Chapter 7

The Addenbrooke’s Cognitive

Examination: Revised and
Supplementary Test Suggestions

Introduction
is chapter describes the use of the revised version of the Addenbrooke’s
Cognitive Examination: ACE-III. e original test was developed in the
Cambridge memory clinic in the 1990s and was shown to be sensitive to early
Alzheimer’s disease (AD) and to di erentiate AD from frontotemporal demen-
tia (FTD). In addition it was shown to be useful in the separation of organic
brain disease from psychiatric states, and in the detection of cognitive dysfunc-
tion associated with the parkinsonian syndromes of progressive supranuclear
palsy (PSP), corticobasal degeneration (CBD), and multiple system atrophy
(MSA) (for references to the ACE see ‘Selected Further Reading’ at the end of
this book). e main weaknesses were the imbalance across domains (espe-
cially the limited range of visuospatial/perceptual tasks), the insensitivity of the
naming component, and the di culty in translation of certain components.
Another goal in developing the ACE-R was to have a version with clearer cogni-
tive subtest scores. e ACE-R went through multiple prototypes before arriv-
ing at the nal version.
e ACE and the ACE-R incorporated the Mini-Mental State Examination
(MMSE) which was included at the time that the MMSE was freely available.
It subsequently came under copyright and was published by Psychological
Assessment Resources, Inc. (PAR). is necessitated changes to the ACE-R to
remove MMSE-speci c items and resulting in the ACE-III was shown to be
highly equivalent to the ACE-R with a correlation of 0.99 between the two ver-
sions of the ACE.
is chapter describes the ACE-III (Fig. 7.1) together with scoring criteria
and normative data followed by suggestions for ‘add-on’ bedside tasks that test
areas not well covered by the ACE-III.
ere is also now an iPad version of the ACE-III which is freely available
via https://www.neura.edu.au/research-clinic/frontier/research/downloads/ as
well as the mini- ACE which is a reduced 30-item version of the ACE-III.
164 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 ACE-III English version and scoring instructions.

Copyright © John Hodges.
 INTRODUCTION 165

Fig. 7.1 (continued)

166 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

 INTRODUCTION 167

Fig. 7.1 (continued)

168 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

 INTRODUCTION 169

Fig. 7.1 (continued)

Fig. 7.1 (continued)
Fig. 7.1 (continued)
172 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

 INTRODUCTION 173

Fig. 7.1 (continued)

Fig. 7.1 (continued)
 INTRODUCTION 175

Fig. 7.1 (continued)

176 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

 INTRODUCTION 177

Fig. 7.1 (continued)

178 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

Fig. 7.1 (continued)
180 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Fig. 7.1 (continued)

Normative Data
For the original version of the ACE we derived normative data from 127 controls
and recommended the use of two cut-o s (<88 and <82) with high sensitivity
and high speci city, respectively. We expected that the revised version might
perform di erently but we were very pleased to nd almost identical results
 NORMATIVE DATA 181

when it was administered to 63 normal volunteer subjects along with 178 clinic
attenders (142 with dementia). Again, a cut-o of 88 produced high sensitivity
(0.94) but lower speci city (0.89) whereas a cut-o of 82 resulted in lower sen-
sitivity (0.84) but great speci city (1.00). In other words, patients with a score
of <88 are at high risk of having an organic brain disease, but at this high cut-o
you will generate false positives. Conversely, at the lower cut-o (<82) you will
almost certainly detect all dementia patients but you will miss cases with early
AD (false negatives). Table 7.1 shows the full sensitivity, speci city, and positive
predictive values (PPV) at di erent base rates of dementia and Table 7.2 shows
cut-o for the various subscores. In order to tackle the ‘grey zone’ between 88
and 82, we devised likelihood ratios for probability of dementia. Table 7.3 illus-
trates that as the ACE-R cut-o falls from 88 to 82, the likelihood ratio rate rises
progressively from 8.4 to 100, which means that a score of 82 is 100 times more
likely to come from a patient with dementia than one without.
One of the aims of the old ACE was to di erentiate AD from FTD. A er vari-
ous attempts we produced a ratio score (VLOM ratio) that re ects the relative
balance of cognitive dysfunction in these two disorders:
VLOM ratio = [verbal fluency and language] / [orientation and memory].

Table 7.1 Sensitivity and specificity of different ACE-R (and the MMSE) cut-off
scores for diagnosing dementia, with corresponding positive predictive values (PPV)
at different rates of dementia prevalence

PPV at different prevalence rates

ACE-R Sensitivity Specificity 5% 10% 20% 40%
cut-off
88 0.94 0.89 0.31 (1.0) 0.48 0.68 0.85 (1.0)

82 0.84 1.00 1.0 (0.96) 1.0 1.0 1.0 (0.90)

Copyright © John Hodges.

Table 7.2 Lower limit of normal (cut-off scores) for total ACE-R and subscores
according to age, showing control mean minus two standard deviations

Age Education Total Attention/ Memory Fluency Language Visuospatial

range (years) ACE-R orientation
(years) score
50–59 12.7 86 17 18 9 24 15

60–69 12.9 85 17 19 8 21 14
70–75 12.1 84 16 17 9 22 14

Copyright © John Hodges.

182 THE ADDENBROOKE’S COGNITIVE EXAMINATION

Table 7.3 Likelihood ratios for probability

of dementia at various ACE-R cut- off scores

ACE-R score Likelihood ratio of dementia

88 8.4

87 11.5
86 14.2
85 18.9
84 27.6
83 52.5
82 100

Copyright © John Hodges.

Using both the old and the revised ACE, a VLOM ratio of >3.2 was found to be
optimal in di erentiating AD from FTD (74% sensitivity and 85% speci city),
whereas a ratio of <2.2 was highly suggestive of FTD (sensitivity 58% and spe-
ci city 95%). Scores between 2.2 and 3.2 were poorly predictive of diagnosis.
A number of other groups have reported similar ndings and that VLOM is a
generally useful adjunct to diagnosis.
In a small validation study of the ACE-III involving a group of 61 patients
with mixed syndromes dementia and 25 controls scores on the ACE-R and
ACE-III were extremely close in all groups suggesting that the cut-o s used for
the ACE-R are applicable to the ACE-III.

Additional Material for Particular Cases

Although we consider the ACE-III to be a good general screening instrument,
especially in the context of a memory or cognitive disorders clinic, it is o en neces-
sary to supplement the instrument with other tasks targeted towards suspected
cognitive dysfunction. A full description of cognitive evaluation is presented in
Chapter 5. e tasks described in the following sections are ‘selected highlights’
which have been chosen because either the ACE-III is de cient in this domain
(e.g. frontal executive function, praxis, face recognition, etc.) or further tests are
frequently required to clarify the nature of the de cit (e.g. aphasia or agnosia).

Remote memory
To assess remote memory we typically ask patients about recent news events,
which should be tailored towards the patient’s interests and cultural setting. In
the UK, the following are helpful in guiding assessment: