A new scoring system and norms for, and the performance of cognitively-unimpaired older adults on the cube copying test
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.246242
Source of Support: None, Conflict of Interest: None
Keywords: Construction, cube copying, dementia screening
Cube copying task is one of the neuropsychological tests suggested by the National Institute of Neurological and Communicative Disorders and Stroke and Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) for the clinical diagnosis of Alzheimer's disease (AD). A number of different perceptual-motor, cognitive, and sociocultural components converge during the act of drawing the cube. Construction involves a receptive component (visuoperceptual) and executive (practical) component. Cube copying also involves selective attention which is affected in AD. Figure-copying tests require analysis of a visual stimulus followed by its reproduction using motor executive skills. For analyzing a visual image, two different streams are involved, namely, a ventral stream projecting to the inferior temporal cortex for object perception and a dorsal stream connected to the parietal cortex for visuospatial control of movement. An impairment in drawing without any compromise in intelligence, visual, or motor capabilities can be caused by parietal lesions involving either of the hemispheres. Therefore, it is fair to infer that the motor executive skills involved in cube copying may be a reflection of parietal lobe function. Prior studies have shown that, along with other factors, the constructional ability in cube-copying is also related to verbal intelligence. In addition, drawing tasks are also a measure of executive function in dementia screening.
The objective of the present study was to present a slight modification of the existing scoring system on the cube copying task to make it more comprehensive, to analyze the performance of cognitively unimpaired elders using it, and to delineate the various factors affecting it. We also aimed to generate normative data for the task using the new scoring system in community-dwelling older adults with diverse literacy and socioeconomic status living in the southern Indian state of Kerala. The performance of the participants on the cube copying test was also compared with their performance on the Addenbrooke's Cognitive Examination (ACE), a global cognition screening battery, which has been adapted to the regional language (Malayalam).
The task consisted of copying the drawing of a three-dimensional (3D) wire-cube. The form of oblique projection used was ‘cabinet oblique’. In this projection, equal length is maintained of the horizontal and vertical lines; however, the oblique lines were half in length. This form of projection is preferred because maintaining the true length in oblique lines results in suppression of elongation in the third dimension. The length of the vertical line was 2.5 cm. Copying was allowed only once but no time limit was set for the attempt.
The wire-cube drawing is a two-dimensional (2D) representation of a three-dimensional (3D) figure. It is the varying combinations of angles, parallel lines, and faces that brings out the 3D effect. Based on these principles, we developed a scoring system by significantly modifying some of the existing scoring systems (Maeshma et al., and Brenner et al.). There are scores for each of the three elements in the figure, i.e., the number of corners (total eight), parallel lines (total twelve), and parallel faces (total six). To capture any inadequacy concerning the 3D concept in the drawing, we introduced scores for a fourth element, i.e., the overall 3D concept in the figure to capture if the information on the depth of the model was conveyed in the figure (total two). A corner was defined as a point at which three lines meet to form a vertex, and lines more than 3 mm off the intersection point were considered as being inaccurate. No scoring for error was done similar to the scoring system of Maeshma et al. The composite score on cube-copying is a sum of the subscores obtained on each of the four elements [Table 1], and the maximum possible composite score is 28. This scoring system provides for a wide range of scores, thus allowing the capture of subtle deficits in the task. For the purpose of this study, every drawing was evaluated independently by two raters (PSM and RM), and figures that were considered distorted by either of the raters were discussed in a consensus meeting. These were excluded from the final analysis only if both the raters agreed that the figure was grossly distorted and not possible to score. Cubes with a score of 0 for the 3D concept were also excluded.
The sampling frame consisted of community-based elders, ≥65 years of age, living in a predefined geographical area as per the 1991 census and the 1999 electoral list in Trivandrum district of Kerala in south India. This cohort participating in the “Cognition in Older Adults in Trivandrum” (COAT) study is being screened and followed up periodically over the last 8 years using elaborate neuropsychological, neurological, and clinical assessments as a part of the COAT study. The neuropsychological assessments included the Malayalam adaptation of the MMSE (m-MMSE) and the Malayalam adaptation of the ACE (m-ACE). The sample for this study consisted of cubes drawn in the m-ACE by 511 randomly selected individuals from among those who qualified as cognitively unimpaired in the COAT evaluation.
The mean age of the study sample was 69 ± 7.2 years, 62% were females, and 99% were right handed. The mean duration of education was 7.74 ± 5.45 years (range 0–25 years). Only 298 (57.84%) of the 511 participants attempted the cube drawing task. The rest refused to draw either due to their physical disability, such as visual impairment, or an inability to hold a pen to draw. Of the attempted cubes, 51 were rejected and excluded from the analysis because they were considered by consensus as having no resemblance to a cube and hence not scorable. These included a single line or no meaningful attempt to draw a cube, as shown in [Figure 1]. We categorized all those who either refused or drew unscorable drawings as “unsuccessful” at the cube-copying task. Thus, 247 participants, whose cubes were considered scorable and were accepted for and included in the analysis, were categorized as “successful” [Table 2]. [Figure 1] shows the selected samples of the cube copied by participants along with their ACE and MMSE scores. For interrater reliability, 50 participants were rated by a second rater (PSM) who was blinded to the results.
The Statistical Package for the Social Sciences (SPSS) [manufacturers: International business machines (IBM), USA]. software for Windows, version 11 was used. To assess the relation between various parameters, partial correlation as well as multinomial logistic regression analysis was used. A P value of <0.05 was considered significant.
Age was comparable across genders though men (with 10.25 ± 5.62 years of education) had significantly (P < 0.0001) more years of formal education than females (6.18 ± 4.72 years of education). Among the successful participants, 182 (74.5%), and among the unsuccessful, 54 (22.3%) had education of ≥9 years. In all, 75.5% of those with ≥9 years of education and 23.5% of those with <9 years of education were categorized as ‘successful’ at cube copying. Compared to the successful participants, age was significantly more (70.16 ± 7.9 vs 67.12 ± 6.01 years; P =<0.001) and education significantly less (4.8 ± 4.4 vs 10.89 ± 4.64 years) in those who were unsuccessful at cube-copying. A significant (P =<0.001) negative correlation was seen between lack of success at cube-copying and female sex even when controlled for age and education. An interrarter reliability of 0.79 (Kendall's tau-b) was seen.
In total, 62% subjects obtained a composite score of ≥66%. The composite score obtained by female subjects (17.90) was significantly lower than those by the male ones (21.57, P = 0.001), although on controlling for education, this difference was no longer significant (P = 0.204). Education significantly (Spearman's rho 0.524, P < 0.001) correlated with the composite score; however, age did not.
Only 59% of those who attempted their drawing obtained the maximum possible values on the concept subscore. Significant (P < 0.0003) correlation was seen between the subscores related to four elements –angles, parallels faces, parallel lines, and 3D concept. After controlling for age and gender, significant (Spearman's rho 0.469–0.546; P < 0.0001) correlation with education was seen for all the subscores. No correlation was seen between age and 3D concept (P = 0.703).
The composite score as well as subscores on cube-copying were significantly (Spearman's rho 0.434–0.529; P < 0.001) correlating with the m-MMSE and the m-ACE composite scores (Spearman's rho 0.502–0.820, P < 0.001). The composite score on cube-copying correlates well with 11 items on the m-ACE (calculation, recall, retrograde memory, naming, reading, construction, and writing) that accounts for 33% of m-MMSE and 57% of m-ACE composite scores. The subscore on corners correlated with that of 14 items on the m-ACE, accounting for 70% of the m-MMSE score and 82% of the m-ACE score.
To be able to compare across the different elements in cube-copying task, for every participant, we calculated the subscore on each of the four elements (parallel lines, parallel faces, corners, and concept) as a percentage of the maximum possible score (PMS) for that element. [Table 3] shows the PMS for each subscore. Across elements, PMS was the highest for parallel faces and the least for corners, suggesting that copying the parallel faces was the easiest, and copying the corners was the most difficult in the figure. Education was the single factor with significant effect on the composite score, and hence, the education-specific cut-off was calculated. [Table 4] shows the education-specific cut-off scores.
To determine the concurrent validity, the scoring system was administered by blinded raters to 190 cubes of patients diagnosed as having dementia, to determine the sensitivity and specificity of the test for detecting dementia. The area under the curve was 0.725 [Figure 1]. A cut-off score of 25 on the composite score offered an area under curve of 0.73 and a sensitivity of 81.9% and specificity of 53.9%. When the fifth percentile was taken as the cut-off value for each educational group, the test had a good sensitivity (83–99.2%) even in the low education group [Table 5].
The test showed a good sensitivity and modest specificity in diagnosing dementia. Sixty-two percent of the participants could obtain a composite score of >66%, as per the modified scoring system. A good interrater reliability was also seen. Copying a line diagram is simple and just involves replication of arrangement of lines. Copying of a rectilinear figure, such as a cube, involves replication of arrangement of lines as well as their orientation. This involves either the ability to replicate angles accurately or to use parallel alignment with the model. One of the difficulties often encountered while drawing a solid figure is in ordering the lines to avoid inclusion of hidden lines, or in arriving at the correct arrangement of intersections of lines. While copying a 3D picture, it has to be perceived that the lines in the drawing can represent a number of features of the objects, such as the intersection of faces, a boundary that marks the true extent of an object part, or an occluding contour that is the connected set of points at which a curved surface become invisible from a given stand point. Mistakes can be in the form of distortion of angles, failure to use parallel construction in complex figures, or difficulty in forming integrated representation of the figure. The scoring system was adopted and modified bearing these principles in mind. The copying-disabled patients may experience particular difficulty in encoding the pictorial structure of a model, which depicts a 3D object.
The cube-copying test using this scoring system showed a good correlation with the global cognition screening tests such as the ACE and MMSE. It has already been shown that geometric copying and handwriting skills decreased in direct proportion to decreased cognitive functioning. Patients with copying disability show greater impairment for the cube than the noncube figure, whereas patients without copying impairment and controls perform equally on the two figures. It has been found that children make more mistakes in copying a line diagram that represents a 3D object, such as a cube, than 2D objects. In one of the studies, the skills most sensitive to small changes in cognition were copying of cube and the pentagons. It has been proposed that quantitatively scored cube-copying tasks can roughly predict the nonverbal intelligence quotient (IQ) in aphasic patients. Failure in copying a cube has been shown to be useful in differentiating between normal and AD participants. More than just screening, the cube analysis has been found to be useful in differentiating the subtypes of dementia as well. Furthermore, visuospatial impairment may be particularly important to identify during the initial examination because this problem may be associated with difficulties in walking and other aspects of activities of daily living.
Forty-eight percent of the candidates could draw the cube successfully. One of the advantages of cube copying as a screening instrument is that it is language independent. The test had good utility among the literate because 80.7% of the literate subjects attempted the task. In one of the earlier studies, 40% of normal participants failed in drawing a cube to command, whereas only 17% failed in copying a cube. Previous studies have shown that gender influences cube copying performance along with other visuospatial parameters, with male subjects showing a better performance. In the current study, however, no significant difference in scores was seen between male and female subjects after controlling for education. Age and education, however, were found to significantly influence performance. These findings are in conformity with prior studies.,,, Hence, age and education specific cut-off values are needed for the appropriate interpretation of performance.
Among various subscores, the subscore on corners correlate with the performance on maximum number of items on global cognitive scales such as the ACE and MMSE. It has been postulated that, while copying, an individual recognizes the line diagram as representing the set of square faces constituting the cube and then replicates the pattern of lines in front of them. This suggests that perhaps copying of faces is easier and replicating angles more difficult in the entire task for the participant, which is confirmed in this study where across the elements, the score on face drawing was the highest and the one on angle replication was the lowest.
One of the strengths of this scoring system is that it introduces credit for the overall 3D concepts, which is an essential component in the drawing of 3D figures. However, this study did not show any added advantage for this additional score. This may be partly because the score for the 3D concept only formed 7% of the total score (2 out of 28). A small limitation of the scoring system, however, is that the alignment of the picture in relation to the paper is not analyzed or scored.
This study suggests that the cube-copying test with this scoring system can be used as a rapid screening test for dementia. In addition, it has the advantage of not being culture- and language-dependent. This may be of utility in situ ations where MMSE or ACE cannot be used. However, it is also important to recognize its limitations in illiterate subjects, as well as its low specificity.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]