U.S. flag

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

  • Publications
  • Account settings
  • Advanced Search
  • Journal List
  • Neuropsychiatr Dis Treat

Addenbrooke’s cognitive examination III in the diagnosis of dementia: a critical review

Diana bruno.

1 Instituto de Investigación en Psicología Básica y Aplicada (IIPBA), Facultad de Filosofía y Humanidades, Universidad Católica de Cuyo, J5400 Rivadavia, San Juan, Argentina, [email protected]

2 Neuropsicología y Rehabilitación Cognitiva, Instituto de Neurociencias Cognitivas y Traslacional (INCyT), Fundación INECO, Universidad Favaloro, CONICET, Buenos Aires, Argentina, [email protected]

Sofia Schurmann Vignaga

Addenbrooke’s cognitive examination III is a screening test that is composed of tests of attention, orientation, memory, language, visual perceptual and visuospatial skills. It is useful in the detection of cognitive impairment, especially in the detection of Alzheimer’s disease and fronto-temporal dementia. The aim of this study is to do a critical review of the Addenbrooke’s cognitive examination III. The different language versions available and research about the different variables that have relationship with the performance of the subject in the ACE-III are listed. The ACE-III is a detection technique that can differentiate patients with and without cognitive impairment, is sensitive to the early stages of dementia, and is available in different languages. However, further research is needed to obtain optimal cutoffs for the different versions and to evaluate the impact of different age, gender, IQ, and education variables on the performance of the test.


The Addenbrooke’s cognitive examination (ACE) was developed by Hodges et al as an extended cognitive screening technique, designed to detect dementia and differentiate Alzheimer dementia from fronto-temporal dementia. 1 It was also developed to overcome the neuropsychological omissions present in the Mini Mental State Examination (MMSE). 2 The aim of ACE was to be a screening technique that evaluates the principal cognitive functions and grants free access to health professionals. 3 In this way, ACE turns into a brief cognitive screening tool, which takes 15–20 minutes to administer and is useful in the detection of dementia syndromes. 4

The ACE is composed of tests of attention, orientation, memory, language, visual perception and visuospatial skills. 3 All of these measures have significant correlations with the classical neuropsychological tests. 4

The aim of this study is to critically review the Addenbrooke’s cognitive examination III.

Description of ACE-III

The ACE-III was developed to remove the MMSE elements from the ACE and ACE-R, as the MMSE was no longer open access in the year 2001. 5 Because of this, recent guidelines have provided alternatives to the MMSE, and the ACE-III has been recommended by the Department of Health and the Alzheimer’s Society in the UK. 6 In this way, the MMSE items present in the ACE-R were substituted for by similar items. 1 For example, in the attention section the spelling of the word “WORLD” backwards was omitted, leaving only the subtraction of serial 7s. In the language section, the written command “close your eyes” was omitted, the denomination of a pencil and clock was replaced by a book and a spoon, and the three-step command was replaced by three single-step commands, due to the lack of sensitivity to cognitive impairment. 1 , 4 Finally, in the same section, the writing of a single sentence was replaced by writing two or more sentences. In the visuospatial section, the intersecting pentagons were replaced with intersecting lemnisci. 3 Hence, with these changes the administration of the ACE-III makes scoring the MMSE void. 4 As the ACE-III is designed to address the weakness of the ACE-R, the verbal repetition item was modified due to the poor performance of this item in healthy adults. 3

As previously described, the ACE-III is composed of five cognitive domains, attention, memory, language, verbal fluency, and visuospatial abilities. The ACE-III takes ~20 minutes to complete ( Table 1 ). Similarly to the ACE-R, the total score of the ACE-III is based on a maximum score of 100, with higher scores indicating better cognitive functioning.

Cognitive domain, tasks, and sub-total score of ACE-III

Abbreviation: ACE-III, Addenbrooke’s cognitive examination III.

The index study of the ACE-III demonstrated high sensitivity and specificity, with cutoffs recommended as for the ACE-R as follows: 1) 88 (sensitivity =1.0; specificity =0.96) and 2) 82 (sensitivity =0.93; specificity =1.0). 4

Correlation of ACE-III with neuropsychological tests

It has been demonstrated that the subtests of the ACE-III have significant correlations with neuropsychological tests in that domain. The memory domain of the ACE correlated with two classical neuropsychological tests of memory, Free and Cued Selective Reminding Test and the Rey Auditory Verbal Learning Test. 4 , 7 The language domain correlated with the Boston Naming Test, the attention domain correlated with tests that evaluate attention and executive functions (the trail making test, memory span, Stroop test), and the fluency scores correlated with executive functions. 7 Therefore, the administration of this screening technique quickly provides the clinician with a neuropsychological profile.

The cutoff points of ACE-III show strong correlations with the cutoff points of the ACE-R, 4 suggesting that this screening technique is capable of differentiating patients with and without cognitive impairment, and mild cognitive impairment (MCI). 3 In addition, the ACE-III performance has broader clinical implications in that it relates to carer reports of functional impairment in most common dementias. 8

Comparison of the ACE-III with other screening techniques

In different studies that compare the ability to discriminate healthy people and people with dementia, the ACE-III showed similar results to other screening techniques 9 (MoCA 10 and RUDAS 11 ).

Like the other screening techniques (MMSE, MOCA, RUDAS), the ACE-III provides the clinician with a quick and brief global cognitive screen of the patient specifying both the overall cognitive profile and measures of each of the evaluated domains. 9 , 10 In this way, ACE-III provides the clinician with a more comprehensive assessment view of the cognitive profile of the patient, helping to provide a differential diagnosis. 12 Moreover, as the ACE-III includes different scores for each domain, in addition of the general score, it allows for tracking the progression of cognitive deficits over time. 13

The MMSE lacks sensitivity to identify fronto-temporal dementias, whereas the ACE-III had demonstrated accuracy for detecting fronto-temporal dementia. 4 An important limitation of the MMSE is the lack of sensitivity for the early stages of dementia, 14 whereas the ACE-III had demonstrated accuracy in detecting MCI. The ACE-III showed better sensitivity for detecting dementia compared to the MMSE. 15 The ACE-III more efficiently identifies everyday functional impairments compared with both the MMSE and MoCA. 16

Despite the above considerations, the MMSE continues to be the preferred screening instrument for many neurologists. For this reason, a conversion table between ACE-III and MMSE has been developed and is used for clinical and research purposes. 17

In addition, a study by Larner investigated the relationship between administration time and diagnostic accuracy in cognitive screening tests. The author reports positive correlations between the accuracy and time of administration of the test and significant correlations between the accuracy and the number of items included in the test. These observations suggest that tests with more items (ie, longer tests) are more accurate. 18 The number of items of the MMSE is 30 compared with the ACE-III, which have 100 items.

Utility of the ACE-III in the detection of cognitive impairment

Dementia has been declared a global challenge, causes a great burden for the families of the patients, and leads to enormous global annual costs, which are expected to increase significantly in the next few decades. 20 – 22 Although several risk factors are implicated, the principal risk factor is age, and wit aging and growing populations dementia is becoming more prevalent. 22 Therefore, it is essential that a sensitive and specific screening tool that not only identifies patients with dementia but also identifies them in the early stages of the disease will be widely used to allow earlier diagnosis and intervention and to postpone dementia. 23

MCI is the prodromal phase associated with brain disorders, including of Alzheimer’s disease, 24 Parkinson’s disease, 25 cerebrovascular disease, 26 and fronto-temporal dementia. 27

The ACE-III has shown high diagnostic accuracy for MCI, being the memory domain the most sensitive in early stages of Alzheimer’s disease patients. 7 Moreover, the ACE-III has demonstrated high diagnostic accuracy in individuals with subjective cognitive impairment. 28

The ACE-III, like its predecessors, was designed for the detection of dementias in early stages. 3 Good levels of sensitivity have been reported in the distinction between healthy controls and patients with some type of dementia in initial stages. 4 , 7

Research reports that ACE-III is one of the most sensitive screening tools for the detection of dementia, compared to other screening tests such as MMSE and MOCA. 3 It has been reported that a cutoff point of 61 on the ACE-III is sensitive for distinguishing mild dementia from moderate dementia. 16

Considering that the ACE-III has properties similar to that of its predecessors, it can be considered to be a useful instrument for longitudinal follow-ups as its predecessors. 29 , 30

In addition, the value of ACE-III for discriminating between Alzheimer’s dementia and fronto-temporal dementia has been reported. 4 , 7 , 28 , 31 Patients with Alzheimer dementia and fronto-temporal dementia showed significant differences in the performance on the different components of the ACE: orientation, attention, and memory were worse in Alzheimer patients, while the fluency with letters, language, and names were worse in patients with fronto-temporal dementia. Mathuranath, using the ACE and the ACE-R, 1 , 3 translated this scoring pattern into an index that is considered useful for the differentiation of both types of dementia (the VLOM ratio). Many different researchers have shown the usefulness of the new version of the ACE. 4 , 7 , 28 , 31

On the other hand, the usefulness of the annualized change rates (ARC) in the total ACE scores was reported. This can be calculated using the total score in the previous and current ACE and the number of months between both evaluations, according to this formula: ARC of ACE = [(last ACE score-baseline ACE score)/(months between evaluation)] × 12. 29 , 30

Stroke can involve physical and cognitive impairments. To the best of our knowledge, there is only one study that studied the utility of the ACE-III in the detection of cognitive impairment after stroke. 19 As an advantage, the ACE-III not only provides the clinician with a cutoff point but also shows an estimated cognitive profile of the patient. 7 In this way, it provides the clinician with useful information about the cognitive functions of the patient. Moreover, the application of a screening tool can accelerate the diagnostic process of cognitive deficit after stroke and implementing cognitive rehabilitation. 32

It is fundamental when interpreting the cutoff points after stroke to understand that because many of the subtests of the ACE-III cannot be evaluated. In this way, the vast majority of patients after stroke score below the cutoff point. 19 This is due to the fact that many patients after the stroke typically present with motor difficulties, which negatively impacts the motor output subtests (example: drawing) and often have difficulties in the with language, many times because they present with aphasia. 33

Currently, there are no studies that have studied the accuracy of ACE-III in the seeking of cognitive impairment in Parkinson disease (PD). Nevertheless, the coping of the wire cube, present in the visuospatial domain of ACE-III, has correlated significantly with a poor performance on other cognitive domains, suggesting that is a sensitive detector of cognitive impairment in PD. 34

Variables to consider in the interpretation of the cutoff points

Previous studies with the ACE-R have shown that the cutoff points are influenced by sociodemographic variables. 35 , 36 In several studies, with the ACE-III in several studies, the influence of demographic variables has been considered as seen to be an important variable to take into account when interpreting the suggested cutoff points and to improve diagnostic accuracy. 9 , 38 , 40 , 44 , 46

Years of education

The years of education are an important variable that must be taken into account in order to correctly interpret the cutoff points of the ACE III. Level of education has been observed to have an effect on the accuracy of this screening test in the diagnosis of dementia 15 , 37 – 40 and may be attributable to the presence of items dependent on the level of education or literacy, 40 such as the use of irregular words, phonemic verbal fluency, 41 naming task, 42 and constructional abilities. 43 Previous investigations have shown that the level of education has a significant impact on both the total score and the scores of the domains. 44 , 45 Thus, different cut points have been proposed depending on years of education 44 and correction factors have been proposed to adjust the raw scores and equivalent scores with cutoff values. 46

It has been found that people over 75 years old score less on the ACE-R in comparison with younger people. 47 , 48 Interpretation of the cognitive profile is thus limited by age, suggesting that age is an independent predictor of performance. 9 , 15 , 40 It has been shown that all sub-scores of the ACE-III were influenced by age, being orientation, repetition of three words, and serial subtraction of the less affected by this variable. 38 Hence, it is essential to ensure appropriate cutoff point for older age groups, because the prevalence of cognitive impairment increases with age. 9 , 38

It has been suggested that the cutoff points for screening techniques should be adjusted depending on the premorbid IQ of the patient, for better sensitivity in the detection of dementia. 49 In previous studies, the cutoff scores of the MMSE 50 and MOCA 49 have been associated with premorbid IQ. Likewise, the ACE-III cutoff points were also affected by variation in premorbid IQ. 40 Therefore, the cutoff points must be adjusted to the premorbid IQ values to ensure correct interpretation.

Translation of different languages

Mirza et al (2017) performed a review of all the reports of translation and cultural adaptation procedures of the cognitive examination of Addenbrooke version III (ACE-III) and its predecessors. 51 In this review, it was reported that the first version of ACE is available in 12 languages, the revised version in 16 languages and the third version in 4 languages. Stott et al (2017) reported that only two studies evaluated the ACE-III, but in these studies the ACE-III showed very similar results to those of the ACE-R and these results could be applied equally to the ACE-R. 40

In Table 2 , the different versions of ACE-III currently available are listed.

Different versions of ACE-III currently available

Abbreviations: ACE-III, Addenbrooke’s cognitive examination III; MCI, mild cognitive impairment.

ACE mobile was designed by Newman et al (2018) to support users of the ACE-III by guiding and automating the administration, rule adherence, scoring, and reporting. The new version of the ACE-III, ACE mobile, is an iPad version. The aim is to support the clinician in capturing accurate measurement with zero measurement error. ACE mobile is very effective at reducing errors when compared with the standard paper-and-pen test. ACE mobile is currently provided as a free tool, with no restrictions for clinical use, available on iTunes. 52

The Mini-Addenbrooke’s Cognitive Examination (M-ACE) is a short version of the ACE and was developed and validated in dementia patients. 3 , 53 The M-ACE consists of 5 items with a maximum score of 30. Hsieh et al (2014) identified two cutoffs: 1) ≤25/30 has both high sensitivity and specificity and 2) ≤21/30 is almost certainly a score to have come from a dementia patient regardless of the clinical setting. It has been found to be superior to the MMSE and MoCA in diagnostic utility. Although relatively good levels of sensitivity have been reported, the use of this tool should be questioned in clinical trials where high specificity and low false positive rates are more desirable. 18 , 24

The ACE-III is a screening technique that is capable of differentiating patients with and without cognitive impairment and is sensitive to the early stages of dementia.

Unlike other screening tests (MMSE, MOCA, RUDAS), the ACE-III provides the clinician with a brief multi-component cognitive profile, since it provides specific scores for different cognitive domains: attention, memory, verbal fluency, language, and visuospatial function. It has been demonstrated that the subtests of the ACE-III have significant correlations with neuropsychological test specific for that domain.

Currently, in addition to the English version there are versions in Spanish, Italian, Chinese, Portuguese, Egyptian Arabic, and Thai.

ACE-III is influenced by demographic variables including age, education, and IQ. All of these are considered to be important variables to take into account when interpreting the suggested cutoff points in order to improve diagnostic accuracy.

Future investigations should investigate the utility of the ACE-III in other neurological and psychiatric pathologies, such as head trauma and mood disorders.

The authors report no conflicts of interest in this work.

People also looked at

Original research article, applicability of the ace-iii and rbans cognitive tests for the detection of alcohol-related brain damage.

ace 111 assessment

Background and Aims: Recent investigations have highlighted the value of neuropsychological testing for the assessment and screening of Alcohol-Related Brain Damage (ARBD). The aim of the present study was to evaluate the suitability of the Addenbrooke’s Cognitive Examination (ACE-III) and the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) for this purpose.

Methods: Comparing 28 participants with ARBD (11 with Korsakoff’s Syndrome and 17 with the umbrella “ARBD” diagnosis) and 30 alcohol-dependent participants without ARBD (ALs) we calculated Area Under the Curve (AUC) statistics, sensitivity and specificity values, base-rate adjusted predictive values, and likelihood ratios for both tests.

Results: High levels of screening accuracy were found for the total scores of both the ACE-III ( AUC = 0.823, 95% CIs [0.714, 0.932], SE = 0.056; optimal cut-off ≤86: sensitivity = 82%, specificity = 73%) and RBANS ( AUC = 0.846, 95% CIs [0.746, 0.947], SE = 0.052; optimal cut-off ≤83: sensitivity = 89%, specificity = 67%) at multiple cut-off points. Removing participants with a history of polysubstance from the samples (10 ALs and 1 ARBD) improved the diagnostic capabilities of the RBANS substantially ( AUC = 0.915, 95% CIs [0.831, 0.999], SE = 0.043; optimal cut-off ≤85: sensitivity = 98%, specificity = 80%), while only minor improvements to the ACE-III’s accuracy were observed ( AUC = 0.854, 95% CIs [0.744, 0.963], SE = 0.056; optimal cut-off ≤88: sensitivity = 85%, specificity = 75%).

Conclusion: Overall, both the ACE-III and RBANS are suitable tools for ARBD screening within an alcohol-dependent population, though the RBANS is the superior of the two. Clinicians using these tools for ARBD screening should be cautious of false-positive outcomes and should therefore combine them with other assessment methods (e.g., neuroimaging, clinical observations) and more detailed neuropsychological testing before reaching diagnostic decisions.


It has been estimated that 50–80% of people who misuse alcohol will experience some degree of cognitive impairment ( Bernardin et al., 2014 ). Deficits are primarily observed in memory, executive abilities, visuospatial processing, speed of processing and, to a lesser extent, attention and general intelligence ( Stavro et al., 2013 ). In chronic and severe cases of alcohol-dependence, the neurocognitive impairment may progress to an extent where more debilitating and permanent damage occurs. In such cases, the person may receive a diagnosis of Alcohol-Related Brain Damage (ARBD; Royal College of Psychiatrists, 2014 ), or one of the more discretely defined diagnoses subsumed within this larger conceptual category such as Korsakoff’s Syndrome (see Heirene et al., 2018 for an overview of ARBD diagnoses).

Prompt recognition of ARBD is crucial to avoid further deterioration and minimize the potentially deleterious effects of cognitive dysfunction on treatment outcomes ( Bates et al., 2006 ). Recent investigations have found that neuropsychological testing is highly effective at identifying individuals with ARBD and distinguishing them from both healthy controls and alcohol-dependent individuals with mild cognitive impairment (ALC). For example, Wester et al. (2014) found that both the Rivermead Behavioral Memory Test (RBMT; Wilson et al., 1989 ) and California Verbal Learning Test (CVLT; Delis et al., 1987 ) were useful in differentiating Korsakoff’s Syndrome (KS)—a chronic form of ARBD characterized by severe episodic memory deficits—from an ALC group, with statistically significant group differences on every index of both tests. Wester et al. (2013a) also found the updated RBMT-3 demonstrated high sensitivity and specificity values when distinguishing between KS and ALC groups, and between the latter group and healthy controls.

Brief cognitive screening tests have also proved useful for this purpose. Wester et al. (2013b) found the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005 ) was able to significantly differentiate between KS and ALC groups and between KS and healthy controls, both with high sensitivity and specificity values. What is more, when all of these participants were ranked according to their RBMT-3 memory score, the MoCA was also able to differentiate between those classified as severely impaired and those deemed unimpaired, between severely and mildly impaired groups, and between mild and unimpaired groups all with good sensitivity and specificity. Oudman et al. (2014) have also compared the ARBD screening properties of the MoCA with the Mini-Mental Status Examination (MMSE; Folstein et al., 1975 ). Comparing KS and controls, both screening tests were able to significantly differentiate between the groups with high sensitivity and specificity, though the MoCA was the superior of the two.

Several other cognitive tests have been used to assess ARBD but are yet to be specifically evaluated for this purpose. In particular, the Addenbrooke’s Cognitive Examination-III (ACE-III; Hsieh et al., 2013 ) and Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) have been used repeatedly to assess cognitive impairments in alcohol-dependent individuals both with (e.g., Spiegel and Jim, 2011 ; Wilson et al., 2012 ) and without (e.g., Green et al., 2010 ; Rao, 2016 ) ARBD. Both the ACE-III and RBANS, whilst comparatively short compared with batteries of global function such as the Wechsler Adult Intelligence Scale, provide a more extensive assessment of cognition than screening tests such as the MMSE or MoCA. The combination of being relatively brief but providing a more thorough assessment of cognitive function may make these tools particularly suitable for use in alcohol treatment services or hospital wards where time restrictions often apply, yet a more detailed assessment is warranted than that provided by a screening tool such as the MMSE or MoCA.

The ability of the previous version of the ACE-III, the ACE-R, to identify cognitive impairments in alcohol-dependent individuals without ARBD diagnoses was recently found to be comparable to the MoCA and superior to the MMSE ( Ridley et al., 2017 ). The RBANS too, appears useful for identifying alcohol-related cognitive impairments. Green et al. (2010) found large effect sizes for comparisons between alcohol-dependent individuals and controls on tests of immediate memory, visuospatial abilities, and the overall test score. However, the authors highlighted the RBANS’s inadequate assessment of executive abilities as a limitation of the test. The same criticism could also be said of the ACE-III, which assess only one function classified under the rubric of executive function: verbal fluency. This may restrict the use of both tests with this population, as varying degrees of executive dysfunction have become recognized as a central feature of ARBD ( van Oort and Kessels, 2009 ; Maharasingam et al., 2013 ). Nonetheless, the accuracy of these tests for ARBD screening remains unknown. Indeed, in a recent systematic review of studies evaluating the value of multiple neuropsychological tests in the assessment of alcohol-related cognitive impairment ( Heirene et al., 2018 ), the authors highlighted the ACE-III and RBANS as two tests requiring further validation for ARBD assessment. Moreover, in a recently completed prevalence study conducted by some of the present authors (under review), the ACE-III and RBANS were the most commonly used cognitive tests in the diagnosis of ARBD in the United Kingdom. As a result, the aim of this study was to evaluate the psychometric and diagnostic validity of the ACE-III and RBANS for ARBD screening, and to compare the relative value of the two tests for this purpose.

In the present study, a group of persons with ARBD diagnoses was compared to alcohol-dependent individuals with no such diagnosis (ALs). This comparison group was selected instead of healthy controls as clinicians involved in the assessment of cognition in alcohol-dependent individuals are likely to be more focused on establishing whether the impairment is clinically significant (i.e., likely to have a substantial impact on the person’s ability to function on a day-to-day basis), as opposed to its absence or presence. The ability to differentiate between these two groups has important clinical implications as those with ARBD may require the addition of cognitive rehabilitation strategies to their treatment ( Svanberg and Evans, 2013 ), as well as the implementation of strategies to compensate for cognitive deficits ( Arts et al., 2017 ). Thus, if the two screening tests can differentiate those with ARBD from those without, as has been found for other commonly used screening measures (i.e., MoCA and MMSE), then they can provide quick and relatively inexpensive methods of identifying those who may require more support than offered by traditional treatments.

Materials and Methods


A total of 60 persons agreed to participate in the study (AL: n = 31, ARBD: n = 29). Prior to data collection, a power analysis was performed using G ∗ Power ( Faul et al., 2007 ) to determine adequate sample size. Studies using the ACE-III and RBANS to assess alcohol-related cognitive deficits are scant, though Green et al. (2010) found large effect sizes ( d = 1.08—1.17) on 3 of the RBANS’ scores between controls and moderate-heavy alcohol consumers. Similarly, comparisons of the ACE-III total score between cognitively impaired substance misusers (mostly alcohol-related) and unimpaired controls in Ridley et al. (2017) produced a very large effect size ( d = 1.42; calculated by the present authors based on descriptive statistics provided by Ridley et al., 2017 ). Based on this demonstrated sensitivity to alcohol-related cognitive deficits, a power calculation for a one-tailed independent pairs t-test with an estimated medium-large effect size (0.7; Cohen’s d ), alpha at 0.5, and power at 0.8 estimated that 26 participants would be required in each group. A satisfactory sample size was therefore achieved for both groups.

All ARBD participants were recruited through the Glasgow specialist ARBD service. AL participants were recruited from community rehabilitation services (i.e., non-profit addiction support agencies; n = 20), hospital day-patient services (e.g., psychiatry, occupational therapy; n = 5), and secondary services (e.g., community addictions services, psychology; n = 5). The ARBD group comprised 11 persons with a diagnosis of KS and 18 with the umbrella diagnostic term “ARBD.” This latter diagnostic conceptualization has been exposited by Wilson (2013) and Wilson et al. (2011), who propound an inclusive and pragmatic approach to the nosology of alcohol-related neurocognitive decline which has now been adopted in United Kingdom clinical practice. According to Wilson and colleagues, a person with ARBD must meet two key criteria: [1] evidence of cognitive impairment (as demonstrated by clinical examination or cognitive testing) and [2] a significant history of alcohol misuse (i.e., a minimum average of 35 standard drinks per week for men and 28 for woman for a period of 5 years). The authors have also proposed several other symptoms and behaviors that may support the presence of ARBD (e.g., neuroimaging evidence of cerebellar atrophy; frequent and/or delayed hospital admissions attributable to their alcohol use or social and/or psychiatric problems), as well as those that may indicate the presence of complicating conditions (e.g., neuroimaging evidence of cortical or subcortical infarction, subdural hematoma or other focal brain pathology).

The origin of participants’ diagnoses varied and therefore the exact procedures used to make diagnostic decisions was unknown. However, in the Glasgow area, clinicians report that ARBD diagnoses are—in line with Wilson and colleagues’ criteria (2011; 13)—typically made according to most or all of the following criteria: [1] chronic and excessive alcohol history, [2] evidence of cognitive deficits typically associated with alcohol-dependence (e.g., impairments in episodic and working memory, verbal fluency, and visuospatial processing), [3] neuroimaging evidence of structural brain change, and [4] psychosocial deterioration. KS diagnoses are typically made based on the ICD-10 criteria for Alcohol-Related Amnesic Syndrome ( World Health Organisation, 1992 ), and also involves a combination of assessing alcohol-use history, neuropsychological testing, neuroimaging, and general clinical examination. All AL participants met ICD-10 criteria for Dependence Syndrome and had no evidence of ARBD. All diagnoses were made independently of results from either the ACE-III or RBANS and made by clinicians who were not part of the research team.

For inclusion in the study, participants were required to be abstinent from alcohol and other substances (excluding caffeine and nicotine) for a minimum of five-weeks at testing and have no serious physical (e.g., severe hepatic disease) or psychological (e.g., schizophrenia) complications. Less severe psychological disorders (e.g., mild anxiety or depression) were not criteria for exclusion. Due to the high incidence of head injuries in this population it was considered unrealistic to exclude individuals who had experienced any head injury. However, evidence of severe brain injury (Glasgow Coma Scale: 3–8 ( Teasdale and Jennett, 1974 ); Post-Traumatic Amnesia >7 days; loss of consciousness: >24 h) or previous cranial surgery were used as exclusion criteria. To meet the requirements of testing, all participants were required to have use of their dominant hand, adequate visual function, and not suffer from receptive or expressive aphasia. Based on these criteria, one AL participant was excluded for not meeting the minimum abstinence requirement and one individual with an ARBD diagnosis was excluded due to neuroimaging evidence of intracranial hemorrhage within both frontal lobes resulting from a traumatic brain injury.

The demographic and clinical characteristics of the final 58 participants are displayed in Table 1 . Groups were approximately matched for gender, medication use, and occupational status distribution. The ARBD group were significantly older, had a significantly longer duration of abstinence, were more likely to have suffered from previous head injuries, and had longer drinking histories on average; although these latter two differences were not statistically significant. The AL group contained significantly more individuals with a history of polysubstance use (defined as using any illicit substance other than cannabis [which was common among both groups] on more than one occasion). Varying types and degrees of other-substance use were reported by polysubstance users, including the use of heroin, amphetamines, crack cocaine, cocaine, ecstasy, and diazepam; still, alcohol was the primary substance used by all.


Table 1. Characteristics of AL and ARBD participants.

The ACE-III tests five cognitive domains: attention, memory, verbal fluency, language, and visuospatial function. Individual sub-test scores can be calculated as well as a total composite score which has a maximum of 100. Two optimal cut-off points for dementia have been identified that produce high levels of sensitivity and specificity, 88 and 82, with the former resulting in superior sensitivity to specificity and the latter the obverse ( Hsieh et al., 2013 ; So et al., 2018 ). Administration and scoring takes between 15 and 20 min and does not require specialist training in psychometric testing; although introductory training and familiarization with the test before use is recommended. The test is currently available for free in 30 languages, in iPad and mobile form, and in a miniature version. The Brain and Mind Centre at the University of Sydney provide all versions of the test for free, along with administration guidance 1 .

The RBANS contains 12 subtests which provide five index scores: Immediate memory, Visuospatial/constructional, Language, Attention, and Delayed memory. Combining these index scores provides an overall performance score. All scores are converted to age-adjusted norm scores which have a mean of 100 and SD of 15. Normative data is available for participants aged 12–89 years and four parallel versions of the test have been developed for repeat testing, with forms A and B adapted to United Kingdom use. Administration of the test typically takes 20–30 min. The test is available in over 20 languages and can be purchased from Pearson Clinical Assessments. Pearson state that the test can be used by allied health or special educational professionals, as well as those with more formal training in psychometric assessment.

Potential participants were provided with written details of the study by a professional contact (e.g., support worker, care manager) and asked to arrange an appointment. All individuals meeting inclusion criteria were fully informed of the study’s procedures and provided written consent to participate and for access to medical records to ensure there were no significant complications which would preclude participation. Individuals with ARBD were assessed within residential units, acute settings, or within their own home. AL participants were all assessed in day-patient settings (e.g., community addictions unit). The order in which the two tests were presented was counterbalanced to avoid order effects. Ethical approval was obtained from Greater Glasgow and Clyde Health Board before commencing the study (IRAS project ID: 155916).

Data Analysis

Analyses were conducted using jamovi ( The jamovi project, 2019 ) and NCSS 2019. To enhance the reproducibility and transparency of the analysis, the code used in jamovi to analyze the data can be accessed via this project’s Open Science Framework (OSF) page 2 and in Supplementary Document 1 . This document also includes the full outcomes of all analyses, including tests of statistical assumptions. Sensitivity and specificity analyses were conducted using NCSS (2019) and the associated AUC graphs were made using GraphPad (version 8), therefore no code is available for these.

A combination of parametric and non-parametric tests was used for between-group comparisons, the latter whenever data were not normally distributed (Shapiro-Wilk’s test). For comparisons using parametric tests, Welch’s t -test was used as opposed to Student’s t as it is more robust to violations of homogeneity of variance and more suitable when sample sizes are uneven (see Delacre et al., 2017 ). As we made multiple comparisons between the AL and ARBD groups on test scores, we attempted to reduce the family-wise error rate by using a Bonferroni correction. Dividing 0.05 by the number of score comparisons ( n = 12) resulted in an adjusted alpha of 0.0042 for these comparisons.

Receiver Operator Characteristic (ROC) analyses were conducted to determine the relative screening accuracy of the ACE-III and RBANS. The ROC analysis provides sensitivity (proportion of those with the disorder correctly identified as impaired on the test) and specificity (proportion of those without the disorder correctly identified as unimpaired on the test) values, as well as an Area Under the Curve (AUC) statistic. The AUC statistic varies between 0.5 and 1, with 1 representing perfect sensitivity and specificity. Positive and Negative Predictive Values (PPV/NPV) were also calculated to further evaluate the clinical utility of the tests. The PPV is the percentage of persons with a “positive” test score (i.e., within the impaired range) who actually have the disorder (ARBD), and NPV is the percentage with a “negative” score (i.e., within the normal range) who do not have the disorder. Oudman et al. (2014) calculated the PPV and NPV for the MMSE and MoCA, finding excellent predictive values for both tools. However, their calculations did not reflect the base-rate (prevalence) of ARBD in clinical settings, which directly influences predictive values. In order to account for this, PPVs and NPVs were calculated for the tests according to estimations of ARBD prevalence (base-rate) within the alcohol-dependent population. The proportion of alcohol-dependent individuals believed to experience some form of major neurocognitive disorder ranges from 12.5% ( Zahr et al., 2011 ) up to 35% ( Cook et al., 1998 ). Accordingly, predictive values were calculated for base-rates of 12.5% and 35% to reflect environments where ARBD diagnoses are likely to be queried. Finally, positive and negative likelihood ratios were calculated, which express the probability of having the condition given a positive test score and not having the condition given a negative test score, respectively.

In accord with Simmons et al. (2012) , we have reported how we determined our sample size, all data exclusions, all manipulations, and all measures used in the study.

Between-group comparisons for all test scores are presented in Table 2 , along with standardized (Cohen’s d , pooled SD used as the standardizer) and unstandardized effect sizes to provide a detailed understanding of findings ( Lakens, 2013 ; Pek and Flora, 2018 ). The ARBD group scored significantly lower than ALs on all test indices apart from the Attention and Visuospatial scores of both tests; although, these differences (excluding that related to the ACE-III visuospatial score) approached our adjusted alpha level of 0.0042. According to Cohen’s (1992) classification of effect sizes (i.e., small: d = 0.2, medium: d = 0.5, large: d ≥ 0.8), large effects were observed on the ACE-III for the Total score, Attention, Memory and Fluency, and small effects for Language and Visuospatial scores. For the RBANS, large effects were observed for Total score, Immediate Memory, and Delayed Memory, while medium effects were found for Visuospatial, Language and Attention.


Table 2. ACE-III and RBANS performance by alcohol-dependent individuals with and without ARBD.

As we identified multiple statistically significant differences between the groups for the variables presented in Table 1 , we checked the robustness of the between-group differences on each test score by running analyses of covariance (ANCOVAs) for each comparison and including age, weeks of abstinence, and polysubstance use history (coded as a categorical variables with “yes” or “no” outcomes) as covariates in the models. All statistically significant comparisons between groups at p < 0.004 reported in Table 2 remained significant at this adjusted alpha level aside from the Fluency ( p = 0.012) and Language ( p = 0.086) scores of the ACE-III. Thus, when accounting for between-group differences in demographic and clinical variables, it appears that discrepancies in memory scores most differentiate the groups on both tests. The full outcomes for all 12 ANCOVAs and the analysis code used to produce them in jamovi are available on OSF (see text footnote 2) and in Supplementary Document 1 .

Several exploratory analyses were conducted on total test scores to explore possible within-group differences. Alpha was set at 0.05 to minimize the risk of false negative outcomes (i.e., Type-II errors; see Hartgerink et al., 2017 ; Witt, 2019 ). The potential adverse consequences associated with false-positive outcomes (i.e., Type-I) in these exploratory analyses was deemed to be low ( Lakens et al., 2018 ). Nonetheless, the following outcomes should be interpreted as reflecting exploratory, preliminary evaluations of the data. First, due to the high number of ALs with a history of polysubstance use, a within-group comparison between those with a history of polysubstance use and those without was undertaken. No significant difference was observed between AL polysubstance users ( n = 10, M = 85.5, SE = 3.5) and ALs who only used alcohol ( n = 20, M = 91.3, SE = 1.6) for the ACE-III total score, t (12.9) = 1.52, p = 0.152, d = 0.68. On the RBANS, however, a significant difference was found between those with a polysubstance use history ( M = 78.2, SE = 4.1) and those without ( M = 95.7, SE = 3.2) for the total score, t (20.12) = 3.36, p = 0.003, d = 1.25. These poorer scores could not be explained by differences in drinking history duration ( U = 83, p = 0.466) or length of abstinence ( U = 75.5, p = 0.289) between the two sub-groups. The ARBD group was also dichotomized for further analysis into individuals with a specific diagnosis of KS and those with the broad “ARBD” diagnosis. No significant difference was found between individuals with KS ( n = 11; ACE-III: M = 77.1, SE = 3.5; RBANS: M = 65.7, SE = 2.7) and those with ARBD ( n = 17; ACE-III: M = 79.5, SE = 2.3; RBANS: M = 71.5, SE = 3.1) for ACE-III, t (18.4) = 0.568, p = 0.577, or RBANS scores, t (25.9) = 1.4, p = 0.173.

ROC curves for the ACE-III and RBANS are displayed in Figure 1 . Table 3 displays test cut-off scores and their corresponding diagnostic values, including the number of AL and ARBD participants classed as impaired using this cut-off, positive and negative predictive values, and positive and negative likelihood ratios. Only cut-off scores that produced optimal sensitivity (≥ 80%) and specificity (≥ 60%) levels are presented, consistent with previous research in this area ( Oudman et al., 2014 ). The higher threshold for sensitivity over specificity is consistent with the view that screening tests should have high levels of sensitivity to maximize disease detection, while subsequent assessments should have high levels of specificity in order to ensure accurate diagnosis and avoid misdiagnosis ( McNamara and Martin, 2018 ).


Figure 1. ACE-III and RBANS ROC curves for differentiating between alcohol-dependent individuals with and without ARBD.


Table 3. Diagnostic validity of the ACE-III and RBANS for differentiating between alcohol-dependent individuals with and without ARBD.

The ACE-III total score was able to significantly differentiate between the AL and ARBD participants ( AUC = 0.823, 95% CIs [0.714, 0.932], SE = 0.056, p < 0.001), with an optimal cut-off score of ≤86 producing a sensitivity of 82% and specificity of 73%. Similarly, the RBANS total score significantly distinguished between AL and ARBD participants ( AUC = 8.46, 95% CIs [0.746, 0.947], SE = 0.051, p < 0.001), with an optimal cut-off score of ≤83 producing a sensitivity of 89% and specificity of 67%. Although the AUC value was larger for the RBANS than the ACE-III, the difference was not statistically significant (discrepancy = 0.023, SE = 0.045, Z = 0.522, p = 0.602). As AL participants with a history of polysubstance misuse scored significantly lower than their alcohol-use-only counterparts on the RBANS, two further exploratory ROC analyses were ran to see how sensitivity and specificity values were affected by the removal of all polysubstance users (see Figure 2 for ROC plot). This removal resulted in minor improvements to diagnostic properties of the ACE-III ( AUC = 0.854, 95% CIs [0.744, 0.963], SE = 0.056, p < 0.001; optimal cut-off = ≤ 88: sensitivity = 85, specificity = 75) and substantial improvements to the RBANS ( AUC = 0.915, 95% CIs [0.831, 0.999], SE = 0.043, p < 0.001; optimal cut-off: ≤85: sensitivity = 96, specificity = 80). Again, the difference between each test’s AUC value was not significant (discrepancy = 0.0611, SE = 0.0511, Z = 1.196, p = 0.232).


Figure 2. ACE-III and RBANS ROC curves for differentiating between alcohol-dependent individuals with and without ARBD (excluding polysubstance users).

Several exploratory correlational analyses were conducted to investigate the relationships between participant characteristics (i.e., duration of drinking history, duration of abstinence, and age) and total test scores (alpha set at 0.05). None of the correlations between the groups’ test scores and drinking histories ( r s range = −0.023 to −0.164, p s ≥ 0.388), age ( r s range = −0.003 to −0.173, p s ≥ 0.378), or length of abstinence ( r s range = −0.253 to 0.033, p s ≥ 0.194) were significant [full outcomes reported on OSF (see text footnote 2) and in Supplementary Document 1 ]. Both groups’ total ACE-III and RBANS scores were significantly and strongly correlated (AL: r s = 0.784, p < 0.001; ARBD r s = 0.700 p < 0.001), supporting the convergent validity of the tests.

The present study aimed to evaluate the suitability of the ACE-III and RBANS for ARBD assessment and their ability to differentiate alcohol-dependent individuals with ARBD from those without. Both measures produced significant between-group differences on total scores and several sub- test scores, although several significant effects for the ACE-III did not remain when covariates were included in analysis models. Effect sizes ( d ) were mostly in the medium-large range, indicating a substantial discrepancy between the groups’ scores on both tests. This was particularly the case for the subtests indexing memory, which produced very large effects on both tests ( d ≥ 1.2) that were robust to the inclusion of covariables in the models. Optimal sensitivity and specificity levels were identified for the total scores of both tools at multiple possible cut-off points. However, it should be noted that while we selected our sensitivity and specificity thresholds of ≥80 and ≥60%, respectively, for consistency with similar studies in this domain ( Oudman et al., 2014 ), other authors have recommended optimal thresholds of ≥80% should be used for both values. Relatedly, although peak sensitivity values were high for both the ACE-III (86%) and RBANS (96%), specificity values peaked at 73% for the former and 70% for the latter, highlighting a risk of false-positives when using such cut-off scores.

Base-rate-adjusted positive and negative predictive values were also calculated for both tests at ARBD prevalence rates of 12.5 and 35%. At 12.5% prevalence, PPVs were low for both the ACE-III (peak: 30.1%) and RBANS (peak: 28.1%), further supporting a cautious interpretation of positive test scores. As would be expected, these values increased considerably when the prevalence rate was increased to 35% (ACE-III peak: 62.4%; RBANS peak: 59.6%). NPVs were high for the tests at 12.5% (ACE-III peak: 97%; RBANS peak: 99.2%) and 35% (ACE-III peak: 89.7%; RBANS peak: 97.1%) prevalence rates, supporting confident interpretations of negative test scores as true negatives. Overall, while predictive values were better overall when using the increased base-rate of 35%, this figure is predicated on the assumption that alcohol-related cerebellar degeneration is part of the same disease process as Wernicke-Korsakoff’s Syndrome ( Cook et al., 1998 ) and may therefore be an overestimation of ARBD prevalence in the alcohol-dependent population.

Comparing the two tests, the diagnostic values produced by the ACE-III appear largely commensurate with those of the RBANS. However, some minor discrepancies between the two are evident. First, while the conflated outcomes of sensitivity and specificity values were approximately equal between the two, the ACE-III produced a higher level of specificity at its optimal cut-off (sensitivity = 82%, specificity = 73%), while the RBANS had higher sensitivity (sensitivity = 89%, specificity = 67%). This difference in ability was reflected in greater PPVs and PLRs for the ACE-III and greater NPVs and NLRs for the RBANS. Thus, the ACE-III was more likely to correctly classify those without the disorder as unimpaired than the RBANS, and the RBANS was more likely to correctly classify those with the disorder as impaired than the ACE-III. The availability of parallel versions of the RBANS may contribute to its value in assessing this population as repeat testing is required to monitor any changes cognitive dysfunction over time and in response to interventions ( Royal College of Psychiatrists, 2014 ; Heirene et al., 2018 ). Thus, the RBANS can be used for the monitoring of ARBD whilst circumventing the issue of practice/learning effects associated with repeated testing. Overall, if deciding between the two, the ACE-III appears suitable when time restrictions are present, though the more extensive RBANS should be considered when time allows.

The diagnostic values of the ACE-III and RBANS found here, while high, were lower than those for the MMSE and MoCA observed by Oudman et al. (2014) . However, the disparity is likely because the authors compared those with ARBD (KS) to healthy individuals – not ALs as was done here. Indeed, when Wester et al. (2013b) compared KS participants with controls the diagnostic values of the MoCA were superior to those observed here, though when KS participants were compared with an ALC group the sensitivity and specificity values did not reach optimal levels (sensitivity = 73, specificity = 75). The AL group in the present study demonstrated clear impairments relative to norm scores on both measures, suggesting similarities with the mildly impaired group studied by Wester et al. (2013b) . Thus, the screening capabilities of the ACE-III and RBANS when comparing mildly versus severely impaired groups may by superior to those of the MoCA; although a direct systematic comparison would be required to confirm the superior test(s).

This is the first study to directly evaluate the screening capabilities of the ACE-III and RBANS for ARBD. Overall, the findings support the use of both tests in clinical assessments of alcohol-users; although caution should be taken to avoid false-positive tests when using the cut-off points identified. Our findings indicate that clinicians should observe individual subtests scores as well as overall scores to best differentiate those with ARBD from those without, with a particular focus on memory scores. The present study also provides a novel understanding of how using neuropsychological testing in a screening capacity for ARBD is affected by a history of polysubstance use. Findings from our exploratory analyses indicated that those with a history of polysubstance use, compared to those without such a history, will perform worse on neuropsychological tests. The poor scores by AL polysubstance users in the present study could not be explained by differences in drinking history duration or length of abstinence, suggesting it was the additional drug-use which compounded their alcohol-related cognitive deterioration; however, we cannot be certain of this from the data collected in this study. Bondi et al. (1998) reported a similar significant decrease in performance on selective tasks by ALs with concurrent polysubstance use compared to those who only used alcohol; still, causation cannot be inferred from these findings. Overall, our findings suggest a consideration of previous drug use should be made when cognitively assessing alcohol users as this may also contribute to impairment. Nonetheless, it is likely the degree of impairment, as opposed to its etiology, that is of interest to clinicians.

This is also one of the first studies to investigate the value of neuropsychological testing in the detection of ARBD, as opposed to more discretely defined diagnoses such as KS. The recent impetus for using ARBD as a broad conceptual diagnostic term has been motivated by heterogeneity within those diagnosed with KS, including varying numbers of individuals with executive ( van Oort and Kessels, 2009 ) and/or intellectual ( Jacobson and Lishman, 1987 ) deficits, as well as a high prevalence of head injuries, liver disease and other factors which can confound neurocognitive impairment and create further inter-person variability. Indeed, Bowden (1990) has argued that the rigid selection criteria implemented by KS researchers may render their samples artifacts of this process which are, as a result, unrepresentative of the heterogeneous presentation more typical of this population. Comparing the two diagnostic sub-divisions of the ARBD group, no statistically meaningful difference was identified between those with KS and those with ARBD on the total scores of the ACE-III or RBANS; although the sample sizes were small, potentially limiting the ability to detect any subtle differences. Due to small sample sizes, no further differences between the sub-groups’ scores were explored. Future research should compare larger samples of persons with KS and ARBD to explore whether differences in cognitive profiles underpin the choice of diagnostic nomenclature in modern clinical settings, thereby evaluating the merit of the distinction.

The primary limitation of this study was the absence of a reference standard assessment to confirm the existing diagnosis in the ARBD population. However, as previously stated, ARBD diagnoses in the study area are made against rigorous criteria by the Glasgow ARBD service which specializes in the diagnosis and treatment of those with the condition. Additionally, the ARBD group had significantly longer drinking histories and consistently poorer scores on both the ACE-III and RBANS, supporting the diagnostic distinction between groups. A second limitation was the significant difference between groups in regards to age, polysubstance use history, and weeks of abstinence. Although, to account for these differences, we ran ANCOVAs for all group comparisons and included these variables as covariates and have transparently reported all outcomes from these in addition to t -test and Mann-Whitney- U outcomes ( Supplementary Document 1 ).

While the ACE-III and RBANS can be used to screen several different neurocognitive disorders (including mild cognitive impairment and various dementias; Karantzoulis et al., 2013 ; Bruno and Schurmann Vignaga, 2019 ), two screening tests have been developed recently specifically for assessing alcohol-related cognitive impairments. The first of these, the BEARNI (Brief Examination of Alcohol-related Neuropsychological Impairments; Ritz et al., 2015 ), was designed to be easily administered by non-specialists and assesses working and episodic memory, visuospatial skills, executive function, and ataxia. The second test, the TEDCA (Test of Detection of Cognitive Impairment in Alcoholism; Jurado-Barba et al., 2017 ), assesses working and episodic memory and visuospatial skills. Both tests may provide equally, if not superior, screening capabilities to the ACE-III and RBANS for ARBD detection, though neither test has been specifically validated for the screening of clinically diagnosed alcohol-related neurocognitive disorders (e.g., KS, ARBD). The BEARNI was found to have very high sensitivity (100%) for detecting individuals with cognitive impairment (as determine by detailed neuropsychological assessment) within a sample of ALs, although the specificity of the test was very poor (4%; Pelletier et al., 2018 ). In the same sample, the MoCA demonstrated 79% sensitivity and 65% specificity. Future research in this domain should focus on evaluating the screening capabilities of both tests for the populations studied here to determine whether they may better replace more generalized tests used in clinical practice (e.g., ACE-III, MoCa etc.).

In sum, the present findings add to a recent body of evidence suggesting that neuropsychological tests can be used effectively to inform the ARBD diagnostic process. The two tests studied here are more extensive in their assessment of cognition than the cognitive screening tests previously investigated for this purpose (i.e., MoCA and MMSE) and can therefore provide a more comprehensive overview of impaired and preserved cognitive abilities, whilst still remaining relatively quick to administer. Nonetheless, screening tests alone should not be used to confirm ARBD diagnoses ( Heirene et al., 2018 ). Informed diagnostic decision making and treatment planning require more thorough assessments of cognition to detail the severity of impairment and the specific skills affected. In particular, further assessment of executive function (e.g., Behavioral Assessment of the Dysexecutive Syndrome) is warranted as both the ACE-III and RBANS lack sufficient testing of this domain. Finally, it is important to note that while neuropsychological testing is an important and informative feature of ARBD assessment, so too are clinical observations, assessments of activities of daily living, nutritional status investigations, and neuroimaging procedures ( Horton et al., 2015 ), and thus each should be used in conjunction whenever possible.

Data Availability Statement

Research data are not shared publicly due to stipulations made by the research ethics committee at the time of approval regarding the storage and confidentially of patient data. This statement is subject to change: we have applied, in retrospect, to the ethics committee for approval to share the anonymized data collected. For updates on this request or for requests to access the data, please contact RH ([email protected]).

Ethics Statement

The studies involving human participants were reviewed and approved by the Greater Glasgow and Clyde Health Board. The patients/participants provided their written informed consent to participate in this study.

Author Contributions

PB and JE devised the research questions and study design. PB was responsible for data collection. RH analyzed the data, produced the figures, and managed the project’s Open Science Framework page. RH and PB drafted the manuscript. All authors contributed to the interpretation of findings and to manuscript revisions, approved the final manuscript for submission, and agreed to be accountable for all aspects of the work.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg.2019.02636/full#supplementary-material

Arts, N. J. M., Walvoort, S. J. W., and Kessels, R. P. C. (2017). Korsakoff’s syndrome: a critical review. Neuropsychiatr. Dis. Treat. 13, 2875–2890. doi: 10.2147/NDT.S130078

PubMed Abstract | CrossRef Full Text | Google Scholar

Bates, M. E., Pawlak, A. P., Tonigan, J. S., and Buckman, J. F. (2006). Cognitive impairment influences drinking outcome by altering therapeutic mechanisms of change. Psychol. Addict. Behav. 20, 241–253. doi: 10.1037/0893-164x.20.3.241

Bernardin, F., Maheut-Bosser, A., and Paille, F. (2014). Cognitive impairments in alcohol-dependent subjects. Front. Psychiatry 5:78. doi: 10.3389/fpsyt.2014.00078

Bondi, M. W., Drake, A. I., and Grant, I. (1998). Verbal learning and memory in alcohol abusers and polysubstance abusers with concurrent alcohol abuse. J. Int. Neuropsychol. Soc. 4, 319–328. doi: 10.3389/fpsyt.2014.00078

Bowden, S. C. (1990). Separating cognitive impairment in neurologically asymptomatic alcoholism from Wernicke-Korsakoff syndrome: is the neuropsychological distinction justified? Psychol. Bull. 107, 355–366. doi: 10.1037//0033-2909.107.3.355

Bruno, D., and Schurmann Vignaga, S. (2019). Addenbrooke’s cognitive examination III in the diagnosis of dementia: a critical review. Neuropsychiatr. Dis. Treat. 15, 441–447. doi: 10.2147/NDT.S151253

Cohen, J. (1992). A power primer. Psychol. Bull. 112, 155–159. doi: 10.1037/0033-2909.112.1.-155

Cook, C. C. H., Hallwood, P. M., and Thomson, A. D. (1998). B vitamin deficiency and neuropsychiatric syndromes in alcohol misuse. Alcohol Alcohol. 33, 317–336. doi: 10.1093/oxfordjournals.alcalc.a008400

Delacre, M., Lakens, D., and Leys, C. (2017). Why psychologists should by default use welch’s t-test instead of student’s t-test. Int. Rev. Soc. Psychol. 30, 92–101. doi: 10.5334/irsp.82

CrossRef Full Text | Google Scholar

Delis, D. C., Kramer, J. H., Kaplan, E., and Ober, B. A. (1987). The California Verbal Learning Test. San Antonio, TX: Psychological Corporation.

Google Scholar

Faul, F., Erdfelder, E., Lang, A.-G., and Buchner, A. (2007). G ∗ Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav. Res. Methods 39, 175–191. doi: 10.3758/bf03193146

Folstein, M. F., Folstein, S. E., and McHugh, P. R. (1975). “Minimental state”. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12, 189–198. doi: 10.1016/0022-3956(75)90026-6

Green, A., Garrick, T., Sheedy, D., Blake, H., Shores, E. A., and Harper, C. (2010). The effect of moderate to heavy alcohol consumption on neuropsychological performance as measured by the repeatable battery for the assessment of neuropsychological status. Alcoholism 34, 443–450. doi: 10.1111/j.1530-0277.2009.01108.x

Hartgerink, C. H. J., Wicherts, J. M., and Van Assen, M. A. L. M. (2017). Too good to be false: nonsignificant results revisited. Collabra Psychol. 3:9. doi: 10.1525/collabra.71

Heirene, R. M., John, B., and Roderique-Davies, G. (2018). Identification and evaluation of neuropsychological tools used in the assessment of alcohol-related cognitive impairment: a systematic review. Front. Psychol. 9:2618. doi: 10.3389/fpsyg.2018.02618

Horton, L., Duffy, T., Hollins Martin, C., and Martin, C. R. (2015). Comprehensive assessment of alcohol-related brain damage (ARBD): gap or chasm in the evidence? J. Psychiatr. Ment. Health Nurs. 22, 3–14. doi: 10.1111/jpm.12156

Hsieh, S., Schubert, S., Hoon, C., Mioshi, E., and Hodges, J. R. (2013). Validation of the addenbrooke’s cognitive examination III in frontotemporal dementia and Alzheimer’s disease. Dement. Geriatr. Cogn. Disord. 36, 242–250. doi: 10.1159/000351671

Jacobson, R. R., and Lishman, W. A. (1987). Selective memory loss and global intellectual deficits in alcoholic Korsakoff’s syndrome. Psychol. Med. 17, 649–655. doi: 10.1017/S0033291700025885

Jurado-Barba, R., Martinez, A., Sion, A., Alvarez-Alonso, M. J., Robles, A., Quinto-Guillen, R., et al. (2017). Development of a screening test for cognitive impairment in alcoholic population: TEDCA. Actas Esp. Psiquiatr. 45, 201–217.

PubMed Abstract | Google Scholar

Karantzoulis, S., Novitski, J., Gold, M., and Randolph, C. (2013). The repeatable battery for the assessment of neuropsychological status (RBANS): utility in detection and characterization of mild cognitive impairment due to Alzheimer’s disease†. Arch. Clin. Neuropsychol. 28, 837–844. doi: 10.1093/arclin/act057

Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Front. Psychol. 4:863. doi: 10.3389/fpsyg.2013.00863

Lakens, D., Adolfi, F. G., Albers, C. J., Anvari, F., Apps, M. A. J., Argamon, S. E., et al. (2018). Justify your alpha. Nat. Hum. Behav. 2, 168–171. doi: 10.1038/s41562-018-0311-x

Maharasingam, M., Macniven, A. B., and Mason, J. (2013). Executive functioning in chronic alcoholism and Korsakoff syndrome. J. Clin. Exp. Neuropsychol. 35, 501–508. doi: 10.1080/13803395.2013.795527

McNamara, L. A., and Martin, S. W. (2018). “Principles of epidemiology and public health,” in Principles and Practice of Pediatric Infectious Diseases , 5th Edn, eds S. S. Long, C. G. Prober, and M. Fischer (Atlanta, GA: Centers for Disease Control and Prevention), 1–9. doi: 10.1016/B978-0-323-40181-4.00001-3

Nasreddine, Z. S., Phillips, N. A., Bedirian, V., Charbonneau, S., Whitehead, V., Collin, I., et al. (2005). The montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment. J. Am. Geriatr. Soc. 53, 695–699. doi: 10.1111/j.1532-5415.2005.53221.x

Oudman, E., Postma, A., Van der Stigchel, S., Appelhof, B., Wijnia, J. W., and Nijboer, T. C. (2014). The montreal cognitive assessment (MoCA) is superior to the mini mental state examination (MMSE) in detection of Korsakoff’s syndrome. Clin. Neuropsychol. 28, 1123–1132. doi: 10.1080/13854046.2014.960005

Pek, J., and Flora, D. B. (2018). Reporting effect sizes in original psychological research: a discussion and tutorial. Psychol. Methods 23, 208–225. doi: 10.1037/met0000126

Pelletier, S., Alarcon, R., Ewert, V., Forest, M., Nalpas, B., and Perney, P. (2018). Comparison of the MoCA and BEARNI tests for detection of cognitive impairment in in-patients with alcohol use disorders. Drug Alcohol Depend 187, 249–253. doi: 10.1016/j.drugalcdep.2018.02.026

Rao, R. (2016). Cognitive impairment in older people with alcohol use disorders in a UK community mental health service. Adv. Dual Diagn. 9, 154–158. doi: 10.1108/ADD-06-2016-0014

Ridley, N., Batchelor, J., Draper, B., Demirkol, A., Lintzeris, N., and Withall, A. (2017). Cognitive screening in substance users: diagnostic accuracies of the Mini-Mental State Examination, addenbrooke’s cognitive examination-revised, and montreal cognitive assessment. J. Clin. Exp. Neuropsychol. 24, 1–16. doi: 10.1080/13803395.2017.1316970

Ritz, L., Lannuzel, C., Boudehent, C., Vabret, F., Bordas, N., Segobin, S., et al. (2015). Validation of a brief screening tool for alcohol-related neuropsychological impairments. Alcoholism Clin. Exp. Res. 39, 2249–2260. doi: 10.1111/acer.12888

Rose, D., Pevalin, D., and O’Reilly, K. (2005). The National Statistics Socio-Economic Classification: Origins, Development and Use. Gosport: Ashford Colour Press Ltd.

Royal College of Psychiatrists, (2014). Alcohol and Brain Damage in Adults. With Reference to High-Risk Groups. London: Royal College of Psychiatrists.

Simmons, J. P., Nelson, L. D., and Simonsohn, U. (2012). A 21 word solution. Dialogue 26, 4–12.

So, M., Foxe, D., Kumfor, F., Murray, C., Hsieh, S., Savage, G., et al. (2018). Addenbrooke’s cognitive examination III: psychometric characteristics and relations to functional ability in dementia. J. Int. Neuropsychol. Soc. 24, 854–863. doi: 10.1017/S1355617718000541

Spiegel, D. R., and Jim, K. J. (2011). A case of probable korsakoff’s syndrome: a syndrome of frontal lobe and diencephalic structural pathogenesis and a comparison with medial temporal lobe dementias. Innov. Clin. Neurosci. 8, 15–19.

Stavro, K., Pelletier, J., and Potvin, S. (2013). Widespread and sustained cognitive deficits in alcoholism: a meta-analysis. Addict. Biol. 18, 203–213. doi: 10.1111/j.1369-1600.2011.00418.x

Svanberg, J., and Evans, J. J. (2013). Neuropsychological rehabilitation in alcohol-related brain damage: a systematic review. Alcohol Alcohol. 48, 704–711. doi: 10.1093/alcalc/agt131

Teasdale, G., and Jennett, B. (1974). Assessment of coma and impaired consciousness: a practical scale. Lancet 304, 81–84. doi: 10.1016/S0140-6736(74)91639-0

The jamovi project, (2019). jamovi . (Version 1.0) [Computer Software]. Available at: https://www.jamovi.org (accessed October 22, 2019).

van Oort, R., and Kessels, R. P. C. (2009). Executive dysfunction in Korsakoff’s syndrome: time to revise the DSM criteria for alcohol-induced persisting amnestic disorder? Int. J. Psychiatry Clin. Pract. 13, 78–81. doi: 10.1080/13651500802308290

Wester, A. J., Roelofs, R. L., Egger, J. I., and Kessels, R. P. C. (2014). Assessment of alcohol-related memory deficits: a comparison between the Rivermead behavioural memory test and the california verbal learning test. Brain Impair. 15, 18–27. doi: 10.1017/BrImp.2014.6

Wester, A. J., van Herten, J. C., Egger, J. I., and Kessels, R. P. C. (2013a). Applicability of the rivermead behavioural memory test - third edition (RBMT-3) in Korsakoff’s syndrome and chronic alcoholics. Neuropsychiatr. Dis. Treat. 9, 875–881. doi: 10.2147/NDT.S44973

Wester, A. J., Westhoff, J., Kessels, R. P. C., and Egger, J. I. M. (2013b). The montreal cognitive assessment (MoCA) as a measure of severity of amnesia in patients with alcohol-related cognitive impairments and Korsakoff syndrome. Clin. Neuropsychiatry 10, 134–141.

Wilson, B. A., Cockburn, J., Baddeley, A., and Hiorns, R. (1989). The development and validation of a test battery for detecting and monitoring everyday memory problems. J. Clin. Exp. Neuropsychol. 11, 855–870. doi: 10.1080/01688638908400940

Wilson, K. (2013). Alcohol-Related Brain Damage in the 21st Century. London: Royal College of Psychiatrists.

Wilson, K., Halsey, A., Macpherson, H., Billington, J., Hill, S., Johnson, G., et al. (2012). The psycho-social rehabilitation of patients with alcohol-related brain damage in the community. Alcohol Alcohol. 47, 304–311. doi: 10.1093/alcalc/agr167

Witt, J. K. (2019). Insights into criteria for statistical significance from signal detection analysis. Meta Psychol. 3, 1–16. doi: 10.15626/MP.2018.871

World Health Organisation, (1992). The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Genevea: World Health Organisation.

Zahr, N. M., Kaufman, K. L., and Harper, C. G. (2011). Clinical and pathological features of alcohol-related brain damage. Nat. Rev. Neurol. 7, 284–294. doi: 10.1038/nrneurol.2011.42

Keywords : ARBD, Korsakoff’s syndrome, ACE-III, RBANS, diagnosis

Citation: Brown P, Heirene RM, Gareth-Roderique-Davies, John B and Evans JJ (2019) Applicability of the ACE-III and RBANS Cognitive Tests for the Detection of Alcohol-Related Brain Damage. Front. Psychol. 10:2636. doi: 10.3389/fpsyg.2019.02636

Received: 06 September 2019; Accepted: 07 November 2019; Published: 28 November 2019.

Reviewed by:

Copyright © 2019 Brown, Heirene, Gareth-Roderique-Davies, John and Evans. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Robert M. Heirene, [email protected]

† ORCID: Pamela Brown, orcid.org/0000-0003-2727-5250 ; Robert M. Heirene, orcid.org/0000-0002-5508-7102 ; Gareth-Roderique-Davies, orcid.org/0000-0002-6446-749X ; Bev John, orcid.org/0000-0002-5520-2385

Similar articles being viewed by others

Slider with three articles shown per slide. Use the Previous and Next buttons to navigate the slides or the slide controller buttons at the end to navigate through each slide.

Screening Accuracy of Mini Addenbrooke’s Cognitive Examination Test for HIV-Associated Neurocognitive Disorders in People Ageing with HIV

04 January 2022

Mattia Trunfio, Davide De Francesco, … Andrea Calcagno

A model of cognitive evaluation battery for diagnosis of mild cognitive impairment and dementia in educated and illiterate Egyptian elderly people

29 September 2020

Nehal ElKholy, Heba Mohamed Tawfik, … Sarah Ahmed Hamza

Frontal Assessment Battery in Early Cognitive Impairment: Psychometric Property and Factor Structure

30 August 2019

Wen Yang Goh, D. Chan, … W. S. Lim

ace 111 assessment

Validation of Addenbrooke’s cognitive examination III for detecting mild cognitive impairment and dementia in Japan

29 April 2019

Shintaro Takenoshita, Seishi Terada, … Norihito Yamada

ace 111 assessment

The items in the Chinese version of the Montreal cognitive assessment basic discriminate among different severities of Alzheimer’s disease

04 November 2019

Yan-Rong Zhang, Yun-Long Ding, … Wan-Li Dong

ace 111 assessment

Test Your Memory (TYM test): diagnostic evaluation of patients with non-Alzheimer dementias

02 July 2019

Jeremy Brown, Julie Wiggins, … James B. Rowe

Improving clarity and transparency in cognitive assessment: conversion of the Cambridge Cognition Examination to the International Classification of Functioning, Disability and Health

15 May 2018

Sam Kirch, Ellen Gorus, … Patricia De Vriendt

ace 111 assessment

The instruments used by the Italian centres for cognitive disorders and dementia to diagnose mild cognitive impairment (MCI)

03 September 2018

Federica Limongi, Marianna Noale, … for the MCI Working Group

ace 111 assessment

Validation of Slovenian version of ADAS-Cog for patients with mild cognitive impairment and Alzheimer's disease

23 August 2021

Janina Ulbl & Martin Rakusa

Reliability of Addenbrooke's Cognitive Examination III in differentiating between dementia, mild cognitive impairment and older adults who have not reported cognitive problems

European Journal of Ageing volume  19 ,  pages 495–507 ( 2022 ) Cite this article

4607 Accesses

5 Citations

12 Altmetric

Metrics details

Diagnosing dementia can be challenging for clinicians, given the array of factors that contribute to changes in cognitive function. The Addenbrooke’s Cognitive Examination III (ACE-III) is commonly used in dementia assessments, covering the domains of attention, memory, fluency, visuospatial and language. This study aims to (1) assess the reliability of ACE-III to differentiate between dementia, mild cognitive impairment (MCI) and controls and (2) establish whether the ACE-III is useful for diagnosing dementia subtypes. Client records from the Northern Health and Social Care Trust (NHSCT) Memory Service ( n  = 2,331, 2013–2019) were used in the analysis including people diagnosed with Alzheimer’s disease ( n  = 637), vascular dementia ( n  = 252), mixed dementia ( n  = 490), MCI ( n  = 920) and controls ( n  = 32). There were significant differences in total ACE-III and subdomain scores between people with dementia, MCI and controls ( p  < 0.05 for all), with little overlap between distribution of total ACE-III scores (< 39%) between groups. The distribution of total ACE-III and subdomain scores across all dementias were similar. There were significant differences in scores for attention, memory and fluency between Alzheimer’s disease and mixed dementia, and for visuospatial and language between Alzheimer’s disease–vascular dementia ( p  < 0.05 for all). However, despite the significant differences across these subdomains, there was a high degree of overlap between these scores (> 73%) and thus the differences are not clinically relevant. The results suggest that ACE-III is a useful tool for discriminating between dementia, MCI and controls, but it is not reliable for discriminating between dementia subtypes. Nonetheless, the ACE-III is still a reliable tool for clinicians that can assist in making a dementia diagnosis in combination with other factors at assessment.

Working on a manuscript?

Avoid the most common mistakes and prepare your manuscript for journal editors.


As people get older, they experience changes in cognitive function some of which are associated with normal ageing. One of the challenges in clinical practice is to differentiate between presentations that are consistent with functional cognitive impairment, mild cognitive impairment (MCI), a dementia or something else. In most cases, those with cognitive impairments will be referred to specialist memory services. Specialist memory services offer timely differential diagnosis, which is beneficial for the person as it allows for better adjustment, slowing of progression and planning ahead, and there are also significant savings to the health economy (Bamford et al. 2004 ; Prince et al. 2011 ; Pratt and Wilkinson 2003 ; Banarjee and Wittenberg 2009 ).

There is no single test for dementia, and diagnosis is made on the basis of excluding other causes for the symptoms and clinical impression. Dementia assessment at a specialist service generally involves a formal assessment of cognitive function, activities of daily living, social, educational and employment history and a collateral history from someone who knows the person well. The person may also be referred for brain imaging. All of this information is reviewed to help make a differential diagnosis. It is most likely that clinicians, through clinical experience, weight the relative contribution of each of the multiple sources of information to come to a decision about diagnosis; however, there is no agreed weighting for this information. There is also no agreed process across services which means that there is significant variability in the assessments used at different services. Taken together, this leaves the potential for diagnostic variability across services.

Cognitive profiles vary across MCI and the different types of dementia. Those living with MCI typically experience cognitive impairment between that of normal ageing and mild dementia (Grundman et al. 2004 ). Amnestic MCI typically presents with predominant impairment in memory with increased likelihood of progression to Alzheimer’s (Grundman et al. 2004 ) while in non-amnestic MCI memory is preserved but one other cognitive domain will be affected. People with Alzheimer’s will have impairment in memory and at least one other cognitive domain such as attention, language, visuospatial ability and fluency. Greater impairment with episodic memory is seen in people with Alzheimer’s compared to those with vascular dementia (Graham and Hodges 2004 ; Karantzoulis et al. 2011 ). In contrast, people with vascular dementia have worsening semantic memory, attention and visuospatial functioning in comparison to people with Alzheimer’s (Graham and Hodges 2004 ).

Cognitive assessment is a key factor in decision making, and part of this process involves screening. The use of screening tools alone will not determine the diagnosis; however, the choice of test is still important. Different screening tools are used across services such as the Montreal Cognitive Assessment (MoCA) (Nasreddine et al. 2005 ), Mini-Mental State Examination (MMSE) (Folstein et al. 1975 ) and the Addenbrooke’s Cognitive Examination III (ACE-III). The original ACE was developed to help detect mild dementia and differentiate between Alzheimer’s disease and frontotemporal dementia (Mathuranath et al. 2000 ). It was initially designed as an extension to the commonly used MMSE, with additional neuropsychological domains incorporated to improve screening performance (Mathuranath et al. 2000 ), and was later revised (ACE-R) with clearly defined subdomain scores (Hodges and Larner 2016 ). The ACE-III was subsequently created to remove elements of the MMSE and address weaknesses of the ACE-R (Hsieh et al. 2013 ). The ACE-III takes around half an hour to complete and is scored out of 100, with higher scores corresponding to better cognitive function. It incorporates five subdomains: attention, memory, fluency, language and visuospatial. The ACE-III has been validated as a screening tool for cognitive deficits in Alzheimer’s disease and frontotemporal dementia and has been translated and validated in other languages including Chinese, Japanese and Spanish (Wang et al. 2017 ; Li et al. 2019 ; Takenoshita et al. 2019 ; Matias-Guiu et al. 2015 ). The ACE-III, MoCA and MMSE have all been recommended by the Department of Health and the Alzheimer’s Society in the UK for inclusion as part of a comprehensive cognitive assessment in memory clinics (Ballard et al. 2015 ). However, the ACE-III can more accurately detect frontotemporal dementia, as well as the earlier stages of dementia compared to MMSE (Hsieh et al. 2013 ; Slachevsky et al. 2004 ). In addition, ACE-III is better at identifying everyday activity impairments in dementia when compared to MMSE and MoCA (Giebel and Challis 2017 ). Thus, the results from these studies would suggest that ACE-III is preferrable when compared to other screening tools. While the ACE-III is commonly used as part of a full clinical assessment, few studies have looked at how reliable the ACE-III is alone for distinguishing between dementia and MCI, and the different types of dementia.

The aim of this study is to assess how reliable the ACE-III assessment is for making a differential diagnosis between dementia, MCI and controls. Therefore, this study seeks to address the following questions: can the ACE-III help to differentiate between dementia, MCI and older adults who have not reported cognitive problems? For a person with dementia, is the ACE-III helpful in differentiating the type of dementia?

Memory service

The Northern Health and Social Care Trust (NHSCT) in Northern Ireland set up a memory service in 2013 to facilitate timely diagnosis and enable people with dementia to access appropriate supports. The NHSCT Memory Service accepts referrals for people who present with symptoms of memory problems and/or behavioural change within a clinical picture suggestive of a dementia. Since 2013, there have been over 6000 referrals to the NHSCT Memory Service, however not everyone who attends for assessment has dementia. Potential outcomes of assessment include (1) diagnosis of a specific type of dementia, (2) diagnosis of MCI, (3) another condition which causes changes in cognitive function treatable or untreatable and (4) no evidence of a physical or mental health condition. The comprehensive assessment process in the NHSCT Memory Services involves a review of the following factors:

Background information (education level, occupational history, family history of dementia, establishing whether the person can recall their own personal history)

Symptoms (do they have insight into their own symptoms, did the person self-report cognitive difficulties, did their carer report cognitive difficulties, onset of symptoms, progression of symptoms, known previous psychiatric history, current mental health problems/ stressors, sleep problems, history of self-harm/ suicidal ideation, hallucinations, delusions, psychosis)

General information (impairment in activities of daily living, can they drive, are they financially independent or do they require support, living situation)

Health information (do they require assistance with medication, any history of falls, mobility/ movement problems, difficulty hearing, problems with eyesight, if they smoke or drink alcohol)

Physical health risk factors (epilepsy, head injury, heart disease, stroke/ cerebrovascular accident/ transient ischaemic attacks, high blood pressure, recurrent infections, diabetes, peptic ulcer, high cholesterol, Asthma/ COPD, neurological)

Tests (ACE-III, Bristol Activities of Daily Living, Zarit Caregiver Burden)

A formal diagnosis can then be made by the psychiatrist based on the assessment of the aforementioned factors. Psychiatrists in the NHSCT Memory Service use ICD-10 criteria to classify dementia type. However, the ICD-10 codes are not recorded in the database.

Data provenance

This study received ethical approval from the Health Research Authority ethics board (ref: 17/NI/0142). Data were obtained from the NHSCT Memory Service. Over 3500 patient records from dementia assessments were digitised from 2013 to 2019. An overview of the person’s journey through the service is shown in Fig.  1 . Most people were referred to the memory service by their GP (93%), while others were referred by other medical professionals or mental health services (7%). Once referred, people attend for a comprehensive dementia assessment where they may receive a diagnosis of dementia, mild cognitive impairment (MCI), other diagnosis or no diagnosis (Fig.  1 ).

figure 1

Overview of the person’s journey with the NHSCT Memory Service, including those who were included and excluded from the study. Additional path shown for control participants who were not referred to the memory service but were recruited to the study (bottom left)

The outcomes of memory assessment are as follows.

No diagnosis This group received no diagnosis of dementia or MCI. These individuals may have one of a range of presentations, for example: mental health problems, CVA/ brain injury, other cognitive impairment, age-related changes in cognitive function, Parkinson’s disease, other conditions or no evidence of physical or mental health difficulty.

Those with a specific diagnosis recorded, including Alzheimer’s disease, vascular dementia, mixed dementia, Lewy body dementia, frontotemporal dementia or MCI

Dementia unspecified These people have presentations consistent with dementia, but at the time of assessment the exact type of dementia was unclear. These people are typically reviewed by a clinician at a later date and given a diagnosis.

Other type of dementia Those diagnosed with another type of dementia, not listed above. The exact type of dementia was not recorded at time of assessment.

Uncertain diagnosis This group did not receive a definitive diagnosis at the time of assessment. In the database, this was recorded as between two or more diagnoses. These people are typically reviewed by a clinician at a later date and given a diagnosis.

Participants and exclusion criteria

Data were filtered to only those that completed the ACE-III in full; therefore, those who partially completed ( n  = 214) or did not complete the assessment ( n  = 362) were excluded from the present study. All individuals in the ‘no diagnosis’ of MCI or dementia category ( n  = 351) were excluded given the range of presentations in this cohort as mentioned above. Those that were diagnosed with Lewy body dementia ( n  = 15) and frontotemporal dementia ( n  = 8) were removed due to small sample sizes. Those in the dementia unspecified category ( n  = 59), other types of dementia ( n  = 24) and those who received an uncertain diagnosis ( n  = 228) were omitted as the exact diagnosis was not recorded at the time of assessment.

Additionally, a group of older adults ( n  = 32; ≥ 65 years) who had not presented to the memory service and did not have a diagnosis of dementia were recruited via social media, word of mouth and posters in drop-in centres or community groups, to provide comparative ‘control’ data. These participants completed the same dementia assessment, including the ACE-III, administered by a trained memory service practitioner using similar protocols to those employed in the memory service. The intention was to recruit 100 individuals with no reported cognitive problems as controls for the study; however, this sample size was not achieved due to the COVID-19 pandemic.

The final study cohort included 2,331 data records on 2,176 people. A total of 2,023 (93%) people used the service only once, of which 151 (6.9%) used the service twice and 2 (0.1%) people attended three times.

Data analysis

R programming language and RStudio (version 3.6.0) were used for all data analyses. Exploratory analysis was carried out to investigate age, sex, total ACE-III score and scores for the ACE-III domains (attention, memory, fluency, language and visuospatial). ACE-III total and domain scores for the diagnostic groups were visualised using boxplots and density plots and were assessed for normality using Shapiro–Wilk tests. For all features, p  < 0.05 which suggested the data was not normally distributed. This was confirmed by visual inspection of histograms/ boxplots, indicating nonparametric testing should be applied. Kruskal–Wallis tests were performed across diagnostic groups for age, ACE-III total and domain scores, with p  < 0.05 considered to be statistically significant. Post hoc pairwise Wilcoxon rank sum tests were carried out using Bonferroni correction for multiple testing. A Chi-squared test was used to compare proportions of gender across diagnostic groups. Violin plots were produced, which combine the boxplot and density plot to better illustrate summary statistics and distribution in one plot (Hintze and Nelson 1998 ). Pairwise comparisons for total ACE-III score and ACE-III domain scores across diagnostic groups were visualised using a tile plot, with statistical significance shown for each comparison.

Plots showing estimated kernel densities were produced to compare total ACE-III score for all dementias (Alzheimer’s disease, vascular dementia and mixed dementia), MCI and controls. Based on the results of the pairwise comparisons that were significant, additional density plots were produced for a subset of the ACE-III domains across dementia diagnoses. The overlapping coefficient, which is the overlapping area under two probability density functions, was calculated for each of these comparisons.

As the maximum score differs for each ACE-III domain, scores were normalised by rescaling the data points between 0 and 100 (Eq.  1 ).

Eq.  1 : Formula for normalisation where x i is a data point (x 1 , x 2 …x n ) and x norm is a normalised data point.

For each domain, the normalised mean scores were visualised using a line plot for comparisons across diagnostic groups. The relative differences between mean scores across each of the ACE-III domains were compared for all dementias, MCI and controls. Kruskal–Wallis tests were performed to compare ACE-III domain scores for all dementias, MCI and controls, with p  < 0.05 considered to be statistically significant. Post hoc pairwise Wilcoxon Rank Sum tests were carried out using Bonferroni correction for multiple testing.

Receiver operating characteristic (ROC) curves were plotted to assess the optimal cut-off points for the detection of MCI and dementia from controls, and dementia from MCI. Maximum Youden index (Youden index = sensitivity + specificity–1) was used to determine the optimal cut-offs.

Over half of people in this study were female (62%), with a smaller proportion of males (38%), and the average age was 79.6 (SD 7.5). Age was significantly different across diagnostic groups ( p  < 0.001, Table 1 ). Post hoc testing revealed that there were no significant differences in mean age across the dementia groups; however, people diagnosed with MCI were significantly younger on average compared to the dementia groups ( p  < 0.05), and the control group were also significantly younger on average ( p  < 0.05) compared to the MCI and dementia groups. Across diagnoses, proportions of gender were significantly different ( p  < 0.001, Table 1 ). Over 70% of people diagnosed with Alzheimer’s disease were female and around 60% of people with mixed dementia, MCI and controls were female (Table 1 ). Vascular dementia was the only group with almost even split of male and female (Table 1 ).

Comparison of total ACE-III and domain scores across diagnoses

ACE-III total and domain scores were significantly different across diagnoses, with higher average scores for the MCI and control groups compared to the dementia groups (Table 1 ). Overall, the scores for the control group were much higher than the MCI and dementia groups across all domains and total ACE-III (Fig.  2 ). The distribution of ACE-III total and domain scores was similar across all dementia groups (Fig.  2 ). People with Alzheimer’s disease displayed higher median scores for total ACE-III, fluency, visuospatial and language compared to those in the mixed and vascular dementia groups (Fig.  2 ). The MCI group attained scores in between those with dementia and the controls (Fig.  2 ). There was some overlap between the lowest recorded scores for those with MCI compared to the dementia groups; however, in general the scores were centred around the upper ranges of the dementia groups (Fig.  2 ). The distribution of scores for ACE-III total and memory for the controls was bimodal while for all other ACE-III domains the distribution was negatively skewed given most of the scores were high (Fig.  2 ).

figure 2

ACE-III total and domain scores across diagnoses presented as violin plots. Centre boxplots indicate median, quartiles, whiskers and outliers

Pairwise comparisons revealed significant differences in total ACE-III and domain scores between MCI and controls; all dementia groups and controls; MCI and all dementia groups; (Fig.  3 , p  < 0.05 for all). Significant differences in scores were present between mixed dementia and Alzheimer’s disease for attention, memory and fluency (Fig.  3 ). Scores for visuospatial and language were significantly different between vascular dementia and Alzheimer’s disease groups (Fig.  3 ).

figure 3

Pairwise comparisons for ACE-III total and domain scores across diagnoses. Significance codes; ns: p  > 0.05, *: p  <  = 0.05, **: p  <  = 0.01, ***: p  <  = 0.001, ****: p  <  = 0.0001. All results adjusted for multiple testing using Bonferroni correction

ACE-III cut-off analysis

The optimal cut-off for differentiating dementia from controls based on the maximum Youden index is 71, with acceptable sensitivity (0.87) and high specificity (0.97) (Fig.  4 , Table 2 ). When distinguishing dementia from individuals with no cognitive impairment, the recommended cut-off of 88 for screening purposes yielded high sensitivity (0.99) but very poor specificity (0.48). At the lower cut-off of 82 recommended for research, specificity improved slightly (0.63) with comparable sensitivity (0.97). The optimal cut-off for distinguishing MCI from controls was 84 with high sensitivity (0.92) but low specificity (0.63) (Fig.  4 , Table 2 ). A cut-off of 61 was identified as optimal for differentiating dementia from MCI; however, the sensitivity and specificity were poor at 0.66 and 0.79, respectively (Fig.  4 , Table 2 ).

figure 4

ROC curve of ACE-III for detecting dementia, MCI and controls

Overlap in ACE-III scores across diagnoses

The overlap in scores was very small (15%) when comparing total ACE-III between all dementias and the control group (Fig.  5 ). Roughly a third of total ACE-III scores overlapped between dementia–MCI (39%) and MCI–controls (35%) (Fig.  5 ). In contrast, the estimated overlap of densities between the dementia subtypes showed a high proportion of similarity (Fig.  6 ). Comparing densities for visuospatial and language, scores overlapped by 73% and 78%, respectively, for Alzheimer’s and Vascular dementia (Fig.  6 ). Similarly, the density plots for attention, memory and fluency between Alzheimer’s and mixed dementia overlapped to an even higher degree at 83%, 89% and 86%, respectively (Fig.  6 ).

figure 5

Kernel density estimations for ACE-III total between all dementias, MCI and controls. The overlap (%) represented by the shaded area is detailed above theplot

figure 6

Kernel density estimations for ACE-III domains between dementia groups. The overlap (%) represented by the shaded area is detailed above each plot

Pattern across ACE-III domain scores

Across all dementias, MCI and controls, the pattern in normalised mean ACE scores is fairly consistent, with all groups scoring lowest in fluency and highest in language (Fig.  7 ). Mean scores were highest for controls, followed by MCI and the three dementia groups (Fig.  7 ). For the control group, the highest average score across domains was language, followed by attention, visuospatial, memory and fluency (Fig.  6 ). In contrast, the order was slightly different for all dementias and MCI in that the highest average score was language, followed by visuospatial, then attention memory and fluency (Fig.  6 ). On average, people with Alzheimer’s disease scored higher on attention, memory and fluency compared to those with mixed dementia (Fig.  7 ). The Alzheimer’s disease cohort scored better on average across visuospatial and language compared to the vascular dementia group (Fig.  7 ), and these differences in scores were statistically significant ( p  < 0.001 visuospatial, p  < 0.05 language, Fig.  3 ) (Table 1 ).

figure 7

Normalised mean scores for each ACE-III domain across diagnostic groups, ordered from highest to lowest for the control group. Rank order from highest to lowest mean score shown for all dementias, MCI and controls (bottom left)

Comparing all dementias and the control group, the largest difference in average score was seen in the memory domain (> 40%) while the smallest difference was in visuospatial (~ 20%) (Fig.  8 ). The same order was evident when comparing MCI to the control group (Fig.  8 ) although the percentage differences were much smaller (~ 10–25%). In contrast, when comparing all dementias to MCI the largest difference in average score was in the attention domain and the smallest difference was in language; however, the percentage difference in average score was small, ranging from ~ 10 to 20% (Fig.  8 ).

figure 8

Difference in normalised mean scores across ACE-III domains for all dementias, MCI and controls. Ordered from highest to lowest for dementia–control comparison)

This study utilised data from the NHSCT Memory Service. This unique resource is the first database of its kind and size and the first to examine the different patterns of performance across the ACE-III within such a diverse range of individuals with a dementia or MCI diagnosis compared with a control group.

In the NHSCT Memory Service, 62% of referrals were female. This figure is similar to the reported prevalence of dementia given 65% of people living with dementia in the UK are female (Prince et al. 2014 ). The gender ratio across different diagnostic categories varied; however, given 70% of people in the NHSCT Memory Service diagnosed with Alzheimer’s disease were female, which is relatively more than the proportional referral rate. Roughly 60% of mixed dementia and MCI diagnoses were female which highlights that there was a consistent proportion of males and females with MCI and mixed dementia. There was an almost even split of males and females diagnosed with vascular dementia, suggesting that males referred to the service are more likely to have vascular dementia than females referred to the service. This is consistent with the findings of the Rotterdam study, a large population-based study which found that regardless of age, vascular dementia was more prevalent in males compared to females (Ruitenberg et al. 2001 ).

The results of the present study demonstrate significant differences in ACE-III scores between people diagnosed with dementia, MCI and controls. People with dementia attending the NHSCT Memory Service perform significantly worse than the control group and people with MCI in terms of ACE-III total score and each of the domains. Roughly one third of total ACE-III scores overlapped between those with dementia compared to MCI and for the MCI and control groups. The overlap in total ACE-III scores between people with dementia and controls was even less (15%). These results suggest that the ACE-III is good at discriminating between dementia, MCI and people with no reported cognitive problems.

All three groups, including people with dementia, MCI and the control participants scored highest in language and lowest in memory and fluency on average. The pattern of normalised mean ACE-III domain scores highlighted an interesting pattern across groups. It was fairly consistent with the exception of attention which seemed to be disproportionately lower in both the dementia and MCI groups compared to controls. This suggests that relative impairment in attention is greater when there is a decline in cognition and this relative decline is greater in dementia than in MCI. This finding has the potential to be clinically helpful and needs further exploration.

Overall, intra-domain analysis revealed that within each of the groups of dementia, MCI and controls, there are similar abilities in the language and visuospatial domains, given there were no significant differences in scores between these domains. All groups also performed significantly lower on fluency than on language and visuospatial domains. Further analysis which looked at the intra-domain pairwise comparisons revealed some distinct patterns across these three groups. People with MCI and controls also have similar ability in the domain of attention in addition to language and visuospatial processing whereas, for people with dementia, ability in attention is significantly different from ability in the domains of language and visuospatial. This finding also has the potential to be clinically helpful and needs further exploration.

These results indicate different profiles of cognition, revealing distinct areas of cognitive decline progressing from a group reporting no cognitive problems (controls), through to those with MCI and people with dementia.

People with Alzheimer’s disease, vascular dementia and mixed dementia had similar distributions on ACE-III total and domain scores. There were statistically significant differences between those with Alzheimer’s and mixed dementia (attention, memory and fluency) and between individuals with Alzheimer’s and vascular dementia (visuospatial and language). While these differences were statistically significant, these results are not clinically significant or relevant, given that there was a high degree of overlap (> 73%) between these domain scores. Elamin and colleagues reported similar findings in their study, noting that between dementia subgroups there were few significant differences in ACE-III scores (Elamin et al. 2016 ).

The recommended total ACE-III cut-offs for differentiating early-onset dementia patients from healthy controls are 82 and 88 for research and screening, respectively (Hsieh et al. 2013 ). A recent review explored the diagnostic test accuracy of the ACE-III for dementia and found that the lower threshold of 82 provided better specificity with acceptable sensitivity (Beishon et al. 2019 ). However, the authors noted that the optimal cut-offs required future work and should be determined across a variety of settings such as secondary care services which would include the NHSCT Memory Service. Jubb and colleagues suggested a lower cut-off of 81 for better sensitivity and specificity in their sample of patients above 75 years of age presenting to a Memory Clinic in England (Jubb and Evans 2015 ). The authors also recommended taking other factors into consideration such as years of education when using ACE-III to aid diagnosis of dementia (Jubb and Evans 2015 ). In the present study, we identified a lower optimal cut-off of 71 for differentiating people with dementia from those with no cognitive impairment with acceptable sensitivity and high specificity. In practice, this means using a cut-off of 71 is highly likely to correctly predict that an individual doesn’t have dementia if dementia is not present. We also calculated an optimal cut-off score of 84 for distinguishing individuals with MCI from the control group with high sensitivity (0.92) but poor specificity (0.63). Given the control participants recruited for this study were individuals with no cognitive problems and otherwise would not normally be attending the NHSCT Memory Service, the cut-offs identified are more applicable for research purposes rather than screening.

As the ACE-III is not good at discriminating between different types of dementia, these results highlight the importance of the range of other factors that are taken into consideration when making a differential diagnosis of dementia. Other variables that have been shown to influence ACE-III scores include age and education level (Bruno and Vignaga 2019 ). Previous work has found that age significantly contributes to overall ACE-III score as those in older age groups perform worse across all domains of the ACE-III (Matias-Guiu et al. 2015 ; Cheung et al. 2015 ). Additionally, individuals with higher levels of education (> 11 years) perform better on the ACE-III compared to those with low levels of education (Matias-Guiu et al. 2015 ; Jubb and Evans 2015 ). Years of education are positively correlated with a person’s cognitive function as they age (Opdebeeck et al. 2016 ) and predict lower risk of dementia late in life (Lövdén et al. 2020 ). Some believe that increasing years of education builds cognitive reserve, which is a concept that the brain develops resilience that acts as a protective factor against loss through ageing and disease. Studies have linked this to better cognitive performance in people with dementia (Stern 2012 ), but also in other neurological diseases including Parkinson’s (Hindle et al. 2014 ) and multiple sclerosis (Santangelo et al. 2019 ).

The data confirm that the ACE-III total score alone cannot be used to diagnose dementia or distinguish between the different types of dementia. In addition, while there are some different patterns of performance across the domains of the ACE-III in the different types of dementia, it is not clear that a consistent pattern emerges to be helpful in making a decision about the specific diagnosis. This study was important because it confirms what clinicians already believed, that the ACE-III is an important tool to highlight that the person may have dementia or MCI but is not comprehensive enough to differentiate between the subtype of dementia. This is not surprising given that the ACE-III was not designed to differentiate between the range of dementias explored in this paper and instead to help differentiate between Alzheimer’s disease and frontotemporal dementia (Hsieh et al. 2013 ).


The ACE-III scores obtained on the test during the diagnostic process are the scores that were compared in the study to assess the reliability of ACE-III, and the test is only administered once. However, diagnosis is made on the basis of a comprehensive assessment as outlined in the Methods. This includes assessment based on clinical presentation, in combination with a medical report which usually comes from the GP that referred the patient to the service. The GP may have administered another cognitive screening tool such as MMSE, but this is not recorded in the NHSCT Memory Service database. Together, all of these factors are used by the psychiatrists to make a formal diagnosis. The analysis carried out did not control for age or education level, both of which have been shown to influence ACE-III scores, which is a significant limitation. The large sample sizes meant that some of the ACE-III domain scores across diagnostic groups were statistically significant; however, these results were not clinically significant. The sample size of the control group ( n  = 32) was considerably smaller than planned due to the coronavirus pandemic. Additionally, the control group were individuals recruited for the study with no cognitive impairment and normally would not be referred to the NHSCT Memory Service and thus may not be representative of people without dementia/ MCI referred to a memory service. Another limitation is the exclusion of data from certain groups. This study only looked at those who fully completed all sections of the ACE-III. There were many reasons people were unable to complete the ACE-III in full, such as tiredness, anxiety, distress, severity of cognitive difficulties, poor hearing or eyesight. A number of other groups were excluded including those with Lewy body dementia; frontotemporal dementia; unspecified diagnosis of dementia; other types of dementia and those with an uncertain diagnosis. These groups were taken out as the group sizes were not large enough for analyses or the exact diagnosis was not known at the time of assessment. Individuals who received no diagnosis of MCI or dementia were excluded as the majority of these people had other co-morbidities affecting cognitive function, and thus, they could not be included in the ‘control’ category. This study obtained data from the NHSCT Memory Service database in Northern Ireland; however, it is still only representative of data from a single location.

This study analysed data from the NHSCT Memory Service database, a unique and comprehensive data repository detailing the outcome of dementia assessments. The aim of the study was to explore the reliability of the ACE-III in differentiating between dementia, MCI and controls, and whether the ACE-III is useful for making a differential diagnosis on the type of dementia. The results of this study suggest that the ACE-III is good for differentiating between dementia and MCI; however, the test is not reliable for discriminating between Alzheimer’s disease, vascular dementia and mixed dementia. Nonetheless, the ACE-III is a useful tool for clinicians that can help to make a dementia diagnosis in combination with other factors at assessment. Future work will involve utilising the NHSCT Memory Service database to analyse ACE-III scores for those groups that were excluded from the present study and determining the impact of factors that have been shown to influence ACE-III such as age, gender and years in education.

Ballard C, Alistar B, Corbett A, et al (2015) Helping you to assess cognition: a practical toolkit for clinicians. https://www.wamhinpc.org.uk/sites/default/files/dementia-practical-toolkit-for-clinicians.pdf . Accessed 9 June 2021

Bamford C, Lamont S, Eccles M et al (2004) Disclosing a diagnosis of dementia: a systematic review. Int J Geriatr Psychiatry 19:151–169

Article   Google Scholar  

Banarjee S, Wittenberg R (2009) Clinical and cost effectiveness of services for early diagnosis and intervention in dementia. Int J Geriatr Psychiatry 24:748–754. https://doi.org/10.1002/gps.2191

Beishon LC, Batterham AP, Quinn TJ et al (2019) Addenbrooke’s cognitive examination III (ACE-III) and mini-ACE for the detection of dementia and mild cognitive impairment. Cochrane Database Syst Rev. https://doi.org/10.1002/14651858.CD013282.pub2

Bruno D, Vignaga SS (2019) Addenbrooke’s cognitive examination III in the diagnosis of dementia: a critical review. Neuropsychiatr Dis Treat 15:441–447. https://doi.org/10.2147/NDT.S151253

Cheung G, Clugston A, Croucher M et al (2015) Performance of three cognitive screening tools in a sample of older New Zealanders. Int Psychogeriatrics 27:981–989. https://doi.org/10.1017/S1041610214002889

Elamin M, Holloway G, Bak TH, Pal S (2016) The utility of the addenbrooke’s cognitive examination version three in early-onset dementia. Dement Geriatr Cogn Disord 41:9–15. https://doi.org/10.1159/000439248

Folstein MF, Folstein SE, McHugh PR (1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12:189–198. https://doi.org/10.1016/0022-3956(75)90026-6

Giebel CM, Challis D (2017) Sensitivity of the mini-mental state examination, montreal cognitive assessment and the addenbrooke’s cognitive examination III to everyday activity impairments in dementia: an exploratory study. Int J Geriatr Psychiatry 32:1085–1093. https://doi.org/10.1002/gps.4570

Graham NL, Hodges JR (2004) Distinctive cognitive profiles in Alzheimer’s disease and subcortical vascular dementia. J Neurol Neurosurg Psychiatry 75:61–71

Google Scholar  

Grundman M, Petersen RC, Ferris SH et al (2004) Mild cognitive impairment can be distinguished from alzheimer disease and normal aging for clinical trials. Arch Neurol 61:59–66. https://doi.org/10.1001/archneur.61.1.59

Hindle JV, Martyr A, Clare L (2014) Cognitive reserve in Parkinson’s disease: a systematic review and meta-analysis. Park Relat Disord 20:1–7

Hintze L, Nelson RD (1998) Violin plots: a box plot-density trace synergism. Am Stat 52:181–184

Hodges JR, Larner AJ (2016) Addenbrooke’s cognitive examinations: ACE, ACE-R, ACE-III, ACEapp, and M-ACE. In: Cognitive Screening Instruments: A Practical Approach. Springer International Publishing, pp 109–137

Hsieh S, Schubert S, Hoon C et al (2013) Validation of the addenbrooke’s cognitive examination III in frontotemporal dementia and alzheimer’s disease. Dement Geriatr Cogn Disord 36:242–250. https://doi.org/10.1159/000351671

Jubb MT, Evans JJ (2015) An investigation of the utility of the addenbrooke’s cognitive examination III in the early detection of dementia in memory clinic patients aged over 75 years. Dement Geriatr Cogn Disord 40:222–232. https://doi.org/10.1159/000433522

Karantzoulis S, Galvin JE, Manzotti E (2011) Distinguishing Alzheimer’s disease from other major forms of dementia. Expert Rev Neurother 11:1579–1591. https://doi.org/10.1586/ern.11.155

Li X, Yang L, Yin J et al (2019) Validation study of the chinese version of addenbrooke’s cognitive examination iii for diagnosing mild cognitive impairment and mild dementia. J Clin Neurol 15:313–320. https://doi.org/10.3988/jcn.2019.15.3.313

Lövdén M, Fratiglioni L, Glymour MM et al (2020) Education and cognitive functioning across the life span. Psychol Sci Public Interes 21:6–41. https://doi.org/10.1177/1529100620920576

Mathuranath PS, Nestor PJ, Berrios GE et al (2000) A brief cognitive test battery to differentiate Alzheimer’s disease and frontotemporal dementia. Neurology 55:1613–1620. https://doi.org/10.1212/01.wnl.0000434309.85312.19

Matias-Guiu JA, Fernández de Bobadilla R, Escudero G, et al (2015) Validation of the Spanish version of Addenbrooke’s Cognitive Examination III for diagnosing dementia. Neurol (English Ed 30:545–551. https://doi.org/10.1016/j.nrleng.2014.05.001

Nasreddine ZS, Phillips NA, Bédirian V et al (2005) The montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53:695–699. https://doi.org/10.1111/j.1532-5415.2005.53221.x

Opdebeeck C, Martyr A, Clare L (2016) Cognitive reserve and cognitive function in healthy older people: a meta-analysis. Aging. Neuropsychol Cogn 23:40–60

Pratt R, Wilkinson H (2003) A psychosocial model of understanding the experience of receiving a diagnosis of dementia. Dementia 2:181–199. https://doi.org/10.1177/1471301203002002004

Prince M, Bryce R, Ferri C (2011) World Alzheimer Report 2011: The benefits of early diagnosis and intervention; World Alzheimer Report 2011: The benefits of early diagnosis and intervention

Prince M, Knapp M, Guerchet M, et al (2014) Dementia UK: Update Second Edition

Ruitenberg A, Ott A, Van Swieten JC et al (2001) Incidence of dementia: does gender make a difference? Neurobiol Aging 22:575–580. https://doi.org/10.1016/S0197-4580(01)00231-7

Santangelo G, Altieri M, Enzinger C et al (2019) Cognitive reserve and neuropsychological performance in multiple sclerosis: a meta-analysis. Neuropsychology 33:379–390. https://doi.org/10.1037/neu0000520

Slachevsky A, Villalpando JM, Sarazin M et al (2004) Frontal assessment battery and differential diagnosis of frontotemporal dementia and Alzheimer disease. Arch Neurol 61:1104–1107. https://doi.org/10.1001/archneur.61.7.1104

Stern Y (2012) Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 11:1006–1012

Takenoshita S, Terada S, Yoshida H et al (2019) Validation of addenbrooke’s cognitive examination III for detecting mild cognitive impairment and dementia in Japan. BMC Geriatr 19:123. https://doi.org/10.1186/s12877-019-1120-4

Wang BR, Ou Z, Gu XH et al (2017) Validation of the Chinese version of addenbrooke’s cognitive examination III for diagnosing dementia. Int J Geriatr Psychiatry 32:e173–e179. https://doi.org/10.1002/gps.4680

Download references


The authors acknowledge the Atlantic Philanthropies, Department of Health and The Executive Office, Northern Ireland, for funding as part of the Dementia Analytics and Research User Group project (Reference Number 17-F-1801). The authors are grateful to the staff from the NHSCT Memory Service in Northern Ireland and control participants for their time, effort and support in participating in this research.

Author information

Authors and affiliations.

Faculty of Computing, School of Computing, Engineering and the Built Environment, Ulster University, Newtownabbey, Coleraine, UK

C. Potts,  R. B. Bond &  M. D. Mulvenna

Memory Service, Northern Health and Social Care Trust, Antrim, UK

J. Richardson, P. Zvolsky, M. Harvey & F. Duffy

Faculty of Life and Health Sciences, School of Biomedical Sciences, Ulster University, Coleraine, UK

R. K. Price & C. F. Hughes

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to C. Potts .

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Responsible Editor: Matthias Kliegel.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and Permissions

About this article

Cite this article.

Potts, C., Richardson, J., Bond, .B. et al. Reliability of Addenbrooke's Cognitive Examination III in differentiating between dementia, mild cognitive impairment and older adults who have not reported cognitive problems. Eur J Ageing 19 , 495–507 (2022). https://doi.org/10.1007/s10433-021-00652-4

Download citation

Accepted : 11 September 2021

Published : 22 September 2021

Issue Date : September 2022

DOI : https://doi.org/10.1007/s10433-021-00652-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative


  1. The Addenbrooke's Cognitive Examination Revised (ACE-R): a brief

    ace 111 assessment

  2. Addenbrooke's cognitive examination III: diagnostic utility for mild cognitive impairment and

    ace 111 assessment

  3. Adam Watson's Edtech Elixirs: ACE Test

    ace 111 assessment

  4. ACE

    ace 111 assessment

  5. What Job Seekers Need to Know About Pre-Employment Testing

    ace 111 assessment

  6. ACE Assessment Report 14 Feb 2014.ppt (1)

    ace 111 assessment


  1. Fibonacci Time and Price Analysis with Carolyn Boroden: Part 1

  2. Free Webinar: Best Practice Succession Planning



    ADDENBROOKE'S COGNITIVE EXAMINATION – ACE-III. English Version A (2012) ... This test should be done if the subject failed to recall one or more items above.

  2. Addenbrooke's cognitive examination III in the diagnosis of dementia

    Addenbrooke's cognitive examination III is a screening test that is composed of tests of attention, orientation, memory, language

  3. Dementia test

    Addenbrooke's Cognitive Examination - III (ACE - III).


    ADDENBROOKE'S COGNITIVE EXAMINATION – ACE-III ... This test should be done if the participant failed to recall one or more items above.

  5. Addenbrooke's Cognitive Examination

    The Addenbrooke's Cognitive Examination (ACE) and its subsequent versions are neuropsychological tests used to identify cognitive impairment in conditions

  6. Addenbrooke's Cognitive Examination–Third Edition Predicts

    ACE-III subscale scores predicted performance on neuropsychological measures assessing similar constructs. However, overall performance on

  7. Applicability of the ACE-III and RBANS Cognitive Tests for ...

    The ACE-III tests five cognitive domains: attention, memory, verbal fluency, language, and visuospatial function. Individual sub-test scores can

  8. Addenbrooke's Cognitive Examination III (ACE‐III) and mini‐ACE for

    The ACE‐III has 21 questions, with a total score of 100. The test is performed with the patient who presented with, or is suspected to have

  9. Addenbrooke's Cognitive Examination-III

    Testing shows that ACE-III cognitive domains correlate significantly with standardized neuropsychological tests used in the assessment of

  10. Reliability of Addenbrooke's Cognitive Examination III in

    Diagnosing dementia can be challenging for clinicians, given the array of factors that contribute to changes in cognitive function.