Falls in people with dementia

From the Department of Community Medicine and Rehabilitation, Physiotherapy and Geriatric Medicine, UmeåUniversity, Sweden

Umeå 2007

Falls in people with dementia

Staffan Eriksson

From the Department of Community Medicine and Rehabilitation Physiotherapy and Geriatric Medicine, Umeå University, Sweden

Falls in people with dementia

Staffan Eriksson

Umeå 2007

Department of Community Medicine and Rehabilitation Physiotherapy and Geriatric Medicine, Umeå University, Sweden

Copyright © Staffan Eriksson

New Series No.1135 – ISSN 0346-6612 – ISBN 978-91-7264-433-5

Printed in Sweden by Larsson & Co:s Tryckeri AB, Umeå 2007

CONTENTS

ABSTRACT ... 7

POPULÄRVETENSKAPLIG SAMMANFATTNING PÅ SVENSKA ...9

ABBREVIATIONS ... 11

ORIGINAL PAPERS ... 12

INTRODUCTION... 13

Consequences of falls...13

The occurrence of falls...14

Risk factors for falls ...14

Interventions regarding falls ...16

Falls among people with dementia...17

Risk factors for falls among people with dementia ...17

Interventions regarding falls among people with dementia...21

Dementia ...22

Statistical methods in fall investigations ...24

Rationale for the thesis...29

AIMS OF THE THESIS ... 30

METHODS ... 31

Settings and participants ...31

Paper 1...32

Paper 2...33

Paper 3...33

Paper 4...33

Ethical approval...34

Study design...34

Data collection ...35

Baseline assessment Paper 1 ...36

Baseline assessment Papers 2, 3 and 4 ...37

The reporting of falls and their circumstances in Papers 1, 2, 3 and 4...39

Statistical methodology ...41

Fall rate...41

Chi-square test and t-test ...41

Poisson and negative binomial regression...42

Regression tree ...43

Partial least squares regression...44

RESULTS... 47

The occurrence of falls and injuries ...47

Fallpredictors in people with and without dementia living in residential care facilities (Paper 1) ...48

Fallpredictors in patients with dementia in a psychogeriatric ward (Paper 2) ....53

Comparison of three statistical methods for analysis of fall predictors (Paper 3) ...57

Negative binomial regression ...57

Regression tree ...58

Partial least squares regression...61

Circumstances surrounding falls in patients with dementia in a psychogeriatric ward (Paper 4)...62

Time, location, signs, and activity...62

Gender ...66

High risk circumstances ...67

DISCUSSION... 69

Walking difficulties...70

Gender...70

Visual perception ...72

Medications...74

Patterns of fall predictors according to PLS and regression tree analysis...75

Circumstances surrounding falls in patients with dementia in a psychogeriatric ward...78

Methodological considerations ...79

Regression tree and PLS ...80

Time to first fall...83

Ethical considerations ...85

Implications for clinical work and future research...86

CONCLUSIONS... 89

ACKNOWLEDGEMENTS ... 90

REFERENCES... 92

Falls and concomitant injuries are common problems among large groups of the elderly population, leading to immobility and mortality.

These problems are even more pronounced among people suffering from dementia. This thesis targets fall risk factors for people with dementia in institutions. The overall aim of this thesis was to investigate risk factors for falls, predisposing as well as related to circumstances surrounding falls, and to do this as efficiently as possible.

In a prospective cohort study including residents of residential care facilities with and without dementia, the fall rate was higher for those with dementia, the crude incidence rate ratio (IRR) was 2.55 (95% CI 1.60–4.08) and the adjusted IRR was 3.79 (95% CI 1.95–7.36). In the group of people suffering from dementia, including 103 residents, a total of 197 falls resulted in 11 fractures during the 6-months follow- up period. From the same baseline measurements 26% and 55%, respectively, of the variation in falls could be explained in the group of residents with and without dementia. Fall predictors significantly and independently associated with an increased risk of falls in the group of people suffering from dementia were the category “man walking with an aid” and the use of more than four drugs.

In a prospective cohort study, including 204 patients in a psychogeriatric ward, a total of 244 falls resulted in 14 fractures. Fall predictors significantly and independently associated with an increased risk of falls were male sex, failure to copy a design, use of clomethiazole, and walking difficulties. Treatment with statins was associated with a reduced risk of falls. With these fall predictors in the negative binomial regression (Nbreg) model, 48% of the variation in falls was explained.

The data from the psychogeriatric ward were also analysed with the

use of partial least squares regression (PLS) and regression tree to be

compared with the results of the Nbreg analysis. PLS and regression

tree are techniques based on combinations of variables. They both

showed similar patterns, that a combination of a more severe level of

dementia, behavioral complications and medication related to these

complications is associated with an increased fall rate. Thirty-two

percent and 38%, respectively, of the variation in fall rate were explained in the PLS and regression tree analysis.

The circumstances surrounding the falls in the psychogeriatric ward were analysed. It was found that the fall rate was equally high during the night and the day. A large proportion of the falls was sustained in the patients’ own room and a small proportion of the falls was witnessed by the staff. This pattern was even more pronounced during the night. The proportion of diurnal rhythm disturbances and activity disturbances was higher for falls at night than for falls during the day. Circumstances associated with an increased risk of falls, as shown by a short time to first fall, were anxiety, darkness, not wearing any shoes and, for women, urinary tract infection. The proportion of urinary tract infection was also higher in connection to falls sustained by women than to falls sustained by men.

This thesis confirms that people suffering from dementia are prone to fall. Walking difficulties, male sex and impaired visual perception are factors that should be considered in the work of reducing falls among people suffering from dementia. Furthermore, falls at night, behavioral complications and medication related to these complications should also be considered in this work, especially as the dementia disease progresses. A larger portion of the variation of the outcome variable was explained by the Nbreg model than the regression tree and PLS. However, these statistical methods, based on combinations of variables, gave a complementary perspective on how the fall predictors were related to falls.

Keywords: Dementia, behavioural disturbances, accidental falls, risk

factors, circumstances, analysis of fall predictors, male sex, walking

difficulties, and visual perception.

SAMMANFATTNING PÅ SVENSKA

Bland stora grupper av den äldre delen av befolkningen är fall och påföljande skador ett vanligt problem. Skadorna leder ofta till så stor nedsättning av rörelseförmågan att det dagliga livet påverkas, ibland även till en för tidig död. Dessa problem är ännu tydligare i gruppen av människor som lider av demenssjukdom. I avhandlingen ligger fokus på riskfaktorer för fall i gruppen av människor med demenssjukdom som bor på särskilt boende eller som är inlagd på en psykogeriatrisk avdelning. Det övergripande syftet var att identifiera riskfaktorer för fall, hos personen såväl som riskfaktorer relaterade till omständigheterna kring fallen, och att göra det så effektivt och ändamålsenligt som möjligt.

I en studie utförd på äldreboenden, som inkluderade personer med och utan demenssjukdom, visade det sig att fallfrekvensen var mer än dubbelt så hög i gruppen med demenssjukdom. Under 6 månaders uppföljning inträffade det 197 fall i den 103 personer stora gruppen som hade en demenssjukdom och 11 av dessa fall resulterade i någon typ av fraktur. Från samma uppsättning baslinjedata kunde 26%

respektive 55% av variationen av fall förklaras i gruppen med respektive utan demenssjukdom. Faktorer som var relaterade till fall i gruppen med demenssjukdom var män som använde gånghjälpmedel och användandet av mer än fyra olika läkemedel.

I en studie på en psykogeriatrisk avdelning som inkluderade 204 patienter med demenssjukdom resulterade 244 fall i 14 frakturer. Vid analys av data med negativ binomial regression (Nbreg) påvisades att manligt kön, oförmåga att kopiera femhörningar, användande av heminevrin och nedsatt gångförmåga var faktorer relaterade till ökad fallrisk. Användande av statiner var relaterat till en minskad risk för fall. Med dessa faktorer kunde 48% av variationen av fall förklaras.

Heminevrin är ett läkemedel som används vid agitations- och förvirringstillsånd, samt vid sömnproblematik. Relationen mellan statiner och en minskad fallrisk är en intressant observation, men bör ses med en viss skepsis.

Data från den psykogeriatriska avdelningen analyserades också med

två andra statistiska metoder, partial least squares regression (PLS)

och regressionsträd, och resultaten av de tre metoderna jämfördes.

Riskfaktorer som identifieras av en analys med en multivariat negativ binomialregression påverkar det man vill undersöka (fallrisk) oberoende av andra faktorer i modellen och de resulterande fallriskfaktorerna är också av betydelse för de flesta personer med riskfaktorn i den grupp som studeras. PLS och regressionsträd är, tvärtemot Nbreg, metoder som bygger på kombinationer av faktorer.

Analyserna med dessa båda metoder resulterade i liknande mönster:

en kombination av allvarligare grad av demenssjukdom, beteende- förändringar och mediciner relaterade till dessa beteendeförändringar innebar en ökad fallrisk. 32 respektive 38 procent av fallfrekvens- variationen förklarades med PLS respektive regressionsträdsanalyser.

Omständigheterna i anslutning till fallen analyserades. Vi fann att fallfrekvensen var ungefär lika stor på natten som på dagen. En stor andel av fallen skedde i patientens eget rum och en liten andel av fallen var bevittnade av personalen, vilket var än tydligare på natten.

Andelen dygnsrytmstörningar och aktivitetsstörningar var högre i anslutning till nattfallen än till dagfallen. Omständigheter relaterade till en ökad risk att falla, mätt med tid till första fall, var ångest, mörker, att inte ha på sig skor och för kvinnor urinvägsinfektion. Andelen fall relaterade till urinvägsinfektioner var också högre bland kvinnor än bland män.

Avhandlingen stärker intrycket av att fall är vanligt bland personer

som lider av demenssjukdom. Gångsvårigheter, manligt kön och

försämrad förmåga att tolka synintryck är faktorer som bör tas i

beaktning vid fallförebyggande arbete bland personer som lider av

demenssjukdom. Vidare bör komplikationer i form av förändrat

beteende och relaterade mediciner, särskilt hos personer som nått en

allvarligare grad av demenssjukdom, beaktas. En större andel av

variationen av utfallsvariabeln förklarades av Nbreg-modellen jämfört

med analys med regressionsträd och PLS. Emellertid gav dessa två

senare statistiska metoderna, baserade på kombinationer av variabler,

ett kompletterande perspektiv på hur riskfaktorer var relaterade till

fall. Att fall är lika vanligt förekommande på natten som på dagen och

att väldigt få av nattfallen bevittnas av personalen bör tas med i

beräkningen vid planeringen av det dagliga arbetet på avdelningen.

AD Alzheimer’s Disease

ADL Activities of Daily Living

ACE inhibitors Angiotensin Converting Enzyme inhibitors

BEHAVE-AD Behavioral Pathology in Alzheimer’s Disease Rating scale BPSD Behavioural and Psychological Symptoms in Patients with

Dementia

CI Confidence Interval

DLB Dementia with Lewy Bodies

DSM-IV Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition

FAST The Functional Assessment Staging scale

FTD Frontotemporal Dementia

IRR Incidence Rate Ratio

MMSE Mini-Mental State Examination

N Number of persons included or with available data for the variable

n Number of persons with the characteristic of a specific variable or included in each node of a regression tree analysis

Nbreg Negative Binomial Regression

OLS Ordinary Least Squares technique

PD Parkinsons’s Disease without Dementia

PDD Parkinson’s Disease with Dementia

PLS Partial Least Squares Regression

PY Person Years

R2 Multiple Coefficient of Determination

SD Standard Deviation

Std Err Standard Error of the Regression Coefficient

TIA Transient Ischemic Attack

USD US Dollars

VD Vascular Dementia

VIP Variable Importance in the Projection

ORIGINAL PAPERS

The thesis is based on the following papers:

1. Eriksson S, Gustafson Y, Lundin-Olsson L. Risk factors for falls in people with and without a diagnosis of dementia living in residential care facilities: A prospective study. Arch Gerontol Geriatr 2007; Jun 27: Epub ahead of print.

2. Eriksson S, Gustafson Y, Lundin-Olsson L. Characteristics associated with falls in patients with dementia in a psychogeriatric ward. Aging Clin Exp Res 2007; 19: 97-103.

3. Eriksson S, Lundquist A, Gustafson Y, Lundin-Olsson L.

Comparison of three statistical methods for analysis of fall predictors in people with dementia –Negative binomial regression, regression tree, and partial least squares regression, Manuscript.

4. Eriksson S, Strandberg S, Gustafson Y, Lundin-Olsson L.

Circumstances surrounding falls in patients with dementia in a psychogeriatric ward, Manuscript.

The original articles have been reprinted with the kind

permission of the publishers, in Paper 1 copyright Elsevier

and in Paper 2 copyright Kurtis.

INTRODUCTION

In Sweden at least twice as many people die of the consequences of a fall than in all traffic accidents [1]. The total cost for society of fall accidents involving older people has been estimated at about 5 billion Swedish crowns each year [2].

Consequences of falls

Falls and concomitant injuries are common problems among large groups of the elderly population. Not all falls lead to injuries, but up to 60% do lead to injuries of varying severity [3, 4]. Aside from actual injuries, fear of falling is itself a problem that impacts on daily life.

Hip fracture is one of the most serious osteoporotic fractures and the vast majority of all hip fractures are associated with falls [5-7]. As hip fractures are not only serious but also common, their consequences are well studied. The risk of sustaining a fracture varies with a number of factors, including geography and sex. Reported life-time risks of sustaining a hip fracture vary between 5 and 20%, with Swedish women at the top of the range [5, 8]. Depending on the particular elderly population studied, somewhere between 2 and 9% of the falls result in a fracture and around 1-2% of the falls result in a hip fracture [3, 9-16]. In many cases the consequences of hip fractures can be the cause of a dramatic decline in physical function and living standard.

The consequences of a fracture depend on factors such as age,

prefracture physical status and sex, therefore, the consequences will

vary with the population studied. Some approximate examples are

that 20% die within 6 months, 15-30% move into a nursing home,

40-80% lose their ability to walk independently, and 20-37% lose their

ability to dress themselves [17-21]. Although not a direct cause of

falls, fear of falling is a problem in itself among older people, and is

associated with reduced quality of life, function and avoidance of

activity [22]. The cost of a hip fracture is substantial and was in 1997

estimated at USD 40,000 for the first year after fracture [23].

Introduction

______________________________________________________

The occurrence of falls

Among community-dwelling people older than 65, the annual proportion sustaining a fall has been reported to be around 30%. The proportion varies between subgroups, and ranges from 13 to 50%

with the highest reported values being for women and groups of greater age [24]. Fall rates range between 0.5 and 1.0 falls/PY (person year), and fall rates are usually considered to be higher for women than men [24].

In residential care facilities the reported annual proportion of residents sustaining at least one fall varies between 30 and 56% and fall rates of 1.1 and 3.3 falls/PY have been reported [4, 11, 16, 24-26].

The results in studies of fall rate conducted in residential care facilities are inconclusive regarding the question of a possible gender difference [11, 13, 25, 27]. Among older patients in hospitals fall rates of between 2.9 and 3.5 falls/PY have been reported [11, 28, 29]. In this context the evidence for a gender difference is also inconclusive [11, 15, 25, 30-32]. People suffering from dementia living in nursing homes, group dwellings for people with dementia and psychogeriatric nursing homes have displayed fall rates of about 4-6 falls/PY [11, 13, 33, 34]. In this context two larger studies showed that the fall rate for men was about twice as high as that for women [11, 33].

Risk factors for falls

Investigations of falls, have used different classification systems of

risk factors, providing diverse perspectives on accidental falls. One

system is to differentiate between predisposing and precipitating risk

factors. Predisposing factors refer to “chronic” factors, or factors

whose effect is studied in a longer-term perspective, and precipitating

factors refer to acute factors whose effect, timewise, is studied in

close relation to falls. Furthermore, predisposing factors increase the

predisposition for falls and precipitating factors elicit falls. Another

system is to attribute falls predominantly to either person-related

factors, or to environmental factors; called intrinsic and extrinsic

factors respectively [35]. Also, the term “situational risk factors” has

been used. It includes factors, internal and external, that are present at

the time of the fall without any judgement being made of whether or

not they are precipitating [36]. The term we use in Paper 4

“circumstances that are hazardous regarding falls” is very similar in meaning to “situational risk factors”.

Falls among older people are often caused by multiple interplaying factors that together cause a person to fall, often both person-related and environmental factors [37]. For example, a threshold becomes a problem in combination with a shuffling gait; or rising from a chair becomes a problem in combination with insufficient leg strength. Due to the multiple interplaying factors, the work of making accurate classifications of falls might be difficult. So far there is no classification system that completely captures this complexity.

Risk factors for falls have been investigated in various populations of older people, but few studies have focused on people with dementia.

Over 400 risk factors for falls have been identified among the older population in general [3]. Those predisposing factors presented below all have moderate to strong evidence supporting their association with falls, as estimated by Lord et.al [24]. The factors age, difficulties with mobility and activities of daily living, (ADL), and a history of falls, have all been associated with falls. Among community-dwelling people, women seem to have a higher risk of falling. However in hospitals and institutions the risk of falling seems to be similar for men and women, or even higher for men [11, 15, 25, 38]. A number of measures of postural instability have been shown to be associated with falls, for example impaired sit-to-stand and low stepping frequency. In general, the more demanding the postural task is, the stronger the evidence for it being a risk factor for falls. Since postural stability relies on sensory and neuromuscular function it is not surprising that many factors related to these are associated with falls, including peripheral sensation, muscle strength and reaction time.

Visual acuity and, in particular, depth perception and contrast sensitivity, are associated with falls. There is evidence for the involvement of three medical factors, all related to impairments of the brain: impaired cognition, stroke and Parkinson’s disease. A recurrent issue in relation to falls is “number of chronic conditions” and the use of multiple medications [24]. Use of psychoactive medication has been associated with falls; including benzodiazepines, antidepressants and anti-psychotics [39]. When sustaining equilibrium becomes a challenge for the individual, limitations in attention becomes a risk factor for falls [24].

Introduction

______________________________________________________

The fields of situational and precipitating factors are more difficult to study and are thus less studied than predisposing factors [40]. The situational and precipitating factors presented below originate from studies conducted in residential care facilities or among older patients admitted to hospital wards. These studies indicate that the fall rate is higher during the day than during the night [11, 13, 29, 41, 42].

Furthermore these studies show that most falls occur in the participants’ own room [11, 13, 41], especially during the night (84%)[11]. The staff have been reported to witness 8-22% of the falls [11, 41]. Common activities at the time of falls appear to be walking, need of toileting, and getting in and out of a chair or a bed [13, 29, 41]. Rubenstein et al [10] reviewed four articles regarding precipitating factors and found that about 15% of the fall accidents were related to external factors. According to this review, the most important internal factors were decreased physical performance, dizziness or vertigo and confusion [10]. Acute disease or symptoms, such as urinary tract infection and confusion, have been judged to be a present or precipitating factor in 23-39% of all falls [13, 41, 43]. In the area of medications, the use of benzodiazepines and neuroleptics has been reported to either precipitate falls or to be used in connection with a large proportion of falls [29, 43].

Interventions regarding falls

Meta-analyses have shown that, among older people in general, it is possible to intervene to prevent falls. The risk of falling has been reduced through multifactorial interventions, usually directed towards areas of physical exercise, drugs, vision, environmental hazards, and orthostatic blood pressure. There are a few factors that, as sole intervention targets, seem to be effective by themselves in reducing the risk of falling. One single factor that has proved effective is exercise. The exercise programs consist of general aerobic exercises as well as specific exercises directed toward balance, gait, and strength [44, 45]. A second single factor that seems to be important is vitamin D supplementation [46, 47]. A third single factor that has proved to reduce the risk of falls, restricted to community-living older people with a previous fall, is modification of home hazards [45].

Falls among people with dementia

There are a number of studies that show a higher risk of falls in people with dementia or impaired cognition compared to those without. Among community-living people, the risk of falling has been reported to be 3-5 times higher [14, 48] and in institutions most studies show a 1.4-2.2 times higher risk [32, 34, 49, 50]. The susceptibility to falls has also been accompanied by a high fracture rate [51, 52]. It has also been shown that about 25% of all people sustaining a femoral neck fracture suffer from dementia [53]. After a hip fracture complications in the form of wound infections and fracture dislocation are more common among people with dementia [54]. In addition, following a hip fracture, dementia has been associated with death, as an independent risk factor [55]. After a hip fracture, people with cognitive impairment benefit from a rehabilitation period for their ADL and motor capacity, but possibly to a lesser extent than people without dementia [56-60]. However, following a hip fracture, it has been shown that it is possible to reduce the rate of new falls substantially among people with dementia with the use of a structured, postoperative rehabilitation program including early mobilization and daily efforts to detect and treat postoperative complications such as delirium, pain, decubital ulcers, and malnutrition [61].

Risk factors for falls among people with dementia As previously described there are a large number of risk factors identified among older people in general and there is reason to believe that most of these are also of importance among older people with dementia. There may also be factors that are of specific interest in this group. Predisposing factors for falls presented below and in Table 1 have all been identified in populations of people with dementia.

Various measures of motor function are often collected in studies.

Impairments in these measures are often associated with falls whether they are coarse or specific, such as walking, tandem gait, stride frequency, stride-length variability, and signs of parkinsonism [62-66].

In the area of behavioural and psychological symptoms wandering

and hyperactivity have been associated with falls [65, 67]. In studies,

where no distinction could be made between the drug and the

behaviours for which the drug had been prescribed, atypical

antipsychotic drugs as well as antipsychotic drugs in general have

Introduction

______________________________________________________

been associated with an increased risk of falling [68, 69]. However, in a randomized, controlled study it was seen that an intermediate dose of risperidone (1.0 mg/day) could reduce the risk of falling for people with a combination of wandering behaviour and other behavioural disturbances, while there was no reduced risk of falling for those receiving a higher dose (2.0 mg/day) or a lower dose (0.5 mg/day).

On the contrary, the risk of falling increased for those receiving a higher dose (2.0 mg/day) if they displayed little or no wandering behaviour [67].

The relationship between the severity of the dementia and the risk of falls is unclear. This uncertainty might arise because of interactions between cognitive level and physical function related to the risk of falls. One study showed a linear relationship between MMSE and the risk of falls, those with the lowest MMSE having the highest risk, which remained after adjustment for balance [70]. Others have shown results where there has been a tendency for intermediate levels of cognition and physical function to have the highest association with falls [33, 65].

Aside from predisposing factors identified among people with

dementia there are other factors that might be of specific interest in

this group, both because they have been identified as risk factors in

other groups of older people and they are common among people

with dementia. Orthostatism has been seen in 40-50% of people with

dementia and has also been identified as a fall predictor in some

studies of older people [71, 72]. A combined measure of orthostatic

hypotension and hypotension has also been associated with falls

among people with dementia [71]. Some aspects of impaired vision

have been associated with falls and they coincide with visual

impairments commonly seen in large groups of people with dementia

(poor visual acuity, contrast sensitivity and depth perception) [24, 73].

Table 1. Overview of studies analyzing predictors that increased or decreased the risk of falling among people with dementia or impaired cognition

Sample Follow up of falls

Multi- factorial analysis

Predisposing factors related to any outcome of falls or injuries falls

Brody et al.

1984 [74]

Institution AD ^a Women N=60 and 49

Prospective Yes A decline in physical vigor from a comparatively high level

Camicioli et al.

2004 [64]

Institution AD N=42

Prospective Yes - Morse fall scale - Parkinsonism

- A higher cadence reduced the risk

Kallin et al.

2005 [65]

Institution Cognitively impaired N=2008

Retrospective Yes - Can rise from a chair - Previous falls - Walked with helper - Hyperactive symptoms Katz et al.

2004 [67]

Institution AD, VD ^b , Mixed AD and VD, BPSD ^g Ambulatory N=537

Prospective Randomized and controlled

Yes - In those with moderate/severe wandering risperidone 1.0 mg/day reduced the risk of falling, but 0.5 and 2.0 mg/day did not. In those with no/mild wandering 2.0 mg/day increased the risk of falling.

- Moderate/severe wandering at baseline

Nakamura et al.

1996 [66]

Institution AD Ambulatory N=97

Prospective Yes Stride length variability

Olsson et al.

2005 [75]

Institution Dementia N=364

Prospective No Visual spatial impairment

Passant et al.

1997 [71]

Institution AD, VD, FTD ^c Able to rise N=151

Prospective No Orthostatic hypotension and hypotension

van Dijk et al.

1993 [33]

Institution Dementia N=240

Prospective No A combined measure of physical and psychological function: Increased fall risk as the function decreased to a certain point, thereafter decreased risk Allan et al.

2005 [76]

Community, institution AD, VD, PDD ^d , DLB ^e , and PD ^f

Retrospective Yes Patients with AD lower fall risk

than others

Introduction

______________________________________________________

Sample Follow up of falls

Multi- factorial analysis

Predisposing factors related to any outcome of falls or injuries falls

Ballard et al.

1999 [63]

Community, institution AD, DLB N=65

Prospective No - Patients with DLB higher fall risk than patients with AD - Previous falls

- Parkinsonism - Poor tandem gait Buchner et al.

1987 [51]

Community, institution AD

N=156 and 78 Ambulatory

Retrospective Yes - Peripheral neuropathy - Muskuloskeletal problem - Muscle weakness - Romberg - Arthritis

- Adverse drug reaction Buchner et al.

1988 [70]

Community, institution AD N=117

Retrospective Yes A higher level of cognition reduced the risk of falling Kenny et al.

2000 [62]

Community, institution Cognitively impaired Fallers N=144

Prospective No - Previous fracture - Abnormal gait

- Number of falls in the past six months

- Vascular dementia Lowery et al.

2000 [77]

Community, institution Dementia N=65

Prospective Yes Parkinsonism

Morris et al.

1987 [48]

Community, institution AD N=44

Prospective No For men, high blood pressure

Asada et al.

1996 [78]

Community Dementia N=86

Prospective (injuries falls)

Yes - More difficult care status related to basic ADL.

- Falls in the past year - Barthel score Horikawa et

al.

2005 [69]

Community AD N=124

Prospective Yes - Periventricular white matter lesions

- Use of neuroleptics Kolanowski et

al.

2006 [68]

Community Dementia N=959

Unclear Yes Neuroleptics in general as well as atypical neuroleptics by it self

a Alzheimer’s disease

b Vascular dementia

c Frontotemporal dementia

d Parkinson’s disease with dementia

e Dementia with lewy bodies

f Parkinsons’s disease without dementia

g Behavioural and psychological symptoms in patients with dementia

To our knowledge there are four studies that focus on the circumstances surrounding falls in the context of people with dementia living in institutions [11, 13, 33, 65]. One interesting finding is that the proportion of falls occurring at night has been found to be twice as large at a psychogeriatric clinic compared to a geriatric clinic and a nursing home [11]. Moreover the fall rate has been found to be high the first week after admission compared to the remaining time [33]. The location of most falls was revealed to be the patient’s own room [11, 13], especially at night [11]. In this context it is also of interest to point out that a large proportion of the falls were not witnessed [11, 33]. Common circumstances in connection with falls, sometimes regarded as their cause, are failures in gait and transfer, slipping on urine, acute disease, behavioural and psychological symptoms, fatigue, reduced insight in one’s own abilities and not wearing shoes [26, 33, 65].

Interventions regarding falls among people with dementia

The effect of interventions against falls for people with cognitive

impairment has been investigated in two well conducted studies

neither of which found any significant results [79, 80]. However, both

showed positive trends. In Jensen et al [79] the sample, comprising

residents of residential care facilities, was analyzed in groups of higher

and lower cognition where a non-significant odds ratio of 0.66 refers

to the intervention in the lower cognition group. Furthermore, in the

higher-cognition group the intervention had a significant effect, even

though a large part of this group probably were also cognitively

impaired (median MMSE 24, interquartile range 21-26) [79]. In the

other study conducted by Shaw and colleagues [80] the trend was less

positive (risk ration 0.92) in spite of an extensive multifactorial

intervention directed at, among other things, cardiovascular

instability. In this study the participants had visited an emergency

department following a fall. There are also other positive trends, a

meta-analysis, including studies with different proportions of

dementia, revealed that the sample proportion of people with

dementia does not seem to have any influence on the outcome of fall

interventions in institutions [81].

Introduction

______________________________________________________

There are a couple of successful fall interventions among people suffering from dementia, but these can be considered as special cases because of the groups studied. One was a fall-intervention study at a hospital ward among people suffering from dementia recovering after surgery of a hip fracture. The researchers and the staff managed to reduce the risk of new falls by using an intensive, structured, postoperative care program, including early mobilization and daily training and efforts to detect and treat postoperative complications such as delirium, pain, decubital ulcers, and malnutrition [61]. The other study was a fall intervention among people suffering from dementia in long-term care facilities. Inclusion criteria restricted the sample to those who displayed behavioral disturbances corresponding to a total score of at least 8 on the BEHAVE-AD scale. In this study treatment with risperidone was used as a single intervention tool.

With a treatment dose of 1.0 mg/day they managed to reduce falls, but not with doses of 0.5 and 2.0 mg/day. The positive effect was seen among people displaying a moderate to severe wandering behavior [67]. There is also a pilot study, with an historical control design, where falls seem to be reduced through rigorous supervision of high-risk patients [82].

Dementia

Dementia is a common syndrome among older people and the

prevalence increases with advanced age. In Europe the prevalence of

dementia is about 5-10% for people aged over 65. The approximate

prevalence for the group of people between 65 and 74 is 1-5%, and

for those aged 85 and over it is 10-40% [83]. Dementia is a syndrome

caused by a disease of the brain and the diagnosis of dementia is

based on a set of criteria. There are several organizations that have

worked out criteria for dementia. Commonly included in these criteria

are impairments, which tend to be progressive, of memory and other

higher cognitive functions, as well as emotional and social

functioning. The DSM-IV is perhaps the most commonly used

criteria for diagnosis of dementia and includes the following: 1)

Memory impairment. 2) At least one of the following cognitive

disturbances: aphasia, apraxia, agnosia, and disturbance in executive

functioning. The cognitive deficits in criteria 1 and 2 should each

cause significant impairment in social or occupational functioning and

represent a significant decline from a previous level of functioning

[84]. There are subtypes of dementia, depending on which disease is

affecting the brain, requiring fulfillment of different criteria for a diagnosis [85, 86]. The subtypes of dementia show different patterns of cognitive deficits. Common forms of dementia are Alzheimer’s disease (~50%), vascular dementia (~15-20%) and dementia with Lewy bodies (~10-20%) [85, 87, 88].

In Alzheimer’s disease memory, abstract thinking, language function, visuospatial ability and executive function are typically affected areas [86, 89, 90]. Vascular dementia constitutes of different subtypes with damage, with a variety of causes, to vessels of the brain and concomitant damage of nerve tissue as their common factor and therefore the impairments are more heterogeneous. Vascular dementia is often attributed of impairments in executive function, psychomotor speed, and attention. Moreover, compared to Alzheimer’s disease the areas of cognition including memory are better preserved [86, 91]. Common features of dementia with Lewy bodies are visual hallucinations, motor features of parkinsonism and variation in cognition and attention. Falls have also been reported to be common in this dementia group [85].

Behavioral and psychological symptoms are very common among people with dementia. In institutions 80-90% have been found to exhibit at least one such disturbance and a majority of these receive psychotropic medication. Common symptoms are: psychosis, depression, activity disturbances, aggression, anxiety and delirium [92- 94].

The reduced ability to maintain equilibrium in everyday life can be

seen in a number of ways in people with dementia. In this context it is

important to note that voluntary movements, even routine ones like

walking, are controlled by a network of frontal cortical and

subcortical areas where integration of sensory information with

previously learned motor programs is required [95]. The risk of

developing gait and balance disorders is higher for people with

Alzheimer’s disease, vascular dementia, Parkinson’s disease with

dementia, and dementia with Lewy bodies compared to controls and

the risk is also higher for all of these other types of dementia

compared to Alzheimer’s disease [76, 96]. Notably, the abilities to

maintain equilibrium and to coordinate repetitive limb movements at

high speed is already reduced in the early stages of Alzheimer’s

disease and with sufficiently demanding tasks and sensitive measures

Introduction

______________________________________________________

it seems that the presence of these impairments can be revealed already in stages of mild cognitive impairment [97, 98]. In people suffering from Alzheimer’s disease the reduced speed of repetitive limb movements has been attributed to executive function and attention [99]. In people with dementia, primitive reflexes are also more common and become even more common with increased severity of the dementia [73].

There are also some indications of difficulties in interacting with obstacles when walking, as shown by small margins, for people with Alzheimer’s disease [100]. Some evidence suggests that Alzheimer’s patients rely less on visual sensory information for maintenance of their equilibrium than others [101]. Among older people in general there is a slowing of motor responses due more to slowing of the integration of sensory information rather than of the movement itself [102]. The slowing of motor responses is even more pronounced among people with Alzheimer’s disease than in older people in general and has been attributed to impairments of sensory information integration as well as motor function itself [103].

Although a reduced walking speed is often reported among patients with dementia it has also been suggested that, related to their physical capacity, people with Alzheimer’s disease walk faster than older people without dementia [104]. Thus, it may be that people with Alzheimer’s disease make errors when they move faster than they can really manage. A dual task, making demands on cognitive function, has been found to reduce performance in everyday tasks, such as walking and standing among people with dementia [105-107]. This has been particularly evident for stride-length variability and the increased variability was found to be associated with a reduced executive function, where attention is thought to play a crucial role [107]. Thus, it seems that aside from impairments in the direct movement-generating function the ability to maintain equilibrium is also reduced by areas of slowed motor response and impaired coordination, related to impairments of sensory information integration and attention deficit, respectively [99, 103, 107].

Statistical methods in fall investigations

In studies concerning falls, methodological difficulties and data

properties must be considered when choosing statistical methods.

Logistic regression has often been used. However, in a logistic regression only the proportion of fallers is analyzed; by ignoring repeated falls, relevant information is discarded [108]. In studies of falls, the observation time of study participants often varies due to different amounts of time spent on a medical ward or to drop-outs from the study. Different observation times are difficult to take into account when using logistic regression. On the other hand, Cox regression models can handle the problem of different observation periods [109]. Poisson regression, negative binomial regression, and some Cox regression models can utilize information from multiple falls and handle different observation times [108]. In Cox regression, it is assumed that the ratio of the risk of an event in two groups is constant over time, but this assumption has been questioned. Another assumption that might be questioned in fall investigations taking place on hospital wards is that the risk of falling is the same for participants regardless of their observation time. Negative binomial regression (Nbreg) is often the best choice and is favoured over Poisson regression because it is appropriate for a more extreme skewed distribution (overdispersion) [108]. Poisson regression and Nbreg are similar models where the outcome is a count of events and the distribution is skewed [108, 110]. In these techniques adjustments for possible confounders and different observation times can be made. In the Poisson technique it is assumed that recurrent events are independent of each other, occur randomly in time and that the mean and variance of the outcome variable are equal. When the variance exceeds the mean of the outcome variable the data is said to be overdispersed.

This is accounted for in Nbreg and hence it can be seen as a generalization of the Poisson regression. The overdispersion procedure in Nbreg is an attempt to adjust for a skewed distribution which, among other things, might originate from dependence of events in the same individual [108].

Regression techniques are used to determine the mean effect of a predictor variable on an outcome variable, that is extracting predictor variables that are important for the entire population [111].

Consequently, predictor variables that are important for a population

subgroup are difficult to detect. Furthermore, when using regression

analysis to analyze a heterogeneous population, the dispersion of the

outcome variable is high. Hence, it is more difficult to detect

predictor variables of importance for the entire population. This is

because a large dispersion is usually attributed to a non-representative

Introduction

______________________________________________________

sample, but if the population is heterogeneous the sample may be representative despite a large dispersion. In the context of fall investigations, a large dispersion is common [108].

In addition to the problem of a heterogeneous group, another problem related to dispersion of the outcome variable is the correlation among predictor variables. Multicollinearity is likely to be a problem when many variables are of possible interest with respect to the dependent variable. If two or more predictor variables are involved in interdependencies, problems with multicollinearity eventually arise as these interdependencies grow stronger [112]. In the area of fall investigations, as many as 400 variables have been suggested as being of interest [3]. In the work of making regression models, multicollinearity leads to an increased standard error of the regression coefficients [112]. Large standard errors result in loss of statistical power [112] and thus a decreased probability of rejecting the null hypothesis when it is untrue. The lack of statistical significance of individual variables does not necessarily mean that they do not add explanatory information to the model. In the case when many predictor variables more or less correlate with each other, or when the population of interest is heterogeneous, lack of statistical significance may merely reflect an insufficient sample size, which increases the risk of making a type II error. Also, with multicollinearity it is difficult to partition the individual effects of predictor variables. Accordingly, in these situations the regression coefficients are more unstable and unreliable [112].

In investigations of falls there are some demands on the statistical method that have to be considered. In these investigations the dispersion of the outcome is high and many, possibly correlated, predictor variables are of interest. Furthermore it might be interesting to investigate subgroups of the population, since they might fall for different reasons. Those demands may not be met by regression techniques previously used in fall investigations, therefore regression tree and partial least squares regression (PLS) are methods that could be considered in fall investigations.

Regression tree analysis is a method suitable for exploration of complex

areas [113]. Among the advantages are the ability of the method to

handle a large number of explanatory variables and the possibility of

detecting patterns between variables and nonlinear relationships [113].

Regression tree is a nonparametric statistical method that identifies mutually exclusive subgroups of a population whose members share common characteristics that influence the outcome variable [111].

The tree is constructed by repeatedly splitting the data from parent to child nodes (Figure 1). From all possible splits of all explanatory variables, the algorithm will select the variable that minimizes the variance within groups [113]. Mathematically, the best split minimizes

2 2

R R L

L s p s

p  , where p _L and p _R are the proportions of elements that go to the left and right child nodes, respectively. Further, s and _L ² s ² _R are the variances in the left and right child nodes, respectively [114].

With this splitting procedure the first split is made from the predictor variable that is most important to the outcome in the entire group.

Two main branches will be created, one mainly at “higher risk,” and one mainly at “lower risk”. The branches will form conditional patterns, so that predictor variables located at a lower level in the tree can only be interpreted as important in the combination formed by variables above.

Figure 1. Schematic picture of a regression tree

Partial least squares regression (PLS) can be considered an extension of

the multiple linear regression model Y=Xβ+ε [115-117]. In multiple

regression, the coefficients β are usually estimated by the Ordinary

Introduction

______________________________________________________

Least Squares (OLS) technique, which yields unbiased estimates (provided that the model is correctly specified). Statistical inference is made by making assumptions about the distribution of Y.

Alternatively, the coefficients can be estimated using PLS. In PLS, no distributional assumptions are made, making PLS a rather non- restrictive method. PLS can also be used in situations where OLS is not feasible for reasons not only related to distributional assumptions.

Examples of such cases are when the number of predictor variables is

close to or exceeds the number of observations or when there are

many predictors and multicollinearity problems. On the other hand,

PLS usually yields slightly biased estimates, and there is no elegant

inferential theory, in comparison to OLS where Y can be assumed to

have a multivariate normal distribution. The objective of PLS is to

find linear combinations of X and Y such that the correlation

between the linear combinations (or components) is maximized, in

comparison with Principal Component Analysis where only linear

combinations of X are considered. PLS is focused on prediction; thus,

it has very good predictive properties. However, due to the prediction

focus, it may not be appropriate to use PLS to investigate the

influence on the outcome of particular predictor variables. Analysis by

PLS results in a set of the original predictor variables, a latent variable,

that all more or less have connections with each other and influence

the predicted variable. These variables describe features that pull on

the same thread, in either a positive or negative direction, influencing

the predicted variable. In this way, the correlation of predictor

variables becomes a useful source of information about groups of

variables, in contrast to an ordinary regression technique [115]. The

latent variable can be thought of as a scale describing a phenomenon

corresponding to an increase or decrease in the outcome. In SIMCA,

the linear combination coefficients regarding the x-variables are

denoted “loadings,” and the coefficients regarding the strength of the

association with y are termed “weights.” Another measure that is

possible to use is variable importance in the projection (VIP), which

is a combination of “loadings” and “weights”. VIP is a summary

measure given to each original variable. However, VIP is an overall

measure, meaning that the VIP value cannot be interpreted for each

latent variable separately, except for latent variable 1. The VIP

measure is normalized so that the mean VIP is equal to one.

Rationale for the thesis

Falls and their consequences are a major health problem among the older population, and the problem is even greater among people with cognitive impairment and dementia. The risk of falling is higher among people with dementia and correspondingly high fracture rates have also been reported. Post-operative complications and mortality after hip fracture are also higher among people with dementia.

Several studies have shown that falls and injuries among certain groups of older people can be prevented with a multi-factorial intervention program. However, when well-planned attempts have been made to reduce falls among people suffering from dementia they have mostly failed. This is unfortunate since these people are already in a vulnerable position and would perhaps benefit most from avoidance of additional problems.

Since these interventions among people with dementia have largely

failed and the risk factors for falls among people with dementia have

not been extensively investigated it seems reasonable to further

explore risk factors in this specific group. To increase the efficiency

of the exploration of risk factors and, ultimately the chance of

reducing falls in this group of people, it might be wise to explore

further the statistical methods used for analysis of risk factors.

Aims

______________________________________________________

AIMS OF THE THESIS

This thesis targets fall risk factors in people with dementia in institutions. The overall aim of this study was to investigate risk factors for falls, predisposing as well as those related to circumstances surrounding falls, and to do this as efficiently as possible.

Specific aims

x To identify fall predictors in older people with and without dementia living in residential care facilities, and to compare the results in the two groups (Paper 1).

x To identify predictors associated with falls in patients with dementia in a psychogeriatric ward (Paper 2).

x To compare the efficacies of three statistical methods for the analysis of fall predictors in people with dementia (Paper 3).

x To explore circumstances that are hazardous regarding falls

among people with dementia in a psychogeriatric ward (Paper

4).

METHODS

Settings and participants

In this thesis the papers originate from two samples (A and B), both collected in the northern part of Sweden, in the city of Umeå (Figure 2). This thesis includes residents living in residential care facilities and patients admitted to a hospital ward. When appropriate members of those two groups will be referred to as participants.

Figure 2. The two study samples in Papers 1-4

Sample A (Paper 1) comprised residents in 4 residential care facilities.

These 4 residential care facilities included senior citizens’ apartments, old people’s homes and group dwellings for people with dementia. In general, in Sweden older adults living in residential care facilities are disabled by cognitive and/or physical impairments and require supervision or functional support and care. In the studied facilities a majority of the residents were able to move independently between bed and chair as well as walk indoors, but only a minority were able to use the stairs and/or leave the facility unsupervised. Some lived in private apartments and others in private rooms with shared dining and living rooms. All those who lived in group dwellings for people with dementia and about one third of those living in the other types of residential care facilities, had a diagnosis of dementia. In all facilities residents had 24-hour daily access to assistance with activities of daily living, household issues, and medical care. A specialist in

Sample B 1 ward

Paper 2 204 patients

Paper 3 192 patients

Paper 4 191 patients

75 fell Sample A

4 facilities

Paper 1

186

residents

Methods

______________________________________________________

geriatric medicine made planned weekly house calls as well as emergency visits.

Sample B (Papers 2, 3 and 4) comprised patients in a psychogeriatric ward at Umeå University Hospital. This 24-bed ward specialized in the assessment and treatment of cognitive impairment, with a focus on behavioural and psychological symptoms in patients with dementia (BPSD). Patients with any dementia diagnosis present in the ward from September 1, 2001 to August 31, 2003 were included in the study. The ward comprised three separate identical units, each with a kitchen/dining room, dayroom, two double patient rooms and four single patient rooms. Attached to each patient room was a combined lavatory and shower room. The concept was to create a home-like atmosphere similar to that in group-dwellings for people with dementia. There were some exceptions to the home-like atmosphere.

For instance there were no carpets. Normally the patients were free to move between the units, but when a calmer atmosphere was desired the units were separated by locking the doors, for example when excessive wandering was a problem on the ward. Breakfast, lunch, dinner and coffee with something light to eat twice a day, were served on the ward. The first meal of the day was breakfast, usually served between 8-9 a.m. and the last meal of the day was coffee with something light to eat at 18.30. The staffing varied around the clock.

Counting nurses and nurses’ aids the staffing was 6-9 people between 7 a.m. and 9 p.m. and 4 people during the staff night shift, 9 p.m. to 7 a.m.

Paper 1

Paper 1 was based on sample A and included 219 residents who were

aged 65 years or older. Eleven residents declined to participate. In

addition 22 residents were excluded due to insufficient data regarding

the diagnosis of dementia, leaving 186 residents. Of these, 103

residents suffered from dementia, 21 (20.4%) men and 82 (79.6%)

women. They had a mean age±SD of 83.6±6.3 years (Table 3). The

remaining 83 residents did not suffer from dementia, 30 (36.1%) men

and 53 (63.8%) women. They had a mean age±SD of 83.5±7.1 years

(Table 3).

Paper 2

Paper 2 was based on Sample B. During a 2-year period 242 patient were registered on the ward. Thirty-eight patients were not included, either because they did not fulfil the criteria for a diagnosis of dementia (N=19), or were admitted to the ward more than two months prior to the study onset (N=8), or were admitted to the ward for a second time (N=11). A total of 204 patients, 126 (61.8%) women and 78 (38.2%) men with a mean age±SD of 78.7±7.4 years, were included in the study (Table 3).

Paper 3

Paper 3 was based on sample B. The sample in Paper 3 was selected to obtain a sample more comparable with the final regression model in Paper 2, which included 192 of the 204 patients from the original sample. Twelve patients lacked data from the FAST scale; thus, to obtain a comparable set of data only 192 patients were included in the current data set. Of the 192 patients, 62% were women. The patients had a mean±SD age of 78.8±7.4 years (Table 3).

Paper 4

The main focus in Paper 4 was the circumstances surrounding falls

and it was based on the falls that were sustained by the patients in

sample B. During a 2-year period, 242 patients were registered on the

ward and 191 remained after exclusions. Fifty-one patients present on

the ward during this time were excluded for one or more of the

following reasons: they did not fulfil the criteria for a diagnosis of

dementia (N=19); were admitted to the ward for a second time

(N=11); were already admitted to the ward at the time when the study

with the follow-up of falls was initiated (N=21). They had a mean age

of 78.6±7.5 years. Seventy-five patients fell at least once and in total

they suffered 229 falls. Of the 75 patients who fell, 37 (49%) were

women with a mean age±SD of 80.5±7.8 and 38 (51%) were men

with a mean age±SD of 80.7±7.2 (Table 3).

Methods

______________________________________________________

Ethical approval

The Ethical Committee of the Medical Faculty at Umeå University has approved the studies (§3/98) and (§438/03). All residents in Paper 1 received written and oral information and gave their informed consent to participation in the study. In the case of residents with severe cognitive impairment informed consent was given by relatives or guardians. In this study the staff of the facilities involved also received information about the study and agreed to participate. The data collection in Papers 2-4 was considered as ward development work by the Ethics Committee.

Study design

Papers 1, 2 and 3 were designed as cohort studies with recordings of baseline measurements and prospective follow up of falls. In Paper 4 there was a prospective collection of falls and their circumstances.

An overview of the study designs in Papers 1-4 is presented in Table 2.

Table 2. Overview of study design, setting, sample and aims in Papers 1-4

Paper 1 Paper 2 Paper 3 Paper 4

Design Prospective cohort study

Prospective cohort sudy

Prospective cohort study

Prospective collection of falls and their circumstances Period of

fall collection

6 month Period of ward stay

Period of ward stay

Period of ward stay

Setting Residential care facilities

Psychogeriatric ward

Psychogeriatric ward

Psychogeriatric ward

Study sample Patients with and without dementia

Patients with dementia

Patients with dementia

Patients with dementia with any fall during the ward stay Short aim Identify fall

predictors in people with and without dementia and to compare the results

Identify fall predictors in patients with dementia

Compare the efficacies of three statistical methods for the analysis of fall predictors

Explore

circumstances

that are

hazardous for

falls in patients

with dementia

Data collection

An overview of baseline characteristics of the participants is presented in Table 3.

Table 3. Baseline characteristics of the participants

Paper 1 Paper 1 Paper 2 Paper 3 Paper 4 Without

dementia N=83

With dementia

N=103

Total N=204

Total N=192

Fallers N=75 Age (years), mean±SD ^a 84 ±7 84 ±6 79 ±7 79 ±7 81 ±7 Male, n (%) 30 (36) 21 (20) 78 (38) 73 (38) 38 (51) MMSE score, mean±SD 24 ±4 12 ±7 17 ±6 17 ±6 16 ±6

BEHAVE-AD score, mean±SD   8 ±6 8 ±6 10 ±6

FAST score, mean±SD   8 ±2 8 ±2 8 ±2

Barthel Index, mean±SD 15 ±5 11 ±6   

Living in ordinary housing 0 (0) 0 (0) 155 (76) 146 (76) 51 (69) Walking ability on level

ground, n (%) ^b

Walking without aid 34 (41) 36 (35) 128 (63) 122 (64) 42 (56) Walking with an aid 33 (40) 53 (52) 69 (34) 65 (34) 32 (43) Unable to walk 16 (19) 14 (14) 5 (2) 3 (2) 1 (1)

Dementia diagnoses, n (%) ^c 0 (0) 103 (100)

Alzheimer’s disease   94 (46) 90 (47) 30 (40)

Vascular dementia   50 (25) 47 (25) 22 (29)

Alzheimer’s disease with cerebrovascular lesion

  11 (5) 11 (6) 6 (8)

Frontotemporal dementia 12 (6) 11 (6) 3 (4)

Lewy body dementia   21 (10) 19 (10) 10 (13)

Dementia with parkinson   2 (1) 2 (1) 0 ( 0)

Alcohol-induced dementia   7 (3) 6 (3) 3 (4)

Dementia without further specification

  10 (5) 9 (5) 1 (1)

Other diagnoses, n (%) ^b

Depression 25 (31) 27 (26)   

Previous stroke/TIA 30 (37) 28 (27) 50 (25) 48 (25) 19 (25)

Medication

Bensodiazepines, n (%) 27 (33) 35 (34) 77 (40) 72 (38) 37 (49) Neuroleptics, n (%) 20 (24) 40 (39) 94 (46) 89 (46) 41 (55) Antidepressants, n (%) 24 (29) 38 (37) 84 (41) 78 (41) 31 (41) Number of drugs, mean±SD 7 ±4 6 ±3 6 ±4 6 ±4 7 ±3

a Standarddeviation

b Valid percent

c Some had two dementia diagnoses

Methods

______________________________________________________

Baseline assessment Paper 1

Data collection was performed during one month prior to the onset of data collection of falls and was done by physiotherapists, physicians and in a few cases by registered nurses.

Diagnoses and medication were registered by 6 specialists in geriatric medicine who were also the residents’ ordinary physicians. Delirium was recorded by a nurse at the facility who knew the residents well, and who answered a written questionnaire about episodes of delirium in the preceding month. A review of the diagnoses of dementia was made after the termination of the study. This review was made, according to the DSM IV criteria [84], through chart review and a review of the baseline assessments in a blinded fashion concerning falls, by a specialist in geriatric medicine. All participants with a Mini- Mental State Examination (MMSE) score of less than 24 with no recorded diagnosis of dementia were included in the review (n=82).

An explanation, other than dementia, for the low MMSE score was explored. The low score could be due to different diagnoses or signs such as depression, stroke, impaired vision, paresis affecting writing, and aphasia. If no alternative explanation regarding the low MMSE score was found, the findings still had to fit the symptom profile of dementia, including memory deficit and social dependency, for the participant to be diagnosed as having dementia. In the re-evaluations, 29 new diagnoses of dementia were determined and 22 residents were excluded due to insufficient data.

Hearing, vision, and cognition were assessed by six physiotherapists who

were employed in the study and specially trained for this task. Hearing

was rated as impaired if the resident, without a hearing aid, could not

hear the physiotherapist when he/she spoke in a normal voice from a

distance of 1 meter. Vision was rated as impaired when the resident,

with or without spectacles, could not read a word written in 5 mm

capital letters at reading distance. The Mini-Mental State Examination

(MMSE) was used to assess cognitive function [118]. The maximum

score is 30 and a score below 24 is generally accepted as an indication

of cognitive impairment [119].