• Parkinson's disease;
  • precision;
  • predictors;
  • psychometric attributes


  1. Top of page
  2. Abstract

Cross-cultural adaptation and independent psychometric assessment of the Scales for Outcomes in Parkinson's disease-Cognition (SCOPA-COG), Brazilian version was performed. Parkinson's disease (PD) patients were evaluated by means of the SCOPA-Motor scale, Hoehn and Yahr staging (HY), Clinical Impression of Severity Index-PD (CISI-PD), Parkinson Psychosis Rating Scale, and Hospital Anxiety and Depression Scale. Cognition was evaluated using the Mini-Mental State Examination (MMSE), Short Portable Mental Status Questionnaire (SPMSQ), and SCOPA-COG. The following attributes were explored: acceptability, scaling assumptions, reliability, precision, and construct validity. One hundred fifty-two patients were assessed (mean age, 63.2 years; disease duration, 7.8 years; median HY stage, 3). Mean SCOPA-COG and MMSE were 18.2 and 25.7, respectively. The internal consistency of the SCOPA-COG (Cronbach's alpha = 0.81; item-total correlation, 0.38–0.62) was satisfactory. While the intraclass correlation coefficient value was 0.80, weighted kappa ranged from 0.30 (dice task) to 0.72 (animal fluency task). The standard error of measurement value for the SCOPA-COG was 3.2, whereas the smallest real difference was 8.9. SCOPA-COG total scores significantly decreased as the HY stage increased (Kruskal-Wallis, P < 0.0001). Age, years of education, and PD duration (all, P < 0.001) were observed to have an independent, significant effect on the SCOPA-COG. The SCOPA-COG is a short, reliable, valid instrument that is sensitive to cognitive deficits specific to PD. © 2007 Movement Disorder Society

According to a recent systematic review of prevalence studies of dementia in Parkinson's disease (PD), 24 to 31% of PD patients have dementia, and 3 to 4% of the dementia population may be attributable to PD.1 About 50% of patients with dementia-free PD are found to have mild cognitive impairment. Significant risk factors for the development of dementia in PD include increasing age, older age at diagnosis, longer disease duration, increasing severity of motor symptoms, and the onset of hallucinations.1–7

Although the Mini-Mental State Examination (MMSE) has been extensively used for cognitive assessment of PD patients, it has some limitations. The MMSE has a significant ceiling effect and fails to cover a full range of cognitive domains, particularly global and executive functions. Alternative cognitive screening methods, such as the Cambridge Cognitive Assessment (CAMGOG),8 have been proposed. However, the CAMCOG is not a specific tool for cognitive evaluation in PD. Recently, a specific cognitive questionnaire for PD, the Scales for Outcomes in Parkinson's disease-Cognition (SCOPA-COG), has been developed.9

The Latin American population is becoming older, and the proportion of people with PD and dementia is expected to increase accordingly.10 In coming decades, population aging may be more pronounced in Brazil, a country with over 180 million people. Nevertheless, there is an absence of standardized and specific neuropsychological tests for the Brazilian PD population. No epidemiological studies addressing the impact of cognition on Brazilian PD patients have been previously performed.

The primary aim of this study was to check the SCOPA-COG's psychometric attributes, using a cross-cultural version adapted to Brazil.


  1. Top of page
  2. Abstract


All patients with diagnosis of PD, consecutively admitted to the Neurology Outpatient Clinic of the Sarah Hospital in Brasilia DF, Brazil, from August 2006 to January 2007, were considered candidates for the study. Inclusion criteria were diagnosis of PD, as per the United Kingdom Parkinson's Disease Society Brain Bank Criteria11 and age ≥40 years at disease onset. Exclusion criteria were as follows: (1) PD patients with disability due to other neurological disorders; (2) concomitant severe illness, acute disorder or injury (e.g., advanced neoplasia, pneumonia, hip fracture), pharmacological effect (e.g., dopamine antagonists), sensorial deficit (e.g., blindness) or sequela status (e.g., hemiplegia) that could interfere with or significantly modify the evaluation of the effects caused by PD; (3) inability of any kind to read or understand Portuguese (e.g., illiteracy, severe dementia); and (4) refusal to participate in the study.


PD patients answered a demographic questionnaire that covered age, gender, marital status, and education, as well as historical information about PD (age at onset, duration of disease, and drug treatment).

Neurologist-based assessments included Hoehn and Yahr staging (HY),12 Scales for Outcomes in Parkinson's disease-Motor scale (SCOPA-MS),13 Clinical Impression of Severity Index-Parkinson's disease (CISI-PD),14 and Parkinson Psychosis Rating Scale (PPRS).15 Cognition was assessed using the Brazilian version16 of Folstein's Mini-Mental State Examination (MMSE),17 the Short Portable Mental Status Questionnaire (SPMSQ),18 and the SCOPA-COG.9 Patients also completed the Hospital Anxiety and Depression Scale (HADS).19

The assessments were conducted by neurologists with competence in movement disorders, during scheduled visits by patients to the clinic. In addition, the SCOPA-COG, CISI-PD, and PPRS were readministered to 52 patients who returned to the clinic ∼10 days after the first evaluation, in order to obtain data for a test–retest reliability analysis.

The HY is usually applied to establish the severity of PD,20 inasmuch as HY 1 is the mildest stage with only unilateral symptoms, and HY 5 is the most severe stage (wheelchair-bound or bedridden).

The SCOPA-MS13 contains three sections, namely, motor impairment (10 items), activities of daily living (ADL) (7 items), and motor complications (4 items). All items score in a range from 0 (normal) to 3 (severe); the higher the score, the greater the severity.

The neurologist-based CISI-PD score14 represents a subjective judgment of global PD severity, and reflects and summarizes the clinical evaluation. The CISI-PD contains the following 4 items: motor signs, disability, motor complications, and cognitive impairment. Overall evaluations of these four domains are then combined to provide the CISI-PD global score, which ranges from 0 (normal) to 24 (severe).

The Parkinson Psychosis Rating Scale15 was designed to assess the severity of psychosis in PD patients. It is administered by the clinician, and consists of 6 items scored in each case from 1 (no symptoms) to 4 (extreme symptoms). The total score ranges from 6 to 24 points.

The MMSE17 measures certain areas of cognitive functioning, including memory, orientation, language, and the ability to follow command. The cutoff point for cognitive impairment was 24/23.

The SPMSQ18 is a useful screening test for moderate-to-severe dementia in both community and hospital settings.21 Levels of severity of intellectual impairment are assigned to the number of errors as follows: mild (3–4 errors), moderate (5–7 errors), and severe (8–10 errors). It allows for scores to be adjusted to subjects' educational level.

The HADS is a self-administered screening tool for mood disorders, which consists of 14 items. Individual item scores can either be summed to calculate a total score, or summed per subscale to produce separate anxiety (HADS-A) and depression (HADS-D) scores.22, 23

The SCOPA-COG9 consists of 10 items with a maximum total score of 43, with higher scores reflecting better performance. The SCOPA-COG has the following four domains: Memory, Attention, Executive functions, and Visuospatial functions. Memory domain items assess both visual and verbal memory, as well as immediate and delayed recall. In the Executive Function domain, items address semantic fluency, set shifting, and motor planning. Two tasks are included in the Attention domain (counting down by threes; and months backward), and one in the area of Visuospatial Function (figure assembly task).

Cross-Cultural Adaptation

Two persons with excellent knowledge of English and Portuguese independently produced the forward translation of the SCOPA-COG, compared the results, and resolved the differences. An English native made the backward translation from this Portuguese version to English (without access to the original English questionnaire). The back-translation was compared to the original, and amendments forward–backward were then made in order to eliminate all discrepancies between the original and back-translated versions. In the process of cross-cultural adaptation no changes were made to the items of the original version. The final joint translation, considered linguistically and conceptually equivalent to the original one, was called the SCOPA-COG Brazilian version.

Data Analysis

The following psychometric attributes were explored for the SCOPA-COG: acceptability, scaling assumptions, reliability, precision, and construct validity. To establish acceptability, observed versus possible score ranges, mean score distance to the median, floor and ceiling effects (acceptable, <15%),24 and skewness (limits: −1 to +1)25 were determined.

Scaling assumptions refer to the correct grouping of items and the appropriateness of their summed score. These were checked using item-total correlation, corrected for overlapping. Values of 0.40 or higher were deemed appropriate.26

Cronbach's alpha27 was used to assess internal consistency, with this being considered acceptable where the alpha value was 0.70 or higher.28 Test–retest reliability was ascertained using weighted kappa (with quadratic weights) for individual items and the intraclass correlation coefficient (ICC) (model 2, single rating) for the total score. Kappa values >0.40 for items and an ICC ≥ 0.70 for total measure scores were taken as the criterion of acceptable stability.26, 29

To determine the scale's precision and potential responsiveness, the standard error of measurement (SEM) and smallest real difference (SRD) were calculated. The SEM, defined as “the standard error in an observed score that obscures the true score” [SEM = SD × √(1 − reliability coefficient)],31 reflects a measure's reliability and precision.28, 30, 31 The SRD indicates 95% confidence of real difference between the true scores of two successive measurements, capturing “the essence of the reproducibility of a measurement instrument” [SRD = 1.96 × √2 × SEM].32 The less a scale's reliability, the greater its SEM and SRD, and the lower its precision and responsiveness.

Convergent validity represents the extent to which a measure is related to other variables and measures for the same construct, and was explored by means of Spearman rank correlation coefficient (rS). It was a priori hypothesized that coefficient values would be high (rS ≥ 0.60) as between the SCOPA-COG and the other applied cognitive instruments (MMSE and SPMSQ), moderate (rS = 0.30–0.59) as between education, PD duration, and PD rating scales, and low (rS < 0.30) as between age and measures of noncognitive mental aspects (psychosis, depression).26, 33 To ascertain the internal validity of the Brazilian SCOPA-COG, inter-domain correlations were determined. Coefficient values of 0.30 to 0.70 were deemed satisfactory.34

Discriminant validity refers to a measure's ability to discriminate between groups at a given point in time. The Mann-Whitney test was used to compare SCOPA-COG scores between patients with an MMSE cutoff point of 23/24. The Kruskal-Wallis test was used to compare SCOPA-COG scores broken down by HY stage and by SPMSQ and CISI-PD Item 4 levels.

To identify predictors of the SCOPA-COG score, a multiple regression model was constructed, with age, years of education, and duration of PD as independent variables. Interaction, collinearity, or transgression of the mathematical assumptions impeded the introduction of other variables. The assumptions made by the model (linear relationship, normality, and homoscedasticity) were upheld.


  1. Top of page
  2. Abstract

During the 6-month study period, 181 PD patients were evaluated as outpatients. Twenty-nine PD patients were excluded from the study by reason of illiteracy (23), severe dementia (3), blindness (2), and severe disability due to concomitant ankylosing spondylitis (1). One hundred and fifty-two patients (60.5% males; mean age 63.2 ± 11.3 years) were included in the study. Disease duration was 7.8 ± 5.1 years (range: 1–23).

Demographic characteristics of PD patients are shown in Table 1. Patients distribution by HY was: stage 1, 15.1%; stage 2, 33.5%; stage 3, 38.2%; stage 4, 9.9%; and stage 5, 3.3% (median = 3; interquartile rank = 2–3). Two thirds of the sample were treated with levodopa, 16.4% with dopamine agonists, and 27.6% received other treatments (selegiline, amantadine, and anticholinergics).

Table 1. Demographic characteristics of patients (n = 152)
Mean age, years (SD)63.2 (± 11.3)
Duration of disease, years (SD)7.8 (± 5.1)
Years of education, mean (SD)8.4 (± 5.1)
Gender (%) 
Marital status (%) 
Occupation (%) 
 Still working12.5

Descriptive statistics of the applied measures are displayed in Table 2. No significant sex-related differences were observed in SCOPA-COG, MMSE, and SPMSQ scores. Thirty-seven patients (24.3%) scored ≤23 on the MMSE.

Table 2. Patients' motor and cognitive evaluation (n = 152)
 MeanSDRangeMaximum possible score
  1. SCOPA-COG, scales for outcomes in Parkinson's disease-cognitive questionnaire; ADL, activities of daily living; CISI-PD, clinical impression severity index-Parkinson's disease; HADS, hospital anxiety and depression scale; IQR, inter-quartile range; SCOPA-MS, scales for outcomes in Parkinson's disease-motor scale.

SCOPA-MS total score23.111.70–5275
 SCOPA-MS, motor impairment12.46.70–3042
 SCOPA-MS, ADL7.74.50–2021
 SCOPA-MS, motor complications3.02.70–1012
CISI-PD total score9.74.70–2124
 CISI-PD, motor signs3.01.30–66
 CISI-PD, disability2.71.60–66
 CISI-PD, motor complications1.91.50–56
 CISI-PD, cognitive function2.01.50–66
 HAD-Anxiety sub-scale8.74.40–2021
 HAD-Depression sub-scale8.74.20–2121
Parkinson psychosis rating scale1.21.90–1018
Mini-mental state examination25.73.611–3030
Short portable mental status questionnaire1.01.30–810
SCOPA-COG total score18.27.24–3743
 Executive function6.72.60–1212
 Visuospatial function2.91.50–55

The mean SCOPA-COG score was 18.25, and the median was 18 (Δ = 0.25 = 0.58% of the maximum possible score). Neither ceiling nor floor effects were observed for the SCOPA-COG (both extreme scores, 0.66%). Skewness (0.31) was within the standard limits. For the SCOPA-COG domains, a ceiling effect was observed for Attention (38.8%) and Visuospatial Function (18.4%).

SCOPA-COG item-total correlation coefficients are shown in Table 3. With the single exception of the “Months backward” item (r = 0.38), all coefficients attained values higher than the criterion value of 0.40. For the SCOPA-COG, Cronbach's alpha was 0.81, and for multi-item domains, it was as follows: Memory, 0.70; Attention, 0.50; and Executive function, 0.51. Also shown in Table 3 are the weighted kappa values for the SCOPA-COG items: these ranged from 0.30 (Dice) to 0.72 (Fluency task). The ICCs were 0.80 for the SCOPA-COG total score, 0.71 for Memory and Attention, 0.62 for Visuospatial function, and 0.61 for Executive function. The SEM and SRD values for the SCOPA-COG total score were 3.22 (95%CI = 6.31) and 8.93, respectively.

Table 3. Item-total correlation and reproducibility of SCOPA-COG items
SCOPA-COGItem-to-scale correlationκw
  1. SCOPA-COG, scales for outcomes in Parkinson's disease-cognitive questionnaire; κw= weighted kappa.

Immediate word recall0.470.57
Digits backward0.550.64
Counting down by threes0.470.65
Months backward0.380.62
Fluency task0.510.72
Figure assembly0.620.62
Delayed word recall0.490.62

SCOPA-COG scores displayed a statistically significant association with age (rS = −0.34; P < 0.0001), years of education (rS = 0.26; P = 0.01), and disease duration (rS = −0.28; P < 0.001). Table 4 shows the convergent validity between the total score and domains of the SCOPA-COG and the other scales. A close association was found between the SCOPA-COG and the MMSE (rS = 0.72), but the correlation with the SPMSQ was moderate (rS = −0.37). Correlation coefficients between total SCOPA-COG score and PD measures ranged from −0.45 (HY) to −0.56 (CISI-PD), except for the Complications subscale of the SCOPA-MS (rS = −0.21). Most of the coefficients of correlation with mood and psychosis scores attained values lower than 0.30. Insofar as SCOPA-COG domains were concerned, the closest correlations were obtained with the MMSE and the weakest with the SCOPA-MS Examination and PPRS. While PD postural instability and gait difficulty phenotype (PIGD), SCOPA-MS postural instability, freezing during ‘on’, gait and walking items) displayed a significant correlation with SCOPA-COG total score (rS = −0.47; P < 0.0001), predominant tremoric profile (SCOPA-MS rest and postural tremor items) failed to do so (rS = 0.05; nonsignificant). Intercorrelation between SCOPA-COG domains ranged from 0.39 (Attention with Executive function) to 0.59 (Executive function with Visuospatial function). Patients scoring ≤23 on the MMSE scored 11.7 on the SCOPA-COG versus 20.4 for those with ≥24 on the MMSE (P < 0.0001). Patients who scored ≤23 on the MMSE also had significantly lower scores on SCOPA-COG domains, as shown in Table 5. The SCOPA-COG total score significantly decreased as HY stage increased (Kruskal-Wallis, P < 0.0001) (Table 6). All SCOPA-COG domains except Attention displayed a similar trend. Differences in SCOPA-COG total scores broken down by SPMSQ and CISI-PD Item 4 levels proved to be significant (Kruskal-Wallis test; both, P = 0.0001).

Table 4. Correlations* between SCOPA-COG and other scales
  • *

    Spearman's rank correlation coefficient: r ≥ 0.16, P < 0.05; r ≥ 0.27, P < 0.001; r ≥ 0.31, P < 0.0001.

  • HY, Hoehn and Yahr stage; SCOPA-MS, scales for outcomes in Parkinson's disease-motor (I = motor impairment, II = ADL; III = complications); CISI-PD, clinical impression severity index-Parkinson's disease; MMSE, mini-mental status examination; SPMSQ, short portable mental status questionnaire; HADS, hospital anxiety and depression scale; HADS-A, anxiety-subscale; HADS-D, depression-subscale; PPRS, Parkinson psychosis rating scale; SCOPA-COG, scales for outcomes in Parkinson's disease-cognition.

SCOPA-COG score−0.45−0.41−0.48−0.21−0.47−0.560.72−0.37−0.23−0.32−0.21
Executive functions−0.45−0.44−0.44−0.22−0.47−0.540.57−0.33−0.19−0.24−0.21
Visuospatial functions0.42−0.35−0.43−0.24−0.41−0.480.57−0.34−0.20−0.31−0.20
Table 5. Breakdown of SCOPA-COG scores by MMSE cut-off
 MMSE < 24MMSE ≥ 24P*
 n = 37n = 115 
  • Mean ± standard deviation.

  • *

    Mann-Whitney test.

  • MMSE, mini-mental status examination; SCOPA-COG, scales for outcomes in Parkinson's disease-cognition.

SCOPA-COG score11.7 ± 4.820.3 ± 6.5<0.0001
SCOPA-COG domains   
 Memory3.4 ± 1.76.8 ± 3.3<0.0001
 Attention1.6 ± 1.52.9 ± 1.5<0.0001
 Executive functions4.9 ± 2.57.3 ± 2.3<0.0001
 Visuospatial functions1.8 ± 1.23.3 ± 1.4<0.0001
Table 6. Breakdown of cognitive scales by Hoehn and Yahr stage
 Hoehn and Yahr stageP*
n = 23n = 52n = 58n = 15n = 5
  • Mean ± standard deviation; ns, non significant.

  • *

    Kruskal-Wallis test.

  • SCOPA-COG, scales for outcomes in Parkinson's disease-cognition; MMSE, mini-mental status examination; SPMSQ, short portable mental status questionnaire.

SCOPA-COG Total21.5 ± 6.121.4±7.015.9 ± 6.213.9 ± 6.110.8 ± 6.1<0.0001
Memory6.4 ± 3.37.6 ± 3.54.9 ± 2.94.8 ± 2.63.6 ± 2.8<0.0001
Attention2.9 ± 1.13.0 ± 1.32.3 ± 1.52.6 ± 1.32.0 ± 2.0ns
Executive functions8.3 ± 2.27.6 ± 2.46.2 ± 2.14.7 ± 2.63.6 ± 3.0<0.0001
Visuospatial functions3.9 ± 1.13.3 ± 1.42.6 ± 1.51.8 ± 1.61.6 ± 0.9<0.0001
MMSE27.5 ± 2.626.6 ± 2.825.3 ± 3.524.0 ± 3.419.6 ± 6.7<0.0001
SPMSQ0.5 ± 0.70.6 ± 0.81.1 ± 1.22.3 ± 1.92.8 ± 2.5<0.0001

Age, years of education, and duration of PD were shown by the multiple regression model to have an independent, significant effect (all, P < 0.001) on the SCOPA-COG (F = 18.85; P < 0.0001). A similar result (F = 25.53; P < 0.0001) was yielded when PD duration was replaced by HY stage.


  1. Top of page
  2. Abstract

A wide range of clinical forms of PD (initial, moderate, and advanced stages) were represented in this study. The Brazilian version of the SCOPA-COG showed no significant floor or ceiling effects, fitted the scaling assumptions, and displayed satisfactory internal consistency. Cronbach's alpha value (0.81) proved to be comparable to that obtained in the original study (0.83).9 While the alpha coefficient exceeded the standard value of 0.70 for the SCOPA-COG Memory domain, both the Attention and Executive functions fell short of the standard. This finding may, at least in part, be related to the number of items per domain, since alpha coefficients have a direct relationship with scale length.35 The Memory dimension includes 4 items, whereas Attention includes two and Executive function three, a fact supporting the previous argument.

Test–retest reliability was satisfactory for the SCOPA-COG, with the ICC attaining a value similar to the original version (0.80 vs. 0.78).9 Insofar as the stability of the individual items was concerned, the extreme weighted kappa values in the present study were similar to those found in the original (minimum: 0.30 vs. 0.40; maximum: 0.72 vs. 0.75), yet the specific items yielding these values did not coincide (Dice task vs. Word recall; Fluency task vs. Delayed word recall, respectively).

Calculation of the SEM and SRD for the study by Marinus et al.9 showed that both were similar to ours (SEM = 3.33 vs. 3.22; SRD = 9.23 vs. 8.92, respectively). According to the classical test theory, the SEM is independent of the sample and remains relatively constant (except at both extremes) across the ability levels of a given population.36 SEM and SRD are indices that capture within-subject variability, a key aspect of reproducibility and responsiveness.32 In addition, they provide information about the threshold of variability that has to be surpassed for a change to be deemed relevant. A relationship has been proposed between SEM, half a standard deviation, and the minimal important difference.36–38 In line with these arguments, and depending on the chosen index (SEM or SRD), a change of 6.5 to 9.0 points (around 15–20% of the maximum possible score, at a 95%CI) in the SCOPA-COG total score could be considered a real change and a minimal important difference.

Evidence of adequate construct validity has been shown for the SCOPA-COG. The construct validity of the SCOPA-COG was supported by the expected correlations with other scales and by differences observed among categories of participants grouped by dementia status and disease severity. As hypothesized, SCOPA-COG total score showed a high correlation with the MMSE (rS = 0.72), in line with the study by Marinus et al.,9 whereas correlation with the raw SPMSQ score was moderate, a finding attributable to differences in content and structure between the two instruments. Whereas the SCOPA-COG also displayed a moderate association with the other scales applied in the study (HY, SCOPA-MS, and CISI-PD), the association with the PPRS and HADS was weaker, thereby demonstrating adequate divergent validity. A low-to-moderate association between the SCOPA-COG and age, years of education, and duration of PD was also detected. Previous findings relating cognitive deterioration in PD to PIGD phenotype39 were supported by the present study, using the same measures as Verbaan et al.40

According to previous studies,9, 40 SCOPA-COG scores were significantly lower for patients shown to be more severely affected by the MMSE, SPMSQ, and HY. Based on the statistical criterion, the internal validity of the scale was adequate. As a whole, and taking into account the inter-relationships among different constructs embedded in PD, the construct validity of the SCOPA-COG proved to be satisfactory. Age, years of education, and disease duration or HY stage were identified as independent predictors of SCOPA-COG.

The neuropsychological profile of PD dementia is often described as a “subcortical” dementia. Persons with PD typically exhibit impairment in complex attention, executive functions, information retrieval, procedural memory, visuoconstruction, verbal fluency, and speed of information processing.41 Most of these cognitive functions are specifically tested by the SCOPA-COG.

Our study expands the current body of knowledge surrounding this scale's psychometric properties9, 40 and suggests that the SCOPA-COG is indeed a useful, short, reliable, sensitive, and valid instrument for assessing cognition in PD patients.


  1. Top of page
  2. Abstract