An economic evaluation of early assessment for Alzheimer's disease in the United Kingdom.
ABSTRACT Diagnosing and treating patients with Alzheimer's disease (AD) at an early stage should improve the quality of life of the patient and caregiver. In the United Kingdom, cost-effectiveness of early assessment of individuals presenting with subjective memory complaints and treating those with AD with donepezil was evaluated.
A discrete event simulation of AD progression and the effect of treatment interventions was developed. Patient-level data from donepezil trials and a 7-year follow-up registry were used to model correlated longitudinal rates of change in cognition, behavior, and function. Other epidemiological and health services data, including estimates of undiagnosed dementia and delays in diagnosis, were based on published sources. Simulated individuals were followed up for 10 years.
In the base-case estimates, 17 patients need to be assessed to diagnose one patient with AD, resulting in an average assessment cost of £4100 ($6000; $1 US = £0.68 UK) per patient diagnosed (2007 cost year). In comparison with a scenario without early assessment or pharmacologic treatment, early assessment reduces health care costs by £3600 ($5300) per patient and societal costs by £7750 ($11,400). Savings are also substantial compared with treatment without early assessment, averaging £2100 ($3100) in health care costs, and £5700 ($8400) in societal costs. Results are most sensitive to estimates of patient care costs and the probability of patients reporting subjective memory complaints. In probabilistic sensitivity analysis, early assessment leads to savings or is highly cost-effective in the majority of cases.
Although early assessment has significant up-front costs, identifying AD patients at an early stage results in cost savings and health benefits compared with no treatment or treatment in the absence of early assessment.
- [Show abstract] [Hide abstract]
ABSTRACT: ABSTRACT Background: The Cambridge Cognitive Examination-Revised (CAMCOG-R) is a sensitive screening tool for the early diagnosis of dementia in older adults. Overall performance on the CAMCOG-R is influenced by educational attainment. Few studies have, however, examined the association between educational attainment and performance on the individual CAMCOG subscales. We aimed to address this question in a sample from a low-and middle-income country (LAMIC), where resource constraints may have compromised access to, and quality of, education for many older adults. Methods: Participants, all over 60 years of age, were 51 cognitively healthy community-dwelling volunteers and 47 individuals diagnosed with mild-moderate stage Alzheimer's disease (AD). Most participants had some high school education. They were administered the CAMCOG-R under standardized conditions. Results: Within both the control and AD patient groups, there were significant associations between years of completed education and CAMCOG-R total score, MMSE score, and CAMCOG-R Language subscale score. In both groups, level of education was not associated with scores on these subscales: in controls, recent memory, R 2 = .21, p = .055, learning memory, R 2 = .16, p = .398, attention/calculation, R 2 = .19, p = .467, and perception, R 2 = .18, p = .984; in AD patients, recent memory, R 2 = .14, p = .340, learning memory, R 2 = .03, p = .680, perception, R 2 = .09, p = .723, and attention/calculation, R 2 = .19, p = .097. Conclusions: Some CAMCOG-R subscale scores were more strongly associated with educational attainment than others. Importantly, however, performance on the recent memory and learning memory subscales was not affected by education. These subscales are sensitive indicators of amnestic mild cognitive impairment (MCI) and early AD. These subscales may therefore remain valid for use as an AD screening tool in resource-poor healthcare settings.International Psychogeriatrics 11/2014; · 1.89 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Phototest is a simple, easy and very brief test with theoretical advantages over available dementia screening tests in Spain. The objective of this study was to estimate the diagnostic accuracy of the Phototest for cognitive impairment and dementia and to compare it with that of the MMSE and the Clock Drawing Test (CDT) in an Argentine population. A phase II cross-sectional study of diagnostic tests evaluation was performed in a sample of 30 controls, 61 with amnestic mild cognitive impairment (a-MCI), and 56 with mild Alzheimer type dementia (DAT). The diagnostic accuracy (DA) was assessed in relation to the clinical diagnosis by calculating the area under the ROC curve (UAC), Sensitivity (Sn), and Specificity (Sp).The DA of the Phototest for a-MCI and DAT (0.93 and 0.97 [UAC]) was higher than that of the MMSE and the CDT. The cut-off points of 27/28 for DAT (Sn = 89.29 [78.1-96.0], Sp = 96.67 [82.8-99.9]) and 30/31 for a-MCI (Sn = 85.25 [73.8-93.0], Sp = 90.00 [73.5-97.9]) maximized the sum of Sn and Sp. Phototest correlates significantly with MMSE and CDT. The Phototest is an efficient instrument for the detection of mild dementia or MCI, with good accuracy and good correlation with tests measuring overall cognitive impairment.The Clinical Neuropsychologist 06/2014; · 1.58 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Considerable advances have been made in modeling Alzheimer's disease (AD), with a move towards individual-level rather than cohort models and simulations that consider multiple dimensions when evaluating disease severity. However, the possibility that disease-modifying agents (DMAs) may emerge requires an update of existing modeling frameworks.PharmacoEconomics 08/2014; · 3.34 Impact Factor
An economic evaluation of early assessment for Alzheimer’s
disease in the United Kingdom
Dennis Getsiosa,*, Steve Blumeb, Khajak J. Ishakc, Grant Maclained, Luis Herna ´ndeze,f
aUnited BioSource Corporation, Lexington, MA, USA
bUnited BioSource Corporation, Bethesda, MD, USA
cUnited BioSource Corporation, Montreal, Quebec, Canada
dEisai Europe Ltd, Hatfield, Hertfordshire, United Kingdom
eUnited BioSource Corporation, Bogota ´, Colombia
fIndustrial Engineering Department, Universidad de los Andes, Bogota ´, Colombia
AbstractBackground: Diagnosing and treating patients with Alzheimer’s disease (AD) at an early stage
should improve the quality of life of the patient and caregiver. In the United Kingdom, cost-
effectiveness of early assessment of individuals presenting with subjective memory complaints
and treating those with AD with donepezil was evaluated.
Methods: A discrete event simulation of AD progression and the effect of treatment interven-
tions was developed. Patient-level data from donepezil trials and a 7-year follow-up registry
were used to model correlated longitudinal rates of change in cognition, behavior, and function.
Other epidemiological and health services data, including estimates of undiagnosed dementia and
delays in diagnosis, were based on published sources. Simulated individuals were followed up for
Results: In the base-case estimates, 17 patients need to be assessed to diagnose one patient with AD,
resulting in an average assessment cost of £4100 ($6000; $1 US 5 £0.68 UK) per patient diagnosed
(2007 cost year). In comparison with a scenario without early assessment or pharmacologic treat-
ment, early assessment reduces health care costs by £3600 ($5300) per patient and societal costs
by £7750 ($11,400). Savings are also substantial compared with treatment without early assessment,
averaging £2100 ($3100) in health care costs, and £5700 ($8400) in societal costs. Results are most
complaints. In probabilistic sensitivity analysis, early assessment leads to savings or is highly cost-
effective in the majority of cases.
Conclusions: Although earlyassessment hassignificant up-front costs,identifyingAD patientsat an
early stage results in cost savings and health benefits compared with no treatment or treatment in the
absence of early assessment.
? 2012 The Alzheimer’s Association. All rights reserved.
Keywords:Donepezil; Cholinesterase inhibitors; Alzheimer’s disease; Cost-effectiveness; Early assessment
Alzheimer’s disease (AD) is a progressive, debilitating
condition with profound effects on patients and their care-
givers, and substantial costs to society. In the United King-
dom, the number of individuals with dementiais
approaching 700,000, with more than half of these suffering
from AD , and the cost of caring for these patients has
been estimated at £17 billion annually . Despite the bur-
den posed on individuals and the health care system, diagno-
sis of AD is clearly suboptimal, with estimates of
undiagnosed AD as high as 80% of all cases . A recent re-
port on dementia services by the National Audit Office esti-
mates that more than half of all cases of AD in the United
Kingdom are undiagnosed .
*Corresponding author. Tel.: 781-960-0206; Fax: 781-761-0147.
E-mail address: email@example.com
1552-5260/$ - see front matter ? 2012 The Alzheimer’s Association. All rights reserved.
Alzheimer’s & Dementia 8 (2012) 22e30
Cholinesterase inhibitors have been shown to be effective in
improving or stabilizing patients’ cognition, function, and
behavioral symptoms [4,5], and most economic evaluations
on the use of these treatments have indicated that they are
highly cost-effective [6,7]. Little research, however, has
been conducted on how earlier detection and treatment of
patients would influence long-term outcomes, and whether
early assessment for AD, which would have significant up-
front costs, might be a cost-effective strategy. Two recent
studies [2,8] have suggested that early identification of
dementia could indeed be cost-effective.In the United King-
program found that over a period of 10 years, the costs of the
program would be more than offset by reductions in public
and private expenditures if the program led to a 10% reduc-
were based on broad assumptions on the effect of interven-
tions. This article presents an economic evaluation of early
assessment for AD in the United Kingdom, and treatment
developed discrete event simulation . The health and eco-
nomic effects of such a strategy are compared with the fol-
lowing: (1) a scenario where patients are treated when
diagnosed with AD in the absence of assessment at an early
stage, and (2) a scenario where patients remain untreated,
even when diagnosed. The latter is relevant in the United
Kingdom, where severe restrictions have been imposed on
Clinical Excellence (NICE) .
The economic analysis is based on a discrete event simu-
lation developed for the evaluation of the cost-effectiveness
of donepezil in the United Kingdom  and adapted to con-
sider the effect of early assessment. The model calculates
both the direct costs of care and the indirect costs of care-
giver time, and tracks individuals over a period of 10 years.
The primary outcome for patients and caregivers is the
quality-adjusted life year (QALY). QALYs are a composite
measure of survival and quality of life, where survival
time is weighed by individuals’ health-related quality of
life, as measured by health utilities. Individuals in perfect
health are assigned a health utility score of 1, with lower
scores representing worse quality of life . Costs are re-
ported in 2007 British pounds, and costs and outcomes are
discounted at 3.5% per annum.
Discrete event simulation models outcomes by creating
simulated individuals who capture the heterogeneity in dis-
ease progression and other outcomes. Furthermore, as AD is
a multifaceted illness, the technique is able to capture corre-
latedchangesonmultipledomains oncontinuousrather than
discrete scales. The approach allows for a compact means of
modeling the complexities associated with AD progression
and avoids the need to oversimplify the disease [6,12].
Although the core model and predictive equations have
been detailed elsewhere , elements related to screening
have not been described previously. Figure 1 provides an
overview of the model flow. The model begins by creating
simulated patients with each patient assigned their own
unique attributes. For ensuring comparable populations, an
identical copy of each patient is then created with the origi-
nal patient assigned to early assessment and treatment with
donepezil 10 mg, and the “cloned” patient assigned to treat-
ment either in the absence of early assessment or to no treat-
Patients are assumed to enter the model at some random
interval between onset of disease and the time they would
have been diagnosed in the absence of early assessment.
For patients who are assigned to early assessment, an assess-
ment cost is incurred and treatment is initiated if their Mini-
Mental State Examination (MMSE) score falls between 10
and 26. If their score is above 26, they are assigned a new as-
sessment time, and treatment is initiated only after they are
reassessed and their MMSE is sufficiently low to warrant
treatment. If, by the time they are reassessed, their MMSE
score is found to be below 10, treatment is not initiated.
For patients treated in the absence of early assessment pro-
gram, treatment is initiated at the pre-assigned time of diag-
Patients are then followed up with their characteristics
updated over time. The simulation measures disease severity
on the basis of cognition (using the MMSE), behavior (using
ing (ADLs) and, instrumental activities of daily living
(IADLs). At each follow-up point in the simulation, the
change in MMSE score of each patient since the last time
the patient’s status was updated is calculated. To ensure
severity, NPI, ADL, and IADL changes are then calculated
sequentially, each accounting for changes in the previously
estimated scores. Changes in all these measures are a func-
tion of patients’ characteristics, previous changes in disease
severity, whether they are on treatment, how long they have
been on treatment, and if they have been on treatment, but
have discontinued therapy, the duration since last treated.
On the basis of a given patient’s disease severity and treat-
ment at each update, costs, health utilities, and caregiver out-
comes are calculated and accumulated over the appropriate
period, with the update process continuing until either the
ment early, or in case when their disease has reached a high
degreeof severity(MMSE,10 inthese analyses). Mortality
is also modeled. Because cholinesterase inhibitors have not
been associated with improvements in survival, time of
death is assigned to each individual before treatment assign-
ment, thereby ensuring that survival is identical in all patient
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e3023
Foreachscenario,20,000patientswere sampled andsim-
2.1. Data sources
2.1.1. Disease progression and treatment effects
Data from the CERAD (Consortium to Establish a Regis-
try for Alzheimer’s disease) registry , seven donepezil
clinical trials spanning mild to severe AD [14e20], and
two open-label extension studies [21,22] were used to
derive individualized predictive equations for progression
of AD and the effect of donepezil on progression .
CERAD data were used to establish the baseline progres-
data than the donepezil trials. A piece-wise linear regression
model was created to calculate the rate of change in MMSE,
defined as annual change in score since previous measure-
ment, from the previous MMSE in CERAD . In this ap-
proach, a different slope is allowed in different intervals of
the MMSE scale, which reflects a different rate of change
at different stages of the disease.
A model similar to that derived from the CERAD data
was fitted to the trial data to quantify a treatment effect.
On the basis of the observed patterns of rate of decline in
the treated and placebo groups, a period of 20 weeks was
identified as a point of change in effect. The estimated treat-
yr in the first 20 weeks of treatment and 2.47 in weeks
20e52. After week 52, continued treatment was assumed
to have no further effect on the predicted rate of disease pro-
gression; however, previous treatment gains were main-
tained and rate of disease progression was the same as that
for untreated patients.
On the basis of the data from the donepezil trials, it was
predicted that the scores obtained from NPI, ADL, and
IADL change from baseline . Changes in NPI scores
were linked to changes in MMSE, but were also found to
be a function of baseline and previous NPI, baseline
MMSE, race, and use of psychiatric medications. Changes
in ADL were also a function of baseline and most current
MMSE scores, baseline and previous ADL, as well as age,
use of psychiatric medications, and treatment. Finally,
changes in IADL were found to be a function of baseline
and most recent IADL, baseline and most recent MMSE,
baseline ADL, and gender.
Premature treatment discontinuation was modeled using
data from a real-world UK study of AD patients treated
with donepezil , and hazard ratios for treatment discon-
tinuation derived from a Cox regression model of the done-
pezil clinical trial data . The hazard ratios were adjusted
for the average rates for disease severity (baseline and
End of Model
Fig. 1. Simplifiedrepresentation of the Alzheimer’s disease simulation flow outlining creation ofindividual patients, disease progression, accumulation of out-
living; IADL 5 instrumental activities of daily living).
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e3024
current) and changes in disease severity, both measured by
Age and gender-specific survival data from the Medical
Research Council’s Cognitive Function and Ageing Study
 were used to estimate survival patterns for the popula-
2.1.4. Direct costs
A daily treatment cost of £3.18 for donepezil 10 mg was
assigned using the British National Formulary. Patients on
active therapy were also assumed to incur costs associated
with biannual visits to their physician, at a cost of £50 per
Direct care costs of the patients, including medical costs
alized patients were taken from the Dementia UK report 
and inflated to 2007 British pounds. For patients living in
the community, the Dementia UK report provided cost esti-
polated to fit the following MMSE severity ranges: mild
(MMSE 25e26), mild-moderate
(15e19), moderate-severe (10e14), and severe (0e10)
(Table 1). For institutionalized patients, a conservative ap-
Because the Dementia UK report only provided an over-
all rate of institutionalization (36.5%) , it was necessary
to combine this information with other data to get severity-
dementia with MMSE scores of 17 or lower, and 21.6% had
dementia with MMSE scores between 18 and 23. This was
used in combination with the severity distribution of AD
in the United Kingdom to derive the proportion of patients
institutionalized in each severity category (Table 1).
2.1.5. Indirect costs
Theonly indirect costsmodeled were the costs associated
with caregiver time by patients’ primary caregivers. The re-
lationship of caregiver time to disease severity parameters
was developed from the two donepezil clinical trials where
these data were available [15,16] using a linear repeated
measure, fixed effects model .
Care minutes per day576:4111:8 Agecc193:02 Malecc
185:56 Malepatient26:47 MMSE
10:58 NPI12:66 ADL12:61 IADL120:55 PsyMed
AgeCGrepresents the caregiver’s age, MaleCGis a dummy
variable for the caregiver’sgender, and Malepatientfor the pa-
tient’s gender. PsyMed is a dummy variable for whether the
patient was on psychiatric medications at baseline.
Caregiver time was valued at the minimum wage preva-
lent in the United Kingdom, which was £5.30 per hour .
2.1.6. Health utilities
Health utilities for patients were estimated on the basis of
a published regression equation  relating patient health
utilities to MMSE and NPI scores, as well as whether they
were institutionalized or living with their caregiver:
Utility50:40810:010 MMSE20:004 NPI
20:159 Institutionalized10:051 Caregiver
As the source utility study used the brief, rather than the full
NPI, the published coefficient for the NPI term was modified
to correspond to the full NPI scale, which is used in the sim-
donepezil trials where caregivers completed the SF-36
[14e16]. The scores were transformed to health utilities,
which were related through a regression model to age and
gender of both patient and caregiver, patient use of
behavioral symptoms .
Cost inputs for the early assessment model
Direct cost per patient living in the
community per monthy
Mild (MMSE: 25e26)
Direct cost per institutionalized patient per
Patients living in institutions
Early assessment cost components
First visit to specialist
Second visit to specialist
Full blood count
Erythrocytes sedimentation rate/C-
Average cost of MRI/CT scan in 5% of
Abbreviation: MMSE, Mini-Mental State Examination.
* Cost values are in 2007 £.
yInpatient care, outpatient care, medication, day hospitals, day centers,
community health services, social care, and respite care.
zIncludes above, as well as cost of accommodation.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e3025
2.1.7. Early assessment parameters
The early assessment scenario evaluated was for annual
assessment of patients reporting memory complaints and
aged 65 to 100 years, with treatment restricted to patients
with MMSE scores between 10 and 26. Early assessment
was assumed to consist of the initial visit to a General Prac-
titioner, two specialist visits, laboratory tests, and, in 5% of
cases, an MRI or CT scan. In the base case, early assessment
was assumed to be 100% accurate, and patients without AD
were assumed to only incur the cost of the initial General
Practitioner visit, and one specialist visit. An assumed delay
between the start of the assessment process and initiation of
treatment with donepezil for patients with AD was set to 10
weeks in the base case.
As mentioned, patients with undiagnosed AD enter the
model (i.e., report to their physician with memory com-
plaints) sometime between onset of the disease and the
time they would have been diagnosed in the absence of early
assessment. Time to diagnosis without early assessment was
set to 36 months based on the mean delay from onset of
symptoms to diagnosis in the United Kingdom as reported
in the Facing Dementia Survey , and assuming a 1-
month period from onset of disease to onset of symptoms.
Characteristics of the simulated patients are created by
sampling from an individual patient data set, with baseline
information from donepezil clinical trials. Each individual
is created by matched sampling from the dataset based on
weights were created using the age and gender distribution
UK report , to ensure that the characteristics of the simu-
lated patients reflect those ofUK patients. The data elements
include patient age, gender, use of psychiatric medications,
MMSE, NPI, ADL scores, and IADL scores, as well as
age and gender of the caregiver.
220.127.116.11. Early assessment costs
For establishing the costs of early assessment, it was nec-
essary to determine who is assessed. The analysis assumes
that only patients with subjective memory complaints un-
dergo early assessment. For this, data from a recent meta-
analysis were used, which indicated that 42.8% of patients
with dementia and 17.4% without dementia reported mem-
ory complaints . This information was used in combina-
tion with data on the prevalence of dementia and AD in the
United Kingdom , and an estimate from a National Audit
Office report on dementia  that 58.5% of AD patients in
the United Kingdom are undiagnosed.
The cost of early assessment and its various components
are outlined in Table 1 [25,31,32].
Patient and caregiver outcomes are reported as QALYs,
and outcomes among the different scenarios are compared,
where appropriate, as incremental cost-effectiveness ratios
As Table 2 indicates, the benefits of early treatment are
substantial, both in terms of health-related and economic
outcomes. Although assessment of AD at an early stage re-
sults in up-front costs averaging almost £4100 perpatient di-
agnosed with AD and treatment-related costs of £2400 per
patient, these are more than offset by savings in patient
care. As compared with no treatment, overall direct costs
are reduced by more than £3600 per patient, with another
£4150 in indirect cost savings attributable to reductions in
Base case results (per patient) for early assessment and treatment of AD with donepezil
Early assessment 1
AnalysisEarly assessment 1
Early assessment costs
Direct care costs (while in community)
Direct care costs (while in institutional care)
Total direct costs
Years in community
QALYS (patient 1 caregiver)
Health care direct cost/QALY (patient)
Societal total cost/QALY (patient
Abbreviation: QALY, quality adjusted life year.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e3026
time that the caregivers spend caring for patients. Treatment
in the absence of early assessment, although beneficial and
leading to savings compared with no treatment, is predicted
to lead to a much smaller benefit than when patients are as-
from £2100 to £5725 per patient in comparison with delayed
treatment, depending on whether indirect costs are included.
These savings are largely derived from reducing the amount
of time patients spend with severe disease and thus increas-
ing the amount of time they are in the community and not in
institutions. Although reduced institutional care is the larg-
est contributortosavings, reduced disease progression while
in the community also leads to savings. As Table 2 shows,
even though screened and treated patients remain longer in
the community, the overall cost of care for these patients
while in the community is lower. As compared with no treat-
ment, early assessment reduced the time patients spend with
MMSE scores below 10 by over 5 months. Delayed treat-
ment in the absence of early assessment reduces time spent
with MMSE ,10 by less than half that amount (about 2.3
3.1. Sensitivity analyses
To evaluate the sensitivity of predicted outcomes to
changes in input parameters, both deterministic one-way
sensitivity analyses, where parameters are assessed based
on using discrete alternative values in the simulation while
holding other parameters constant, and probabilistic sensi-
tivity analyses, where all input parameters are simulta-
neously varied by sampling from probability distributions
for each parameter , were conducted. Inputs on early as-
sessment costs, costs of care, indirect caregiver costs, health
utilities, the prevalence of undiagnosed AD, rates of self-
reported memory complaints, early assessment sensitivity,
treatment effectiveness, treatment persistence, treatment du-
ration, time horizon, and the delay to diagnosis in the ab-
sence of early assessment were assessed.
Table3summarizes theone-way sensitivityanalysisfind-
ings for early assessment versus no treatment. In most sce-
narios, early assessment with treatment led to savings as
compared with no treatment and delayed treatment. The
two exceptions were when the cost of caring for patients
was reduced by 50%, and when rates of self-reported mem-
ory complaints were set as equal for patients with and with-
out dementia. In each of these cases, however, early
assessment was still associated with an attractive cost-
effectiveness ratio, with incremental costs per QALY from
the health care payer perspectivewell below £30,000. These
same scenarios were associated with savings from the soci-
The probabilistic sensitivity analyses vary all the critical
input parameters simultaneously and give a percent confi-
dence that the cost-effectiveness is below a given threshold
(Figs. 2, 3; 500 sets of parameters were drawn randomly
and 1000 patients modeled for each set). In both
approximately 40% probability of overall savings from the
health care payer perspective, and a 70% probability from
the societal perspective. More importantly, incremental
cost-effectiveness ratios were below £30,000 per QALY,
the cost-effectiveness threshold adopted by NICE in the
United Kingdom, in 70% to 90% of replications, depending
on the comparison being made and the perspective being
adopted. There was, nevertheless, considerable variability
in outcomes revealed by these analyses. For example, in
was associated with
One-way sensitivity analysis results for early assessment versus no treatment
% of undiagnosed dementia reduced by 50%
Caregiver cost reduced by 25%
Caregiver cost reduced by 50%
Diagnosis without early assessment reduced
to 9 months
Diagnosis without early assessment reduced
to 18 months
Patient cost reduced by 25%
Patient cost reduced by 50%
Early assessment costs increased by 25%
Early assessment costs increased by 50%
Sensitivity of early assessment reduced to
Prevalence of subjective memory complaints
equal for demented and non-demented
Prevalence of subjective memory complaints
much higher for demented
Abbreviation: QALYs, quality adjusted life years.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e30 27
the comparison to no early assessment and no treatment, dis-
counted QALYs gained with early assessment and treatment
ranged from 0.03 per patient to 0.52 per patient, with an av-
erage across all 500 sets of replications of 0.17. Similarly, in
28% of replications, early assessment with treatment was
predicted to lead to incremental costs rather than savings.
In the United Kingdom, the use of cholinesterase inhibi-
tors has been restricted to patients in whom the disease has
reached moderate severity, effectively eliminating any phar-
macologic treatment options for patients in whom the dis-
ease is still in its mild stages. This decision was based
largely on the results of an economic analysis suggesting
that treatment was not cost-effective for patients with mild
disease . It disagrees with European clinical guidelines
which recommend treatment with cholinesterase inhibitors
on diagnosis of AD, irrespective of whether the disease is
in its mild or moderate stage . The decision and the eco-
nomic analysis on which it was based have been much de-
bated. The economic model on which the analyses was
based has many limitations: only considering two health
ing differences in patients’ baseline characteristics, progres-
sion of the disease or treatment effects; restricting estimates
of treatment effect to 6 month trials and ignoring random-
ized placebo-controlled trials of longer duration; and assum-
ing perfect treatment persistence, with no clinical stopping
rules in place, contrary to previous and current guidelines
for use of cholinesterase inhibitors issued by NICE itself
. In previous work, we developed a discrete event simu-
lation to address many of these shortcomings, incorporating
the most recent available data and individual simulation
techniques, and found that treatment with donepezil was
highly cost-effective, and indeed more economically effi-
cient when treatment was initiated over the mild stages of
the disease . The analyses described here extend the use
of this simulation to evaluate the effect and cost-
effectiveness of early assessment followed by treatment
with cholinesterase inhibitors, provide further support for
the argument that early treatment is beneficial to patients
and their caregivers, and have important economic conse-
quences. In base case estimates, evenwhen the costs of early
assessment are factored in, early treatment reduces costs,
leading to significant savings both to the health care system
and society. Although probabilistic sensitivity analyses indi-
cate a fair degree of uncertainty around the issue of cost sav-
ings, they nevertheless show that even if early assessment
were to lead to incremental costs, the health benefits associ-
ated with earlier treatment are large enough that the program
would very likely be cost-effective.
As with any model, this study has limitations. Effective-
ness data for donepezil were limited to trials of up to
ment would result in maintenance of benefits attained over
the first 12 months. If treatment beyond 1 year was associ-
ated with rapidly diminishing effectiveness, or if the benefits
observed in clinical trials could not be reproduced in actual
practice, economic results could be worse than predicted by
the simulation. As sensitivity analyses indicate, the eco-
nomic attractiveness of early assessment could also be com-
promised if many individuals without dementia report
memory complaints and undergo early assessment. Of fur-
ther note, although unlike previous modeling work in this
area, the current simulation considers progression of AD
ture all aspects of patient characteristics that could poten-
tially influence outcomes. For example, comorbidities,
are not considered. Furthermore, although the simulation
does track changes in the overall severity of behavioral
symptoms through the NPI, it does not examine changes in
specific behavioral or psychiatric symptoms such as depres-
sion, hallucinations, or aggression.
The simulation findings, however, indicate that early as-
sessment of AD would be cost-effective, over a wide range
£0 £20,000£40,000 £60,000£80,000 £100,000
o l e
o i t a
c i l
Societal Perspective Healthcare Perspective
Fig. 3. Cost-effectiveness acceptability curve for early assessment plus
treatment with donepezil versus treatment in the absence of early assess-
£0 £20,000£40,000 £60,000£80,000 £100,000
o l e
o i t a
c i l
Societal PerspectiveHealthcare Perspective
Fig. 2. Cost-effectiveness acceptability curve for early assessment plus
treatment with donepezil versus no treatment.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e3028
of inputs, suggesting that it should be carefully considered.
Our findings are consistent with a recently published eco-
nomic forecast model of early dementia diagnosis and treat-
ment interventions in the United Kingdom . The analyses
from that model suggested that a 10% reduction in nursing
ment would lead to overall savings. Our analyses show re-
duced time spent in nursing homes of 9.5% compared with
no treatment, and 7.0% compared with treatment in the ab-
sence of early assessment.
Whether these results would translate to other countries
clearly depends on several factors including the prevalence
of undiagnosed AD and average delays to diagnosis, the
costs of implementing an early assessment program, and
more, the relative cost-effectiveness of new emerging treat-
ments could be different depending on the costs of those
treatments and the particular losses in long-term outcomes
associated with their delayed use.
In final conclusion, our analyses suggest that there are
substantial benefits in terms of both patient and economic
outcomes, for a program of early assessment and early treat-
ment of AD in the United Kingdom.
This work was supported by an unrestricted grant from
Eisai Ltd. The authors would like to thank Ye Tan for biosta-
tistical support and Sunning Tao for running the economic
analyses and assisting in the production of the manuscript.
Messrs. Getsios, Blume, Ishak and Herna ´ndez are em-
ployees of United BioSource Corporation, a consulting re-
biotechnology, and medical device manufacturing, some of
whom have competing therapies for Alzheimer’s disease.
Dr. Maclaine is a paid employee of the study sponsor.
 Knapp M, Prince M, Albanese E, Banerjee S, Dhanasiri S,
Fernandez J-L, et al. For Kings’ College London and London School
of Economics. London: Alzheimer’s Society, Gordon House. Avail-
 Weimer DL, Sager MA. Early identification and treatment of Alz-
heimer’s disease: social and fiscal outcomes. Alzheimers Dement
 National Audit Office. Improving services and support for peoplewith
dementia. London: The Stationery Office. Available at: http://www.
nao.ork.uk; July 2007.
 Birks J. Cholinesterase inhibitors for Alzheimer’s disease. Cochrane
Database Syst Rev 2006:CD005593.
 Hansen RA, Gartlehner G, Webb AP, Morgan LC, Moore CG,
Jonas DE. Efficacy and safety of donepezil, galantamine, and rivastig-
mineforthetreatmentofAlzheimer’sdisease:a systematic reviewand
meta-analysis. Clin Interv Aging 2008;3:211e25.
 Green C, Modelling disease progression in Alzheimer’s disease. A re-
view of modelling methods used for cost-effectiveness analysis. Phar-
 Cohen J, Neumann P. Decision analytic models for Alzheimer’s disease:
 Banerjee S, Wittenberg R. Clinical and cost effectiveness of services
for early diagnosis and intervention in dementia. Int J Geriatr Psych
 Getsios D, Blume S, Ishak KJ, Maclaine G. The cost-effectiveness of
updated UKevaluation using discrete event simulation. Pharmacoeco-
zil, galantamine, rivastigmine (review) and memantine for the treat-
ment of Alzheimer’s disease. London: MidCity Mace. Available at:
http://www.nice.org.uk; November 2006.
 Dasbach EJ, Insinga RP, Elbasha EH. Cost-utility analysis: a case
study of a quadrivalent humn papillomavirus vaccine. In: Arnold Re-
ne ´e JG, ed. Pharmacoeconomics: From Theory to Practice. Boca Ra-
ton, FL: CRC Press; 2009. p. 119e32.
 Getsios D, Migliaccio-Walle K, Caro JJ. NICE cost-effectiveness ap-
the best tools used? Pharmacoeconomics 2007;25:997e1006.
 CERADdan overview. Consortium to establish a registry for Alz-
heimer’s disease. Available at: http://cerad.mc.duke.edu/Default.htm.
Retrieved March 6, 2009.
 Mohs RC, Doody RS, Morris JC, Ieni JR, Rogers SL, Perdomo CA,
et al. A 1-year, placebo-controlled preservation of function survival
study of donepezil in AD patients. Neurology 2001;57:481e8.
 Winblad B, Engedal K, Soininen H, Verhey F, Waldemar G, Wimo A,
et al. A 1-year, randomized, placebo-controlled study of donepezil in
patients with mild to moderate AD. Neurology 2001;57:489e95.
week, randomized, double-blind study of donepezil in moderate to se-
vere Alzheimer’s disease. Neurology 2001;57:613e20.
 Rogers SL, Farlow MR, Doody RS, Mohs R, Friedhoff LT, Donepezil
pezil in patients with Alzheimer’s disease. Neurology 1998;50:136e45.
 Rogers SL, Doody RS, Mohs RC, Friedhoff LT, Donepezil Study
Group. Donepezil improves cognition and global function in Alz-
heimer disease: a 15-week, doubleblind, placebo-controlled study.
Arch Intern Med 1998;158:1021e31.
 Black SE, Doody R, Li H, McRae T, Jambor KM, Xu Y, et al. Done-
pezil preserves cognition and global function in patients with severe
Alzheimer disease. Neurology 2007;69:459e69.
 Winblad B, Kilander L, Eriksson S, Minthon L, Batsman S,
Wetterholm AL, et al. Donepezil in patients with severe Alzheimer’s
disease: double-blind, parallel-group, placebo-controlled study. Lan-
 DoodyRS,GeldmacherDS, GordonB, PerdomoCA, Pratt RD. Open-
label,multicenter,phase3extensionstudyofthe safetyandefficacy of
donepezil in patients with Alzheimer disease. Arch Neurol 2001;
 Winblad B, Wimo A, Engedal K, Soininen H, Verhey F, Waldemar G,
et al. 3-year study of donepezil therapy in Alzheimer’s disease: effects
of early and continuous therapy. Dement Geriatr Cogn Disord 2006;
 Lyle S, Grizzell M, Willmott S, Benbow S, Clark M, Jolley D. Treat-
ment of a whole population sample of Alzheimer’s disease with done-
pezil over a 4-year period: lessons learned. Dement Geriatr Cogn
 Xie J, Brayne C, Medical Research Council Cognitive Function and
Ageing Study Collaborators. Survival time in people with dementia:
analysis from population based cohort study with 14 year follow-up.
consultation lasting 17.2 min. Unit Costs of Health and Social Care.
Kent, United Kingdom: PSSRU; 2007.
 Macdonald A, Cooper B. Long-term care dementia services: an im-
pending crisis. Age Ageing 2007;36:16e22.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e30 29
 Department for business enterprise & regulatory reform. National
minimum wage. Available at: http://www.berr.gov.uk/whatwedo/
 Jonsson L, Andreasen N, Kilander L, Soininen H, Waldemar G,
Nygaard H, et al. Patient- and proxy-reported utility in Alzheimer dis-
ease using the EuroQoL. Alzheimer Dis Assoc Disord 2006;
 Bond J, Stave C, Sganga A, O’Connell B, Stanley RL. Inequalities in
dementia care across Europe: key findings of the facing dementia sur-
vey. Int J Clin Pract 2005;59(Suppl 146):8e14.
 Mitchell AJ. The clinical significance of subjective memory com-
plaints in the diagnosis of mild cognitive impairment and dementia:
a meta-analysis. Int J Geriatr Psychiatry 2008;23:1191e202.
 Department of Health. NHS reference costs 2006-07. Available at:
 Department of Health. NHS reference costs 2005-06. Available at:
 PostmaMJ. Sensitivity analysis. In: ArnoldRene ´e JG, ed. Pharmacoe-
conomics: From Theory to Practice. Boca Raton, FL: CRC Press;
2009. p. 163e74.
 Waldemar G, Dubois B, Emre M, Georges J, McKeith IG, Rossor M,
et al. Recommendations for the diagnosis and management of Alz-
heimer’s disease and other disorders associated with dementia:
EFNS guideline. Eur J Neurol 2007;14:e1e26.
D. Getsios et al. / Alzheimer’ s & Dementia 8 (2012) 22e30 30