Belinda Ann Dridan, School of Psychology, La Trobe University, Bundoora, Vic. 3086, Australia. Fax: +61 5428 1469; email: email@example.com
The Simple Copy Task (SCT) is a figure copying test with inherent appeal due to its short administration time, graded task difficulty, varied stimuli, and potential to eliminate floor effects. Despite this, there is a lack of data regarding its construct validity and diagnostic utility. The present study compared SCT performance of schizophrenia (n = 29), dementia (n = 64), and movement disorder (n = 12) groups with that of unmatched healthy control participants (n = 49). Movement disorder patients committed a high degree of misplacement errors, whereas dementia patients tended to omit items, add extraneous detail, and perseverate. The schizophrenia group was most similar to the dementia group in their performance on the SCT, committing primarily omission and perseveration errors. The SCT was most closely related to the Rey Complex Figure Task (r = 0.68, p < .01) and the Block Design Task (r = 0.62, p < .01). Age (r = −0.14, p < .01) and education (r = 0.35, p < .01) effects were present; however, there was no impact of gender and handedness. Taken together, these findings provide support for the utility of the task and directions for clinical interpretation.
1Schizophrenia, dementia, and movement disorder patients have demonstrable difficulty with figure copying tasks.
2Examination of error types on figure copying tasks can provide information regarding the source of poor quantitative scores.
3Various figure copying tasks have been reported to be affected by details other than neurological diagnosis, including demographic variables such as age and education.
What this paper adds
1The Simple Copy Task overall score is better than the Rey Complex Figure Task at distinguishing between healthy control and schizophrenia groups; however, both tasks have difficulty distinguishing between groups where there is relatively high impairment in higher order visual processing (i.e., dementia and movement disorders).
2Misplacement type errors on the Simple Copy Task are most common in movement disorder patients.
3Schizophrenia patients perform similarly to dementia patients on the Simple Copy Task.
Higher order visual processing—the collective of visuoperceptual, spatial, and constructional skills—is the cognitive domain that enables one to process spatial location, integrate visual components into a gestalt, and interact effectively with the visual environment (Benton & Tranel, 1992; Lezak, Howieson, & Loring, 2004; Sadock & Sadock, 2005). Neuropsychologists have long recognised the potential for figure copying tasks to provide valuable information regarding higher order visual processing (Lezak et al., 2004).
In schizophrenia, the underlying mechanisms leading to impaired figure copying performance are considered secondary to impairments in executive functioning, including planning (Howanitz et al., 1998) and working memory (Lee & Park, 2005). Despite this, there is some evidence to suggest that even in relatively stable, outpatient, medicated schizophrenia patients, there are demonstrable deficits in lower level visual processing tasks, such as shape discrimination and identification of spatial location (Tek et al., 2002). Thus, while it is clear that patients with schizophrenia have more difficulty than controls on higher order visual processing tasks (including figure copying), and that the impairment becomes more profound with increases in design complexity, the aetiology of the deficit is less clear.
This is likely owing to the manner in which figure copying tasks have been assessed in schizophrenia. Specifically, past research has focused on comparing an overall score on figure copying tasks between control and schizophrenia groups (Howanitz et al., 1998; Jogems-Kosterman et al., 2001). Such a comparison fails to address the qualitative nature of the errors in schizophrenia—if the groups were instead compared on error patterns (e.g., sizing, spatial placement, inattention type errors), the source of the deficit may be more clear.
Vascular dementia is similar to AD in the type of impairment in higher order visual processing skills (Looi & Sachdev, 1999). Both groups are equally impaired on perceptual and constructional tasks, and both exhibit poor scores on simple and complex figure copying (Looi & Sachdev, 1999). A further similarity between the two groups is that of perseveration, a failure to terminate a specific behaviour when appropriate (Hotz & Helm-Estabrooks, 1995). Figure reproductions have been successful in eliciting perseverative behaviour in both AD (Ryan et al., 1995) and vascular dementia patients (Lamar et al., 1997). Despite differences in the type of memory impairment seen in AD and vascular dementia groups, there are similarities in their higher order visual processing performance (Graham, Emery, & Hodges, 2004), which are pronounced on figure copying tasks.
Movement Disorders and Figure Copying
Although movement disorder patients, such as those with a diagnosis of Huntington's disease (HD) and Parkinson's disease (PD), demonstrate impaired higher order visual processing skills, they often differ from dementia groups in the types of errors made on figure copying tasks (Rouleau, Salmon, Butters, Kennedy, & McGuire, 1992). In particular, when copying a simple figure, HD patients make more errors of misplacement, imprecision, and size minimisation, whereas AD patients tend towards errors of perseveration and size enlargement (Rouleau et al., 1992).
Interestingly, the graphic errors seen in HD patients occur above and beyond that explained by physical aspects of the disability (e.g., chorea, motor disability; Rouleau et al., 1992), suggesting that the impairment is, in fact, due to higher order visual processing difficulties rather than due to a difficulty in drawing per se. In fact, HD patients tend to show impairments on complex visual processing tasks early in the disease process, and with disease progression, their scores demonstrate further decline (Gómez-Tortosa, del Barrio, Barroso, & García Ruiz, 1996).
Although HD and PD patients represent distinct types of movement disorders, they both share similar patterns of impairment on tasks measuring perceptual functioning (Lawrence, Watkins, Sahakian, Hodges, & Robbins, 2000). Deficits in higher order visual processing have been reflected in unmedicated PD patients' poor copies of a three-dimensional figure (Maeshima, Itakura, Nakagawa, Nakai, & Komai, 1997). Interestingly, the simple act of copying a cube has been effectively used in PD patients to predict functional impairment (Maeshima et al., 1997). A further frequently documented feature of PD is the presence of micrographia, which is present in figure reproductions even when patients are well controlled with medication (Kim, Lee, Park, Lee, & Na, 2005). This again parallels HD patients in that their figure reproductions tend to be notably smaller than the target figure (Rouleau et al., 1992).
In sum, it is clear that higher order visual processing difficulties are present in various clinical groups including dementia (AD and vascular dementia), movement disorder (HD and PD), and psychiatric groups (schizophrenia). These difficulties differ in nature between clinical groups, and tend to manifest as distinct error patterns on figure copying tasks. To date, there is little research directly comparing the error patterns of these specific clinical groups.
Problems With Existing Figure Copy Research
Existing research in neurological samples has generally focused on the copying of a simple, individual figure, such as a clock (Rouleau et al., 1992) or a cross (Guest & Fairhurst, 2002); a single complex figure (Freeman et al., 2000; Looi & Sachdev, 1999; Maeshima et al., 1997); or a group of difficult figures (Hécaen & Assal, 1970). There are several problems in assessing figure reproductions in these ways. First, administering figure/s with a limited range of difficulty (i.e., figures that are too simple or too complex) produces potential ceiling and floor effects. Consider the example of the Rey Complex Figure Task (RCFT; Osterrieth, 1944; Rey, 1941), a complicated copying task that has been documented to load on executive functioning, and in particular planning abilities (Smith & Zahka, 2006). Failure on the RCFT may not necessarily provide information regarding the patients' higher order visual processing deficit but may actually represent a decline in executive functioning (Smith & Zahka, 2006). Second, administration of an individual figure (e.g., the RCFT) does not allow for the clinician to observe a “pattern” of errors. The occurrence of a particular error type (e.g., perseveration) is much less compelling when observed in isolation than multiple times across a task. In the past, clinicians have employed tasks that have attempted to overcome these problems by including multiple figures that increase in difficulty (e.g., Bender-Gestalt test, Pascal & Suttell, 1951). However, these tasks have fallen out of favour in clinical practice due to their lengthy administration time (the Bender-Gestalt test is documented to take approximately half an hour to administer; Dibner & Korn, 1969). In sum, an effective figure copying task should (a) feature more than one figure, (b) consist of items that are graded in difficulty, and (c) be fast and easy to administer.
With these considerations in mind, the Simple Copy Task (SCT) may provide an attractive alternative to more commonly studied measures. The SCT is a figure copying test that requires the patient to reproduce a series of pictures presented in increasing levels of difficulty, beginning with simple geometric shapes and progressing on to three-dimensional and complex designs (see Figure 1). The origins of the task are unclear; however, it can be attributed in part to Australian neuropsychologist Dr Kevin Walsh, who demonstrated the effectiveness of many of the figures in eliciting higher order visual processing deficits (Walsh, 1985, 1987). The task is potentially appealing to both clinicians and patients for several reasons. First, it is inherently simpler than some of the more popular figure copying tests promoted in neuropsychology such as the RCFT, and may lessen the load on executive functioning and eliminate floor effects. Further, it provides a mix of simple and more difficult figures, allowing the examination of clinical error patterns to occur across figures. Finally, it is shorter than other commercially available tasks (e.g., the Bender-Gestalt test).
Surprisingly, despite the appeal of the SCT, there is no published data regarding the task. In particular, research is yet to be conducted on the SCT regarding construct validity, its ability to distinguish among various neurological groups, its robustness to potentially confounding demographic characteristics (e.g., gender, age), and normative data. Further, the SCT has yet to be explored as a tool for comparing higher order visual processing deficits in clinical groups (e.g., schizophrenia, dementia, and movement disorder patients). Given this, the aims of the current study pertaining to the SCT were (a) to examine construct validity, (b) to determine the impact of demographic variables (i.e., age, gender, education) on task performance, and (c) to examine differences in task performance across various neurological groups.
There were two sources of data used in the current study: hospital file records and a sample of healthy adults. Data for the clinical groups were gathered from the Bendigo Hospital John Lindell Rehabilitation Unit neuropsychological filing storage. In total, there were 1,137 files, dating from 1980 to 2010. Of these, there were 449 cases in which the SCT was used. For classification purposes, diagnoses were attained from the summary section of the neuropsychological report, which was formulated by the neuropsychologist based upon the following: neuropsychological test performance, imaging, medical records, patient history, and/or corroborative history. Note that in certain cases (e.g., AD), diagnostic classification is considered “probable” until confirmed at autopsy. In order to increase confidence in diagnosis for inclusion in the current study, unclear and multifactorial cases were excluded. Other exclusion criteria included the following: clinician-identified insufficient effort/malingering, age under 18 years, or rare presentations (fewer than five cases of a particular diagnostic category). In total, 79 cases were excluded, leaving 370 cases remaining, with an age range of 18–89.
The resulting clinical group comprised 18 different diagnoses, including alcohol-related brain injury (n = 13), AD (n = 33), anxiety disorder (n = 15), depression (n = 19), depression plus anxiety disorder (n = 29), frontotemporal dementia (FTD) (n = 11), HD (n = 5), hypoxia (n = 12), intellectual disability (n = 18), multiple sclerosis (n = 17), mixed Alzheimer's and vascular dementia (n = 23), schizophrenia (n = 30), PD (n = 9), vascular cognitive impairment (n = 20), vascular dementia (n = 18), right hemisphere stroke (n = 37), left hemisphere stroke (n = 23), and traumatic brain injury (n = 38).
For clarity of analysis and in order to attain sufficient statistical power, specific groups were combined into larger diagnostic categories of dementia (vascular dementia, AD, and mixed vascular and AD), movement disorders (PD and HD), and schizophrenia. Note that there was not sufficient theoretical background to justify including FTD patients into the dementia group. In particular, some research has reported AD patients are more impaired on spatial tasks than FTD patients (Mendez et al., 1996; Miller et al., 1997; Razani, Boone, Miller, Lee, & Sherman, 2001), while other studies have reported no difference between the groups (Grossi et al., 2002; Kramer et al., 2003). For this reason, the FTD group was not included in the dementia group.
An additional unmatched control group (n = 49) was recruited via community groups in the Melbourne metropolitan area. Participants were selected on the basis that they did not have a positive history of neurological or psychological disorder, and had normal (or corrected) vision and hearing.
Demographic variables including age, gender, handedness, and years of education for the four groups are presented in Table 1. The groups differed in terms of age, F (3, 163) = 67.10, p < .01, education, F (3,155) = 9.12, p < .01, and gender, x2(3, n = 167) = 11.31, p = .01, but not handedness, x2(3, n = 155) = 1.13, p = .77. Tukey's post hoc analyses indicated that the healthy control group had a higher level of education than did the dementia (p < .01), schizophrenia (p < .01), and movement disorder groups (p < .01). In addition, the healthy control group was older than the schizophrenia group (p < .01), and younger than the dementia group (p < .01), but not different with respect to age from the movement disorder group. Gender differences were not statistically significant, except between the movement disorder and dementia group (p = .02). Given the presence of between-group demographic differences, subsequent statistical analyses controlled for demographic variables where appropriate.
Table 1. Mean (M), Standard Deviation (SD), and Range of Demographic Characteristics of the Control and Clinical Subgroups
Note. Due to use of retrospective data, occasionally data were unavailable for one or more of the demographic variables.
The rationale for selection of neuropsychological tests was based upon two criteria, namely that (a) the test was part of the standard battery administered at Bendigo Health (as data were collected retrospectively), and (b) that the test measured non-verbal abilities or a construct that could have an impact upon neuropsychological test performance (e.g., depression or anxiety).
The measures included the SCT, the Austin Maze, the RCFT, the Clock Drawing Task, and performance subtests from the Wechsler Adult Intelligence Scale (WAIS, 3rd ed.; Wechsler, 1997), including block design, digit symbol coding, and the Picture Completion Task. The Depression Anxiety Stress Scale, short version (DASS-21; Lovibond & Lovibond, 1995) was also included in the battery to screen control participants for psychiatric symptoms that may have influenced the results. These measures were collected from the files of the retrospective cases, and administered to the control participants in a 45-min session, which took place in their homes.
Figures comprising the SCT are presented in Figure 1. Patients were required to reproduce, as accurately as possible, the figures presented by drawing them freehand. They were allowed to refer back to the target figures as much as they chose during the copying task, thus ruling out demands on memory. The current researchers designed a scoring system based upon systems used in other copying tasks, such as RCFT (Taylor, 1959), the visual reproduction task (Wechsler, 1981), and the Benton visual retention test (Benton, 1992). The total score is comprised of two subtotals: one that is figure-specific, and therefore requires inclusion of all the relevant parts that comprise each figure; and one that is general, and measures clinical errors that can occur on any figure regardless of the design. The clinical errors included absence of items, inclusion of extraneous items, perseveration (i.e., multiple retracing of a single line), misplacement of items, size distortions, rotation, and closing-in (i.e., overlapping the target figure). Possible attainable scores ranged from 0 to 68, with higher scores indicating better performance. Scoring for the study was completed by the principal researcher, with a subset (n = 20) rescored by a neuropsychology intern to ensure objectivity of the system. In the current sample, inter-rater reliability (r = 0.93) and internal consistency (Cronbach's α = 0.89) were both high.
The Austin Maze, designed by Milner (1965), is a test of visuospatial abilities, including visuospatial memory and learning (Crowe et al., 1999). The task was administered via the standard guidelines as presented in Walsh (1985). The outcome measure used in the study was the total number of errors across trials, with a lower score indicating better performance.
The RCFT can be described as a test of visual perception, visuospatial organisation, and motor functioning (Strauss, Sherman, & Spreen, 2006). Both the immediate copy condition and the 30-min delay condition were administered in the current study. The original scoring system featured in Taylor (1959) was used to score all RCFT figure reproductions. Total attainable scores ranged between 0 and 36. Inter-rater reliability of this system has been found an acceptable 0.88, while intra-rater reliability has been demonstrated to be an impressive 0.96 (Liberman, Stewart, Seines, & Gordon, 1994).
The WAIS-III subtests including the block design, picture completion, and Digit Symbol Coding Tasks were administered via the standard procedure outlined in the administration and scoring manual (Wechsler, 1997). Age-scaled scores were used in the analysis. The block design subtest is primarily considered a measure of constructional functioning; however, task completion can also be affected by processing speed and planning (Lezak et al., 2004). The picture completion subtest is a time-dependent measure of visual attention (Kurachi et al., 1994). The Digit Symbol Coding Task requires a sequence of coordinated motor responses (Goldstein, Johnson, & Minshew, 2001), and loads upon processing speed and incidental memory (Strauss et al., 2006). The matrix reasoning subtest was not included in the study as the data were sparsely available in the retrospective records.
Clock Drawing Task
In the Clock Drawing Task (CDT; Borod, Goodglass, & Kaplan, 1980), the participant is required to draw a clock on a blank piece of A4 paper. The participant is instructed to “draw the face of the clock, with all the numbers in the right place, and set the time to ten minutes past eleven” (Lezak et al., 2004). The task provides a measure of spatial and constructional functioning, but also loads on executive functioning ability (Strauss et al., 2006). Clock drawings were scored using the Rouleau et al. (1992) system, which relates to the correct placement of hands, presence and sequencing of the numbers, and quality of the clock face. The CDT has a total attainable score ranging from 0 to 10, with low scores indicating poor performance. Scoring with the Rouleau et al. system has been shown to have a 0.70 correlation with driving ability (Freund, Gravenstein, Ferris, Burke, & Shaheen, 2005), and has been effectively used to differentiate between mild Alzheimer's and intact older adults (Esteban-Santillan, Praditsuwan, Ueda, & Geldmacher, 1998), thus demonstrating predictive and ecological validity.
Ethical approval was obtained from Bendigo Health and the La Trobe University Ethics Committee. For the clinical group, data were transcribed and grouped as explained in the participants section. For the control group, informed consent was obtained and participants were tested in a 45-min session in their homes on the battery of neuropsychological tests described in the Measures section. Control participants were administered the tests in the following order: RCFT (copy condition), Austin Maze, DASS-21, Block Design Task, Digit Symbol Coding Task, RCFT (30-min delay), CDT, and the SCT. Data from the picture completion subtest were not available in the healthy group; however, it was obtainable from the Bendigo Health filing storage, and therefore is presented only for the clinical group.
The Effect of Gender, Age, Handedness, and Education on the SCT
The relationship among gender, age, handedness, education, and SCT scores was examined using Pearson's product moment correlation coefficients (see Table 2 for descriptive statistics and correlations between the variables). Preliminary analysis revealed the SCT score was negatively skewed, and thus it was treated with the reflect and square-root transformation. As such, Table 2 presents correlations in the reverse direction of their true meaning (for clarity, the true direction of the correlations is reported hereafter in text). Education had a medium strength positive relationship with SCT scores, indicating that as education increases, so too do scores on the SCT. Age had a small negative correlation with the SCT. Gender and handedness had very weak correlations with the SCT, which were not statistically significant.
Table 2. Intercorrelations and Descriptive Statistics of Simple Copy Task (SCT) Score and Demographic Variables
Note. M = mean; SCT = Simple Copy Task; SD = standard deviation.
Reflected and square-root transformed.
p < .001. The reflect transformation reverses the direction of the correlations with SCT. Figures presented in parenthesis represent untransformed descriptive statistics.
A standard multiple regression was conducted to further analyse the effect of age and education on the SCT. The analysis used all control cases, as well as all clinical cases with a complete SCT (n = 402) in order to gain sufficient statistical power. Preliminary analyses revealed the SCT total was negatively skewed, thus it was treated with the reflect and square-root transformation. Age and education significantly predicted SCT total score, R2 = 0.14, F (2, 384) = 29.77, p < .001 (see Table 3). Education (sr2 = 0.12, p < .001) made a larger contribution than age (sr2 = 0.02, p = .02).
Table 3. Standard Multiple Regression of Age and Education Predicting Simple Copy Task Total Score
Note. The Simple Copy Task total score was treated with the reflect and square-root transformation.
Construct Validity of the SCT
Correlations were investigated between SCT and other non-verbal measures of spatial ability (see Table 4). Again, to attain sufficient statistical power, all control participants and all clinical cases with a complete SCT were included in the analysis. Note that several of the measures violated the assumption of normality, and thus were treated with transformations (see Table 4 for details). Results indicated that the SCT correlated strongly with the RCFT copy condition, as well as the Block Design Task, indicating that patients with high scores on SCT would also have high scores on these tasks. The SCT was found to have moderate correlations with the Picture Completion Task, RCFT delay, CDT, and digit symbol coding. The correlation between the SCT and the Austin Maze task was small and, of note, it was the only measure that failed to attain a statistically significant relationship with the SCT (p = .26).
Table 4. Intercorrelations Between the Simple Copy Task (SCT) and Additional Neuropsychological Measures of Non-Verbal Ability
Note. SCT = Simple Copy Task; Pic Comp = Picture Completion Task; Rey1 = Rey Complex Figure Task, copy condition; Rey 2 = Rey Complex Figure Task, 30-min delay condition; Clock = Clock Drawing Task; Blocks = Block Design Task; Austin = total number of errors on the Austin Maze; Coding = Digit Symbol Coding Task. The following measures were negatively skewed and treated with the reflect transformation, thus reversing the true direction of the correlation: SCT, Rey1, Rey2, Clock. Picture Completion Task scores were positively skewed and corrected with a square-root transformation.
Given that the healthy control, schizophrenia, dementia, and movement disorder groups differed in terms of age and education (see Method section), and that age and education both had significant effects on SCT performance, it was necessary to control for these variables when analysing between-group differences on SCT score. For this reason, a one-way between-groups analysis of covariance (with age and education employed as the covariates) was conducted to compare SCT score among healthy control, schizophrenia, dementia, and movement disorder groups. After controlling for age and education, the groups were found to differ in their mean SCT scores, F (3, 148) = 24.77, p < .01, η2 = 0.33. Table 5 provides the sample sizes and descriptive statistics for SCT scores for each of the groups. Post hoc Tukey pairwise comparisons indicated that the healthy control group had a significantly higher mean score than the schizophrenia (p < .01), dementia (p < .01), and movement disorder (p < .01) groups. The schizophrenia group performed better than the movement disorder group (p = .03) but did not significantly differ from the dementia group (p = .52). Likewise, the movement disorder group and dementia group did not significantly differ in mean SCT scores (p = .39).
Table 5. Simple Copy Task Performance by Group
Note. CI = confidence interval; M = mean; SD = standard deviation.
It was also of interest to determine whether the RCFT was better able to distinguish between groups (i.e., dementia vs movement disorder; dementia vs schizophrenia) than the SCT. Again, given the differences in age and education between the groups, and the potential for demographic variables to impact upon figure copy task scores, it was necessary to conduct a one-way between-groups analysis of covariance (with age and education as covariates) to determine the difference in RCFT scores between groups (i.e., healthy control, schizophrenia, dementia, and movement disorders). Table 6 provides sample sizes and descriptive statistics for SCT score for each of the groups. There was a statistically significant difference in RCFT scores between groups after controlling for age and education, F (3, 141) = 10.21, p < .01, η2 = 0.18. Tukey pairwise comparisons revealed that the RCFT had many similar characteristics to the SCT: The mean scores (as demonstrated in Table 6) differed between healthy control and dementia groups (p < .01); healthy control and movement disorder groups (p < .01); schizophrenia and movement disorder groups (p = .04), but not the dementia and movement disorder group (p = .71), or the schizophrenia and dementia group (p = .18). However, unlike the SCT, the mean RCFT scores did not differ between healthy control or schizophrenia groups (p = .73). The RCFT total score was, therefore, no better than the SCT in distinguishing between neurological conditions (i.e., dementia, movement disorder, and schizophrenia), and was in fact worse at distinguishing between the schizophrenia and healthy control groups.
Table 6. Rey Complex Figure Task (RCFT) Performance by Group
Note. CI = confidence interval; M = mean; SD = standard deviation.
Error Patterns on the SCT
Analysis of error patterns on the SCT was conducted to examine whether there were underlying differences between the clinical groups that were obscured by using the total score. Initially, frequencies were inspected for each of the error types (i.e., absence of items, inclusion of extra items, perseveration, misplacement of items, distortions in size, rotation, and closing-in) among the different groups. Table 7 presents the percentage of people in each group who committed one or more of each of the errors on the SCT. Upon visual inspection, the neurological groups committed a higher frequency of all the error types than did the healthy controls.
Table 7. Percentage of People in the Healthy Control, Dementia, Schizophrenia, and Movement Disorder Groups Committing One or More Clinical Errors on the Simple Copy Task
Absent items (%)
Extra items (%)
Size distortion (%)
In order to discover the most important error types in predicting a diagnosis of various neurological disorders (i.e., movement disorder, dementia, and schizophrenia) while controlling for demographic differences between groups, three hierarchical logistic regressions were performed to compare the healthy controls with each of the clinical groups. In each case, where there were demographic differences between groups (see Method section), these variables were entered at the first step, and then the seven error types entered at the second step.
First, a hierarchical logistic regression analysis was conducted to predict membership to the healthy control (n = 49) or movement disorders group (n = 12). The final model, χ2 (3) = 37.48, p < .001, had 90% prediction success. After controlling for education and gender, the misplaced items error type was the single significant predictor of group membership (Wald statistic = 9.95, p < .01, eB = 12.63, 95% confidence interval [CI] [2.61, 61.10]). Figure 2 illustrates the characteristic misplacement errors in the movement disorder group.
Next, a hierarchical stepwise logistic regression was conducted to predict membership of the healthy control (n = 49) or dementia group (n = 69). The final model, χ2 (6) = 76.66, p < .001, had 83% prediction success. After controlling for age and education, there were significant contributions from absence of items (Wald statistic  = 7.19, p < .01, eB = 2.58, 95% CI [1.29, 5.17]), inclusion of extra items (Wald statistic  = 6.11, p= .01, eB = 4.09, 95% CI [1.34, 12.49]), and perseveration (Wald statistic  = 3.78, p = .05, eB = 5.29, 95% CI [0.99, 28.42]). Figure 3 demonstrates the error types that were common in the dementia group.
Finally, a hierarchical logistic regression analysis was conducted to predict membership to healthy control (n = 49) or schizophrenia group (n = 30). The model, χ2 (4) = 70.06, p < .001, had 92% prediction success. After controlling for age and education, there were two significant predictors of group membership: absence of items (Wald statistic  = 6.79, p < .01, eB = 4.95, 95% CI [1.49, 16.50]) and perseveration (Wald statistic  = 7.42, p = < .01, eB = 3.99, 95% CI [1.31, 5.91]). Figure 4 demonstrates the error types that were common in the schizophrenia group.
The Impact of Demographic Variables on the SCT
The results of the current study indicated that those who were younger and had higher levels of education tended to perform better on the SCT. However, these demographic variables explained only a relatively small amount of difference in the total score. Education was the strongest predictor, with age contributing only a very small portion of variance. The current study revealed that the SCT is unaffected by both gender and handedness.
In keeping with the findings of the current study, education effects have been found on a cube copying task, with research indicating that 6 years of formal education is required to complete an accurate reproduction of the three-dimensional figure (Shimada et al., 2006). Interestingly, Shimada et al. (2006) did not find that age had a significant effect on cube copying ability; however, their sample was restricted to older adults, and thus may not have encapsulated the tendency for younger adults to perform better on simple copying tasks. In a sample comprised of a broad age range, complex figure copying has been found to be affected by both age and education, but not gender (Caffarra, Vezzadini, Dieci, Zonato, & Venneri, 2002). Earlier research has indicated that age-related decrements on figure copying tasks emerge only after the age of 70 years (Boone, Lesser, Hill-Gutierrez, Berman, & D'elia, 1993). The apparent late decline in copying abilities that occurs with age helps explain why this sample, with a broad age range, was sensitive to the effects of age on the SCT.
Overall, these results indicate that the SCT may be diagnostically useful in a neurological population of both left- and right-handed males and females, with minimal effects of age. However, poor results on the task should be interpreted with caution when the patient's education is very low, particularly if they are of advanced age.
SCT as a Measure of Perceptual, Spatial, and Constructional Abilities
Examination of the construct validity of the SCT indicated that it was most similar to the copy administration of the RCFT. This is perhaps not surprising due to the similarities between the tasks—both involve direct reproduction of a target figure, attention to visual detail, and correct placement of items. The SCT may be attractive to clinicians in situations where the RCFT is too difficult for the client, or where it is unclear whether failure on the RCFT is due to poor planning rather than to higher order visual processing difficulties.
The next most closely related task to the SCT was the block design subtest of the WAIS, a measure typically considered by neuropsychologists to tap into constructional abilities (Lezak et al., 2004). The SCT was related, to a lesser extent, to the picture completion subtest of the WAIS, a measure commonly thought to measure perceptual skills, visual attention, and concentration (Groth-Marnat, 2003). These results were consistent with expectations, as past research using a quantitatively scored cube copying task has been found to have particularly high correlations with visual subtests of the WAIS, and in particular the Block Design Task (Maeshima et al., 2002).
Taken together, these results indicate the SCT may be a useful measure of higher order visual processing abilities in a neurological population. Aside from merely correlating well with WAIS perceptual reasoning measures, it may add useful information when used in conjunction with these measures. Specifically, the SCT is not confounded by psychomotor speed, it provides both two and three-dimensional items (unlike the Block Design Task which consists solely of three-dimensional models), and it allows for delineation of error types rather than merely providing a solitary total score.
Group Differences on the SCT
The total SCT score was useful in distinguishing between healthy and non-healthy participants, including the schizophrenia group. Despite this, there were some difficulties in using the total score to differentiate between the neurological groups themselves (i.e., dementia vs movement disorder; schizophrenia vs dementia). Difficulties using copying tasks to distinguish between neurological groups are not unique to the SCT. In fact, in the current study, the RCFT had the same difficulty distinguishing between the neurological groups, and was actually worse than the SCT in distinguishing between the healthy control group and patients with schizophrenia.
The utility of the SCT in distinguishing between control and psychiatric patients most likely lies in the broad measurement of error types provided by the SCT scoring system. The RCFT, using the commonly employed Taylor (1959) scoring method, rates figure reproductions based upon two criteria: inclusion and correct placement of items. Contrasting this, the quantitative scoring method used on the SCT measures a broader range of error types, including missing items, inclusion of extra items, perseveration, misplacement of items, size distortions, rotations, and closing-in on the target figure. Thus, the broader range of error types likely provides more opportunity for subtle deficits to be detected.
The current study indicated that the total score of the SCT can be used as a rough guide to distinguish between “impaired” and “intact” performance; however, this score does not appear to differ between movement disorder and dementia groups, nor dementia and schizophrenia groups. These results were expected, given that the task is intended for use as a brief screen. In order to yield richer clinical information, it is important to examine the frequency of individual error types.
Examination of Error Types
Overall, each of the clinical errors tended to occur more frequently in the neurological groups rather than the healthy control group. These errors helped provide more information regarding underlying differences obscured by the total score. For example, using the quantitative score alone, the movement disorder and dementia groups did not differ in terms of SCT total score; however, some differences did emerge when comparisons were made using error analysis between the groups. In particular, there was a high degree of spatial misplacement occurring in the movement disorder group; each patient in this diagnostic category committed at least one misplacement error across the task.
The dementia and schizophrenia groups performed similarly, both in total score and underlying error commission (i.e., absence of items and perseveration). The overlap in error commission between the schizophrenia and dementia groups were not surprising as the error types are thought to reflect underlying cognitive constructs (i.e., attention, higher order visual processing, executive functioning) that are not expected to be pathognomonic to any one clinical population. Indeed, executive functioning and higher order visual processing difficulties have been widely demonstrated in both schizophrenia (for review, see Stip, Lecardeur, & Sepehry, 2008) and dementia (Looi & Sachdev, 1999; Reed et al., 2007).
The one difference between the schizophrenia and dementia groups was that the inclusion of extra items (i.e., confabulation) error type was present in the model for dementia but not schizophrenia. Confabulation in figure reproductions has been proposed to relate to memory and executive compromise common in various forms of dementia (Pelati et al., 2011). Given that the SCT involves direct copy of the target figure (thus reducing demands placed on memory), the mechanism underlying the deficit in this case appears executive in nature. For example, inclusion of extra items on the SCT (e.g., addition of spokes or a horn on the bicycle) may represent poor attention to the target figure, or poor inhibition.
Directions for Future Work
The present study provides evidence for the utility of the SCT in distinguishing between patients with and without higher order visual processing deficits. Identifying individual predictors of a large range of more specific disorders (e.g., vascular dementia vs AD) was unfortunately not possible due to the participant numbers required for analysis; however, it remains a worthwhile area to be explored in future work.
Additionally, the sample featured a moderately sized group of healthy volunteers as a basis for comparison to the clinical group; however, a full analysis of healthy control performance on the SCT was outside the scope of this study. The next logical step to aid clinicians in interpreting performance is to develop comprehensive normative data for the task, especially given the finding of education and age effects on SCT scores.
The SCT is a measure of visual attention, spatial cognition, and constructional skills. The task is relatively resistant to the effects of gender and handedness; however, low scores in patients of advanced age or particularly low education should be interpreted with caution. The total score of the SCT is actually more useful than the RCFT at distinguishing between healthy control and schizophrenia groups. Although the SCT total score itself cannot accurately distinguish between movement disorder and dementia patients, error analysis focusing on the frequency of errors made between the groups indicates that certain groups have a propensity towards particular error types. Specifically, movement disorder patients tend to misplace items in their figure reproductions, while the dementia group leave out details, perseverate, and confabulate. Taken together, these results are impressive given the brief nature of the SCT. The findings indicate that the SCT is a useful screening tool for visual attention, spatial, and constructional difficulties, providing support that further research and development of the SCT is warranted.
Acknowledgements are provided to Bendigo Health Care Group for providing the clinical data utilised in the current study.