The Tuokko version of the Clock Drawing Test: A validation study in the Greek population.

The present study aims to be the first to validate the Tuokko version of the Clock Drawing Test (CDT) and estimate its cutoff score after its translation into the Greek language and administration in the Greek population. One hundred and thirty-two individuals participated in this study [60 with Good Cognitive Health (GCH), 24 with Parkinson's Disease (PD), 24 with Parkinson's Disease Dementia (PDD) and 24 with Alzheimer's Disease (AD)]. The CDT was administered to all participants. Additionally, the cognitive and mental status of the sample were estimated through the use of the Mini Mental State Examination (MMSE), Abbreviated Mental Test Score (AMTS), Arizona Battery for Communication Disorders of Dementia (ABCD), Instrumental Activities of Daily Living (IADL), the Neuropsychiatric Inventory (H-NPI) and the Geriatric Depression Scale −15 (GDS-15). Statistically significant differences were found between all groups on the CDT, with AD patients having lower scores than all subgroups in the study. The CDT showed a high internal consistency (Cronbach's alpha = 0.832). The ROC analysis provided a cutoff point equal to 4.00 (AUC: 0.821, p < 0.001) between the Cognitively Unimpaired Group (CUG: GCH and PD group) and the Cognitively Impaired Group (CIG: PPD and AD patients), 5.00 (AUC: 0.845, p < 0.001) between the GCH group and the PDD group, and 4.00 (AUC: 0.780, p < 0.001) between the GCH group and the AD group. Finally, the cutoff point between the PD group and the PDD group was 4.00 (AUC: 0.896, p < 0.005), and 3.00 (AUC: 0.899, p < 0.001) between the PD group and the AD group. Significant positive Pearson's correlations were observed between CDT and MMSE (r = 0.808, p < 0.001), CDT and AMTS (r = 0.688, p < 0.001), CDT and ABCD (r = 0.770, p < 0.001), CDT and the ABCD Visuospatial Construction subdomain (r = 0.880, p < 0.001); while a negative correlation was found between CDT and IADL (r = −0.627, p < 0.001) between the CUG and the CIG groups. Given the results obtained, the CDT appears to be a clinically valid screening instrument for the assessment of visuospatial abilities, with high reliability in Greek populations with cognitive impairment.

Keywords: Alzheimer's disease; cognitive impairment; Clock Drawing Test; cutoff scores; Parkinson's Disease; Parkinson's Disease Dementia; test validation; visuospatial abilities

The Movements Disorder Society (MDS) task force (Litvan et al., [32]), the National Institute on Aging-Alzheimer's Association (NIA-AA; McKhann et al., [33]) and the International Psychogeriatric Association (IPA; Katona et al., [27]) have underlined the need for clearly defined areas of assessment, screening tools and full battery tests that can be used in the characterization of cognitive abilities and in the diagnosis of cognitive impairment. One of the areas in need of assessment in cognitively impaired groups is that of visuospatial abilities. Clock Drawing Tests (CDT; Park et al., [36]; Reiner et al., [40]) are among the tests developed for evaluation of visuospatial abilities and are recommended by both the MDS task force (Litvan et al., [32]) and the NIA-AA (McKhann et al., [33]). Additionally, IPA has also highlighted the need for the development of culturally appropriate instruments so that different populations can be efficiently tested (Katona et al., [27]). The above is also applicable in Greek clinical practice, since there are few validated and standardized tests in Greek for the assessment of cognitive abilities (e.g., Fountoulakis et al., [14]; Tafiadis et al., [50]).

The CDT are valid and psychometrically robust tools (Bozikas et al., [4]; Caffarra et al., [7]; Shulman, [45]) that quantify visuospatial abilities for patients according to the severity of their cognitive status. They are also among the most frequently used cognitive screening instruments for aging populations (Nishiwaki et al., [34]), as well as for populations with dementia (Park et al., [36]; Reiner et al., [40]; Shulman, [45]). Their popularity with clinicians and researchers is related to the fact that: (a) they are simple to administer and easy to score, (b) they are well accepted by patients (Shulman, [45]) and (c) they are relatively free of cultural influences (Borson et al., [3]; Ismail et al., [24]; Parker & Philp, [37]; Shulman, [45]; Steward et al., [46]), making them highly appropriate as gross screening tests. Consequently, they have been used in different clinical populations, showing good discriminatory abilities between patients and cognitively healthy individuals, and have been translated for a variant of languages such as Greek (Bozikas et al., [6]), Brazilian (De Oliveira et al., [9]) and Chinese (Lin et al., [31]). Analytically, CDT have been administered to patients with AD (Bozikas et al., [6]; Freedman et al., [16]; Heinik et al., [21]; Rouleau et al., [42]), patients with Huntington's disease (HD; Rouleau et al., [43]), patients with schizophrenia (Bozikas et al., [5], [6]), patients with vascular diseases (Heinik et al., [21]), stroke patients (Lieberman et al., [30]) and patients with visuospatial neglect (Di Pellegrino, [10]). In all the above research, patient groups were reported to have significantly lower CDT scores than control groups, a finding which underlines the discriminatory validity of CDT as a gross screening test.

On the other hand, it is still unclear whether these tests are envisaged as gross cognitive screening tools (in which case sensitivity to all kinds of decline is the desirable outcome) or only as assessments of visuospatial function in AD and PDD, meaning that CDT should be specific to visuospatial deficits and have stronger correlations with other visuospatial tests than with tests in other domains of general function (Park et al., [36]; Shulman, [45]; Sunderland et al., [47]; Watson et al., [55]; Wolf-Klein et al., [56]). In recent research, it appears to be used as gross screenings for the visuospatial abilities only (Park et al., [36]).

There are several versions of the CDT, for example, those of Freedman et al. ([16]), and Shulman ([45]), and many scoring methods have been introduced over the years, with research evidence supporting good discriminant validity of all of them (Shulman, [45]). In Greek, Bozikas and colleagues used the Freedman et al. ([16]) method to investigate the function of CDT in cognitively unimpaired as well as cognitively impaired populations (Bozikas et al., [4], [5], [6]). Specifically, Bozikas and colleagues administered the Freedman et al. ([16]) CDT method to two groups (cognitively healthy participants and patients with schizophrenia) to examine its discriminatory ability (Bozikas et al., [5]). They concluded that the Freedman's CDT method has the ability to discriminate between patients with schizophrenia and cognitively healthy participants (Bozikas et al., [5]). In another study, Bozikas et al. ([6]) also administered the Freedman et al. ([16]) of the CDT method to three groups (cognitively healthy participants, patients with schizophrenia and patients with AD) and they concluded that it appears to be as sensitive in differentiating patients with AD, as well as those with schizophrenia, from cognitively healthy individuals (Bozikas et al., [6]). The above research reveals the ability of the CDT to differentiate between cognitively impaired and unimpaired groups of participants, indicating a good discriminating power in order to be used as a gross assessment in clinical settings. Finally, Bozikas et al. ([4]) administered the same CDT method to Greek cognitively healthy individuals in order to provide normative data and examine the effects of age and education on the performance in the CDT. In their latter work, the researchers reported that both age and educational level do not significantly affect CDT performance. This finding is in line with a review conducted by Ismail et al. ([24]) reporting that CDT tests are " ... decreasing susceptibility to cultural and educational biases ...." In the same line were also two more studies (Park et al., [36]; Shulman, [45]) that concluded that the CDT tests are free of cultural influences.

In 1992, Tuokko, Hadjistavropoulos, Miller and Beattie developed a short version of the CDT. This version is an easy-to-administrate scale, which originally used solely to assess visuo-constructive abilities (Jouk & Tuokko, [26]; O'Rourke et al., [35]; Tuokko et al., [53], [54]). Additionally, it is a reliable and time-saving assessment tool (Jouk & Tuokko, [26]; O'Rourke et al., [35]; Tuokko et al., [53], [54]). Previous relevant studies (Jouk & Tuokko, [26]; O'Rourke et al., [35]; Tuokko et al., [53], [54]) revealed this version's ability to differentiate dementia from normal age-related memory decline, as well as its usefulness as an assistive tool in distinguishing cognitive impairment from cases of memory loss due to depression (pseudodementia; McKhann et al., [33]). The above stress that Tuokko's version of CDT can be a valid and quick screening tool that can be used efficiently in different Greek clinical settings.

On the basis of previous data, the CDT has proved a useful screening tool for the evaluation of visuospatial abilities. However, to the best to our knowledge, no previous study has attempted to validate the psychometric properties of any CDT versions and scoring systems. Additionally, no previous study has attempted to examine the psychometric properties of the Tuokko CDT in any language. The novelty of the present work is that it constitutes the first attempt to validate the Tuokko CDT for Greek-speaking individuals with cognitive impairment due to AD and PD, after translation of the instrument by our research team. Additionally, by means of Receiver Operating Characteristics (ROC) analysis, we aim to examine the ability of the test to discriminate between Greek-speaking cognitively unimpaired and cognitively impaired populations and provide cutoff points for the CDT mean total score.

One hundred and thirty-two individuals participated in the study. Of these, 60 participants with good cognitive health (GCH). This group was recruited from the "National Open Care Centre for the Elderly," a Greek institution that fosters the health prevention of the elderly and provides them with psycho-emotional support, social care, medical and hospital care, as well as physiotherapy and/or occupational therapy.

The clinical group consisted of 24 patients with PD without cognitive impairment, 24 patients with PDD and 24 patients with AD. The last two groups, PPD and AD patients, formed the cognitively impaired group (CIG). The GCH group and the PD patients formed the cognitively unimpaired group (CUG).

All participants of the clinical groups were recruited from a neurological outpatient clinic of the University Hospital of Ioannina in the area of Epirus. A neurologist specialized in neurodevelopmental disorders gave a formal diagnosis for PD or AD, according to clinical neurological examination and Magnetic Resonance Imaging (MRI; Jack et al., [25]; Litvan et al., [32]; Postuma et al., [39]). The same neurologist also determined PD patient staging in accordance with Hoehn and Yahr scale references (Goetz et al., [19]; Hoehn & Yahr, [23]). For PD and PDD participants, the categorization of their cognitive status was determined on the basis of the MDS task force criteria (Goetz et al., [19]; Hoehn & Yahr, [23]; Litvan et al., [32]; Postuma et al., [39]). Additionally, the level of cognitive impairment associated with PD was determined according to clinical diagnostic criteria suggested by Emre et al. ([12]) as well as Goetz et al. ([18]). The level of cognitive impairment for the AD patients was determined according to NIA-AA guidelines (Jack et al., [25]; McKhann et al., [33]). All participants with a history of neurological impairment (e.g., any kind of brain injury and/or surgery, epilepsy or stroke) and/or psychiatric disorders (e.g., mood disorders, schizophrenia) that could affect the assessment procedures were excluded from the current study. Moreover, those who had prior deficits in cognitive abilities according to their medical record were also excluded.

Finally, all subgroups were comprised of monolingual Greek speakers and matched in terms of age, educational background, and marital status. Informants in the PD and PPD group were regulated with dopamine medications, while most of the participants (patient and non-patient) lived in urban areas at the time of testing.

The categorization of AD patients was made in accordance with the criteria suggested by NIA-AA (Jack et al., [25]; McKhann et al., [33]). These determine that cognitive impairment must be diagnosed based on the combination of medical history information (from the patient and his/her caregivers) and from cognitive testing or a neuropsychological assessment. All recorded symptoms must negatively affect the activities of daily living and functioning at work and must have insidious onset and a clear history of cognitive worsening, and not be justified by delirium or major psychiatric disorders.

For the categorization of PD and PDD Patients, the MDS task force criteria (Emre et al., [12]; Goetz et al., [18]; Litvan et al., [32]; Postuma et al., [39]) were used. These criteria can be used to classify PD patients into two diagnostic categories "Probable PDD" and "Possible PDD" (Emre et al., [12]).

After the approval of this study by the Ethical Committee of the Medical School of the University of Ioannina, all participants and their caregivers were informed of the purpose of the study and asked to sign a consensus form. All participants were administered the following tests:

The MDS task force (Litvan et al., [32]) and NIA-AA (Jack et al., [25]; McKhann et al., [33]) suggested the MMSE (Folstein et al., [13]) as an instrument for assessing cognitive impairment in PD and AD patients, respectively (Emre et al., [11]; Jack et al., [25]; McKhann et al., [33]). In this study, the Greek version of the MMSE (Fountoulakis et al., [14]) was administered, and a threshold score of <24 was used for the formation of the groups in this study. A score below 24 meant inclusion of a participant in the patient group, while a score over 26 meant inclusion of a participant in the non-clinical group.

The AMTS was introduced (Hodkinson, [22]) as a quick 10-items test (estimated administration time 3 minutes) for the detection of dementia in geriatric populations. The Greek version of AMTS (Tafiadis et al., [50]) was used in the present study, with a threshold of <6.5 being used to categorize patients with or without cognitive impairment.

The change of everyday activities was suggested as a criterion for dementia by the NIA-AA (Jack et al., [25]; McKhann et al., [33]). The IADL evaluates the level of functionality of a patient with cognitive impairment (Lawton & Brody, [28]). The Greek version of the IADL scale (Theotoka et al., [51]) was used for this study.

The GDS-15 scale is a short scale designed to assess depression in geriatric patients by means of fifteen closed "yes" or "no" questions (Fountoulakis et al., [15]; Yesavage, [57]). The GDS-15 was standardized in Greek and the threshold indicating probable occurrence of depression was set at 7 (Fountoulakis et al., [15]). Participants who had a GDS score under 7 were considered eligible for the patient group and included in this study.

The NPI scale was designed to assess behavioral disturbances that are common in dementia (Cummings et al., [8]). The Greek version of NPI (Politis et al., [38]) was used in this study for the purposes of excluding participants with psychiatric disturbances.

The ABCD has five domains of assessment: i) Mental Status, ii) Episodic Memory, iii) Language Expression, iv) Language Comprehension and v) Visuospatial Construction of an individual (Bayles & Tomoeda, [2]). This battery test can categorize patients into four diagnostic groups (PD without dementia, mild AD, mediocre AD and PDD). The preliminary Greek version of the ABCD was used to classify patients according to their cognitive-communication abilities (Tafiadis et al., [48], [49]).

The Tuokko version (Tuokko et al., [53]) of the CDT scoring system and its instructions were translated into the Greek language by our research team according to the "Minimal Translation Criteria" (Medical Outcomes Trust, [52]) and administered to all participants. The scoring of the Tuokko CDT is determined by specific error categories (Omissions, Perseverations, Rotations, Misplacements, Distortions, Substitutions and Additions). The maximum score, which is seven, is the result of the addition of the sub-scores of Clock drawing, Clock Setting and Clock Reading (Tuokko et al., [53]). The Clock Drawing is attributed a maximum of 3 points (2 points for the right placement of number 12 and 1 point for the right placement of the rest of the numbers), the Clock Setting is given a maximum of 2 points (1 point for the correct placement of each hand and 1 point for indicating the correct relative lengths of the hour and minute hands), while the Clock Reading is given a maximum of 2 points. In Figure 1, representative clock drawings are shown for the clinical group and the group with good cognitive health.

Graph: Figure 1. Representative Examples of CDT Score Criteria for Good Cognitive Health Group (A), PD Group (B), PDD Group (C) and AD Group (D).

The Kolmogorov-Smirnov and Shapiro-Wilk tests were used to test for variable distribution. All the study variables, which were non-skewed, were expressed with mean (M) and standard deviations (SD). The independent sample's t-test was used to compare the mean scores between the GCH and the CΙG groups. The Student t-test was used to compare mean scores in all possible subgroup pairs (i.e., pairwise comparisons). To compare the CDT mean scores across all subgroups, the One-Way ANOVA test was calculated. For the estimation of CDT cutoff points, a ROC curve analysis was performed between a) the GCH and the CIG groups, b) the GCH and PDD group, (c) the GCH and the AD group, (d) the PD group and PDD group (e) the PD group and AD group, (f) PDD group and the AD group and (g) GCH and non-demented PD group. Cronbach's alpha coefficient and the split-half reliability coefficient were used to evaluate the internal consistency of the Greek CDT for the Cognitively Unimpaired Group (CUG) and the Cognitively Impaired Group (CIG). To determine the sensitivity of the Greek CDT, a Pearson's correlation was calculated between the CDT and Greek MMSE, the CDT and the Greek version of AMTS, the CDT and the Greek version of ABCD mean total scores for the Cognitively Unimpaired Group (CUG) and the Cognitively Impaired Group (CIG). Additionally, a Pearson's correlation was calculated between the CDT and the ABCDs visuospatial construction domain, which includes a clock test for the Cognitively Unimpaired Group (CUG) and the Cognitively Impaired Group (CIG). Furthermore, a Pearson's correlation was computed to explore if age or educational level of the participants are correlated to CDT total score. Also, a Logistic Regression Analysis was performed between the CHG and the CIG groups in order to examine any influences of sex, age, educational level (in years) and medical condition (participants having or not any pathology affecting their mental status). Finally, effect sizes were computed using Cohen's d coefficient for the independent samples t-test and partial Eta square coefficient for analysis of variance (ANOVA) models. The statistical significance was set at p < 0.05 and all reported p values were two-tailed. The analysis was conducted using SPSS statistical software (version 19.0, Armonk, NY, USA).

The non-clinical and the clinical groups were similar in age, sex and years of education. The patient' groups were equivalent in duration of disease, while the PD and PDD groups did not differ in the Hoehn-Yahr staging. An analysis of the participants' demographic data is presented in Table 1.

Table 1. Demographics of the samples

1 Abbreviations: GCH, Good Cognitive Health; PD, Parkinson's Disease; PDD, Parkinson's Disease Dementia; AD, Alzheimer's Disease; H-Y Staging, Hoehn – Yahr Staging.

The internal consistency of the CDT was estimated at Coefficient alpha = 0.838, which is excellent. The Correlation and Reliability Measures of the CDT, by subdomain according to the Cronbach analysis, were, for Clock Drawing, equal to 0.816, for Clock Setting, equal to 0.890, and for Clock Reading, equal to 0.795. Additionally, the split-half reliability technique also demonstrated very good CDT internal consistency (split-half reliability coefficient = 0.802).

We subsequently ran a Pearson's correlation between the CDT total score and the total score of the other tests used in this study to determine the sensitivity of the former against an external validity criterion. Specifically, the analysis returned a strong positive correlation between the CDT total mean score and the MMSE total score (r =.80, p <.001), and a positive correlation between both the CDT and the AMTS total score (r =.68, p <.001), and the CDT and the ABCD (r =.77, p <.001). A strong positive correlation was also detected between the CDT and the ABCD Visuospatial Construction subdomain score (r =.88, p <.001). Moreover, a negative correlation was detected between the CDT total score and the IADL total score (r = −.62, p <.001). Finally, in order to explore if age or educational level of the participants are correlated to CDT total score a Pearson's correlation was computed. A non-significant negative correlation was detected between the age and CDT total score (r =.11, p =.898). Likewise, a non-significant but positive correlation was detected between years of education and CDT total score (r =.15, p =.890).

Turning to the comparisons in the CDT test an independent sample t-test analysis was computed. A statistically significant difference was observed in the total scores between the GCH and the CIG group [t (106) = 7.59, p <.001] with the GCH group scoring significantly higher and with medium effect size results (Cohen's d = 0.71, 95% CI [−7.64, 9.06]). Likewise, statistically significant differences were observed for the 3-items/subdomains (Clock Drawing, Clock Setting, Clock Reading) between the GCH and the CIG group, as follows: a) in the "Clock Drawing" [t (106) = 5.14, p <.001] with medium effect size results (Cohen's d = 0.45, 95% CI [−2.45, 3.34]), b) in the "Clock Setting" [t (106) = 3.24, p <.005] with medium effect size results (Cohen's d = 0.45, 95% CI [−1.09, 1.88]), and c) in the "Clock Reading," [t (106) = 8.12, p <.001] with medium effect size results (Cohen's d = 0.46, 95% CI [−1.11, 1.94]). (Table 2).

Table 2. Good Cognitive Health Group and Cognitive Impaired Group Comparisons on the CDT Total Mean Score and Subdomain Score

2 Abbreviations: GCH, Good Cognitive Health; CIG, Cognitive Impaired Group; SD, standard deviation; CDT, Clock Drawing Test.

Moving on to the pairwise comparisons of means between the GCH group and every clinical subgroup on the CDT total score and subdomain scores (Clock Drawing, Clock Setting, Clock Reading), the analysis returned statistically significant differences between the GCH group and AD patients in the CDT total score [t (82) = 8.66, p <.001] with medium effect size results (Cohen's d = 0.72, 95% CI [−8.87, 10.31]) and in all subdomain scores (see, Table 3). Similarly, statistically significant differences were found between the GCH group and the PDD group in the CDT total score [t (82) = 9.85, p <.001] with medium effect size results (Cohen's d = 0.67, 95% CI [−9.67, 11.02]) and in all the subdomain scores. Moreover, statistically significant differences were detected between AD and PD patients in the CDT total score [t (46) = 8.55, p <.001] with medium effect size results (Cohen's d = 0.71, 95% CI [−6.41, 7.82]) and in all the subdomain scores. Likewise, statistically significant differences were detected between PD and PDD patients in the CDT total score [t (46) = 7.17, p <.001] with medium effect size results (Cohen's d = 0.65, 95% CI [−7.17, 8.47]) and in all subdomain scores. (Table 3)

Table 3. CDT Mean Pairwise Comparisons (GCH, PD, PDD, AD)

3 Abbreviations: GCH, Good Cognitive Health; PD, Parkinson's Disease; PDD, Parkinson's Disease Dementia; AD, Alzheimer Disease; SD, standard deviation; CDT, Clock Drawing Test; *p <.005

In contrast, the analysis returned no statistically significant differences between AD and PDD patients and the GCH group and the PD patients either in the CDT total score or any of subdomain scores in the CDT total score. Detailed information is presented in Table 3.

In terms of group effects, the one-way Anova method was used, the analysis returned a main group effect in the ABCD total scores F (4, 117) = 63.73, p <.001; ηp2 = 0.66, in the AMTS total score [F (4, 117) = 63.91, p <.001; ηp2 = 0.55], in the IADL total score [F (4, 117) = 24.40, p <.001; ηp2 = 0.46] and in the MMSE total score F (4, 117) = 41.20, p <.001; ηp2 = 0.57. Proceeding to main group effect results in the CDT total score, the One-way ANOVA analysis revealed a significant group effect showing that all the clinical subgroups scored significantly lower than the non-clinical group [F (3, 128) = 25.12, p <.001; ηp2 = 0.66]. Generally, in all measurements, the AD and PDD patients had the lower scores (Table 4).

Table 4. Group effects (GCH, PD, PDD, AD) on the ABCD, AMTS, IADL, MMSE and CDT total scores

4 Abbreviations: GCH, Good Cognitive Health; PD, Parkinson's Disease; PDD, Parkinson's Disease Dementia; AD, Alzheimer's Disease; AMTS, Abbreviated Mental Test Score; MMSE, Mini Mental State Examination; IADL, Instrumental Activities of Daily Living; ABCD, Arizona Battery for Communication Disorders of Dementia; **p <.001.

Additionally, a logistic regression analysis was performed to examine whether the CDT total score adds any unique information to predicting group membership after investigating the effect of all demographic variables on CDT performance. The analysis indicated that only the variable of "Medical Condition," which refers to the different cognitive profiles of the participants with or without pathology, significantly predicted the CDT total score, β = −1.143, t (1) = 3.163, p <.001. The results indicated that there was a significant association only between Medical Condition and CDT score χ2(8) = 40.701, p <.001.

As to the estimation of the CDT total score cutoff points, the ROC analysis revealed that a statistically significant positive discrimination was observed between: (a) the GCH group and the CIG group [AUC 0.821, p <.001] with the cutoff point being 4.00 (out of 7) (sensitivity: 1.000 and specificity 0.534; Figure 2); (b) the GCH group and PDD group [AUC 0.845, p <.001] with the cutoff point being 5.00 (sensitivity 0.983 and specificity 0.583); (c) the GCH group and AD group [AUC 0.780, p <.001] with the cutoff point being 4.00 (sensitivity 0.983 and specificity 0.545); (d) the PD group and PDD group [AUC 0.764, p <.001] with the cutoff point being 4.00 (sensitivity 0.983 and specificity 0.583) and (e) the PD group and AD group [AUC 0.899, p <.001] with the cutoff point being 3.00 (sensitivity 0.983 and specificity 0.528). Finally, no valid and statistically significant cutoff points were calculated between the PDD group and the AD group [AUC 0.451, p =.537] or between the GCH group and the non-demented PD [AUC 0.348, p =.742]. Detailed information is presented in Table 5.

Table 5. ROC data on the Discrimination between Groups

5 Abbreviations: GCH, Good Cognitive Health; CIG, Cognitive Impaired Group; PD, Parkinson's Disease; PDD, Parkinson's Disease Dementia; AD, Alzheimer's Disease; AUC, Area Under Curve; CI, Confidence Interval.

PHOTO (COLOR): Figure 2. Receiver Operating Characteristics (ROC) curve for the CDT – Good Cognitive Health Group and the Group with Cognitive Impairment Total Score (CDT-T).

The present study constitutes the first attempt to validate the Greek version of the Tuokko CDT and to examine the ability of the test to discriminate between Greek-speaking cognitively unimpaired from cognitively impaired participants. This preliminary study supports findings from previous work in the field and emphasizes the validity of the Tuokko CDT as a means to assess mental status (Jouk & Tuokko, [26]; Lessig et al., [29]; Lin et al., [31]; O'Rourke et al., [35]; Park et al., [36]; Reiner et al., [40]; Rouleau et al., [43]; Shulman, [45]; Sunderland et al., [47]; Tuokko et al., [53], [54]; Watson et al., [55]). Additionally, this study provides evidence for the ability of the test to assess the level of cognitive impairment caused by different medical conditions (Jouk & Tuokko, [26]; Lessig et al., [29]; Lin et al., [31]; O'Rourke et al., [35]; Park et al., [36]; Reiner et al., [40]; Rouleau et al., [43]; Shulman, [45]; Sunderland et al., [47]; Tuokko et al., [53], [54]; Watson et al., [55]).

Previous research has proven that different versions of the CDT were successful in assessing the level of cognitive impairment in AD patients (Bozikas et al., [6]; Freedman et al., [16]; Heinik et al., [21]; Rouleau et al., [42], [43]), PD patients (Freedman et al., [16]), and HD patients (Rouleau et al., [43]) or patients with schizophrenia (Bozikas et al., [5]). Our findings agree with previous research since the Greek version of the Tuokko CDT proved suitable to assess the level of cognitive impairment in PDD and AD patients. Additionally, previous relevant studies that focus on simplified versions of the Tuokko CDT also attest to its clinical utility and discriminatory ability (Jouk & Tuokko, [26]; O'Rourke et al., [35]; Tuokko et al., [53], [54]; Watson et al., [55]). The findings of the present study point in the same direction and corroborate relevant previous research, suggesting that the Greek version of the Tuokko CDT can be used to discriminate between cognitively healthy and cognitively impaired groups, with PPD and AD patients having the lowest scores (Jouk & Tuokko, [26]; Lessig et al., [29]; Lin et al., [31]; Rouleau et al., [43]; Shulman, [45]; Sunderland et al., [47]; Watson et al., [55]). Furthermore, the statistical analysis showed that the Greek version of the Tuokko CDT has good discriminatory validity and can, therefore, discriminate cognitive unimpaired from cognitive impaired individuals, with the patients with dementia scoring significantly lower than cognitively unimpaired participants, a finding which is in line with previous research (Jouk & Tuokko, [26]; Lessig et al., [29]; Lin et al., [31]; O'Rourke et al., [35]; Park et al., [36]; Reiner et al., [40]; Rouleau et al., [43]; Shulman, [45]; Sunderland et al., [47]; Tuokko et al., [53], [54]; Watson et al., [55]). The same discriminatory ability was observed for CDT in studies that examined medical conditions other than PDD and AD (Heinik et al., [20]). Moreover, previous studies reported that the CDT can detect differences in the cognitive level of cognitive healthy individuals in comparison to AD and HD patients, with the AD patients showing lower scores (Rouleau et al., [43]) as in this study did. Likewise, a comparison between cognitively healthy individuals and patients with schizophrenia and AD patients was conducted and again the AD patients had achieved the lower scores (Bozikas et al., [5], [6]). Similar results were found between cognitively unimpaired individuals and patients with AD (Lin et al., [31]; Shulman, [45]; Sunderland et al., [47]).

Previous studies determined the cutoff points of the CDT for populations with dementia using a ROC analysis (Jouk & Tuokko, [26]; Lessig et al., [29]; Lin et al., [31]; Rouleau et al., [43]; Shulman, [45]; Sunderland et al., [47]; Watson et al., [55]). Specifically, to date, two studies have estimated a total cutoff point of 1.00 (AUC 0.750 and 0.690; Jouk & Tuokko, [26]; Lin et al., [31]) and three studies a cutoff point of 2.00 (AUC 0.700, 0.700 and 0.730; Rouleau et al., [43]; Tuokko et al., [53]). Additionally, Watson et al. ([55]) estimated a cutoff point equal to 3.00 (AUC = 0.630) for patients with cognitive impairment, which approximates the cutoff point suggested by the Greek version of the CDT which also targeted patients with cognitive impairment (i.e., 4.00). The above convergent data suggest that, given it is accurately and faithfully translated, the CDT can sustain its properties and serve its purpose in any language across different language typologies and families. At this point, it must be emphasized that the reported cutoff points can differentiate non-cognitive impaired from cognitive impaired individuals. These cutoff scores must be used in tandem with proper minor invasive or noninvasive assessment.

Regarding the psychometric properties of the CDT examined here, our results are in agreement with previous research findings (Rouleau et al., [43]; Tuokko et al., [53]). The Greek Tuokko CDT presents very good internal consistency in line with the data reported in the original study (Tuokko et al., [53]) as well as in studies that examined the consistency of various versions of the CDT (i.e., adaptations of the test in various languages; Lin et al., [31]). Τhe intra-class correlation coefficient of the Greek CDT was excellent, a result also reported by previous relevant literature, showing that the Tuokko CDT is a reliable and valid tool (Jouk & Tuokko, [26]; Rouleau et al., [43]; Tuokko et al., [53]).

Despite significant variability in administration procedures and scoring systems, the CDT performance appears to correlate positively with performance in the MMSE test (Folstein et al., [13]), which shows the validity of the CDT against an external criterion. Specifically, statistically significant positive correlations were reported between MMSE scores and CDT total score, with r ranging from.32 to.59 (Adunsky et al., [1]; Heinik et al., [21]; De Oliveira et al., [9]; Richardson & Glass, [41]; Seigerschmidt et al., [44]). Our results are in line with the majority of the previous data, with the exception of the work of Shulman ([45]) and Richardson and Glass ([41]), since they reported negative correlations between the MMSE and CDT different scoring systems. We also found positive correlations between the Greek version of the Tuokko CDT and the MMSE, the AMTS, the ABCD total score and Visuomotor subdomain. These data show the sensitivity of the Greek version of the Tuokko CDT, as well as its validity against an external criterion.

Furthermore, we found a negative correlation between the Tuokko CDT total scores and the IADL total scores. Conversely, Fukui and Lee ([17]) found that IADL-R total scores were positively correlated to CDT total score (rs=.480, p <.050) in a sample of 64 patients with probable AD. However, they concluded that clock tasks are suitable for the evaluation of the functional status in AD populations, since visuoperception, visuoconstruction, and executive functions, as well as global cognition, may modulate functional status in AD populations, which we think agrees with the main findings of our study.

Finally, it is important to underline that the Tuokko CDT appears to be relatively free of cultural bias, age, and educational level, notwithstanding the "Medical Condition" factor, which was expected. This can be attributed to the fact that most of the individuals are familiar with the clock both as an object and as a means for telling time (Borson et al., [3]; Parker & Philp, [37]; Steward et al., [46]). Moreover, the results of the study revealed that the Tuokko CDT demonstrated sensitivity in differentiating cognitively impaired from non-cognitively impaired individuals. Specifically, Borson et al. ([3]) reported that the sensitivity of the CDT in detecting dementia in a multi-ethnic group was found to be better that of the MMSE, while the specificity measures of the tests were equivalent.

Sample size was a limitation for this study, which could weaken the generalizability of the results, albeit not substantially. On the other hand, one of the strengths of this study are the robust characterization-selection criteria of the sample. Another strength is that this is the first attempt to validate psychometrically the Tuokko version of the CDT in the Greek population.

This study is the preliminary validation of the Tuokko version of the CDT in a Greek sample. The CDT is a tool that can be used for the assessment of patients with cognitive impairment. The Greek version of Tuokko CDT was shown to have a good validity. The findings as regards the instrument's validity are in agreement with previously presented CDT results across different languages and versions. As the Tuokko test has fewer items than other CDT, it could easily be employed by Primary Health Care specialists in outpatient clinics as a reliable and time-saving assessment tool. However, it should be stressed that this instrument can be used only for screening purposes; thus, it is suggested it be employed alongside other clinical assessment tools rather than as the sole instrument.

We thank PRO-ED, Inc for giving us the research rights for the ABCD test (Copyright © 1993 PRO – ED, Inc. Arizona Battery for Communication Disorders of Dementia, translated by permission of the publisher. All rights reserved)

No potential conflict of interest was reported by the author(s).