https://orcid.org/0000-0002-9237-3749
Figures
Abstract
The Addenbrooke’s Cognitive Examination III is one of the most widely used tests to assess cognitive impairment. Although previous studies have shown adequate levels of diagnostic utility to detect severe impairment, it has not shown sensitivity to detect mild decline. The aim of this study was to evaluate the psychometric properties of Addenbrooke’s Cognitive Examination III in a large sample of elderly people through Item Response Theory, due to the lack of studies using this approach. A cross-sectional study was conducted with 1164 people from the age of 60 upwards, of which 63 had a prior diagnosis of Alzheimer dementia. The results showed that, globally, the Addenbrooke’s Cognitive Examination III possesses adequate psychometrics properties. Furthermore, the information function test shows that the subscales have different sensitivity to different levels of impairment. These results can contribute to determining patterns of cognitive deterioration for the adequate detection of different levels of dementia. An optimized version is suggested which may be an economic alternative in the applied field.
Citation: Calderón C, Beyle C, Véliz-García O, Bekios-Calfa J (2021) Psychometric properties of Addenbrooke’s Cognitive Examination III (ACE-III): An item response theory approach. PLoS ONE 16(5): e0251137. https://doi.org/10.1371/journal.pone.0251137
Editor: Francesca Chiesi, University of Florence, ITALY
Received: October 7, 2020; Accepted: April 21, 2021; Published: May 6, 2021
Copyright: © 2021 Calderón et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: Fondo Nacional de proyectos de investigación y desarrollo en salud (FONIS). Ministerio de Salud-Conicyt. Chile.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Dementias are one of the main causes of malfunction and dependency in adulthood [1]. According to the World Health Organization [2], in 2015 50 million people suffered from some kind of dementia, and the number is expected to triple by 2030. This neurocognitive impairment affects not only those who suffer from it but also their caregivers, family members, the community, and general society.
Previous evidence has presented the benefits of early detection and treatment of dementia. These benefits are not only of a clinical nature [3] but also economic [4–6], social [7], and in terms of public health [8,9], both for the patient and their relatives and caregivers. Therefore, it is crucial that professionals, both specialists and non-specialists, have access to accurate, time-saving, and easy-to-use screening tools that allow early detection of dementia.
The Classical Test Theory (CTT) traditionally used for psychometric analysis of cognitive tests, uses the sum of raw scores both for diagnosis and in order to estimate the level of cognitive decline in people. This method is easy to apply and is based on the assumption that all items reflect a common trait or ability from which the level of impairment can be inferred [10].
However, that assumption can be seen as unrealistic, since it assumes that all items have the same precision and difficulty. The direct sum of the raw scores does not consider the differences in the metric properties of the items, and the information provided by the different response patterns remains unknown. In other words, two people with the same number of correct responses could have different levels and/or paths of decline depending on the specific items they have hit.
To overcome these and other limitations of CTT, the Item Response Theory (IRT) was presented several years ago [11–16]. Unlike the CTT—which associates the item’s response to the total score—the IRT relates the probability of success of each item with the estimation of the latent trait through a logistic function called Item Characteristic Curve (ICC). There are different IRT models depending on the number of parameters that are estimated to define the ICC. The most widely used is the 2-parameter logistic model. This ICC model is defined by two parameters: difficulty (b-parameter) and discrimination (a-parameter). These parameters provide useful information about technical properties of the ICC.
The b-parameter corresponds to the level of the latent trait at which the probability of success is 50% (p = .5). On the other hand, a-parameter informs on the slope of the curve at the point of the trait continuum where the b-parameter is located. It informs about how well the item manages to differentiate between subjects on a particular area of the trait.
The overall performance of the test can be evaluated through the information function. It shows the test’s acuteness/discernment capability, which can be estimated for each level of the latent trait. Therefore, IRT provides information for determining deterioration sequences/paths and at the same time increasing the test’s sensitivity to certain impairment levels [17–20]. Furthermore, IRT provides global and item related goodness-of-fit indices, allowing an adequate evaluation of the measures.
In order to establish a hierarchy of items and scales of difficulty it could be helpful to determine the sequence of deterioration or the level of dementia severity based on the patient’s answers. Following different impairment trajectories could help to diagnose different types of dementia [19]. Finally, knowing the discrimination capability of the items allows to examine their sensitivity at certain levels of impairment.
Despite the mentioned advantages, there is little prior evidence of the psychometric properties of tests that evaluate cognitive impairment using IRT’s framework. Existing research has evaluated a small number of tests [10,18–24], or have used non-parametric IRT models [25–31].
Cognitive screening for dementia faces many difficulties, especially in early stages of impairment. Considering that cognitive performance change is a key symptom within the diagnostic criteria [32,33], it is necessary for tests to possess adequate sensitivity at different stages of cognitive deterioration in order to detect unusual patterns in cognitive performance. A competent cognitive assessment must have the capability to distinguish between age-related decline and dementia-related impairment, and differentiating mild from more severe stages of the disease [33]. Additionally, it should help to define a cognitive performance profile that provides information on the different cognitive domains, allowing the detection of different paths of deterioration for diagnosis and treatment [34].
There is currently a large number of cognitive assessment tests to screen for dementia. Among these, the Mini-Mental State Examination [35] is the most widely used. Despite this, several studies report its limitations as a screening test, among which the following can be found: it presents low sensitivity to detect dementia in the initial stages [36,37], shows little diagnostic utility in primary care [36], presents difficulties in detecting dementias other than Alzheimer’s [37], it has imprecise cut-off scores that lead to inaccurate diagnostic classifications [38] and has limited evidence of generalization validity, being affected by sociocultural elements [39].
One of the tests that attempts to overcome these limitations is the Addenbrooke’s Cognitive Examination (ACE) [40]. The third version of the test (ACE-III) has shown adequate diagnostic effectiveness for detecting Alzheimer’s and other types of dementia [41,42]. It provides information from five cognitive domains (attention, memory, language, verbal fluency, and visuospatial abilities), which makes it an excellent alternative to obtain a deterioration profile [43]. Despite several studies evaluating the validity of ACE-III and its adaptation to a large number of languages and cultural contexts [44], most of these studies focus on analyzing its diagnostic utility and its relation to other tests or variables [45–54]. At present, few studies have conducted a psychometric analysis of ACE-III from parametric IRT approaches [55].
Therefore, the aim of this study is to perform a psychometric evaluation of the ACE-III from a 2-parameter IRT model. The results of this research will provide evidence about the metric properties of the test. The results of the estimation of the IRT model could help to determine levels of impairment progression, to determine the most sensitive items and subscales to certain levels of impairment, and to develop profiles of progression.
Materials and methods
The study was approved by the Scientific Ethics Committee of the Universidad Católica del Norte, under the resolution 004/2018. Informed consent and/or consent was obtained in writing from all research participants.
Participants
A cross-sectional study was conducted with 1164 people from the age of 60 upwards. Writing and reading were evaluated. Regarding sex, 288 (26.2%) participants were men and 810 were women (73.8%). The mean age was 71.8 years (SD = 7.9) and the mean years of schooling was 10.9 (corresponding approximately to complete secondary education). 63 participants had a prior diagnosis and treatment of Alzheimer dementia, confirmed in secondary specialty care (geriatrics, neurology, and/or psychiatry) that was periodically registered in their medical history. The diagnosis was established by means of scores equal to or greater than 4 points in the Global Deterioration Scale (GDS). The time of evolution of the disease was not considered due to differences in the time of diagnosis and the course of the disease. The rest of the participants did not present a medical diagnosis that could affect cognitive functioning. Exclusion criteria considered the absence of severe sensory disturbances (without correction), consciousness, and other medical conditions different from dementia that could affect cognitive functioning. A suspension evaluation criterion was considered in three cases in which comprehension of the task’s instructions was affected (different from alteration of consciousness).
Measures
The ACE-III is a hetero-applied test for optimal performance. The Chilean version of the test was used [53] which consists of 81 items and is composed of five subscales: orientation-attention (18 items and 18 points), memory (25 items and 26 points), verbal fluencies (2 items and 14 points), language (27 items and 26 points), and visuospatial skills (11 items and 16 points), with 100 being the maximum total score. According to its latest report [53], the test presents an α-Cronbach = 0.87, establishing 86 points (98.5% sensitivity and 82% specificity) as a cut-off score for dementia detection.
Procedures
Recruitment was carried out in the primary healthcare network (PHC) in social groups of older people, and through a public notice through the National Service for the Elderly of Chile. The evaluations were carried out by a team of three psychologists and a neuropsychologist with training in cognitive evaluation.
Data analysis
To contrast the assumptions of unidimensionality and local independence necessary for the adequate estimation of the IRT models, the total scale and the subscales have been analyzed using Confirmatory Factor Analysis (CFA). In all analyses we have specified a one-factor model. To estimate these models, we have used the Weighted Last Square with mean and variance adjusted (WLSMV) estimation method available in the Mplus software (v.8) which is recommended for dichotomous variables. Once the unidimensionality of the data had been evaluated, the estimation of the 2-parameter IRT models was carried out with the Marginal maximum likelihood estimation method and the MHRM algorithm was used. Finally, for the goodness-of-fit evaluation of the models, we have used the fit indices commonly used in research practice. For the global fit, we have used the M2 statistic and the derived CFI, TLI and RMSEA indices. In the case of the local fit of the items, we have used the S−χ2 and RMSEA statistics. All analyses were carried out on the total sample of participants.
Results
As previously mentioned, the study attempted to evaluate the psychometric properties of the ACE-III test from an IRT framework. The results are presented in three sections. First, the results of the goodness-of-fit of each subscale to a one-factor model through Confirmatory Factor Analysis (CFA). The goodness-of-fit to data to a one-factor model allows us to check the assumptions of unidimensionality and local independence, which are needed for the proper estimation of IRT models. Second, the results of the estimation of parameters for the IRT models, together with the goodness-of-fit indices of the items, and the global fit indices of the subscales. Additionally, a revised version for each subscale is presented, which have been obtained by selecting the items that presented an adequate fit to the IRT model.
Confirmatory factor analysis
As mentioned above, we have evaluated the unidimensionality assumption through a CFA for each subscale. The Weighted Least Square with mean and variance adjusted estimation method (WLSMV) was used for the estimation of the models, which is appropriate for categorical variables, using the statistical software Mplus v.8. In order to evaluate the parameters fit, in addition to the χ2 statistic, the most used goodness-of-fit indices in psychometry (RMSEA, TLI and CFI) were obtained. The results are presented in Table 1.
The results of Orientation and Attention subscales are presented separately, because a two-factor model shows a significant improvement in the fit compared to the one-factor model. These findings have been presented in previous studies [51]. Goodness-of-fit indices show excellent fit to all subscales. Only one of the indices in the Memory subscale shows a poor fit (RMSEA = .085). Despite this, the results allow us to assume the unidimensionality of the subscales, since previous studies have shown that IRT models are robust enough to non-excessive violations of this assumption [56,57].
Item response theory
To assess each subscale from the IRT framework, a logistic model of 2-parameters was adopted. The Marginal maximum likelihood (ML) estimation method [58], available in the Mirt package for R software [59]. Item fit was assessed obtaining the S−χ2 statistic, which tests the hypothesis of equality between the model’s predicted response probabilities and the observed response frequencies. Probabilities associated with S−χ2 higher than .05 indicated an adequate fit of the item parameters. Additionally, the RMSEA related to S−χ2 was obtained in order to evaluate the magnitude of lack of fit. To obtain a revised version, items with lack of fit were deleted. For both the original and the revised version, we present the global goodness-of-fit statistics based on the M2 [60]. The results of the revised version are presented below. the reader can consult the supplements for the results of the full versions of the subscales.
a) Orientation subscale.
The S−χ2 statistic shows that “date” and “commune”(In Chile, the smallest administrative division governed by a mayor assisted by a municipal council) (see S1 Table) present a significant lack of fit (p<.05). In this regard, it might be suggested that failure in both items could be more related to educational and occupational differences than to cognitive impairment (see discussion). Based on the above, these items have been removed from the revised version. The Table 2 shows the estimated parameters for the proposed Orientation subscale.
The values of a-parameters show these items with a high discrimination, mostly over 2. The b-parameters indicate the highest discrimination of the items between -2.5 and -1.0 of the trait level (θ). Thereby, the Orientation subscale reaches its maximum discrimination within the 25% of people with lower performance, which is expected in a test that evaluates cognitive impairment.
As observed, values for this revised version are very similar to the original. All items present statistical fit to the model (p>.05).
Table 3 displays the fit indices of all the subscales to the IRT model. The first two rows show the fit of the original version and the proposed revised version of the Orientation subscale. Although both models present good fit indicators, the revised version shows a significant fit improvement (Δχ2 = 1838.979; p<.05).
b) Attention subscale.
Similarly as the Orientation subscale, the Attention subscale items present high levels of discrimination and difficulty parameters between -2.5 and 0 (see Table 4). Further, the S−χ2 statistic identifies three items (key, rest1, and rest4) with lack of fit to data. Removing these items would produce the subscale to contain an insufficient number of items for an adequate estimation of the person parameter. Additionally, two of these correspond to items of progressive presentation (rest1 and rest 4: progressive subtraction), which would advise against their deletion. Finally, the global fit indices show a clear lack of goodness-of-fit (RMSEA>.08; CFI < .95; TLI < .95). Overall, the Attention subscale shows no evidence of global goodness-of-fit.
c) Memory subscale.
The goodness-of-fit indices shows that 5 out of 26 items of Memory subscale present lack of fit (see S2 Table). Assuming that the items Declarative memory, Episodic free recovery, and Episodic recognition are related to genuine memory deficits [61], they represent stronger evidence of the specific memory processes in the context of dementia screening [61–63], we evaluated a revised version of the scale deleting the Working memory item and not scoring Episodic encoding items (Table 5).
Table 5 shows the high discrimination of this subscale. Regarding their difficulty, the items Episodic free recognition and Declarative memory reach values around -1, while the difficulty of Episodic free recovery are closer to the mean (θ = 0). The Table 5 show that, with the exception of one item (“recognition 68”), all the items show adequate fit. Although both versions of the subscale (original and revised) have adequate global goodness-of-fit, the revised version might be seen as a more parsimonious model.
Based on the values of the estimated b-parameters of the different tasks, the results show that items of the Episodic free recovery task have greater sensitivity in mild levels of impairment (b = ±0), while items in the tasks of Declarative memory and Episodic recognition have greater sensitivity at more severe levels of deterioration (b = −1). These findings support the possibility of establishing performance-based impairment levels among different memory tasks.
d) Language subscale.
The estimates for the revised version of Language subscale are shown in Table 6. The results of the full version are presented in the supplements.
Because three items scored 0, 1 or 2 points, they have been recoded into dummy variables (repeat the sentence, write a sentence and repeat words). The a-parameters show that all items have high discrimination, while b-parameters indicate maximum discrimination between trait levels from -2.5 to 0. Regarding the fit of the model, five items show lack of fit: “repeat words 1”, “kangaroo”, “rhino”, “accordion” and “repeat word 4” (see supplement). The lack of fit of “repeat words 1” suggests that conceding full score in case of success, and zero points in case of at least one failure, is more appropriate than using partial scores. For the rest of the items, a cultural/educational effect is likely to affect their fit (e.g. limited knowledge of the animals, or the accordion and bandoneon lookalike) and acquiescence effect might be found (a hippopotamus appears in a previous task, which is a common error for the rhino item). By deleting these items, the fit indices improve reaching an adequate goodness-of-fit, both to item and model levels (Table 3).
e) Viso-construction subscale.
Table 3 shows the global lack of fit of the subscale (CFI>.95; TLI>.95; RMSEA < .06). As the items of the subscale involve two well-differentiated tasks (visual recognition and visual construction), we have specified a different model for both sets of items. The estimated parameters of both models are presented in the Table 7.
Two important results may be highlighted. First, all the Viso-construction items present a lack of fit to the model. In contrast, in the Visual Recognition task, four out of eight items show a lack of fit, but none of them present an excessive misfit (RMSEA < .06). Second, the global fit indices show that only Visual Recognition presents an adequate fit. These results suggest that a revised version of this subscale should comprise only Visual recognition items, in order to avoid the Viso-construction items’ lack of goodness-of-fit.
f) Information function.
In order to determine the sensitivity of the ACE-III’s revised subscales for the different levels of cognitive decline, the information function of each subscale has been estimated. The information function corresponds to an indicator of the test’s level of precision (reliability) or discrimination capacity. Once the information function is calculated for each interval of the trait level, the information function curve is obtained. Fig 1 compares the information curves of the Orientation, Language, Memory and Visual recognition subscales.
Two findings may be highlighted. First, the Memory subscale presents higher levels of sensitivity at mild levels of the trait. In addition, it is the subscale that presents a greater curve amplitude compared to the rest of the subscales. This last result was mainly due to the differences in the difficulty parameters between the Episodic free recovery task and the Declarative memory and Episodic recognition tasks. These findings support the use of the memory subscale for the detection of mild cognitive impairment. Second, the location of the information curve peaks with respect to the continuum of trait level may be indicating a decline pattern, which begins with memory decline, then with language and orientation simultaneously, and finally ending with visual recognition decline.
Discussion
The purpose of this study was to evaluate the psychometric properties of ACE-III from the framework of IRT in a large sample of older people. This objective was set due to two reasons. First, the lack of studies analyzing the psychometric properties of the measuring of cognitive impairment in general, in the ACE-III. Second, the advantages that IRT has over classic approaches. We were interested in the possibility of determining patterns and/or sequences of deterioration from the set of parameters of the subscales, contributing with evidence towards the development of indicators for the detection of mild cognitive deterioration, as well as the possibility of constructing profiles which can contribute towards the development of indicators for the differential diagnosis of different types of deterioration.
Concerning the overall results, three findings are noteworthy. First, the presence of two well-differentiated latent factors for the orientation/attention subscale. This result suggests that it would be best to obtain scores for the orientation tasks separately for an adequate evaluation of cognitive performance.
Additionally, the analysis of the attention tasks’ items shows a lack of model fit. From a theoretical point of view, it is likely that lack of model fit is related to the complexity of an adequate neuropsychological assessment of attention [64–66]. This subscale is mainly composed of subtraction-related items which involve arithmetic skills that depend on neural pathways other than those related to attention [67,68]. Additionally, the remaining three items (repeat lemon-key-door) is a very small number of items to estimate the subjects’ trait level adequately. Therefore, it is recommended that the subscale of attention in the ACE-III be used with caution until there are works that propose more adequate forms for assessing attention.
Second, in relation to the pattern of cognitive deterioration, the information curves by subscales show a clear pattern that begins with the deterioration of memory capacity (especially vis-à-vis the tasks of episodic free recovery) followed by the deterioration in orientation and language capacity and, finally, problems in visual recognition. Associated with these findings, the greater sensitivity shown by the memory subscale in the middle levels of the latent trait suggest the use of this subscale—comparing it with the performance in the rest of the subscales—as an indicator to differentiate mild impairment from more severe states. More empirical evidence is needed to support the presence of this possible pattern of impairment, as well as to support the diagnostic utility of the memory scale in detecting mild cognitive impairment.
Regarding the results by subscale and the proposal of corrected versions, a detailed analysis is necessary.
Third, the location of the information curve peaks with respect to the continuum of trait level may be indicating a decline pattern, which begins with memory decline, then with language and orientation simultaneously, and finally ending with the visual recognition decline.
Orientation
As for the Orientation scale, two of its items have malfunctioned. Although this could be explained from an occupational and/or cultural point of view, their exclusion does not diminish but rather increases the fit of the subscale. As for the item "date”, due to the participants’ demographic characteristics and age, it is possible that in most of their occupational activities there is no demand for the use of this information. As for "commune," since it is an administrative district, it requires some local administrative policy knowledge. Therefore, this item may have differential functioning due to educational levels. In both cases, not getting one of the items right could not necessarily be associated with a loss of orientation due to cognitive deterioration. We recommended not to consider these items in the total score of the subscale.
Memory
We propose an alternative model by eliminating the items of the Working memory and Episodic coding tasks considering, on the one hand, the theoretical proposals [61–63] and, on the other, the results of the analysis. They showed a significant improvement of the global fit and a misfit was found in only one of its items.
According to the proposal that the tasks of episodic free recovery, declarative memory, and recognition correspond to indicators more suitable for memory function assessment [61–63], and to the possibility that the tasks of working memory and episodic coding could contribute to the lack of adjustment of the model, we have assessed a modified version of the memory subscale for our analyzes, composed only of the Episodic free recovery, Declarative memory, and Episodic recognition tasks.
As for the evocation of the three words (lemon-key-door), the characteristics of the task do not allow us to differentiate whether the person’s response depends on his or her working memory or on the evocation itself. On the contrary, on the items related to the repetition of name and address (Miguel González), the instructions involve sequential processes of repetition, similar to how other tests are widely used to evaluate episodic memory do [61]. From a statistical point of view, the model presents a significant improvement when the lemon-key-door items and the first item score of Miguel González are excluded. Nevertheless, we recommend maintaining the application of the repetition phase of Miguel González without scoring it, since it is required to assure the information codification process for its later evocation evaluation.
As we have previously mentioned, the values of the b-parameters indicate the point of the trait level at which the probability of hitting becomes greater than .5. Because the b-parameter is found in the trait continuum metric (theta), it is possible to determine the level of cognitive impairment in which people fail in their response [17–20].
Based on the values of the estimated b-parameters of the different tasks, the results show that the items of the Episodic free recovery task have greater sensitivity in mild levels of impairment (b = 0, mean of trail level), while items in the tasks of Declarative memory and Episodic recognition have greater sensitivity at more severe levels of deterioration (b = −1, one standard deviation below the mean). These findings support the possibility of establishing performance-based impairment levels against different memory tasks.
Language
As for the language subscale, only 4 of the 25 items showed a lack of fit. In the case of the word repetition task, we recommend modifying its score by coding it in a dichotomous way (all correct words versus failing in at least one) since the intermediate score does not contribute to discrimination in evaluating the trait. We recommend that the remaining three items (rhinoceros, kangaroo and accordion) be eliminated or modified, given the possible cultural effect on their functioning.
Visuospatial
The Visuospatial subscale showed an apparent global lack of fit. Given that the tasks of visoconstruction and recognition, although related, correspond to clearly differentiated tasks [69], we decided to specify both models separately. The results showed a lack of fit on the visoconstructive task, contrary to the recognition tasks, which showed an adequate fit. Developing studies that can improve the functioning of visoconstructive tasks and evaluate their contribution to visuospatial performance assessment is recommended.
There are some limitations to this study that are possible to highlight. Despite the use of a large sample of participants and that people with a diagnosis of dementia correspond to a similar percentage to the prevalence of dementia in the general population, it is possible that the small number of people with severe deterioration may partially affect the results. More studies are needed to consider a larger number of participants with severe levels of impairment. Given that this study only considered participants with a diagnosis of Alzheimer’s type dementia, subsequent studies should consider other types of dementias for the psychometric evaluation of the proposed version of the test under different conditions, and evaluate the possibility of detecting differential patterns of deterioration.
Although this study’s results are not conclusive, they present significant contributions to improving the overall functioning of the ACE-III. The findings presented can be beneficial for modifying and/or eliminating certain items from the subscales. Special attention is needed in the Attention subscale and the tasks of construction of the visuospatial subscale.
A special contribution are the findings regarding the supposed presence of a pattern of deterioration shown in the analysis of information function curves and the possibility of using the Memory subscale, especially the task of episodic free recovery, which can become an alternative for the detection of mild cognitive impairment. Additionally, obtaining scores by subscales can become a good alternative for elaborating deterioration profiles that can be transformed into indicators for the differential diagnosis for different types of courses of dementia. Studies of diagnostic utility are necessary to accumulate evidence to support this type of diagnostic use.
Supporting information
S1 Table. Parameters estimated and item fit of full version of orientation subscale.
https://doi.org/10.1371/journal.pone.0251137.s001
(DOCX)
S2 Table. Parameters estimated and item fit of full version of memory subscale.
https://doi.org/10.1371/journal.pone.0251137.s002
(DOCX)
S3 Table. Parameters estimated and item fit of full version of language subscale.
https://doi.org/10.1371/journal.pone.0251137.s003
(DOCX)
S4 Table. Parameters estimated and item fit of full version of visual construction subscale.
https://doi.org/10.1371/journal.pone.0251137.s004
(DOCX)
References
- 1. Hildreth KL, Church S. Evaluation and management of the elderly patient presenting with cognitive complaints. Medical Clinics of North America. 2015; 99 (2): 311–335. pmid:25700586
- 2. Dua T, Seeher KM, Sivananthan S, Chowdhary N, Pot AM, Saxena S. World Health Organization’s global action plan on the public health response to dementia 2017–2025. Alzheimer’s & Dementia. 2017; 13(7): 1450–1451.
- 3. Panegyres P., Berry R., & Burchell J. (2016). Early Dementia Screening. Diagnostics, 6(1), 6. pmid:26838803
- 4. Gil Gregorio P. Criterios diagnósticos. Beneficios del diagnóstico precoz. Revista Española de Geriatría y Gerontología. 2016; 51(1): 7–11.
- 5. Maresova P, Mohelska H, Kuca K. Social and family load of Alzheimer’s disease. Applied Economics. 2015; 48(21): 1936–1948.
- 6. Barnett J., Lewis L., Blackwell A., & Taylor M. (2014). Early intervention in Alzheimer’s disease: a health economic study of the effects of diagnostic timing. BMC Neurology, 14(1). pmid:24885474
- 7. Kenigsberg P., Aquino J., Bérard A., Gzil F., Andrieu S., & Banerjee S. et al. (2015). Dementia beyond 2025: Knowledge and uncertainties. Dementia, 15(1), 6–21. pmid:25740575
- 8. Llibre J, Gutierrez R. Demencias y enfermedad de Alzheimer en América Latina y el Caribe. Revista Cubana Salud Pública [Internet]. 2014; (40)3: 378–87. Available from: http://scielo.sld.cu/scielo.php?script=sci_arttext&pid=S0864-34662014000300008&lng=es&nrm=iso&tlng=es.
- 9.
Global action plan on the public health response to dementia 2017–2025. Geneva: World Health Organization; 2017. Licence: CC BY-NC-SA 3.0 IGO. [Ebook].
- 10. Wouters H, van Gool WA, Schmand B, Lindeboom R. Revising the ADAS-cog for a More Accurate Assessment of Cognitive Impairment. Alzheimer Disease & Associated Disorders. 2008; 22(3): 236–244. pmid:18580585
- 11.
Hambleton RK, Swaminathan H. Item Response Theory. Netherlands: Springer; 1985.
- 12. Drasgow F, Parsons CK. Application of unidimensional item response theory models to multidimensional data. Applied Psychological Measurement.1983; 7(2): 189–99.
- 13.
Lord FM. Applications of Items Response Theory to Practical Testing Problems. New Jersey: LEA; 1980.
- 14.
Lord FM, Novick MR. Statistical Theories of mental tests score. Reading, MA., Addison-Wesley; 1968.
- 15.
Rasch G. Probabilistic models for some intelligence and attainments tests. Copenhagen: Danish Institute for Educational Research; 1960.
- 16.
Van der Linden WJ, Hambleton RK. Handbook of Modern Item Response Theory. New York: Springer; 1997.
- 17. McGrory S, Doherty JM, Austin EJ, Starr JM, Shenkin SD. Item response theory analysis of cognitive tests in people with dementia: a systematic review. BMC Psychiatry. 2014; 14(1). pmid:24552237
- 18. Mungas D, Reed B. Application of item response theory for development of a global functioning measure of dementia with linear measurement properties. Statistic in Medicine. 2000; 19(11): 1631–1644. pmid:10844724
- 19. Mungas D, Reed BR, Kramer JH. Psychometrically matched measures of global cognition, memory, and executive function for assesment of cognitive decline in older persons. Neuropsychology. 2003; 17(3): 380–392. pmid:12959504
- 20. McGrory S, Shenkin SD, Austin EJ, Starr JM. Lawton IADL scale in dementia: can item response theory make it more informative? Age and Ageing. 2013; 43(4): 491–495. pmid:24212917
- 21. Ashford JW, Kolm P, Colliver JA, Bekian C, Hsu L-N. Alzheimer Patient Evaluation and the Mini-Mental State: Item Characteristic Curve Analysis. Journal of Gerontology. 1989; 44(5): 139–146. pmid:2768772
- 22. Gibbons LE, van Belle G, Yang M, Gill C, Brayne C, Huppert FA, Paykel E, Larson E. Cross-cultural comparison of the Mini-Mental State Examination in United Kingdom and United States participants with Alzheimer’s disease. International Journal of Geriatric Psychiatry. 2002; 17(8): 723–728. pmid:12211121
- 23. Tsai C-F, Lee W-J, Wang S-J, Shia B-C, Nasreddine Z, Fuh J-L. Psychometrics of the Montreal Cognitive Assessment (MoCA) and its subscales: validation of the Taiwanese version of the MoCA and an item response theory analysis. International Psychogeriatrics. 2011; 24(4): 651–658. pmid:22152127
- 24. Lindeboom R, Schmand B, Holman R, de Haan RJ, Vermeulen M. Improved brief assessment of cognition in aging and dementia. Neurology. 2004; 63(3): 543–546. pmid:15304591
- 25. Galindo-Garre F, Hendriks SA, Volicer L, Smalbrugge M, Hertogh CMPM, van der Steen JT. The Bedford Alzheimer Nursing-Severity Scale to Assess Dementia Severity in Advanced Dementia. American Journal of Alzheimer’s Disease & Other Dementiasr. 2013; 29(1): 84–89. pmid:24085256
- 26. Bagnasco A, Watson R, Zanini M, Rosa F, Rocco G, Sasso L. Preliminary testing using Mokken scaling of an Italian translation of the Edinburgh Feeding Evaluation in Dementia (EdFED-I) scale. Applied Nursing Research. 2015; 28(4): 391–396. pmid:26608444
- 27. Hsieh S, McGrory S, Leslie F, Dawson K, Ahmed S, Butler CR, et al. The Mini-Addenbrooke’s Cognitive Examination: A New Assessment Tool for Dementia. Dementia and Geriatric Cognitive Disorders. 2015; 39(1–2): 1–11. pmid:25227877
- 28. Liu W, Watson R, Lou F. The Edinburgh Feeding Evaluation in Dementia scale (EdFED): cross-cultural validation of the simplified Chinese version in mainland China. Journal of Clinical Nursing. 2013; 23(1–2): 45–53. pmid:23387397
- 29. Watson R, van der Ark LA, Lin L-C, Fieo R, Deary IJ, Meijer RR. Item response theory: How Mokken scaling can be used in clinical practice. Journal of Clinical Nursing. 2011; 21(19–20): 2736–2746. pmid:21883577
- 30. Crane PK, Belle G van, Larson EB. Test bias in a cognitive test: differential item functioning in the CASI. Statistics in Medicine. 2004; 23(2): 241–256. pmid:14716726
- 31. Watson R. The Mokken scaling procedure (MSP) applied to the measurement of feeding difficulty in elderly people with dementia. International Journal of Nursing Studies. 1996; 33(4): 385–393. pmid:8836763
- 32. Robinson L, Tang E, Taylor J-P. Dementia: timely diagnosis and early intervention. BMJ. 2015. pmid:26079686
- 33. Sheehan B. Assessment scales in dementia. Therapeutic Advances in Neurological Disorders. 2012; 5(6): 349–358. pmid:23139705
- 34. Guthrie DM, Davidson JGS, Williams N, Campos J, Hunter K, Mick P, et al. Combined impairments in vision, hearing and cognition are associated with greater levels of functional and communication difficulties than cognitive impairment alone: Analysis of interRAI data for home care and long-term care recipients in Ontario. PLOS ONE. 2018; 13(2): e0192971. pmid:29447253
- 35. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research; 1975; 12(3): 189–198. pmid:1202204
- 36. Contador Castillo I., Fernández Calvo B., Ramos Campos F., Tapias Merino E., & Bermejo Pareja F. (2010). El cribado de la demencia en atención primaria. Revisión crítica. Revista De Neurología, 51(11), 677. pmid:21108230
- 37. Devenney E., & Hodges J. (2016). The Mini-Mental State Examination: pitfalls and limitations. Practical Neurology, 17(1), 79–80. pmid:27903765
- 38. Chapman K., Bing-Canar H., Alosco M., Steinberg E., Martin B., & Chaisson C. et al. (2016). Mini Mental State Examination and Logical Memory scores for entry into Alzheimer’s disease trials. Alzheimer’s Research & Therapy, 8(1). pmid:26899835
- 39. Rojas-Gualdrón D., Segura C., A., Cardona A., D., Segura C., Á., & Garzón D., M. (2017). Análisis Rasch del Mini Mental State Examination (MMSE) en adultos mayores de Antioquia, Colombia. Retrieved 5 January 2021, from: http://www.scielo.org.co/pdf/cesp/v10n2/2011-3080-cesp-10-02-00017.pdf.
- 40. Mioshi E, Dawson K, Mitchell J, Arnold R, Hodges JR. The Addenbrooke’s Cognitive Examination Revised (ACE‐R): a brief cognitive test battery for dementia screening. International Journal of Geriatric Psychiatry. 2006; 21(11): 1078–1085. pmid:16977673
- 41. Mathuranath PS, Nestor PJ, Berrios GE, Rakowicz W, Hodges JR. A brief cognitive test battery to differentiate Alzheimer’s disease and frontotemporal dementia. Neurology. 2000; 55(11): 1613–1620. pmid:11113213
- 42. So M, Foxe D, Kumfor F, Murray C, Hsieh S, Savage G, et al. Addenbrooke’s Cognitive Examination III: Psychometric Characteristics and Relations to Functional Ability in Dementia. Journal of the International Neuropsychological Society. 2018; 24(8):854–863. pmid:30189909
- 43. Matías-Guiu JA, Fernández-Bobadilla R, Fernández-Oliveira A, Valles-Salgado M, Rognoni T, Cortés-Martínez A, et al. Normative data for the Spanish version of the Addenbrooke’s Cognitive Examination III. Dementia and geriatric cognitive disorders. 2016; 41(5–6): 243–250. pmid:27240425
- 44. Mirza N, Panagioti M, Waheed MW, Waheed W. Reporting of the translation and cultural adaptation procedures of the Addenbrooke’s Cognitive Examination version III (ACE-III) and its predecessors: a systematic review. BMC Medical Research Methodology. 2017: 17(1). pmid:28903725
- 45. Alexopoulos P, Ebert A, Richter-Schmidinger T, Schöll E, Natale B, Aguilar CA, et al. Validation of the German Revised Addenbrooke’s Cognitive Examination for Detecting Mild Cognitive Impairment, Mild Dementia in Alzheimer’s Disease and Frontotemporal Lobar Degeneration. Dementia and Geriatric Cognitive Disorders. 2010; 29(5): 448–456. pmid:20502019
- 46. Bajpai S, Upadhyay A, Sati H, Pandey R, Chaterjee P, Dey AB. Hindi Version of Addenbrooke’s Cognitive Examination III: Distinguishing Cognitive Impairment Among Older Indians at the Lower Cut-Offs. Clinical Interventions in Aging. 2020; 15: 329–339. pmid:32184582
- 47. Larner AJ, Mitchell AJ. A meta-analysis of the accuracy of the Addenbrooke’s Cognitive Examination (ACE) and the Addenbrooke’s Cognitive Examination-Revised (ACE-R) in the detection of dementia. International Psychogeriatrics. 2014; 26(4): 555–563. pmid:24423470
- 48. Li X, Yang L, Yin J, Yu N, Ye F. Validation Study of the Chinese Version of Addenbrooke’s Cognitive Examination III for Diagnosing Mild Cognitive Impairment and Mild Dementia. Journal of Clinical Neurology. 2019; 15(3): 313. pmid:31286702
- 49. Hsieh S, Schubert S, Hoon C, Mioshi E, Hodges JR. Addenbrooke’s Cognitive Examination-III. PsycTESTS Dataset. 2013. pmid:23949210
- 50. Muñoz-Neira C, Henríquez CF, Ihnen J, Sánchez M, Flores P, Slachevsky A. Propiedades psicométricas y utilidad diagnóstica del Addenbrooke’s Cognitive Examination-Revised (ACE-R) en una muestra de ancianos chilenos. Rev. méd. Chile. 2012; 140(8): 1006–1013. pmid:23282773
- 51. Veliz-García O, Calderón C, Beyle C. Propiedades Psicométricas del Addenbrooke’s Cognitive Examination III: Estructura Factorial, Análisis TRI y Utilidad Diagnóstica para la detección de demencia en atención primaria de salud. Rev. Méd. Chile. 2020; 148(9): 1279–1288. pmid:33399703
- 52. Yoshida H, Terada S, Honda H, Ata T, Takeda N, Kishimoto Y, et al. Validation of Addenbrooke’s cognitive examination for detecting early dementia in a Japanese population. Psychiatry Research. 2011. 185(1–2): 211–214. pmid:20537725
- 53. Bruno D, Slachevsky A, Fiorentino N, Rueda D, Bruno G, Tagle A, et al. Validación argentino-chilena de la versión en español del test Addenbrooke’s Cognitive Examination III para el diagnóstico de demencia. Neurología. 2020; 35(2): 82–88. pmid:28865943
- 54. Matias-Guiu JA, Fernández de Bobadilla R, Escudero G, Pérez-Pérez J, Cortés A, Morenas-Rodríguez E, et al. Validación de la versión española del test Addenbrooke’s Cognitive Examination III para el diagnóstico de demencia. 2015; 30(9): 545–551. pmid:25002342
- 55. Junco JI, Prieto G. Análisis del test neuropsicológico Addenbrooke’s Cognitive Examination mediante el Modelo de Rasch. Revista de Psicología. 2014; 23(1):
- 56.
Cuesta M. Unidimensionalidad. In: Muñiz J, editor. Psicometría. Madrid: Pirámide; 1996. pp. 239–292.
- 57.
Muñiz J, Cuesta M. El problema de la dimensionalidad en la medición psicológica. In: Fons M, Anguera MT, editors. Aportaciones recientes a la medición psicológica. Barcelona: PPU; 1993. pp 51–70.
- 58. Cai L. Metropolis-Hastings Robbins-Monro Algorithm for Confirmatory Item Factor Analysis. Journal of Educational and Behavioral Statistics. 2010; 35(3): 307–335.
- 59. Chalmers RP. mirt: A Multidimensional Item Response Theory Package for the R Environment. Journal of Statistical. 2012; 48(6).
- 60. Maydeu-Olivares A, Joe H. Limited Information Goodness-of-fit Testing in Multidimensional Contingency Tables. Psychometrika. 2006; 71(4): 713–732.
- 61. Grober E, Buschke H. Genuine memory deficits in dementia. Developmental Neuropsychology. 1987; 3(1): 13–36.
- 62. Weissberger GH, Strong JV, Stefanidis KB, Summers MJ, Bondi MW, Stricker NH. Diagnostic accuracy of memory measures in Alzheimer’s dementia and mild cognitive impairment: A systematic review and meta-analysis. Neuropsychology Review. 2017; 27(4): 354–388. pmid:28940127
- 63. Boraxbekk C-J, Lundquist A, Nordin A, Nyberg L, Nilsson L-G, Adolfsson R. Free recall episodic memory performance predicts dementia ten years prior to clinical diagnosis: Findings from the betula longitudinal study. Dementia and Geriatric Cognitive Disorders Extra. 2015; 5(2): 191–202. pmid:26078750
- 64. Crawford JR. Introduction to the Assessment of Attention and Executive Functioning. Neuropsychological Rehabilitation. 1998; 8(3): 209–211.
- 65. Tucha L, Tucha O, Laufkötter R, Walitza S, Klein HE, Lange KW. Neuropsychological assessment of attention in adults with different subtypes of attention-deficit/hyperactivity disorder. Journal of Neural. 2008; 115(2): 269–278. pmid:18200435
- 66.
Mirsky AF, Fantie BD, Tatman JE. Assessment of Attention across the Lifespan. In: Mapou RL, Spector J, editors. Clinical Neuropsychological Assessment. Boston: Springer; 1995. pp 17–48. https://doi.org/10.1007/978-1-4757-9709-1_2
- 67. Mizuhara H, Wang L-Q, Kobayashi K, Yamaguchi Y. Long-range EEG phase synchronization during an arithmetic task indexes a coherent cortical network simultaneously measured by fMRI. NeuroImage. 2005; 27(3): 553–563. pmid:15921932
- 68. Micheloyannis S, Sakkalis V, Vourkas M, Stam CJ, Simos PG. Neural networks involved in mathematical thinking: evidence from linear and non-linear analysis of electroencephalographic activity. Neuroscience Letters. 2005; 373(3): 212–217. pmid:15619545
- 69. Förster S, Teipel S, Zach C, Rominger A, Cumming P, la Fougere C, et al. FDG-PET mapping the brain substrates of visuo-constructive processing in Alzheimer´s disease. Journal of Psychiatric Research. 2010; 44(7): 462–469. pmid:19875130