 681; 251 admitted), Neuropsychiatric Inventory score (hazard ratio per s.d. disadvantage 1.21, 95% CI 1.08–1.36) was identified as an independent predictor of admission to hospital.

Conclusions

Neuropsychiatric symptoms in dementia, measured using the Neuropsychiatric Inventory, predict non-psychiatric hospital admission of people with dementia. Further studies are merited to test whether interventions to reduce such symptoms might reduce unscheduled admissions to acute hospitals.

Type
Papers
Copyright
Copyright © Royal College of Psychiatrists, 2015 

Dementia is a major and increasing public health concern. There are currently over 800 000 people with dementia in the UK, Reference Luengo-Fernandez, Leal and Gray1 and approximately 6% of these individuals are estimated to be in hospital at any one time, ten times the rate in older adults without dementia. Reference Russ, Shenkin, Reynish, Ryan, Anderson and MacLullich2 People with dementia in acute general hospital wards are extremely vulnerable and suffer from increased mortality, Reference Andersen, Lolk, Martinussen and Kragh-Sorensen3Reference Rait, Walters, Bottomley, Petersen, Iliffe and Nazareth5 as well as other adverse outcomes including longer length of stay, increased rates of delirium and new institutionalisation. 6 There is general agreement that unscheduled admission to hospital should be avoided or prevented in this group, where possible. Reference Lyketsos7,Reference Phelan, Borson, Grothaus, Balch and Larson8 However, little attention has been paid to the factors which might influence the risk of a person with dementia being admitted to hospital, as highlighted by a recent systematic review on the topic. Reference Toot, Devine, Akporobaro and Orrell9 Candidate risk factors include age, gender, function, behaviour, medication use and comorbidity. Reference Toot, Devine, Akporobaro and Orrell9Reference Voisin, Sourdet, Cantet, Andrieu and Vellas13 Published studies on risk of admission are either based on clinical samples, Reference Rudolph, Zanin, Jones, Marcantonio, Fong and Yang10,Reference Andrieu, Reynish, Nourhashemi, Shakespeare, Moulias and Ousset12 which limits the generalisation of their findings to the general population, or are limited in size, duration of follow-up or the detail in which participants are characterised at baseline. Reference Voisin, Andrieu, Cantet and Vellas11Reference Gillette-Guyonnet, Nourhashemi, Andrieu, Cantet, Micas and Ousset14 Here we present a prospective cohort study of the association between baseline health, cognitive and functional status and admission to general hospital in a well-characterised national sample of community-dwelling older adults with dementia from across Scotland.

Method

The Scottish Dementia Research Interest Register (SDRIR) comprises people from a variety of demographic backgrounds who have been diagnosed by a clinician as having dementia or a related cognitive disorder (and carers, who were not included in this study) and who have consented (or in cases where the person lacked capacity, through his or her legal representative) to the storage of information on demographic, cognitive, functional and behavioural measures and have expressed an interest in being approached for participation in future research studies. Reference Law, Connelly, Randall, McNeill, Fox and Parra15 At the time of this study people had been enrolled from eight areas of Scotland covering 75% of the Scottish population including both urban and rural areas. Additional details of the SDRIR and recruitment to it are available in a published report. Reference Law, Connelly, Randall, McNeill, Fox and Parra15 Approximately 95% of SDRIR registrants consented to record linkage.

Assessment

An individual joining the register is visited by a clinical studies officer who assesses and records cognition using the Addenbrooke’s Cognitive Examination – Revised (ACE-R). Reference Mioshi, Dawson, Mitchell, Arnold and Hodges16 Function and behaviour are rated, through interviewing the carer, using the Instrumental Activities of Daily Living (IADL) scale, the Physical Self-Maintenance Scale (PSMS) and the Neuropsychiatric Inventory (NPI) including the caregiver distress scale. Reference Lawton and Brody17,Reference Cummings, Mega, Gray, Rosenberg-Thompson, Carusi and Gornbein18 A global assessment of severity, the Clinical Dementia Rating scale (CDR), is completed by interviewing the patient and carer. Reference Hughes, Berg, Danziger, Coben and Martin19,Reference Morris20 The ACE-R is scored out of 100 and the forms used allow direct calculation of the Mini-Mental State Examination (MMSE) score. Reference Folstein, Folstein and McHugh21 The IADL and PSMS are scored out of 8 and 6 respectively, with one point awarded for each domain of activity in which the individual can function independently; higher scores indicate greater independence. The NPI was recorded as the product of frequency and severity scores for each domain (ranging from 1 to 12) with caregiver distress rated separately. Reference Connor, Sabbagh and Cummings22 Total NPI and caregiver distress scores are the sum of scores for all ten domains. The CDR consists of global ratings in six domains or ‘boxes’ (memory, orientation, judgement/problem-solving, community affairs, home and hobbies, and personal care) with a rating of 0 (none), 0.5 (very mild), 1 (mild), 2 (moderate) or 3 (severe) being given for each. Reference Hughes, Berg, Danziger, Coben and Martin19 Details of any illnesses or conditions that are also present and prescribed medication are recorded, as are health-related behaviours (including smoking) and the number of carers. To overcome coding problems associated with initial data entry into the SDRIR and to correct obvious misclassifications (primarily ‘young-onset dementia’ and ‘early dementia’), clinical diagnoses were independently verified by two clinicians (two of T.C.R., P.J.C. and J.M.S.) reviewing the data recorded on the SDRIR with disagreement resolved by discussion or, if necessary, consultation with the third clinician. The medications recorded for each person were scrutinised and individuals prescribed cholinesterase inhibitors or antipsychotics were separately flagged. Additionally, any potentially inappropriate medication for use in older adults was identified according to the updated Beers criteria. Reference Campanelli23

Informed consent for potential data linkage was obtained from all participants at the time of registration on the SDRIR or, if they were incapable of consent, from their legal representative. Ethical approval for the SDRIR and the data linkage was granted by the Scotland A Research Ethics Committee and the record linkage was approved by the Privacy Advisory Committee of the Information Services Division of National Health Service (NHS) National Services Scotland.

Statistical analysis

Data from consenting registrants on the SDRIR were linked, using the unique Community Health Index number, with Scottish Morbidity Records 01 and 04 which respectively record all admissions to general and psychiatric hospitals in Scotland. The date of the admission and all diagnostic codes recorded on admission and discharge were recorded. For this analysis admission to a general hospital was the outcome of interest; data for individuals admitted to a psychiatric hospital were considered censored at that point. The SDRIR data are held by the Health Informatics Centre (HIC) at the University of Dundee and the anonymised, linked data-set was accessed through and analysed in the HIC safe haven using IBM SPSS Statistics version 21.

We calculated the time in calendar days from the date of entry to the SDRIR until the first admission to a general hospital for any reason. For people who were not admitted to hospital, data were censored either at 2 June 2012 or at the date of admission to a psychiatric hospital (n = 25), whichever came first. After confirming that the proportional hazards assumption was valid, we used Cox regression models to produce hazard ratios with accompanying 95% confidence intervals for the association between individual baseline covariables and admission to hospital. Reference Cox24 Covariables examined comprised dementia subtype, the presence of vascular risk factors or comorbidity, having a carer, MMSE, IADL and PSMS scores, total NPI score, NPI caregiver distress score, CDR rating and CDR sum of boxes. Given the possibility that cholinesterase inhibitor use may be associated with reduced risk of entry to nursing home care, Reference Lopez, Becker, Wahed, Saxton, Sweet and Wolk25 and the concerns surrounding the use of antipsychotic drugs, Reference Banerjee26 these variables were also included, as was being prescribed a potentially inappropriate medication. For this analysis continuous variables were standardised so that hazard ratios reflected 1 s.d. disadvantage regardless of the original direction of the scale. There was no effect modification by gender and therefore data for men and women were pooled. All models included age and gender as these were thought to be of importance a priori. Next, having considered bivariate associations, we noted that some variables were moderately correlated (for example, cognition and activities of daily living). Since we wished to include the key independent variables predicting hospital admission we constructed a multivariable model using forward stepwise entry.

A recent systematic review suggested that the NPI might be important in predicting admission to hospital, Reference Toot, Devine, Akporobaro and Orrell9 and so a number of sensitivity analyses examining this association in more detail were carried out. First, we classified total NPI score into three categories, based on published recommended thresholds, Reference Ballard, Thomas, Fossey, Lee, Jacoby and Lana27 in order to examine the shape of the relationship between NPI and hospital admission; second, we investigated the association between individual items and admission to hospital. Finally, individuals with complete data for all variables were compared with individuals with missing data for one or more variable. Missing data were accounted for by repeating the multivariable model using five multiply imputed data-sets.

Results

From an initial sample of 762 we excluded 26 individuals with mild cognitive impairment and 6 individuals with data errors (e.g. duplicated records, missing identifiers or impossible dates of birth), giving an analytic sample of 730 (47.8% female; mean age 76.3 years, s.d. = 8.2, range 50–94). Figure 1 summarises the derivation of the analytic sample. During a follow-up period of 1.2 years (s.d. = 0.8, range 2 days to 3.3 years), 274 (37.5%) of the 730 SDRIR registrants were admitted to a general hospital for any reason. Online Table DS1 summarises the number of people who had various common conditions mentioned in their records on discharge from hospital, including falls or collapse (15.3%), ischaemic heart disease (10.6%) and urinary tract infection (10.2%). Out of 274 admissions, 146 (53.3%) had dementia correctly recorded on discharge, although 6 of these were recorded as having mild cognitive disorder (ICD-10 code F06.7). 28 In contrast, all of the 25 individuals admitted to a psychiatric hospital (not included as admissions in the current analyses) had dementia correctly recorded on their discharge documentation.

Baseline characteristics of the sample are shown in Table 1, stratified into those who were admitted to hospital and those who were not. Just under half of the participants were women, the majority had mild to moderate dementia at baseline and the mean age was in the mid-70s for both groups, although the range spanned several decades. Approximately 80% had Alzheimer’s disease or mixed dementia and about three-quarters were treated with a cholinesterase inhibitor.

Bivariate analyses

Men were more likely to be admitted to hospital than women (HR = 1.32, 95% CI 1.04–1.68; P = 0.022), but age was not associated with admission rates at conventional levels of statistical significance (per 5 years older HR = 1.07, 95% CI 0.99–1.16; P = 0.07). Online Table DS2 shows the association between other baseline covariables and subsequent admission to hospital. There was no increased risk of admission for any diagnostic subtype, apart from the category of ‘other dementia’, incorporating

Fig. 1 Flow chart of participants from initial pooled sample through to analytic sample showing subsequent admission to hospital. MCI, mild cognitive impairment.

Table 1 Baseline characteristics of study participants according to whether or not an individual was admitted to hospital: longitudinal analysis of 730 men and women from the Scottish Dementia Research Interest Register

Characteristic Admitted to hospital (n = 274) Not admitted (n = 456) P
Length of follow-up, years: mean (s.d.) 0.8 (0.6) 1.4 (0.8) < 0.001
maximum 2.8 3.3
Age, years: mean (s.d.) 76.5 (8.1) 76.1 (8.3) 0.51
range 52–94 50–93
Gender, female: n (%) 120 (43.8) 229 (50.2) 0.09
Diagnosis, n (%) 0.66
 Early-onset Alzheimer’s disease 25 (9.1) 49 (10.7)
 Late-onset Alzheimer’s disease 155 (56.6) 264 (57.9)
 Mixed 38 (13.9) 70 (15.4)
 Vascular 32 (11.7) 47 (10.3)
 Other 34 (12.4) 26 (5.7)
Vascular risk factors, n (%)Footnote b 140 (51.1) 236 (51.8) 0.03
Any comorbidityFootnote c 175 (63.9) 228 (50.0) < 0.001
No carers 19 (6.9) 29 (6.4) 0.44
Prescribed cholinesterase inhibitor, n (%) 197 (71.9) 347 (76.1) 0.21
Prescribed antipsychotic, n (%) 14 (5.1) 20 (4.4) 0.65
Prescribed potentially inappropriate medication, n (%)Footnote d 25 (9.1) 44 (9.6) 0.81
MMSE score, mean (s.d.)Footnote e 21.3 (6.2) 21.3 (5.9) 0.94
IADL score, mean (s.d.)Footnote e 3.5 (2.3) 3.7 (2.3) 0.27
PSMS score, mean (s.d.)Footnote e 4.1 (2.0) 4.3 (2.0) 0.23
NPI total score, mean (s.d.)Footnote e 12.2 (15.7) 10.9 (15.2) 0.30
NPI carer distress score, mean (s.d.)Footnote e 6.4 (7.9) 6.3 (8.0) 0.87
CDR overall rating, n (%)Footnote e 0.09
 0 7 (2.7) 18 (4.1)
 0.5 104 (40.2) 181 (41.1)
 1 96 (37.1) 165 (37.5)
 2 45 (17.4) 51 (11.6)
 3 7 (2.7) 25 (5.7)
CDR sum of values, mean (s.d.)Footnote e 5.8 (3.8) 5.7 (4.0) 0.85

a. ‘Other dementia’ comprises dementia with Lewy bodies, Parkinson’s disease dementia, frontotemporal dementia, and other (not specified).

b. Vascular risk factors include hypertension, hypercholesterolemia, diabetes and smoking.

c. Comorbidity includes cancer, respiratory disease, and diabetes.

d. Potentially inappropriate medication according to updated Beers criteria. Reference Connor, Sabbagh and Cummings22

e. Ratings of cognition, function and behaviour were not available for the complete sample, missing data as follows: Mini-Mental State Examination (MMSE), missing n = 44; Instrumental Activities of Daily Living (IADL) scale, missing n = 67; Physical Self-Maintenance Scale (PSMS), missing n = 29; Neuropsychiatric Inventory (NPI), missing n = 34; Clinical Dementia Rating (CDR) scale, overall rating missing n = 31, sum of values missing n = 53.

dementia with Lewy bodies, Parkinson’s disease dementia, frontotemporal lobar degeneration and ‘other dementia, not specified’. The suggestion that people with non-Alzheimer’s, non-vascular dementia subtypes might be more likely to be admitted to hospital owing to a higher frequency of neuropsychiatric symptoms in those conditions was explored in a Cox model incorporating age and gender with total NPI score and diagnostic grouping selected by forward conditional stepwise entry. Total NPI score remained in the model (per s.d. disadvantage HR = 1.22, 95% CI 1.09–1.37; P<0.001), but having a diagnosis of ‘other’ dementia was not statistically significant at conventional levels in this model.

There was no statistically significant association between individual vascular risk factors (hypertension, hypercholesterolaemia, diabetes or smoking) or individual comorbidities (including cancer, respiratory disease and diabetes) and so these were separately pooled into composite variables for reporting in Table 2. Vascular risk factors were not associated with future admission to hospital, but having any comorbidity was (HR = 1.28, 95% CI 1.00–1.65; P = 0.05). None of the three medication variables investigated – cholinesterase inhibitors, antipsychotics and potentially inappropriate medications – were associated with an increased risk of hospital admission. Baseline cognition was not associated with admission to hospital, nor was impaired functional ability as measured by the IADL scale. However, impaired self-care, measured by the PSMS, was associated with an increased risk of admission (per s.d. disadvantage HR = 1.18, 95% CI 1.04–1.33; P = 0.01). Total NPI score (HR = 1.22, 95% CI 1.09–1.37; P<0.001) and the NPI carer distress rating (HR = 1.14, 95 CI 1.02–1.28; P = 0.03) were both associated with an increased risk of admission to hospital. People with more advanced dementia based on CDR score were not more likely to be admitted to hospital than people with milder dementia (P trend = 0.45). The CDR sum of boxes was similarly not statistically significantly associated with hospital admission. Reference Morris20

Multivariable model

Constructing a multivariable model using forward conditional stepwise entry highlighted NPI score as the only statistically significant independent predictor of admission to hospital (per s.d. disadvantage HR = 1.21, 95% CI 1.08–1.36; P = 0.001; based on 251 admissions out of 681 individuals).

Further analyses

To examine the association between NPI score and hospital admission in more detail, we divided NPI scores into three groups based on recommended thresholds: those scoring 0 (the referent), 1 to 14 and greater than or equal to 15. Reference Ballard, Thomas, Fossey, Lee, Jacoby and Lana27 There was a dose-response association between increasing NPI score and hospital admission (v. NPI score of 0: NPI score 1–14 HR = 1.29, 95% CI 0.95–1.76; NPI score ≥15 HR = 1.82, 95% CI 1.30–2.56; P trend = 0.002), as shown in Fig. 2. Examining individual neuropsychiatric symptoms (NPI items) in a post hoc multivariable model using forward stepwise conditional entry highlighted agitation (per s.d. increase HR = 1.28, 95% CI 1.14–1.43;

Table 2 Age- and gender-adjusted hazard ratios (95% confidence intervals) for the association between baseline characteristics and subsequent admission to hospital for any reason: longitudinal analysis of 730 men and women from the Scottish Dementia Research Interest Register

N hospitalised N total HR (95% CI) P
Diagnostic subtype
AD v. all others 274 730 0.86 (0.67–1.11) 0.24
Mixed dementia v. all others 274 730 0.86 (0.61–1.22) 0.41
Vascular dementia v. all others 274 730 1.21 (0.84–1.75) 0.31
‘Other dementia’Footnote a v. all others 274 730 1.67 (1.09–2.56) 0.02
Vascular risk factorsFootnote b 274 730 1.03 (0.81–1.30) 0.82
Any comorbidityFootnote c 274 730 1.28 (1.00–1.65) 0.05
Having any carers v. none 274 730 1.03 (0.64–1.64) 0.91
Prescribed cholinesterase inhibitor 274 730 0.87 (0.67–1.14) 0.32
Prescribed antipsychotic 274 730 1.32 (0.77–2.27) 0.31
Prescribed potentially inappropriate medicationFootnote d 274 730 1.24 (0.82–1.88) 0.30
Mini-mental state examinationFootnote e 263 686 1.00 (0.88–1.14) 0.97
Instrumental Activities of Daily Living ScaleFootnote e 235 663 1.08 (0.95–1.22) 0.26
Physical Self-Maintenance ScaleFootnote e 259 701 1.18 (1.04–1.33) 0.01
Neuropsychiatric InventoryFootnote e (total) 257 696 1.22 (1.09–1.37) <0.001
Neuropsychiatric InventoryFootnote e (carer distress) 257 696 1.14 (1.02–1.28) 0.03
Clinical Dementia Rating Scale, overall rating 259 699 0.45 (for trend)
 1 v. 0/0.5 1.06 (0.81–1.40)
 2/3 v. 0/0.5 1.24 (0.89–1.72)
Clinical Dementia Rating Scale, sum of valuesFootnote e 251 677 1.04 (0.92–1.17) 0.57

a. ‘Other dementia’ comprises dementia with Lewy bodies, Parkinson’s disease dementia, frontotemporal dementia, and other (not specified).

b. Vascular risk factors include hypertension, hypercholesterolemia, diabetes and smoking.

c. Comorbidity includes cancer, respiratory disease, and diabetes.

d. Potentially inappropriate medication according to updated Beers criteria.22

e. Standardised so that hazard ratio is per standard deviation disadvantage.

P<0.001) as the most important aspect of the NPI in terms of predicting admission to hospital.

Data were missing for one or more variables in 21.1% of the sample (n = 154). People with missing data were slightly younger, more likely to be female, less likely to have vascular risk factors but more likely to have a comorbidity of any sort, were more likely to have no carer and to have had more severe dementia, according to CDR and MMSE, but scored slightly better on IADL, PSMS and both scales of the NPI (online Table DS2). Thus, individuals with

Fig. 2 Kaplan-Meier curves for time to general hospital admission in people with dementia stratified by Neuropsychiatric Inventory (NPI) score.

missing data did not uniformly have less favourable levels of risk factors. Using multiple imputation to account for missing data in the multivariable models did not alter our conclusions (per s.d. disadvantage in NPI score HR = 1.20, 95% CI 1.07–1.35; P = 0.003; ‘other dementia’ HR = 1.56, 95% CI 1.02–2.40; P = 0.043).

Discussion

Our main finding was that a person with dementia being admitted to hospital over a mean of 1.2 years was predicted by their total NPI score, with risk increasing as category of severity rises. Further analyses suggested that agitation was the most important predictor within the NPI scale. Among the modifiable factors associated with dementia, bivariate analyses identified a relationship between risk of admission and comorbidity, carer distress and self-care, but these relationships did not persist in multivariable analyses. The risk of admission was significantly lower in the low and middle categories of NPI score than the highest category.

Observing this association gives us little information about the possible mechanism. One could speculate that more neuropsychiatric symptoms at baseline could reflect a higher burden of physical illness. Alternatively, the exacerbation of these symptoms, when already present, might be a common trigger for a general practitioner to seek admission. Whatever the mechanism, these symptoms are potentially modifiable and the important question is raised whether modifying them reduces unscheduled hospital admissions.

Neuropsychiatric symptoms are often important predictors of nursing home admission, Reference Zuidema, Derksen, Verhey and Koopmans29 and some consideration of these findings may be helpful. Persistent agitation or aggression early in dementia diagnosis may be associated with subsequent depressive symptoms in caregivers. Reference Ornstein, Gaugler, Devanand, Scarmeas, Zhu and Stern30 Furthermore, time-varying measures of caregiver burden fully mediated the relationship between four behavioural disturbances (episodes of combativeness, property destruction, repetitive questions and reliving the past) and nursing home admission. Reference Gaugler, Wall, Kane, Menk, Sarsour and Johnston31 Caregivers who did not indicate a care recipient’s dangerous behaviour initially but did so subsequently (an ‘incident’ behaviour problem) were more likely to experience increases in burden (P<0.0026). Reference Gaugler, Wall, Kane, Menk, Sarsour and Johnston32 Caregivers who indicated greater emotional stress, a desire for the care recipient to enter institutional care and feelings of being ‘trapped’ in care responsibilities were more likely to admit people with dementia to nursing homes. However, demographic variables, incontinence and service use did not consistently predict nursing home admission. Reference Gaugler, Yu, Krichbaum and Wyman33

Thus, there are important parallels with factors that influence admission to care homes, particularly as behavioural changes in people with dementia are common and persistent. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey34 Our data are in keeping with the hypothesis that persistence of behaviour problems increases the risk of home placement breaking down and point to focusing on interventions reducing neuropsychiatric symptoms, especially agitation, which is among the most persistent of these problems, as those are potentially likely to decrease the risk of a person with dementia being admitted to a general hospital.

People with dementia are frequently admitted to hospital, Reference Russ, Shenkin, Reynish, Ryan, Anderson and MacLullich2 and the issue of people with dementia in the general hospital is an important one; 6 where possible, unscheduled admissions should be avoided in this patient group. Reference Lyketsos7,Reference Phelan, Borson, Grothaus, Balch and Larson8 However, it is surprising that, despite growing recognition of the public health importance of dementia, particularly in the setting of the general hospital, just over half of these people with confirmed dementia had their diagnosis correctly recorded on discharge from hospital. Recognising that a patient newly admitted to the general hospital has dementia is vital, perhaps most importantly as these people are at high risk of developing delirium which may be partially preventable, but also because the process of discharge planning should be informed by their diagnosis. Reference Russ, Shenkin, Reynish, Ryan, Anderson and MacLullich2

Rarer forms of dementia, including frontotemporal lobar degeneration, are often associated with more prominent neuropsychiatric symptoms. This group was more likely to be admitted to hospital during follow-up in the bivariable models, but it seems that it might be the NPI score itself rather than diagnostic group per se which is driving this association. Although the mean NPI score was higher in the ‘other dementia’ category (17.9, s.d. = 17.8, v. 10.9, s.d. = 15.1; P = 0.012) the maximum scores were much higher in the Alzheimer’s disease, mixed and vascular dementia group (105 points v. 61 points). This is confirmed by the result of the Cox regression model including both the ‘other’ diagnostic category and total NPI score – only total NPI score remained statistically significant at conventional levels.

Comparison with other studies

Dementia is associated with an increased risk of hospital admission but few studies have investigated the impact of individual characteristics of people with dementia on their risk of admission. Reference Phelan, Borson, Grothaus, Balch and Larson8 A recent systematic review of risk factors for people with dementia being admitted to hospital, comprising ten studies, reported that behavioural disturbance was associated with an increased risk of admission. Reference Toot, Devine, Akporobaro and Orrell9 The results of our study echo this and our finding that agitation may be the most important symptom is consistent with the findings reported in this systematic review.

There is little published research prospectively studying cohorts of people with dementia to identify baseline risk factors for admission to hospital. An American teaching hospital cohort study of 827 men and women with Alzheimer’s disease found that 66% were admitted to hospital at least once over a median follow-up period of 3 years, compared with 41% over a mean 1.2 years in our study. Reference Rudolph, Zanin, Jones, Marcantonio, Fong and Yang10 They identified five independent risk factors: higher comorbidity, previous hospital admission, older age, male gender and shorter duration of dementia. These are similar to our findings apart from the effect of age (details of illness duration and previous admission to hospital were not available on the SDRIR), but they did not include measures of neuropsychiatric symptoms or function. The Réseau sur la Maladie d’Alzheimer Français (REAL.FR) study followed 686 patients with Alzheimer’s disease from all over France for 2 years and found that 29% were admitted to hospital at least once in that period. Reference Voisin, Andrieu, Cantet and Vellas11,Reference Voisin, Sourdet, Cantet, Andrieu and Vellas13,Reference Gillette-Guyonnet, Nourhashemi, Andrieu, Cantet, Micas and Ousset14 They identified three independent risk factors for hospital admission: functional impairment, polypharmacy and greater NPI score. Reference Voisin, Andrieu, Cantet and Vellas11 However, they did not present data on antipsychotic or inappropriate medication use within the polypharmacy risk they identified. Another French study following 134 patients with Alzheimer’s disease from a memory clinic found that 23% were admitted to hospital in the following year. Reference Andrieu, Reynish, Nourhashemi, Shakespeare, Moulias and Ousset12 They identified that an inability to bathe independently and lower educational attainment were independent risk factors for admission to hospital, but had limited data on comorbidities and medication, and a small, unrepresentative sample.

Strengths and limitations

This is the largest general population prospective study of risk factors for hospital admission in dementia resulting in adequate power to detect fairly small effect sizes, including identifying a dose-response association between NPI score and hospital admission. A post hoc power calculation suggested that our sample size was powered to detect hazard ratios of approximately 1.2 with 80% power and a significance level of 0.05. It was possible to follow these individuals up for up to 3.3 years during which period a large number were admitted to acute hospitals. The diagnostic categories were based on clinical diagnoses and were independently verified by two clinicians and so are likely to be robust. These people were at a relatively early stage of their illness at baseline, as demonstrated by the MMSE and CDR scores, and all were then community residents, so it may not be possible to extrapolate these results to all people diagnosed with dementia. However, the fact that this sample consisted mainly of people with mild to moderate dementia does not make the results any less important. The supplementary analysis examining individual NPI items was not specified a priori, but the fact that it supports findings from a previous systematic review gives us some confidence in this result. Reference Toot, Devine, Akporobaro and Orrell9

Since the hospital admissions were identified from national surveillance data-sets, it is likely that all hospital admissions were identified, that these data are robust, and thus the times to admission or censoring from entry to the register are accurate. The mean time to admission was 0.8 years and so the symptoms might not have still been present at the time of admission. However, neuropsychiatric symptoms have been shown to be relatively persistent, for example up to 67% for agitation over a period of 2 years. Reference Haupt, Kurz and Janner35 Comprehensive baseline data were available for participants, collected in a standardised manner by trained, skilled clinical studies officers from the Scottish Dementia Clinical Research Network. Furthermore, item-level data were available, allowing us to examine, for example, individual NPI items. The findings from the models using multiple imputation were similar to the main results, suggesting that missing data did not have a substantial effect on our conclusions.

Implications and further research

This study investigated a common and important condition and an important setting – the general hospital. Indeed, people with dementia in general hospitals were one of the main foci of the first Scottish Dementia Strategy. 36 Our data were drawn from across Scotland, which avoids any problems with factors affecting hospital admission at a regional level such as local policies. Although the sample covers much of the country, it is not representative of the general population of people with dementia. Since these individuals are selected by expressing an interest in participating in dementia research, it is likely that they are healthier than the general population of people with dementia. This will affect the generalisability of our findings, but it is likely that conclusions drawn from studying this community-based sample can be broadly applied to the wider population of people with dementia.

If we accept our finding that total NPI score predicts acute general hospital admission in people with dementia, careful consideration is needed regarding the implications of this. Further research is needed to identify whether these admissions could be avoidable through interventions to treat neuropsychiatric symptoms. If it is the case that interventions to reduce neuropsychiatric symptoms and thus NPI score also reduce admission rates to hospital in this vulnerable group, many of the adverse consequences for people with dementia of being in hospital might be avoided. Our results suggest that such interventions should be targeted at individuals with higher levels of neuropsychiatric symptoms (NPI score ≥15). Moreover, because neuropsychiatric symptoms correlate positively with carer strain, Reference Bradshaw, Goldberg, Schneider and Harwood37 interventions that reduce neuropsychiatric symptoms might shorten admissions since carers under less strain might be more willing to accept discharge back home. Furthermore, risk stratification tools to predict unplanned hospital admissions, 38 for example those used in the National Health Service in England & Wales, do not currently incorporate mental health variables. Our results suggest that, at least in people with dementia, these variables may be among the most important in predicting unscheduled hospital admissions.

Clinical trials to examine whether a reduction in neuropsychiatric symptoms in people with dementia reduces unscheduled hospital admissions are now warranted. If this proved to be a modifiable risk factor for such admissions, it could have substantial clinical and public health impact.

Footnotes

Currently employed by the University of Edinburgh and NHS Lothian, from 2009 to 2013, T.C.R. was supported by Alzheimer Scotland. T.C.R., M.A.P., A.E.L. and J.M.S. are members of the Alzheimer Scotland Dementia Research Centre at the University of Edinburgh. T.C.R., M.A.P. and J.M.S. are members of the University of Edinburgh Centre for Cognitive Ageing & Cognitive Epidemiology, part of the cross-council Lifelong Health and Wellbeing Initiative (G0700704/84698); funding from the Biotechnology and Biological Sciences Research Council, the Engineering and Physical Sciences Research Council, the Economic and Social Research Council and the Medical Research Council is gratefully acknowledged for the latter. M.A.P.'s work is supported by Alzheimer's Society grant AS-R42303. The Scottish Dementia Clinical Research Network is funded by the Chief Scientist Office of the Scottish Government.

Declaration of interest

None.

References

1 Luengo-Fernandez, R, Leal, J, Gray, A. Dementia 2010: The Economic Burden of Dementia and Associated Research Funding in the United Kingdom . A Report Produced by the Health Economics Research Centre, University of Oxford for the Alzheimer's Research Trust. Alzheimer's Research Trust, 2010.Google Scholar
2 Russ, TC, Shenkin, SD, Reynish, E, Ryan, T, Anderson, D, MacLullich, AM. Dementia in acute hospital inpatients: the role of the geriatrician. Age Ageing 2012; 41: 282–4.CrossRefGoogle ScholarPubMed
3 Andersen, K, Lolk, A, Martinussen, T, Kragh-Sorensen, P. Very mild to severe dementia and mortality: a 14-year follow-up – the Odense study. Dement Geriatr Cogn Disord 2010; 29: 61–7.CrossRefGoogle ScholarPubMed
4 Larson, EB, Shadlen, MF, Wang, L, McCormick, WC, Bowen, JD, Teri, L, et al. Survival after initial diagnosis of Alzheimer disease. Ann Intern Med 2004; 140: 501–9.CrossRefGoogle ScholarPubMed
5 Rait, G, Walters, K, Bottomley, C, Petersen, I, Iliffe, S, Nazareth, I. Survival of people with clinical diagnosis of dementia in primary care: cohort study. BMJ 2010; 341: c3584.CrossRefGoogle ScholarPubMed
6 Working Group for Liaison Mental Health Services for Older People, Faculty of Old Age Psychiatry, Royal College of Psychiatrists. Who Cares Wins. Improving the Outcome for Older People Admitted to the General Hospital: Guidelines for the Development of Liaison Mental Health Services for Older People. Royal College of Psychiatrists, 2005.Google Scholar
7 Lyketsos, CG. Prevention of unnecessary hospitalization for patients with dementia: the role of ambulatory care. JAMA 2012; 307: 197–8.CrossRefGoogle ScholarPubMed
8 Phelan, EA, Borson, S, Grothaus, L, Balch, S, Larson, EB. Association of incident dementia with hospitalizations. JAMA 2012; 307: 165–72.CrossRefGoogle ScholarPubMed
9 Toot, S, Devine, M, Akporobaro, A, Orrell, M. Causes of hospital admission for people with dementia: a systematic review and meta-analysis. J Am Med Dir Assoc 2013; 14: 463–70.CrossRefGoogle ScholarPubMed
10 Rudolph, JL, Zanin, NM, Jones, RN, Marcantonio, ER, Fong, TG, Yang, FM, et al. Hospitalization in community-dwelling persons with Alzheimer's disease: frequency and causes. J Am Geriatr Soc 2010; 58: 1542–8.CrossRefGoogle ScholarPubMed
11 Voisin, T, Andrieu, S, Cantet, C, Vellas, B. Predictive factors of hospitalizations in Alzheimer's disease: a two-year prospective study in 686 patients of the REAL.FR study. J Nutr Health Aging 2010; 14: 288–91.CrossRefGoogle ScholarPubMed
12 Andrieu, S, Reynish, E, Nourhashemi, F, Shakespeare, A, Moulias, S, Ousset, PJ, et al. Predictive factors of acute hospitalization in 134 patients with Alzheimer's disease: a one year prospective study. Int J Geriatr Psychiatry 2002; 17: 422–6.CrossRefGoogle ScholarPubMed
13 Voisin, T, Sourdet, S, Cantet, C, Andrieu, S, Vellas, B. Descriptive analysis of hospitalizations of patients with Alzheimer's disease: a two-year prospective study of 686 patients from the REAL.FR study. J Nutr Health Aging 2009; 13: 890–2.CrossRefGoogle ScholarPubMed
14 Gillette-Guyonnet, S, Nourhashemi, F, Andrieu, S, Cantet, C, Micas, M, Ousset, P, et al. The REAL.FR research program on Alzheimer's disease and its management: methods and preliminary results. J Nutr Health Aging 2003; 7: 91–6.Google ScholarPubMed
15 Law, E, Connelly, PJ, Randall, E, McNeill, C, Fox, HC, Parra, MA, et al. Does the Addenbrooke's Cognitive Examination – Revised add to the Mini-Mental State Examination in established Alzheimer disease? Results from a national dementia research register. Int J Geriatr Psychiatry 2013; 28: 351–5.CrossRefGoogle ScholarPubMed
16 Mioshi, E, Dawson, K, Mitchell, J, Arnold, R, Hodges, JR. The Addenbrooke's Cognitive Examination Revised (ACE-R): a brief cognitive test battery for dementia screening. Int J Geriatr Psychiatry 2006; 21: 1078–85.CrossRefGoogle ScholarPubMed
17 Lawton, M, Brody, E. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969; 9: 179–86.CrossRefGoogle ScholarPubMed
18 Cummings, JL, Mega, M, Gray, K, Rosenberg-Thompson, S, Carusi, DA, Gornbein, J. The Neuropsychiatric Inventory comprehensive assessment of psychopathology in dementia. Neurology 1994; 44: 2308–14.CrossRefGoogle ScholarPubMed
19 Hughes, CP, Berg, L, Danziger, WL, Coben, LA, Martin, RL. A new clinical scale for the staging of dementia. Br J Psychiatry 1982; 140: 566–72.CrossRefGoogle Scholar
20 Morris, JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 1993; 43: 2412–4.CrossRefGoogle ScholarPubMed
21 Folstein, MF, Folstein, SE, McHugh, PR. ‘Mini-Mental State’: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189–98.Google Scholar
22 Connor, DJ, Sabbagh, MN, Cummings, JL. Comment on administration and scoring of the Neuropsychiatric Inventory in clinical trials. Alzheimers Dement 2008; 4: 390–4.CrossRefGoogle ScholarPubMed
23 Campanelli, CM. American Geriatrics Society updated Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc 2012; 60: 616–31.Google Scholar
24 Cox, DR. Regression models and life-tables. J R Stat Soc Ser B Stat Methodol 1972; 34: 187220.Google Scholar
25 Lopez, OL, Becker, J, Wahed, A, Saxton, J, Sweet, R, Wolk, D, et al. Long-term effects of the concomitant use of memantine with cholinesterase inhibition in Alzheimer disease. J Neurol Neurosurg Psychiatry 2009; 80: 600–7.CrossRefGoogle ScholarPubMed
26 Banerjee, S. The Use of Antipsychotic Medication for People With Dementia: Time for Action. A Report for the Minister of State for Care Services. Department of Health, 2009.Google Scholar
27 Ballard, CG, Thomas, A, Fossey, J, Lee, L, Jacoby, R, Lana, MM, et al. A 3-month, randomized, placebo-controlled, neuroleptic discontinuation study in 100 people with dementia: the neuropsychiatric inventory median cutoff is a predictor of clinical outcome. J Clin Psychiatry 2004; 65: 114–9.CrossRefGoogle ScholarPubMed
28 World Health Organization. International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10). WHO, 2010.Google Scholar
29 Zuidema, SU, Derksen, E, Verhey, FRJ, Koopmans, RTCM. Prevalence of neuropsychiatric symptoms in a large sample of Dutch nursing home patients with dementia. Int J Geriatr Psychiatry 2007; 22: 632–8.CrossRefGoogle Scholar
30 Ornstein, KA, Gaugler, JE, Devanand, DP, Scarmeas, N, Zhu, CW, Stern, Y. Are there sensitive time periods for dementia caregivers? The occurrence of behavioral and psychological symptoms in the early stages of dementia. Int Psychogeriatr 2013; 25: 1453–62.CrossRefGoogle ScholarPubMed
31 Gaugler, JE, Wall, MM, Kane, RL, Menk, JS, Sarsour, K, Johnston, JA, et al. Does caregiver burden mediate the effects of behavioral disturbances on nursing home admission? Am J Geriatr Psychiatry 2011; 19: 497506.CrossRefGoogle ScholarPubMed
32 Gaugler, JE, Wall, MM, Kane, RL, Menk, JS, Sarsour, K, Johnston, JA, et al. The effects of incident and persistent behavioral problems on change in caregiver burden and nursing home admission of persons with dementia. Med Care 2010; 48: 875–83.CrossRefGoogle ScholarPubMed
33 Gaugler, JE, Yu, F, Krichbaum, K, Wyman, JF. Predictors of nursing home admission for persons with dementia. Med Care 2009; 47: 191–8.CrossRefGoogle ScholarPubMed
34 Aalten, P, de Vugt, ME, Jaspers, N, Jolles, J, Verhey, FR. The course of neuropsychiatric symptoms in dementia. Part I: findings from the two-year longitudinal Maasbed study. Int J Geriatr Psychiatry 2005; 20: 523–30.CrossRefGoogle ScholarPubMed
35 Haupt, M, Kurz, A, Janner, M. A 2-year follow-up of behavioural and psychological symptoms in Alzheimer's disease. Dement Geriatr Cogn Disord 2000; 11: 147–52.CrossRefGoogle ScholarPubMed
36 Scottish Government. Scotland's National Dementia Strategy. Scottish Government, 2010.Google Scholar
37 Bradshaw, LE, Goldberg, SE, Schneider, JM, Harwood, RH. Carers for older people with co-morbid cognitive impairment in general hospital: characteristics and psychological well-being. Int J Geriatr Psychiatry 2013; 28: 681–90.CrossRefGoogle ScholarPubMed
38 Academy Health. 2008 HSR Impact Awardee: Improving the Financing and Delivery of Health Care with Risk-Based Predictive Modeling. Academy Health, 2008.Google Scholar

Russ et al. supplementary material

Supplementary Table S1-S2

PDF 45 KB
Submit a response

eLetters

Predicting non-psychiatric admissions: historical considerations

26 February 2015
Claire Hilton, Old Age Psychiatrist,
Clinicians recognised decades ago that behavioural difficulties in dementia predicted non-psychiatric admissions. Some historical data are presented here, although there is a risk that a short letter cannot do justice to a historical narrative. Aubrey Lewis and Helen Goldschmidt in 1943 observed that for older people non-medical crises often precipitated admission to hospital.(1) In 1948,... More
Conflict of interest: None Declared

Altmetric attention score

Full text views

Full text views reflects PDF downloads, PDFs sent to Google Drive, Dropbox and Kindle and HTML full text views.

Total number of HTML views: 92
Total number of PDF views: 461 *
View data table for this chart

* Views captured on Cambridge Core between 02nd January 2018 - 20th January 2021. This data will be updated every 24 hours.

Access
Hostname: page-component-76cb886bbf-tvlwp Total loading time: 0.809 Render date: 2021-01-20T03:47:05.552Z Query parameters: { "hasAccess": "1", "openAccess": "0", "isLogged": "0", "lang": "en" } Feature Flags: { "shouldUseShareProductTool": true, "shouldUseHypothesis": true, "isUnsiloEnabled": true, "metricsAbstractViews": false, "figures": false, "newCiteModal": false }

Send article to Kindle

To send this article to your Kindle, first ensure no-reply@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

Note you can select to send to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Prediction of general hospital admission in people with dementia: Cohort study
Available formats
×

Send article to Dropbox

To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

Prediction of general hospital admission in people with dementia: Cohort study
Available formats
×

Send article to Google Drive

To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

Prediction of general hospital admission in people with dementia: Cohort study
Available formats
×
×

Reply to: Submit a response


Your details


Conflicting interests

Do you have any conflicting interests? *