Abstract

The Clock Drawing Test has been systematically used to assess visuospatial deficits related to the parietal lobes, but we now acknowledge its much more complex relation with other cognitive abilities. Despite its common use in clinical and investigational practices, no study has developed normative data for the Portuguese population. We present the distribution of clock drawing scores using three scoring systems in a representative community sample of cognitively healthy subjects. We found that the systems were well correlated with each other and with cognitive screening tests widely used and had good psychometric properties. Normative data for the three scoring systems were developed considering age and education. These results allow a more rigorous interpretation of the test performance in clinical context and are especially relevant for epidemiological research.

Introduction

The Clock Drawing Test (CDT) was created as a simple and ecological way to assess visuoperceptual functions related to the parietal lobes. It has been incorporated in several neuropsychological assessment instruments and test batteries as an indication of heminegligence and visuospatial processes both related to several medical conditions and different age groups (Strauss, Sherman, & Spreen, 2006). Later studies suggested its relation with other cognitive measures, namely symbolic and graphomotor representation, auditory linguistic abilities, hemiattention, semantic memory, conceptual abilities, and executive function (organization, planning, and parallel processing) (Cosentino, Jefferson, Chute, Kaplan, & Libon, 2004; Freedman et al., 1994; Libon, Malamut, Swenson, Prouty Sands, & Cloud, 1996; Mendez, Ala, & Underwood, 1992; Rouleau, Salmon, Butters, Kennedy, & McGuire, 1992; Shulman, 2000; Strauss et al., 2006). These functions can also be associated with a large spectrum of pathologies and aging strata, including numerous neurological disorders that can affect younger people. In fact, several studies have shown the potential of the CDT in these populations (e.g., Huntington's disease, schizophrenia, cortical lesions, depression; Strauss et al., 2006). However, in the last 20 years, the CDT has been mostly used with the elderly as a brief cognitive test to differentiate cognitively normal subjects from subjects with Mild Cognitive Impairment (MCI) and dementia (Cahn-Weiner et al., 2003; Lowery et al., 2003; Parsey & Schmitter-Edgecombe, 2011). In this context, some of CDT's characteristics should be stated: (a) the CDT capacity to evaluate multidomain impairment, with a special emphasis in frontal and temporoparietal involvement typical of Alzheimer's disease (AD; Freedman et al., 1994; Strauss et al., 2006) and that may be undetected by other cognitive screening instruments such as the Mini-Mental State Examination (MMSE; Brodaty & Moore, 1997; Folstein, Folstein, & McHugh, 1975); (b) the CDT relative independence of verbal abilities (Strauss et al., 2006; Sunderland et al., 1989) that makes it especially useful in patients who present expressive verbal impairment or aphasia; (c) quick and economic instrument, therefore, easy to administer to the elderly (Shulman, 2000); (d) its good test–retest reliability confirmed by many studies (Mendez et al., 1992; Strauss et al., 2006), as well as high intra- and inter-rater reliabilities that justifies the inclusion of CDT in several neuropsychological cognitive screening batteries (Mendez et al., 1992; Rouleau et al., 1992; Strauss et al., 2006; Sunderland et al., 1989).

Most studies with the CDT have been focusing on the development and standardization of simple and easy to interpret scoring methods (seeGarcía-Caballero et al., 2006; Hubbard et al., 2008), resulting in several quantitative and qualitative scoring criteria for the different investigational universes (Fisher & Loring, 2004; Shulman, 2000). The Rouleau and colleagues (1992) scoring system was the first to compare the qualitative aspects of CDT performance (size, graphic quality, “pull to stimulus,” conceptual deficit, spatial organization, and perseveration) among different types of dementia. It is a 10-point quantitative system that encompasses the three major clock components: clock face, numbers, and hands. The Cahn and colleagues (1996) scoring system is a more complex system as it combines the Rouleau quantitative score and a qualitative analysis of the eight types of errors most commonly found in clock drawing as described by Freedman and colleagues (1994). Hubbard and colleagues (2008) compared the Mendez and colleagues (1992), Freund and colleagues (2005), and Cahn and colleagues (1996) scoring systems and concluded that they are all easy and quick to score (less than a minute), produce well-correlated results, and so the choice for a certain system should be based on the assessment purposes (Hubbard et al., 2008). Yamamoto and colleagues (2004) compared the Sunderland and colleagues (1989), Rouleau and colleagues (1992), and Cahn and colleagues (1996) scoring systems and concluded that the Cahn system is more likely to detect MCI (Yamamoto et al., 2004). This conclusion was also supported by the American Academy of Neurology, which recommended the use of the CDT Cahn scoring system as a complement of short cognitive instruments such as the MMSE (Petersen et al., 2001).

Several studies have developed normative data for the CDT in different countries. The main results have shown the negative effect of age in the CDT performance (Bozikas, Giazkoulidou, Hatzigeorgiadou, Karavatos, & Kosmidis, 2008; Caffarra et al., 2011; Hubbard et al., 2008; Menon, Hall, Hobson, Johnson, & O'Bryant, 2011). Regarding the effect of education, some studies seem to confirm its positive effect: higher education correlates with higher CDT scores (Bozikas et al., 2008; Hubbard et al., 2008; Lourenço, Ribeiro-Filho, Moreira, Paradela, & Miranda, 2008; Menon et al., 2011). The Bozikas and colleagues (2008) study explored the effect of gender in CDT performance and found significant differences in one of the forms of presentation (pre-drawn circle and “6:05” as pre-determined hour). Nonetheless, this effect was not considered for normative data calculation (Bozikas et al., 2008). The Caffara and colleagues (2011) Italian study also explored the effects of age, education, and gender in the normative population and found only a significant effect of age (Caffarra et al., 2011). The multiethnic study conducted by Menon and colleagues (2011) with bilingual rural elderly subjects found a significant negative effect of age and a positive effect of education and female gender in the non-Hispanic sample. The clock drawing performance of the Hispanic sample was only negatively affected by age (Menon et al., 2011).

There are no CDT normative data for the Portuguese population, and the potential effect of demographic and geographical variables has never been explored in our country. The Portuguese population is characterized by a low mean educational level in older ages and a big disparity regarding the geographical distribution of the population. Rural areas, mostly located on the inland, have older and less educated people, whereas coastal areas present a predominance of urban areas with younger, highly educated people.

In this study, we explored the effect of several relevant sociodemographic and geographical variables on the performance of the CDT, using a representative sample of Portuguese population. Although the CDT is more commonly used with the elderly, in this study, we investigated a broad range of ages in adults in order to explore the potentialities of the CDT in a large spectrum of neurological pathologies, including those affecting younger strata. This information was further used to perform the normative study of the CDT for the Portuguese population, according to three scoring systems—Rouleau and colleagues (1992), Cahn and colleagues (1996), and Babins, Slater, Whitehead, and Chertkow (2008). Even though it is suggested that the Babins scoring system is more sensitive to MCI and AD and there is correlation between all the systems, the Rouleau and Cahn scoring systems excel in the literature in different pathologies and studies. Moreover, we know that they are commonly used in Portugal hence adequate Portuguese normative data should be developed.

Methods

Participants and Procedures

We intended to obtain a sample of subjects living across all Portugal continental areas, whose distribution should be representative of the Portuguese population according to geographical variables, gender and education. For age stratification purposes, we used pre-established groups with an approximate fixed number of 215 subjects in each age category.

Initially, we assessed a community-based sample of 936 subjects. The subjects were recruited at the local primary healthcare services and at daycare centers by indication of their general physician or the institutional technical directors, respectively. A smaller percentage of subjects volunteered themselves but the information given was corroborated with a family member or the general practitioner. Several inclusion criteria were considered: age of 25 years or older; Portuguese as mother language and having attended school in Portugal; evidence of no motor, visual, or auditory deficits that could act as confounding variables throughout the neuropsychological assessment; cognitively healthy adults—all subjects should have their capacity to perform activities of daily living intact; no history of alcohol or substance abuse; no recent history of psychiatric or neurologic disorders; as well as instable chronic systemic diseases, significant depressive symptoms or medication intake with a known impact on cognition. An initial interview was conducted by a clinical psychologist to insure that all inclusion criteria were fulfilled. The interview was based on a common script that included a complete demographic questionnaire, medical history, drinking habits, and present health status. For the older subjects, all gathered information was corroborated with the general physician, community center directors, and/or an informant, preferably a close family member or currently cohabitating with the participant. Of the 936 total subjects, 194 (20.73%) were excluded following the initial interview due mostly to the personal history of neurological or psychiatric disorders and the history of alcohol abuse.

The second inclusion phase was based on the subject's performance on the several tests that constituted the neuropsychological battery specially built for this study, as well as functional and depression scales. Ninety-two (9.83%) subjects were excluded due to their performance on the assessment battery—their performances suggested the presence of cognitive impairment or depressive symptoms according to Portuguese cutoff scores. Additionally, 20 (2.14%) subjects were excluded due to inability to complete the neuropsychological tests battery.

The sample selection and stratification process culminated on the final sample of 630 subjects. This sample was stratified according to six sociodemographic variables: age, gender, education, geographic region, geographical localization, and residential area. Age considered subdivided into three groups: 25–49 years; 50–64 years; and 65 years or older. Regarding education, we established four levels according to the total number of years of education that the subject had successfully completed: 1–4 years; 5–9 years; 10–12 years; and more than 12 years. The general variable geographic localization encompassed both the distribution of the Portuguese population across the country's different regions and proximity to sea. As for geographic localization by regions, we used the level II territorial units (NUTS II; INE, 2010) that consider the following continental Portuguese regions: North, Center, Lisbon, Alentejo, and Algarve. We also considered a second subdivision into coastal and inland areas. As for the residential area classification, we considered the typology that integrates three levels of classification: predominantly urban areas (PUAs); moderately urban areas (MUAs); and predominantly rural areas (PRAs) (INE, 2010). The global recruitment process was guided by the previous stratification by sociodemographic variables and we stopped the recruitment when the sample (630 subjects) could be considered representative of the Portuguese population according to all the selected variables (age, education, gender, geographic region, geographical localization, and inland/coastal areas).

This study protocol was approved by the Ethics Committee of the Faculty of Medicine of the University of Coimbra. All study purposes and procedures were thoroughly explained to each subject and/or informant by one of the team members, and their informed consent was obtained. All study subjects were evaluated by two psychologists with experience in neuropsychological assessment.

Materials

Several instruments were used in order to account for as many relevant areas of functioning as possible: As for the three selected scoring systems, we followed the scoring instructions proposed by the authors. The Rouleau and colleagues (1992) is a 10-point quantitative system that encompasses the three major clock components: clock face (2 points), numbers (4 points), and hands (4 points). The Cahn and colleagues (1996) scoring system combines the Rouleau quantitative score and a qualitative analysis of the eight most common types of errors: to the quantitative score (maximum 10 points), one should subtract the number of qualitative errors found in the subject's performance (maximum 8 points) to achieve the final score—10 (best) to −8 (worst). The Babins and colleagues (2008) is an 18-point quantitative scoring system with three main components: (a) assessment of circle integrity (2 points); (b) number placement and sequencing (6 points); and (c) placement and size of the hands (6 points). Additionally, there are two points for representation of the clock's center and two points for general gestalt.

  • A demographic questionnaire and a medical history and habits' inventory.

  • Subjective Memory Complaints (SMC; Ginó et al., 2008; Schmand, Jonker, Hooijer, & Lindeboom, 1996) was used to exclude relevant SMC.

  • The Irregular Word Reading Test (Teste de Leitura de Palavras Irregulares, TeLPI; Alves, Simões, & Martins, 2009) as a premorbid intelligence estimate instrument—we analyzed the effect of TeLPI raw score of the total number of errors in a subsample of 363 subjects; these analyses of TeLPI results assume an exploratory nature as there are still no normative data available.

  • The MMSE (Folstein et al., 1975; Guerreiro, 1998)—Portuguese normative values were used for sample inclusion (possible cognitive decline if MMSE ≤22 for subjects from 1 to 11 years of education, and ≤27 for those with >11 years of education).

  • The Montreal Cognitive Assessment (MoCA; Freitas, Simões, Alves, & Santana, 2011; Nasreddine et al., 2005) was used applying the cutoff scores proposed for the Portuguese population according to age and education. Subjects were excluded if their score was below 1 SD. For subjects with age ranging from 25 to 49 years, cutoff points of 21, 24, 26, and 28 were used according to educational levels 1–4 (primary), 5–9 (middle), 10–12 (high), and >12 (superior), respectively. Subjects with ages ranging from 50 to 64 were excluded if MoCA scores were under 19 (primary), 23 (middle), 24 (high), and 25 (superior). In the older group of subjects aged ≥65 years, the following MoCA cutoff values were adopted: 18 (primary), 22 (middle), 23 (high), and 25 (superior).

  • The autonomy in daily life activities was assessed through an interview with the subject and with information obtained from the general practitioners, community center directors, and/or an informant—in subjects over 49 years, the information was supplemented with the Clinical Dementia Rating (CDR) Scale (Garrett et al., 2008; Hughes, Berg, Danziger, Coben, & Martin, 1982).

  • The depressive complaints were measured through the clinical interview and the Geriatric Depression Scale, 30 items (GDS-30; Barreto, Leuschner, Santos, & Sobral, 2008; Yesavage et al., 1983); subjects with a score of 20 or more points were excluded.

  • The CDT was applied to all subjects in the spontaneous drawing modality. The verbal instruction was as follows: “I would like you to draw a round clock. Place all the numbers on the clock and, once you're finished, set the time for 11 hours and 10. Let me know once you're done.” The selected time setting was “11:10” as recommended by several authors (Goodglass & Kaplan, 1983; Kaplan, 1988). It has the advantage of requiring one hand on each hemispace, on the upper quadrants, considered the temporal area. This time setting involves the executive functions, allowing the assessment of abstract processing abilities and the patient's tendency to process information on a more perceptive level (and not semantic). It can also result in a “stimulus-bound” response in which the subjects place the hands for “10:50” (the minute hand is placed toward the number 10 instead of 2). This is one of the errors described by Freedman and colleagues (1994) that has shown discriminative power in AD cases. The verbal instruction condition was selected as it involves memory capacity to remember the visual model of the clock and the instructions regarding the time as well as executive functions. This condition is, therefore, more sensitive to temporal lobe and frontal dysfunction. The instruction deliberately withheld the word “hand” as it could be an auxiliary for patients with abstraction difficulties.

Statistical Analysis

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 19.0. The study sample was characterized through descriptive statistics. Psychometric properties of the three scoring systems were analyzed with the Cronbach's α for internal consistency and Pearson's correlation coefficient for concurrent and discriminant validity. Student's t-test, analysis of variance (ANOVA), and Tukey post hoc test were used to analyze differences between stratified groups. The influence of gender and geographical variables in the CDT performance was analyzed with the analysis of covariance (ANCOVA) controlling the effects of education and age. The partial eta squared index (forumla) was used as a measure of effect size (Cohen, 1988). Linear multiple regression, through the enter method, was performed to assess whether the influence of age (in years) and education (in total number of years completed successfully) on CDT performance. Multicolinearity was explored through tolerance and variance inflation factor (Meyers, Gamst, & Guarino, 2006). The determination coefficient (R2) was considered in the analysis of the regressions' effect size (Cohen, 1988). Finally, CDT normative data for the three scoring systems were stratified and determined according to the demographic variables significantly associated with CDT performance. The normative data are represented as the mean ± standard deviation (SD), and the distribution is represented as mean minus 1 SD, 1.5 SD, and 2 SD.

Results

The final sample had 630 healthy control subjects, 63.7% women, mean age of 55.96 years (±15.30 years; range 25–91 years), and a mean education of 8.08 (±4.58 years; range 1–17). Data related to the demographic and geographical variables are presented in Table 1. One of the essential challenges of this study was to obtain the best representativeness of the real distribution of the Portuguese population throughout the selected variables. According to the results presented in Table 1, we considered that this goal was achieved.

Table 1.

Demographic characterization of the study sample (n = 630)

Variable Levels Sample (n [%]) Portuguese population (n [%]) 
Age (years) 25–49 203 (32.2) — 
50–64 211 (33.5) — 
≥65 216 (34.3) — 
Gender Women 401 (63.7) 3,946 (52.6) 
Men 229 (36.3) 3,559 (47.4) 
Education (years) 1–4 251 (39.8) 2,426 (36.6) 
5–9 165 (26.2) 2,280 (34.4) 
10–12 106 (16.8) 960 (14.5) 
>12 108 (17.1) 956 (14.5) 
NUTS II North 243 (38.6) 2,722 (36) 
Center 172 (27.3) 1,794 (24) 
Lisbon 156 (24.8) 2,091 (28) 
Alentejo 44 (7.0) 577 (8) 
Algarve 15 (2.4) 321 (4) 
Geographical localization Coast 526 (83.5) 6,379 (85) 
Inland 104 (16.5) 1,126 (15) 
Residential area PUA 428 (67.9) 5,103 (68) 
MUA 112 (17.8) 1,200 (16) 
PRA 90 (14.3) 1,200 (16) 
Variable Levels Sample (n [%]) Portuguese population (n [%]) 
Age (years) 25–49 203 (32.2) — 
50–64 211 (33.5) — 
≥65 216 (34.3) — 
Gender Women 401 (63.7) 3,946 (52.6) 
Men 229 (36.3) 3,559 (47.4) 
Education (years) 1–4 251 (39.8) 2,426 (36.6) 
5–9 165 (26.2) 2,280 (34.4) 
10–12 106 (16.8) 960 (14.5) 
>12 108 (17.1) 956 (14.5) 
NUTS II North 243 (38.6) 2,722 (36) 
Center 172 (27.3) 1,794 (24) 
Lisbon 156 (24.8) 2,091 (28) 
Alentejo 44 (7.0) 577 (8) 
Algarve 15 (2.4) 321 (4) 
Geographical localization Coast 526 (83.5) 6,379 (85) 
Inland 104 (16.5) 1,126 (15) 
Residential area PUA 428 (67.9) 5,103 (68) 
MUA 112 (17.8) 1,200 (16) 
PRA 90 (14.3) 1,200 (16) 

Notes: PUA = predominantly urban area; MUA = moderately urban area; PRA = predominantly rural area. Reference values for the Portuguese population are represented in thousands and correspond to the Portuguese population of 25 years and older living in Portugal continental areas (INE, 2010).

The Rouleau and Cahn systems showed inadequate values of internal consistency (α = 0.492 and 0.487), whereas the Babins system had good internal consistency (α = 0.878). The three systems had good concurrent validity, showing a moderate correlation with the MMSE and a high correlation with the MoCA. Depressive symptoms presented significant negative correlations with CDT scores. The total number of errors in the TeLPI presented moderate significant correlations with the three scoring systems, as well as with the MMSE, MoCA, and age, and a high correlation with education (Table 2).

Table 2.

Correlation coefficients of the CDT scoring systems

 Age MMSE MoCA GDS TeLPI Rouleau and colleagues (1992) Cahn and colleagues (1996) Babins and colleagues (2008) 
Education −.451** .508** .661** −.194** −.688** .405** .421** .463** 
MMSE −.382** — .649** −.172** −.434** .453** .453** .499** 
MoCA −.516** .649** — −.222** −.617** .604** .606** .637** 
GDS .072, ns −.172** −.222** — .136** −.132** −.129** −.134** 
TeLPI .233** −.434** −.617** .136** — −.425** −.447** −.442** 
Rouleau and colleagues (1992) −.377** .453** .604** −.132** −.425** — .974** .951** 
Cahn and colleagues (1996) −.377** .453** .606** −.129** −.447** .974** — .930** 
Babins and colleagues (2008) −.416** .499** .637** −.134** −.442** .951** .930** — 
 Age MMSE MoCA GDS TeLPI Rouleau and colleagues (1992) Cahn and colleagues (1996) Babins and colleagues (2008) 
Education −.451** .508** .661** −.194** −.688** .405** .421** .463** 
MMSE −.382** — .649** −.172** −.434** .453** .453** .499** 
MoCA −.516** .649** — −.222** −.617** .604** .606** .637** 
GDS .072, ns −.172** −.222** — .136** −.132** −.129** −.134** 
TeLPI .233** −.434** −.617** .136** — −.425** −.447** −.442** 
Rouleau and colleagues (1992) −.377** .453** .604** −.132** −.425** — .974** .951** 
Cahn and colleagues (1996) −.377** .453** .606** −.129** −.447** .974** — .930** 
Babins and colleagues (2008) −.416** .499** .637** −.134** −.442** .951** .930** — 

Notes: CDT = Clock Drawing Test; MMSE = Mini-Mental State Examination; MoCA = Montreal Cognitive Assessment; GDS = Geriatric Depression Scale; TeLPI = Teste de Leitura de Palavras Irregulares [Irregular Word Reading Test]; ns = non significant.

**p ≤ .001.

The analysis of the relation between the CDT and the demographic and geographical variables was initially explored through univariate ANOVA. Different significant effects were found according to the scoring system selected: (a) men scored higher than women according to the Rouleau system, t(628) = 2.393, p ≤ .05, the Cahn system, t(628) = 2.663, p ≤ .01, and the Babins system, t(628) = 2.517, p ≤ .05; (b) age had a significant effect in the Rouleau, F(2, 627) = 42.582, p ≤ .001, Cahn, F(2, 627) = 40.369, p ≤ .001, and Babins scoring systems, F(2, 627) = 51.458, p ≤ .001, with younger subjects scoring higher than older subjects (Table 3); (c) a significant effect was also found for education in the Rouleau, F(3, 626) = 45.838, p ≤ .001, Cahn, F(3, 626) = 50.224, p ≤ .001, and Babins scoring systems, F(3, 626) = 64.846, p ≤ .001 (Table 3), with higher education resulting in higher CDT scores; (d) as for geographic variables, the geographic region had a significant effect on the Babins scoring system, F(4, 625) = 2.961, p < .05, and the residential area influenced the results in the Cahn system, F(2, 627) = 4.557, p < .05. No significant effects were found for the remaining variables (Table 3).

Table 3.

Effect of demographic variables on CDT scores (post hoc analyses)

Variables Rouleau (M ± SDt/F Post hoc Cahn (M ± SDt/F Post hoc Babins (M ± SDt/F Post hoc 
Age 
 A: 25–49 9.22 ± 1.373  A ≠ B, C, D 8.73 ± 1.940  A ≠ B, C, D 16.44 ± 2.329  A ≠ B, C, D 
 B: 50–64 8.65 ± 1.690 F(3, 626) = 32.959, p ≤ .001 B ≠ C, D 7.93 ± 2.316 F(3, 626) = 31.434, p ≤ .001 B ≠ C, D 14.99 ± 3.101 F(3, 626) = 38.181, p ≤ .001 B ≠ C, D 
 C: 65–74 7.89 ± 2.284   6.99 ± 2.885   13.62 ± 4.232   
 D: ≥75 6.96 ± 2.457   5.76 ± 3.267   12.08 ± 4.540   
Education 
 A: 1–4 7.47 ± 2.329  A ≠ B, C, D 6.36 ± 2.996  A ≠ B, C, D 12.71 ± 4.238  A ≠ B, C, D 
 B: 5–9 8.72 ± 1.709 F(3, 626) = 46.605, p ≤ .001 B ≠ D 8.04 ± 2.308 F(3, 626) = 51.144, p ≤ .001 B ≠ C, D 15.36 ± 2.967 F(3, 626) = 65.625, p ≤ .001 B ≠ C, D 
 C: 10–12 9.29 ± 1.112   8.87 ± 1.598   16.59 ± 1.766   
 D: >12 9.54 ± 0.877   9.24 ± 1.261   17.03 ± 1.330   
Gender 
 Women 8.32 ± 2.054 t(628) = 2.330, p ≤ .05 — 7.51 ± 2.739 t(628) = 2.585, p ≤ .01 — 14.53 ± 3.748 t(628) = 2.466, p ≤ .05 — 
 Men 8.71 ± 1.861   8.08 ± 2.454   15.28 ± 3.477   
Geographic region 
 A: North 8.48 ± 1.955 F(4, 625) = 2.106, p = .079 — 7.69 ± 2.712 F(4, 625) = 1.130, p = .341 — 14.82 ± 3.568 F(4, 625) = 2.961, p ≤ .05 C ≠ E 
 B: Center 8.44 ± 1.932   7.73 ± 2.452   14.49 ± 3.767   
 C: Lisbon 8.65 ± 1.954   7.94 ± 2.670   15.42 ± 3.395   
 D: Alentejo 8.30 ± 2.195   7.45 ± 2.857   14.55 ± 3.915   
 E: Algarve 7.13 ± 2.722   6.53 ± 2.973   12.53 ± 4.984   
Geographic location 
 Coast 8.48 ± 2.004 t(628) = −0. 388, p = .698 — 7.78 ± 2.613 t(628) = −1.369, p = .172 — 14.81 ± 3.721 t(628) = −0.108, p = .914 — 
 Inland 8.39 ± 1.943   7.39 ± 2.826   14.77 ± 3.397   
Residence area 
 A: PUA 8.58 ± 1.912 F(2, 627) = 2.599, p = .075  7.90 ± 2.558 F(2, 627) = 4.557, p ≤ .05  15.03 ± 3.526 F(2, 627) = 2.489, p = .084  
 B: MUA 8.12 ± 2.155  — 7.06 ± 3.014  A ≠ B 14.35 ± 3.876   
 C: PRA 8.34 ± 2.126   7.66 ± 2.496   14.31 ± 3.985   
Variables Rouleau (M ± SDt/F Post hoc Cahn (M ± SDt/F Post hoc Babins (M ± SDt/F Post hoc 
Age 
 A: 25–49 9.22 ± 1.373  A ≠ B, C, D 8.73 ± 1.940  A ≠ B, C, D 16.44 ± 2.329  A ≠ B, C, D 
 B: 50–64 8.65 ± 1.690 F(3, 626) = 32.959, p ≤ .001 B ≠ C, D 7.93 ± 2.316 F(3, 626) = 31.434, p ≤ .001 B ≠ C, D 14.99 ± 3.101 F(3, 626) = 38.181, p ≤ .001 B ≠ C, D 
 C: 65–74 7.89 ± 2.284   6.99 ± 2.885   13.62 ± 4.232   
 D: ≥75 6.96 ± 2.457   5.76 ± 3.267   12.08 ± 4.540   
Education 
 A: 1–4 7.47 ± 2.329  A ≠ B, C, D 6.36 ± 2.996  A ≠ B, C, D 12.71 ± 4.238  A ≠ B, C, D 
 B: 5–9 8.72 ± 1.709 F(3, 626) = 46.605, p ≤ .001 B ≠ D 8.04 ± 2.308 F(3, 626) = 51.144, p ≤ .001 B ≠ C, D 15.36 ± 2.967 F(3, 626) = 65.625, p ≤ .001 B ≠ C, D 
 C: 10–12 9.29 ± 1.112   8.87 ± 1.598   16.59 ± 1.766   
 D: >12 9.54 ± 0.877   9.24 ± 1.261   17.03 ± 1.330   
Gender 
 Women 8.32 ± 2.054 t(628) = 2.330, p ≤ .05 — 7.51 ± 2.739 t(628) = 2.585, p ≤ .01 — 14.53 ± 3.748 t(628) = 2.466, p ≤ .05 — 
 Men 8.71 ± 1.861   8.08 ± 2.454   15.28 ± 3.477   
Geographic region 
 A: North 8.48 ± 1.955 F(4, 625) = 2.106, p = .079 — 7.69 ± 2.712 F(4, 625) = 1.130, p = .341 — 14.82 ± 3.568 F(4, 625) = 2.961, p ≤ .05 C ≠ E 
 B: Center 8.44 ± 1.932   7.73 ± 2.452   14.49 ± 3.767   
 C: Lisbon 8.65 ± 1.954   7.94 ± 2.670   15.42 ± 3.395   
 D: Alentejo 8.30 ± 2.195   7.45 ± 2.857   14.55 ± 3.915   
 E: Algarve 7.13 ± 2.722   6.53 ± 2.973   12.53 ± 4.984   
Geographic location 
 Coast 8.48 ± 2.004 t(628) = −0. 388, p = .698 — 7.78 ± 2.613 t(628) = −1.369, p = .172 — 14.81 ± 3.721 t(628) = −0.108, p = .914 — 
 Inland 8.39 ± 1.943   7.39 ± 2.826   14.77 ± 3.397   
Residence area 
 A: PUA 8.58 ± 1.912 F(2, 627) = 2.599, p = .075  7.90 ± 2.558 F(2, 627) = 4.557, p ≤ .05  15.03 ± 3.526 F(2, 627) = 2.489, p = .084  
 B: MUA 8.12 ± 2.155  — 7.06 ± 3.014  A ≠ B 14.35 ± 3.876   
 C: PRA 8.34 ± 2.126   7.66 ± 2.496   14.31 ± 3.985   

Notes: CDT = Clock Drawing Test; M = mean; SD = standard deviation; PUA = predominantly urban area; MUA = moderately urban area; PRA = predominantly rural area.

We performed an ANCOVA to assess if the differences found in the CDT scores due to gender and geographic variables persisted once we had controlled the effects of age and education. The results showed that gender maintained its significant effect across all scoring systems but with a low effect size: Rouleau, F(1, 626) = 7.431, p < .01, forumla = .012, Cahn, F(1, 626) = 9.268, p < .01, forumla = .015, and Babins, F(1, 626) = 9.069, p < .01, forumla = .014 (Cohen, 1988). A similar effect was found for the residential area in the Cahn system, F(2, 622) = 3.497, p = .031, forumla=0.011. As for the Babins scoring system, the results showed that there was no significant effect of geographic region when we use age and education as covariates, F(4, 623) = 0.677, p = .608 (Table 4).

Table 4.

Analysis of group differences in the CDT scores (with covariates controlled)

Scoring system  Variables  Covariates  ANCOVA  Effect size 
Rouleau  Age  Education  F(2, 626) = 17.108, p ≤ .001  Small
forumla = 0.052 
Education  Age  F(3, 625) = 21.714, p ≤ .001  Small
forumla = 0.094 
Gender  Age
Education 
F(1, 626) = 7.668, p ≤ .01  Small
forumla = 0.012 
Cahn  Age  Education  F(2, 626) = 14.580, p ≤ .001  Small
forumla = 0.045 
Education  Age  F(3, 625) = 25.005, p ≤ .001  Small
forumla = 0.107 
Gender  Age
Education 
F(1, 626) = 9.556, p ≤ .01  Small
forumla = 0.015 
Babins  Age  Education  F(2, 626) = 18.327, p ≤ .001  Small
forumla = 0.055 
Education  Age  F(3, 625) = 32.505, p ≤ .001  Small
forumla = 0.135 
Gender  Age
Education 
F(1, 626) = 9.411, p ≤ .01  Small
forumla = 0.015 
Scoring system  Variables  Covariates  ANCOVA  Effect size 
Rouleau  Age  Education  F(2, 626) = 17.108, p ≤ .001  Small
forumla = 0.052 
Education  Age  F(3, 625) = 21.714, p ≤ .001  Small
forumla = 0.094 
Gender  Age
Education 
F(1, 626) = 7.668, p ≤ .01  Small
forumla = 0.012 
Cahn  Age  Education  F(2, 626) = 14.580, p ≤ .001  Small
forumla = 0.045 
Education  Age  F(3, 625) = 25.005, p ≤ .001  Small
forumla = 0.107 
Gender  Age
Education 
F(1, 626) = 9.556, p ≤ .01  Small
forumla = 0.015 
Babins  Age  Education  F(2, 626) = 18.327, p ≤ .001  Small
forumla = 0.055 
Education  Age  F(3, 625) = 32.505, p ≤ .001  Small
forumla = 0.135 
Gender  Age
Education 
F(1, 626) = 9.411, p ≤ .01  Small
forumla = 0.015 

Multiple linear regression analysis, enter method, was used to compare the effects of age and education in the CDT score and to examine the additional contribution of significant variables such as depressive symptoms and their interaction. Gender was excluded from this level of analysis due to its low effect sizes in the three scoring systems. We performed preliminary analyses to check the linearity, multicolinearity and homoscedasticity assumptions. Results confirmed that education and age had a significant contribution for the prediction of CDT scores according to the Rouleau scoring system, F(2, 627) = 82.598, p ≤ .001. The same result was found for the Cahn system, F(2, 627) = 87.223, p ≤ .001, and for the Babins system, F(2, 627)=112.199, p ≤ .001. Depressive symptoms did not have a significant effect on the prediction of CDT scores on the Rouleau, β = −0.034, t(614) = −0.908, p = .364, Cahn, β = −0.024, t(614) = −0.628, p = .530, and Babins, β = −0.022, t(614) = −0.602, p = .547, scoring systems. The adjusted R2 value was .209 for the Rouleau scoring system, .218 for the Cahn system, and .261 for the Babins system, indicating that approximately 21%, 22%, and 26% of the scores' variance in the selected scoring systems was explained by this model (Table 5). TeLPI results were not included in these models due to its high correlation with education, thus increasing the risk of multicolinearity. An exploratory second model included TeLPI raw score and age as predictors. The results showed that both age and TeLPI scores have a significant effect on the prediction of CDT scores on the Rouleau, F(2, 360) = 62.202, p ≤ .001, Cahn, F(2, 360) = 68.672, p ≤ .001, and Babins, F(2, 360) = 72.673, p ≤ .001, scoring systems. The adjusted R2 value was .253 for the Rouleau scoring system, .272 for the Cahn system, and .284 for the Babins system, indicating that approximately 25%, 27%, and 28% of the scores' variance in the selected scoring systems was explained by this model (Table 6).

Table 5.

Results of multiple linear regression: influence of education and age in predicting CDT score

Scoring system Variable B Standard error β Adjusted R2 
Rouleau and colleagues (1992) Education Age 0.121 −0.032 0.0170.005 0.288**−0.248** 0.209 
Education 0.176 0.016 0.405** 0.163 
Age −0.049 0.005 −0.377** 0.141 
Cahn and colleagues (1996) Education Age 0.171 −0.042 0.0220.007 0.307**−0.240** 0.218 
Education 0.243 0.021 0.421** 0.176 
Age −0.065 0.006 −0.372** 0.141 
Babins and colleagues (2008) Education Age 0.260 −0.063 0.0300.009 0.337**−0.265** 0.261 
Education 0.371 0.028 0.463** 0.213 
Age −0.100 0.009 −0.416** 0.172 
Scoring system Variable B Standard error β Adjusted R2 
Rouleau and colleagues (1992) Education Age 0.121 −0.032 0.0170.005 0.288**−0.248** 0.209 
Education 0.176 0.016 0.405** 0.163 
Age −0.049 0.005 −0.377** 0.141 
Cahn and colleagues (1996) Education Age 0.171 −0.042 0.0220.007 0.307**−0.240** 0.218 
Education 0.243 0.021 0.421** 0.176 
Age −0.065 0.006 −0.372** 0.141 
Babins and colleagues (2008) Education Age 0.260 −0.063 0.0300.009 0.337**−0.265** 0.261 
Education 0.371 0.028 0.463** 0.213 
Age −0.100 0.009 −0.416** 0.172 

Note: CDT = Clock Drawing Test.

**p ≤ .001.

Table 6.

Results of multiple linear regression: influence of age and TeLPI total number of errors in predicting CDT score

Scoring system Variable B Standard error β Adjusted R2 
Rouleau and colleagues (1992) TeLPI Age −0.083 −0.035 0.0110.006 −0.359**−0.284** .253 
TeLPI −0.098 0.011 −0.425** .178 
Cahn and colleagues (1996) TeLPIAge −0.120−0.047 0.0150.008 −0.381**−0.284** .272 
TeLPI −0.141 0.015 −0.447** .198 
Babins and colleagues (2008) TeLPIAge −0.152−0.068 0.0190.010 −0.369**−0.313** .284 
TeLPI −0.182 0.019 −0.442** .193 
Scoring system Variable B Standard error β Adjusted R2 
Rouleau and colleagues (1992) TeLPI Age −0.083 −0.035 0.0110.006 −0.359**−0.284** .253 
TeLPI −0.098 0.011 −0.425** .178 
Cahn and colleagues (1996) TeLPIAge −0.120−0.047 0.0150.008 −0.381**−0.284** .272 
TeLPI −0.141 0.015 −0.447** .198 
Babins and colleagues (2008) TeLPIAge −0.152−0.068 0.0190.010 −0.369**−0.313** .284 
TeLPI −0.182 0.019 −0.442** .193 

Notes: CDT = Clock Drawing Test; TeLPI = Teste de Leitura de Palavras Irregulares [Irregular Word Reading Test].

**p ≤ .001.

The multiple regression results confirmed the need to consider both age and education in the development of normative data for the Portuguese population. The values were determined and stratified according to the distribution of each variable—we considered the subjects' education, divided into four levels (1–4 years, 5–9 years, 10–12 years, and more than 12 years) and three age groups: 25–49, 50–64, and over 65 years; additionally, subjects with the lowest educational level (1–4 years) were divided into two age groups, 65–74 and over 75 years. This subdivision was made for the lower educational category because 63% of subjects over 65 years had achieved 1–4 years of education when compared with 42% of subjects with 50–64 years and 12% of subjects with 25–49 years. The results are expressed as the mean ± SD; values below 1 SD, 1.5 SD, and 2 SD (Tables 7–9).

Table 7.

Normative data for Rouleau and colleagues (1992) CDT scoring system according to age and education

Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N 25 89 — 87 49 250 
1–4 8.44 ± 1.938 8.06 ± 1.927 7.15 ± 2.518 6.49 ± 2.408 7.47 ± 2.329 
SDa 7, 6, 5 6, 5, 4 5, 3, 2 4, 3, 2 5, 4, 3 
N 63 58 44 — — 165 
5–9 9.02 ± 1.591 8.78 ± 1.545 8.23 ± 1.987 8.72 ± 1.709 
SDa 7, 7, 6 7, 7, 6 6, 5, 4 7, 6, 5 
N 55 33 18 — — 106 
10–12 9.35 ± 1.142 9.30 ± 1.015 9.11 ± 1.231 9.29 ± 1.112 
SDa 8, 8, 7 8, 8, 7 8, 7, 7 8, 8, 7 
N 60 31 18 — — 109 
>12 9.65 ± 0.755 9.45 ± 1.121 9.33 ± 0.767 9.54 ± 0.877 
SD 9, 9, 8 8, 8, 7 9, 8, 8 9, 8, 8 
N 203 211 216 — — 630 
All education 9.22 ± 1.373 8.65 ± 1.690 7.56 ± 2.381 8.46 ± 1.993 
SDa 8, 7, 7 7, 6, 5 5, 4, 3 7, 6, 5 
Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N 25 89 — 87 49 250 
1–4 8.44 ± 1.938 8.06 ± 1.927 7.15 ± 2.518 6.49 ± 2.408 7.47 ± 2.329 
SDa 7, 6, 5 6, 5, 4 5, 3, 2 4, 3, 2 5, 4, 3 
N 63 58 44 — — 165 
5–9 9.02 ± 1.591 8.78 ± 1.545 8.23 ± 1.987 8.72 ± 1.709 
SDa 7, 7, 6 7, 7, 6 6, 5, 4 7, 6, 5 
N 55 33 18 — — 106 
10–12 9.35 ± 1.142 9.30 ± 1.015 9.11 ± 1.231 9.29 ± 1.112 
SDa 8, 8, 7 8, 8, 7 8, 7, 7 8, 8, 7 
N 60 31 18 — — 109 
>12 9.65 ± 0.755 9.45 ± 1.121 9.33 ± 0.767 9.54 ± 0.877 
SD 9, 9, 8 8, 8, 7 9, 8, 8 9, 8, 8 
N 203 211 216 — — 630 
All education 9.22 ± 1.373 8.65 ± 1.690 7.56 ± 2.381 8.46 ± 1.993 
SDa 8, 7, 7 7, 6, 5 5, 4, 3 7, 6, 5 

Notes: CDT = Clock Drawing Test; SD = standard deviation.

aCDT values below 1 SD, 1.5 SD, and 2 SD.

Table 8.

Normative data for Cahn and colleagues (1996) CDT scoring system according to age and education

Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N1–4SDa 257.48 ± 2.5685, 4, 2 897.04 ± 2.6154, 3, 2 — 876.06 ± 3.0743, 2, −1 495.08 ± 3.2332, 0, −1 2506.36 ± 2.9963, 2, 0 
N5–9SDa 638.40 ± 2.2116, 5, 4 588.10 ± 2.0236, 5, 4 447.43 ± 2.6975, 3, 2 — — 1658.04 ± 2.3087, 6, 5 
N10–12SDa 558.96 ± 1.6337, 7, 6 338.82 ± 1.5707, 7, 6 188.67 ± 1.6097, 6, 6 — — 1068.87 ± 1.5987, 7, 6 
N>12SD 609.38 ± 1.1668, 8, 7 319.19 ± 1.470, 7, 6 188.83 ± 1.1508, 7, 7 — — 1099.24 ± 1.2618, 7, 7 
NAlleducationSDa 2038.73 ± 1.9407, 6, 5 2117.93 ± 2.3166, 5, 3 2166.56 ± 3.0724, 2, 0 — — 6307.72 ± 2.6515, 4, 2 
Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N1–4SDa 257.48 ± 2.5685, 4, 2 897.04 ± 2.6154, 3, 2 — 876.06 ± 3.0743, 2, −1 495.08 ± 3.2332, 0, −1 2506.36 ± 2.9963, 2, 0 
N5–9SDa 638.40 ± 2.2116, 5, 4 588.10 ± 2.0236, 5, 4 447.43 ± 2.6975, 3, 2 — — 1658.04 ± 2.3087, 6, 5 
N10–12SDa 558.96 ± 1.6337, 7, 6 338.82 ± 1.5707, 7, 6 188.67 ± 1.6097, 6, 6 — — 1068.87 ± 1.5987, 7, 6 
N>12SD 609.38 ± 1.1668, 8, 7 319.19 ± 1.470, 7, 6 188.83 ± 1.1508, 7, 7 — — 1099.24 ± 1.2618, 7, 7 
NAlleducationSDa 2038.73 ± 1.9407, 6, 5 2117.93 ± 2.3166, 5, 3 2166.56 ± 3.0724, 2, 0 — — 6307.72 ± 2.6515, 4, 2 

Notes: CDT = Clock Drawing Test; SD = standard deviation.

aCDT values below 1 SD, 1.5 SD, and 2 SD.

Table 9.

Normative data for Babins and colleagues (2008) CDT scoring system according to age and education

Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N1–4 SDa 2514.88 ± 3.37012, 10, 8 8913.57 ± 3.57010, 8, 6 — 8712.10 ± 4.5888, 5, 3 4911.12 ± 4.3957, 5, 2  25012.71 ± 4.2389, 6, 4 
N 5–9 SDa 6315.92 ± 2.842 13,12, 10 5815.41 ± 2.47113, 12, 11 4414.48 ± 3.54711, 9, 7 — — 16515.36 ± 2.96712, 11, 9 
N10–12SDa 5516.91 ± 1.44416, 15, 14 3316.33 ± 1.84815, 14, 13 1816.11 ± 2.34914, 13, 11 — — 10616.59 ± 1.76615, 14, 13 
N>12SD 6017.22 ± 1.19516, 15, 15 3116.84 ± 1.69515, 14, 14 1816.72 ± 0.95816, 15, 15 — — 10917.03 ± 1.33016, 15, 14 
NAll educationSDa 20316.44 ± 2.32914, 13, 12 21114.99 ± 3.10112, 10, 9 21613.08 ± 4.3939, 7, 4 — — 63014.80 ± 3.66711, 10, 8 
Education (years) Age
 
All ages 
25–49 50–64 ≥65 65–74 ≥75 
N1–4 SDa 2514.88 ± 3.37012, 10, 8 8913.57 ± 3.57010, 8, 6 — 8712.10 ± 4.5888, 5, 3 4911.12 ± 4.3957, 5, 2  25012.71 ± 4.2389, 6, 4 
N 5–9 SDa 6315.92 ± 2.842 13,12, 10 5815.41 ± 2.47113, 12, 11 4414.48 ± 3.54711, 9, 7 — — 16515.36 ± 2.96712, 11, 9 
N10–12SDa 5516.91 ± 1.44416, 15, 14 3316.33 ± 1.84815, 14, 13 1816.11 ± 2.34914, 13, 11 — — 10616.59 ± 1.76615, 14, 13 
N>12SD 6017.22 ± 1.19516, 15, 15 3116.84 ± 1.69515, 14, 14 1816.72 ± 0.95816, 15, 15 — — 10917.03 ± 1.33016, 15, 14 
NAll educationSDa 20316.44 ± 2.32914, 13, 12 21114.99 ± 3.10112, 10, 9 21613.08 ± 4.3939, 7, 4 — — 63014.80 ± 3.66711, 10, 8 

Notes: CDT = Clock Drawing Test; SD = standard deviation.

aCDT values below 1 SD, 1.5 SD, and 2 SD.

Discussion

This study intended to explore the potentialities of the CDT in a broad spectrum of adulthood aging strata and to analyze the effect of sociodemographic variables in the CDT performance of Portuguese cognitively healthy subjects, recruited in the community. Our decision to select the 25 years as the lower age limit was based in the literature, as it is considered by several authors the beginning of adulthood and the age at which we supposedly reach the full development of cognitive abilities (as measured by Wechsler Adult Intelligence Scales). This study population was community-based because it was developed with Portuguese habitants who resided in the community, had their lives completely integrated in the community, and were autonomous in their activities of daily living. As we referred, we started by recruiting the potential test subjects in national healthcare services and, in order to accomplish a rigorous selection, we previously established an agreement with the directors of the healthcare centers involved and all the physicians were instructed to select patients who fulfilled the inclusion/exclusion criteria. All the criteria were again revised by the neuropsychologists and, when it was possible, all information, including medical history, was revised with a “caregiver” (spouse, son, or other relative in cohabitation) in order to assure that the person was indeed completely autonomous, especially for the older subjects. Portugal has a high rate of older population and a high percentage of that population lives alone. Therefore, there are many older people who consciously choose to benefit from daycare centers facilities, despite being autonomous. A special attention was given to subjects recruited in such centers as we only selected autonomous subjects who spent only part of the day in the institution developing social activities. Once again, all inclusion/exclusion criteria were respected and confirmed by the daycare centers' directors to assure the inclusion of only cognitively healthy subjects. We also used a rigorous methodology in order to exclude participants with comorbidities that could influence cognitive performance as well subjects with objective cognitive impairment on the assessment battery. This explains the high rate of inclusion failures: of the 936 total subjects, 194 (20.73%) were excluded following the initial interview due mostly to the personal history of neurological or psychiatric disorders and 112 participants (±12%) were further excluded due to impairment or inability to complete the neuropsychological tests. Despite this high rate of initial exclusions, we consider that the process of recruitment did not influence the representativeness of the final sample or the interpretation of normative data. First, because the global recruitment was always guided by previous stratification, and second, the process was maintained until we obtained a representative sample of the Portuguese population according to all the selected variables (age, education, gender, geographic region, geographic localization, and inland/coastal areas).

Regardless of the absence of normative data for the CDT in our country, the test is very popular in Portugal and the three scoring systems studied are commonly used, justifying an adequate exploration of their psychometric properties and the development of Portuguese normative data for all of them. These scoring systems presented different values of internal consistency: Rouleau and Cahn systems showed inadequate internal consistency, whereas the Babins system showed good internal consistency. Although it is not the original purpose of the CDT, the scoring systems presented high correlations with global measures of cognition, which supported its potential as a cognitive screening instrument (Strauss et al., 2006) and moderate correlations with the Irregular Word Reading Test, which has been found in several studies with the CDT (e.g., Crowe, Allman, Triebel, Sawyer, & Martin, 2010; Hubbard et al., 2008).

The Babins and colleagues (2008) 18-point scoring system was the one that revealed better psychometric properties in all parameters evaluated among the selected systems. There are no other available studies in which the Babins scoring system was used with cognitively healthy subjects, a fact that limits the corroboration of our data. Nonetheless, our results are a new argument in favor of the use of this scoring system, which revealed a high specificity and sensitivity in differentiating AD patients from cognitively normal subjects as well as in the identification of MCI subjects who developed dementia. As referred, this system takes into account more details than the Rouleau system, including the assessment of circle integrity, number placement and sequencing, size and placement of the hands. According to the authors, the time representation task, which encompasses three main actions—correct placement of both hands, correct representation of their asymmetry, and correct orientation of the hours' hand, proved to be the most discriminative task among the four analyzed groups. In fact, it was able to differentiate MCI subgroups (progressors vs. non-progressors) which could be a good indicator of future cognitive decline. The discriminative capacity of the CDT in MCI patients remains controversial (Ehreke, Luppa, König, & Riedel-Heller, 2010); nonetheless, comparatively with the Cahn and colleagues (1996) scoring system, the Babins system appeared to be more informative and revealed higher discriminative capacity between MCI patients who progressed to dementia and those who remained stable (Babins et al., 2008; Duro, Freitas, Alves, Simões, & Santana, 2011; Freitas & Simões, 2010).

The use of a representative sample, stratified according to several levels of each demographic and geographic variable and with a very close distribution of the Portuguese population, allowed us to take more robust conclusions from the obtained data. Education and age were the variables that contributed more significantly to predict CDT scores according to the Babins and colleagues (2008) scoring system, explaining 26% of the total variance. According to Cohen (1988), this is considered a medium effect. The other scoring systems revealed proximate results, with age and education explaining 21% of the total variance for the Rouleau and colleagues (1992) system and 22% for the Cahn and colleagues (1996) system.

There was a linear positive relation between the CDT total score and the total number of years of education completed successfully. On the other hand, there was a linear negative relation with age. The influence of age has been systematically found in several studies published with the CDT (Bozikas et al., 2008; Caffarra et al., 2011; Hubbard et al., 2008), confirming a negative effect on the subject's performance. The influence of education has been less explored but appears to be equally relevant, presenting an inverse pattern to a highest education corresponds a better performance (Bozikas et al., 2008; Hubbard et al., 2008; Lourenço et al., 2008). In the present study, the results of multiple linear regression supported the need to considerate both age and education for the establishment of normative data. These results can be related to the existence of an important segment of the study sample with a low educational level, typical of the Portuguese population. The effect of gender on the CDT performance, although described in the literature (Strauss et al., 2006), has not been explored in recent studies (Bozikas et al., 2008; Caffarra et al., 2011). In the present study, gender had a significant effect, with men scoring higher than women. Nonetheless, this was considered a minor effect (Cohen, 1988) and it justified its non-inclusion as a criterion for the establishment of normative data for the Portuguese population. The analysis of TeLPI results showed that, together with age, it significantly explained CDT scores in the Portuguese population: the model explained approximately 25%, 27%, and 28% of the total variance for the Rouleau, Cahn, and Babins systems, respectively. It will be interesting to compare these same results in future studies, especially with well-defined clinical samples. There are very few Portuguese studies that explored the influence of geographic variables in cognitive screening tests. Recently, Freitas and colleagues (2011) concluded that geographic region (NUTS-II classification: North, Center, Lisbon, Alentejo, and Algarve), geographical localization (coastal and inland areas), and residential area (according to the Types of Urban Areas: PUA, MUA, and PRA) have a non-significant or reduced effect on MoCA scores after controlling for the effect of age and education (Freitas et al., 2011). In the present study, the variables considered, namely geographic region, geographical localization, and residential area, did not show a significant influence in the CDT score according to the three scoring systems selected once we controlled the effects of age and education. Considering the clinical variables, depressive symptoms presented significant and negative correlations with CDT scores; nonetheless, according to the results of multiple linear regression analysis, there was no significant effect of depressive symptoms in our sample. However, these data should be interpreted with caution as they result from the analysis of performance of cognitively healthy and non-depressed to mildly depressed individuals. Therefore, these findings should not be generalized to individuals with clinical conditions.

To establish normative data, we considered the subjects' education, divided into four levels, and three age groups: 25–49, 50–64, and over 65 years. Additionally, subjects with the lowest educational level (1–4 years) were subdivided into two age groups, 65–74 and over 75 years. We presented the values for scores less than 1 SD, 1.5 SD, and 2 SD. These values found cannot be corroborated, since there are no other Portuguese studies that established normative data for the three CDT scoring systems used. Moreover, there are no publications of international studies comparing all the three systems. Regarding the Cahn and colleagues (1996) scoring system, other authors have demonstrated it to be a valid and reliable measure in several normative studies, where normative data were also established based on age and education (Hubbard et al., 2008). The Hubbard and colleagues (2008) study developed normative data for the Cahn and colleagues (1996) scoring system in a sample of 207 cognitively normal elderly aged 55–98. The results found on our study are relatively close to the ones found in the white sample of subjects when they accounted for both age and education in the development of norms. The Babins and colleagues (2008) study presented results for the Rouleau and colleagues (1992) 10-point system and their 18-point system. Our results for the older, highly educated subjects are slightly lower than the ones presented for the “normal” group; nevertheless, differences in sample size, age ranges, and mean education limit the comparison of such data. The Yamamoto and colleagues (2004) study also compared the Rouleau and Cahn scoring systems and determined seven points as the optimal cutoff score for detecting MCI patients with the Cahn system. Our findings suggest that there is a great variability in CDT scores across different populations as many of our normal subjects would be misclassified as cognitively impaired using previously published cutoffs.

Several studies have shown a strong association between reading ability and cognitive test performance that has led other authors to develop normative data for the CDT based on the reading ability scores (Crowe et al., 2010; Hubbard et al., 2008). Although we found a strong association between TeLPI raw scores and CDT scores, in this study, normative data were based on age and education because the normative study of TeLPI for the Portuguese population is not yet available.

This study has some limitations. Although every effort was made to select community-based cognitively healthy subjects, the test battery selected for this study enabled us to rule out cases of moderate to severe dementia but it is possible that very mild dementia cases as well as MCI cases were included in the normative sample due to lack of sensitivity of the instruments. The use of MoCA's Portuguese cutoff scores allows us to state with a certain degree of confidence that the sample did not include subjects with cognitive impairment; nonetheless, that possibility cannot be completely discarded.

The inclusion of illiterate subjects was not considered, which limits the application of the CDT with this segment of the population. The decision was supported by the literature, given the strong possibility of floor effects. The low performance of illiterate subjects appears to be related to a low exposure to bi-dimensional representations and abstract graphic representation (Howieson, Loring, & Hannay, 2004); also, illiterate subjects tend to have poorer performances in several cognitive domains, particularly in tests that require pen or pencil to make drawings. The Portuguese population was characterized by a high prevalence of older people with low educational level or illiterates, whether young people had a mean level of education. As an illustrative example, 30 years ago, according to the 1981 Census (Instituto Nacional de Estatística, 2010), the percentage of illiteracy in Portugal was close to 26.4% and only 1.6% of the population held a university degree. In the last decades, the educational scenery has rapidly changed as a result of the reorganization of the school system and the imposition of higher obligatory educational plateaus. These changes are already reflected in the younger strata of the population studied; however, the older group continues to be characterized by a very low education level. The sample distribution could not completely eliminate this discrepancy; nonetheless, the distribution achieved is relatively close to the real one. We were also not able to subdivide the “over 65 years” group in the four educational levels as there were not enough subjects in each group. Such division was performed only for the lowest educational category (1–4 years) but our results showed that there were no significant differences between the two subgroups (65–74 and ≥75 years).

We intend to further explore the data regarding the Irregular Word Reading Test as several papers have shown a significant effect of reading ability in clock drawing performance in different countries (e.g., Crowe et al., 2010; Hubbard et al., 2008). Additionally, the findings about depressive symptoms would benefit from better operationalization of such variables. The inclusion of more specific and descriptive instruments may shed greater light on the influence of the variables on CDT performance. Furthermore, these findings must be complemented with studies that consider patients with cognitive impairment and those with depression.

These normative results will be useful in population or epidemiological studies to be developed in Portugal, allowing the use of the CDT as a brief cognitive screening test. Furthermore, the confirmed influence of education and age will be relevant for the future application of this scoring system to clinical groups with cognitive deterioration.

Funding

This work was supported by the Lundbeck Foundation.

Conflict of Interest

None declared.

Acknowledgements

We thank Ana Pedroso for the English revision of our manuscript.

References

Alves
L.
Simões
M. R.
Martins
C.
Teste de Leitura de Palavras Irregulares (TeLPI) [Irregular Word Reading Test]
 , 
2009
Coimbra, Portugal
Serviço de Avaliação Psicológica da Faculdade de Psicologia e de Ciências da Educação da Universidade de Coimbra [Psychological Assessment Department, Faculty of Psychological and Educational Sciences]
Babins
L.
Slater
M-E.
Whitehead
V.
Chertkow
H.
Can an 18-point clock-drawing scoring system predict dementia in elderly individuals with mild cognitive impairment?
Journal of Clinical and Experimental Neuropsychology
 , 
2008
, vol. 
30
 
2
(pg. 
173
-
186
)
Barreto
J.
Leuschner
A.
Santos
F.
Sobral
M.
Escala de Depressão Geriátrica [Geriatric Depression Scale]
Escalas e testes na demência [Scales and tests in dementia]
 , 
2008
Lisboa, Portugal
GEECD
(pg. 
69
-
72
Grupo de Estudos de Envelhecimento Cerebral e Demências [Study Group on Brain Aging and Dementia] (Eds.)
Bozikas
V. P.
Giazkoulidou
A.
Hatzigeorgiadou
M.
Karavatos
A.
Kosmidis
M. H.
Do age and education contribute to performance on the clock drawing test? Normative data for the Greek population
Journal of Clinical and Experimental Neuropsychology
 , 
2008
, vol. 
30
 
2
(pg. 
199
-
203
)
Brodaty
H.
Moore
C. M.
The Clock Drawing Test for dementia of the Alzheimer's type: A comparison of three scoring methods in a memory disorders clinic
International Journal of Geriatric Psychiatry
 , 
1997
, vol. 
12
 (pg. 
619
-
627
)
Caffarra
P.
Gardini
S.
Zonato
F.
Concari
L.
Dieci
F.
Copelli
S.
, et al.  . 
Italian norms for the Freedman version of the Clock Drawing Test
Journal of Clinical and Experimental Neuropsychology
 , 
2011
, vol. 
33
 
9
(pg. 
982
-
988
)
Cahn
D. A.
Salmon
D. P.
Monsch
A. U.
Butters
N.
Wiederholt
W. C.
Corey-Bloom
J.
Screening for dementia of Alzheimer type in the community: The utility of the Clock Drawing Test
Archives of Clinical Neuropsychology
 , 
1996
, vol. 
2
 (pg. 
529
-
539
)
Cahn-Weiner
D. A.
Williams
K.
Grace
J.
Tremont
G.
Westervelt
H.
Stern
R. A.
Discrimination of dementia with Lewy bodies from Alzheimer disease and Parkinson disease using the clock drawing test
Cognitive and Behavioral Neurology
 , 
2003
, vol. 
16
 
2
(pg. 
85
-
92
)
Cohen
J.
Statistical power analysis for the behavioral sciences
 , 
1988
2nd ed.
Hillsdale, NJ
Lawrence Erlbaum Associates
Cosentino
S.
Jefferson
A.
Chute
D.
Kaplan
E.
Libon
D. J.
Clock drawing errors in dementia: Neuropsychological and neuroanatomical considerations
Cognitive and Behavioral Neurology
 , 
2004
, vol. 
17
 
2
(pg. 
74
-
84
)
Crowe
M.
Allman
R. M.
Triebel
K.
Sawyer
P.
Martin
R. C.
Normative performance on an executive clock drawing task (CLOX) in a community-dwelling sample of older adults
Archives of Clinical Neuropsychology
 , 
2010
, vol. 
25
 (pg. 
610
-
617
)
Duro
D.
Freitas
S.
Alves
L.
Simões
M. R.
Santana
I.
O Teste do Desenho do Relógio na população portuguesa: Dados preliminares [The Clock Drawing Test in the Portuguese population: Preliminary data]
Sinapse
 , 
2011
, vol. 
11
 
2
pg. 
67
 
Ehreke
L.
Luppa
M.
König
H-H.
Riedel-Heller
S. G.
Is the Clock Drawing Test a screening tool for the diagnosis of mild cognitive impairment? A systematic review
International Psychogeriatrics
 , 
2010
, vol. 
22
 
1
(pg. 
56
-
63
)
Fisher
J. S.
Loring
D. W.
Lezak
M. D.
Howieson
D.
Loring
D. W.
Construction
Neuropsychological assessment
 , 
2004
4th ed.
New York
Oxford University Press
(pg. 
531
-
568
)
Folstein
M.
Folstein
S.
McHugh
P.
Mini mental state: A practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
 , 
1975
, vol. 
12
 (pg. 
189
-
198
)
Freedman
M.
Leach
L.
Kaplan
E.
Winocur
G.
Shulman
K. I.
Delis
D. C.
Clock drawing: A neuropsychological analysis
 , 
1994
, vol. 
Vol. 6
 
New York
Oxford University Press
Freitas
S.
Simões
M. R.
Teste do Desenho do Relógio: Utilidade e validade como instrumento de rastreio cognitivo [Clock Drawing Test: Utility and validity as a cognitive screening instrument]
Psicologia, Educação e Cultura [Psychology, Education and Culture]
 , 
2010
, vol. 
XIV
 
2
(pg. 
319
-
338
)
Freitas
S.
Simões
M. R.
Alves
L.
Santana
I.
Montreal Cognitive Assessment (MoCA): Normative study for the Portuguese population
Journal of Clinical and Experimental Neuropsychology
 , 
2011
, vol. 
33
 
9
(pg. 
989
-
996
)
Garcia-Caballero
A.
Recimil
M. J.
Darcía-Lado
I.
Gayoso
P.
Cardaso-Suárez
C.
González-Hermida
J.
, et al.  . 
ACE clock scoring: A comparison with eight standard correction methods in a population of low educational level
Journal of Geriatric Psychiatry and Neurology
 , 
2006
, vol. 
19
 (pg. 
216
-
219
)
Garrett
C.
Santos
F.
Tracana
I.
Barreto
J.
Sobral
M.
Fonseca
R.
Avaliação Clínica da Demência [Clinical Dementia Rating]
Escalas e Testes na Demência [Scales and tests in dementia]
 , 
2008
Lisboa, Portugal
GEECD
(pg. 
17
-
32
Grupo de Estudos de Envelhecimento Cerebral e Demências [Study Group on Brain Aging and Dementia] (Eds.)
Ginó
S.
Mendes
T.
Ribeiro
F.
Mendonça
A.
Guerreiro
M.
Garcia
C.
Escala de Queixas de Memória
Escalas e testes na demência [Scales and tests in dementia]
 , 
2008
Lisboa, Portugal
GEECD
(pg. 
117
-
120
Grupo de Estudos de Envelhecimento Cerebral e Demências [Study Group on Brain Aging and Dementia] (Eds.)
Goodglass
H.
Kaplan
E.
The assessment of aphasia and related disorders
 , 
1983
Philadelphia
Lea & Febiger
Guerreiro
M.
Contributo da neuropsicologia para o estudo das demências [Contribution of Neuropsychology for dementia study]
1998
Lisbon
Portugal Lisbon University
 
Unpublished Doctorship Thesis
Howieson
D. B.
Loring
D. W.
Hannay
J.
Lezak
M. D.
Howieson
D. B.
Loring
D. W.
Neurobehavioral variables and diagnostic issues
Neuropsychological assessment
 , 
2004
4th ed.
New York
Oxford University Press
(pg. 
286
-
334
)
Hubbard
E. J.
Santini
V.
Blankevoort
C. G.
Volkers
K. M.
Barruo
M. S.
Byerly
L.
, et al.  . 
Clock drawing performance in cognitively normal elderly
Archives of Clinical Neuropsychology
 , 
2008
, vol. 
23
 (pg. 
295
-
327
)
Hughes
C. P.
Berg
L.
Danziger
W. L.
Coben
L. A.
Martin
R. L.
A new clinical scale for the staging of dementia
British Journal of Psychiatry
 , 
1982
, vol. 
140
 (pg. 
566
-
572
)
Instituto Nacional de Estatística
Portal de Estatísticas Oficiais [Official Statistics Portal]
2010
 
Retrieved June 22, 2010, from http://www.ine.pt
Kaplan
E.
Bull
T.
Bryant
B. K.
A process approach to neuropsychological assessment
Clinical neuropsychology and brain function: Research, measurement, and practice
 , 
1988
Washington, DC
American Psychological Association
(pg. 
129
-
167
)
Libon
D. J.
Malamut
B. L.
Swenson
R.
Prouty Sands
L.
Cloud
B. S.
Further analyses of clock drawings among demented and nondemented older subjects
Archives of Clinical Neuropsychology
 , 
1996
, vol. 
11
 
3
(pg. 
193
-
205
)
Lourenço
R.
Ribeiro-Filho
S.
Moreira
I.
Paradela
E.
Miranda
A.
The Clock Drawing Test: performance among elderly with low educational level
Revista Brasileira de Psiquiatria
 , 
2008
, vol. 
30
 
4
(pg. 
309
-
315
)
Lowery
N.
Giovanni
L.
Mozley
L. H.
Arnold
S. E.
Bilker
W. B.
Gur
R. E.
, et al.  . 
Relationship between Clock-Drawing and neuropsychological and functional status in elderly institutionalized patients with schizophrenia
American Journal of Geriatric Psychiatry
 , 
2003
, vol. 
11
 
6
(pg. 
621
-
628
)
Mendez
M.
Ala
T.
Underwood
K.
Development of scoring criteria for the clock drawing task in Alzheimer's disease
Journal of the American Geriatrics Society
 , 
1992
, vol. 
40
 (pg. 
1095
-
1099
)
Menon
C.
Hall
J.
Hobson
V.
Johnson
L.
O'Bryant
S. E.
Normative performance on the executive clock drawing task in a multi-ethnic bilingual cohort: A project FRONTIER study
International Journal of Geriatric Psychiatry
  
.
Meyers
L. S.
Gamst
G.
Guarino
A. J.
Applied multivariate research: Design and interpretation
 , 
2006
Thousand Oaks, CA
Sage
Nasreddine
Z.
Phillips
N. A.
Bédirian
V.
Charbonneau
S.
Whitehead
V.
Collin
I.
, et al.  . 
The Montreal Cognitive Assessment, MoCA: A brief screening tool for Mild Cognitive Impairment
American Geriatrics Society
 , 
2005
, vol. 
53
 
4
(pg. 
695
-
699
)
Parsey
C. M.
Schmitter-Edgecombe
M.
Quantitative and qualitative analyses of the Clock Drawing Test in Mild Cognitive Impairment and Alzheimer Disease: Evaluation of a modified scoring system
Journal of Geriatric Psychiatry and Neurology
 , 
2011
, vol. 
24
 
2
(pg. 
108
-
118
)
Petersen
R. C.
Stevens
J. C.
Ganguli
M.
Tangalos
E. G.
Cummings
J. L.
DeKosky
S. T.
Practice parameter: Early detection of dementia: Mild cognitive impairment (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology
Neurology
 , 
2001
, vol. 
56
 
9
(pg. 
1133
-
1142
)
Rouleau
I.
Salmon
D. P.
Butters
N.
Kennedy
C.
McGuire
K.
Quantitative and qualitative analyses of clock drawings task
Journal of Neurology, Neurosurgery, and Psychiatry
 , 
1992
, vol. 
64
 (pg. 
588
-
594
)
Schmand
B.
Jonker
C.
Hooijer
C.
Lindeboom
J.
Subjective memory complaints may announce dementia
Neurology
 , 
1996
, vol. 
46
 
1
(pg. 
121
-
125
)
Shulman
K.
Clock-drawing: Is it the ideal cognitive screening test?
International Journal of Geriatric Psychiatry
 , 
2000
, vol. 
15
 (pg. 
548
-
561
)
Strauss
E.
Sherman
E.
Spreen
O.
A compendium of neuropsychological tests: Administration, norms, and commentary
 , 
2006
3rd ed.
New York
Oxford University Press
Sunderland
T.
Hill
J. L.
Mellow
A. M.
Lawlor
B. A.
Gundersheimer
J.
Newhouse
P. A.
, et al.  . 
Clock drawing in Alzheimer's disease: A novel measure of dementia severity
Journal of the American Geriatric Society
 , 
1989
, vol. 
37
 (pg. 
725
-
729
)
Yamamoto
S.
Mogi
N.
Umegaki
H.
Suzuki
Y.
Ando
F.
Shimokata
H.
, et al.  . 
The Clock Drawing Test as a valid screening method for mild cognitive impairment
Dementia and Geriatric Cognitive Disorders
 , 
2004
, vol. 
18
 (pg. 
172
-
179
)
Yesavage
J. A.
Brink
T. L.
Rose
T. L.
Lum
O.
Huang
V.
Adey
M.
, et al.  . 
Development and validation of a geriatric depression screening scale: A preliminary report
Journal of Psychiatric Research
 , 
1983
, vol. 
17
 
1
(pg. 
37
-
49
)