Journal of Clinical Neuroscience 21 (2014) 980–987
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Validation of the Chinese version of the NUCOG cognitive screening tool in patients with epilepsy, dementia and other neurological disorders Lan Gao a, Shu-Chuen Li a,⇑, Li Xia b, Songqing Pan b, Dennis Velakoulis c,d, Mark Walterfang c,d a
School of Biomedical Sciences and Pharmacy, The University of Newcastle, Callaghan, NSW, Australia Neurology Department, Renmin Hospital of Wuhan University, Wuhan, Hubei, China c Neuropsychiatry Unit, Royal Melbourne Hospital, Melbourne, VIC, Australia d Melbourne Neuropsychiatry Centre, Melbourne Health and University of Melbourne, VIC, Australia b
a r t i c l e
i n f o
Article history: Received 15 April 2013 Accepted 24 September 2013
Keywords: Cognition Dementia Epilepsy Mini-Mental State Examination Neuropsychiatry Unit Cognitive Assessment Tool Screening
a b s t r a c t We aimed to develop and validate a Chinese version of the Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG) for use in Chinese-speaking subjects internationally. Patients and healthy controls were recruited from two hospitals between July and October 2012. Receiver operating characteristic (ROC) curves were utilized to test criterion validity. Convergent validity was assessed via correlations between NUCOG and the Mini-Mental State Examination (MMSE). Reliability was measured by internal consistency (Cronbach’s a). Patients with epilepsy (n = 144), neurological diseases (n = 81), dementia (n = 44), and controls (n = 260) completed the NUCOG and the MMSE. Overall, both NUCOG and MMSE scores differed significantly across the four groups with the highest scores in the control group and the lowest in the dementia group (p < 0.0001). The NUCOG scores could differentiate between patients with certain seizure types, stroke and transient ischemic attack. Compared to the MMSE, the NUCOG exhibited a higher area under the ROC curve. The convergent validity was substantially correlated, and internal consistency was very high (0.922). The Chinese version of NUCOG was demonstrated to be a sensitive and reliable screening tool for cognitive impairment in a Chinese-speaking population not only in China, but also in countries where there is a sizeable population of ethnic Chinese. Additionally, our study also showed the NUCOG could better differentiate cognitive function in patients with certain seizure types, stroke and transient ischemic attack than the MMSE. This potentially expands the clinical usefulness of NUCOG, enabling clinicians to measure the cognitive profile of patients with epilepsy and ischemic cerebrovascular diseases. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction The identification of cognitive impairment in at-risk individuals, including patients with neurological disorders, relies on the synthesis of information obtained from history-taking, physical and mental state examination, and investigative testing of cognition. Screening patients for cognitive assessment in busy medical settings is most commonly undertaken through the use of standardized cognitive screening tools (CST), which combine a number of individual test items to test a broad range of cognitive functions. Most of these tools were initially developed as English-language tools, with some being translated and validated in a limited subset of non-English languages. A small number of CST are available to test patients whose chief language is Chinese, including the MiniMental State Examination (MMSE) [1], the Beijing version of the
⇑ Corresponding author. Tel.: +61 2 4921 5921; fax: +61 2 4921 2044. E-mail address: [email protected] (S.-C. Li). 0967-5868/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2013.09.020
Montreal Cognitive Assessment (BJ-MoCA) [2], and the Saint Louis University Mental Status Examination (SLUMS) [3]. Balancing the need for sensitivity, breadth of assessment and brevity of administration in the construction of CST is difficult, and many available tools have significant limitations. For instance, the MMSE may be insensitive to early cognitive impairment, and lacks spatial recall and executive function testing. It also has a floor effect in patients with severe cognitive deficits [4]. While the MoCA was specifically developed to screen patients with mild cognitive impairment and mild dementia, the Chinese-language version may less successfully discriminate between these two disorders [5,6] and may be less successful in detecting cognitive impairment in neurological disorders [2]. To our knowledge the validity of the Chinese version of the SLUMS is yet to be reported [3]. Given the limitations of these instruments, the availability of another valid and portable Chinese-language screening tool for cognitive impairment may provide clinicians with an additional option to reliably detect cognitive impairment in Chinese-speaking patients. Such patients can be found in mainland Chinese settings, in addition
L. Gao et al. / Journal of Clinical Neuroscience 21 (2014) 980–987
to populations of ethnic Chinese living in many Southeast Asian countries and the increasing number of Chinese immigrants who have relocated to non-Chinese-speaking countries or regions (such as the USA, Canada, Australia, Europe, and Singapore) in recent years [7–9]. The availability of such a tool, allowing cognitive assessment to be carried out in the mother tongue of the respondent, minimizes comprehension difficulties as a confounding variable in the interpretation of test results. The Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG) was developed with the intention to address some of the limitations of existing screening tools. It includes extensive executive and spatial function testing, and produces a multidimensional cognitive score. The NUCOG comprises five domains: attention, visuoconstructional function, memory, executive function, and language, with a maximum score of 20 points in each domain [10]. The utility and validity of NUCOG has been demonstrated in a variety of neurological disorders including Alzheimer’s disease, frontotemporal and subcortical dementia, major mood and psychotic illness, epilepsy, stroke, Parkinson’s disease and head injury [10–12]. According to these validation studies, the NUCOG is capable of differentiating between patients with neurological and psychiatric illness associated with comorbid cognitive impairment and controls, and reliably discriminates between demented and non-demented individuals. Like a number of other cognitive tools, the NUCOG appears to have a greater specificity and sensitivity than the MMSE in the detection of significant cognitive impairment [10,11]. Additionally, a Persian translation of the NUCOG essentially demonstrated near-identical validity and reliability in a large mixed neurological, psychiatric and dementia clinical population [13]. This study aimed to translate the NUCOG for clinical use in Chinese patients with epilepsy, non-dementia neurological disease, and dementia, and to test its validity and reliability in these clinical disorders compared to a group of healthy controls. We also aimed to compare the performance of the Chinese-language versions of the NUCOG and MMSE to determine if the NUCOG demonstrated a superior capacity to detect cognitive impairment in these populations.
2. Methods The study was approved by the Ethics Committees of the two hospital study sites in China. All the participants provided informed consent prior to their inclusion in the study.
2.1. Subjects Adult patients from both inpatient and outpatient clinics from the Renmin Hospital of Wuhan University, and the Fifth Hospital of Wuhan (Wuhan, Hubei, China) were randomly selected by attending physicians or consultant neurologists/epileptologists between July and October 2012. The patients to be recruited included patients diagnosed with epilepsy, non-dementia neurological diseases, and those diagnosed with dementia according to the Diagnostic and Statistical Manual revision IV (DSM-IV) criteria. The epilepsy group was divided into four distinct subgroups of patients with predominantly simple partial, complex partial, secondary generalized seizures and primary generalized seizures. Adults without a history of illness associated with cognitive deficit, or any other neurological or psychiatric disease, were deemed as meeting the inclusion criteria for the healthy control group. These subjects were primarily recruited from medical school students, relatives and caregivers of aforementioned patients, and hospital general staff.
981
All the interviews were administered by L.G. (who was trained by the tool’s developers M.W. and D.V. at the Royal Melbourne Hospital, Australia) and L.X. 2.2. Measures 2.2.1. MMSE The MMSE is a well-established and widely-used screening tool for cognitive impairment. The Chinese version of the MMSE has been extensively deployed for a variety of disorders, including those disorders examined in this study [3,14–17]. 2.2.2. NUCOG The NUCOG is a 21 item tool which examines a comprehensive range of cognitive function and requires no instrumentation. Administration of NUCOG results in a total score for a respondent out of a possible 100, comprising scores out of 20 on its five cognitive domains. Performance on the English language version of the NUCOG has been demonstrated to correlate strongly with MMSE scores and formal neuropsychological testing [11]. 2.3. Translation process 2.3.1. Forward and backward translation Two Chinese physicians (L.G. and L.X.) independently translated the English version of NUCOG into simplified Chinese. The two Chinese versions of NUCOG were synthesized into one via thorough discussion between the two translators and using the input of the third consultant neurologist (S.Q.P.). Then another two bi-lingual physicians who were blinded to the original NUCOG as well as the study design performed the back-translation process. Lastly, the two back translations were submitted for review by the NUCOG developers for consistency with the original item content of the tool (M.W. and D.V.). 2.3.2. Cultural and linguistic adaptations The initial Chinese version of NUCOG was sent to a consultant neurologist, an epileptologist, and a neuropsychiatrist to evaluate the equivalence with the original version. Due to linguistic differences between English and Chinese languages, two key changes were proposed by these experts to achieve the equivalent difficulty and depth of the test. 1. Repetition: ‘‘banana, artillery, constitutional’’ were replaced by ‘‘WuDangShan, TianXiaWuDi, XiYangWuXianHao’’ for the equivalent pronounced length of these words. 2. Writing: ‘‘The boy’s aunt made a large pie out of steak and dough’’ was replaced by a Chinese language version of ‘‘Mother rode a horse to buy mangos’’ to match the same syntactical difficulty. 2.4. Pilot testing Forty subjects, including hospital general staff (n = 20), interns (n = 12), nurses (n = 4), and outpatients of a Neurology Clinic (n = 4) were recruited to complete the draft Chinese NUCOG. From their qualitative input, wording and phrasing were changed accordingly to avoid confusion in understanding. 2.5. Statistical analysis Demographic data (age and education) were compared using the Kruskal-Wallis test, while the chi-squared test was used to analyze sex proportion. One-way analysis of variance was employed to analyze cognitive data. Correlations between
982
L. Gao et al. / Journal of Clinical Neuroscience 21 (2014) 980–987
demographic variables and cognitive data were undertaken using Spearman’s correlation coefficient. Receiver operator characteristic (ROC) curves were constructed to examine criterion validity. Spearman’s correlations between NUCOG and MMSE were assessed to evaluate the convergent validity. Reliability was tested by internal consistency using Cronbach’s a. A comparison of NUCOG subscales across diagnostic groups was conducted using repeated-measures analysis of variance (RMANOVA), with within-subject effects tested using the Greenhouse-Geisser test, controlling for age and years of education. Furthermore, the performance of the NUCOG and MMSE across diagnostic groups was compared via multivariate ANOVA, using age and years of education as covariates. Statistical analyses were performed using the Statistical Package for the Social Sciences version 20.0 software (SPSS, Chicago, IL, USA). 3. Results In total 529 subjects were enrolled in this study, including patients diagnosed with epilepsy (n = 144), non-dementia neurological diseases (n = 81), dementia (n = 44) and healthy controls (n = 260) (Table 1). Among the patients with non-dementia neurological diseases, the diagnoses included stroke (n = 26), dizziness/vertigo (n = 26), transient ischemic attack (TIA; n = 13), headache (n = 2), meningitis (n = 7), myasthenia gravis (n = 1), Parkinson’s disease (n = 1), peripheral facial neuritis (n = 4), and Hunt syndrome (n = 1). All neurological patients were enrolled from an inpatient unit within the first 7 days of admission. The dementia group included patients with senile dementia of the Alzheimer type (SDAT, diagnosed according to USA National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association, [NINCDS-ADRDA] criteria), vascular dementia (VD, diagnosed according to the USA National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences [NINDS-AIREN] criteria) and other forms of dementia (OD, diagnosed according to DSM-IV criteria for dementia and clinician consensus). 3.1. Demographic data The demographic variables differed across the subject groups. The patient groups differed significantly with respect to age, with the neurological patient group the oldest, followed by the dementia group, and then the control and epilepsy groups whose mean age did not differ (p > 0.05). For level of education, the control group had the highest mean years of education, followed by the neurological and epilepsy groups (which did not differ), with the least-educated patients belonging to the dementia group
(p < 0.0001). Sex also varied among groups, with the control group having the lowest proportion of male participants with the dementia group including the most (Table 1). 3.2. Relationship between demographic data and cognitive data 3.2.1. NUCOG Age negatively correlated with NUCOG scores in the epilepsy (r = 0.234, p = 0.005), non-dementia neurological (r = 0.266, p = 0.017) and control groups (r = 0.401, p < 0.0001). In contrast, educational level positively correlated with the total NUCOG score in the non-dementia neurological (r = 0.328, p = 0.003) and control groups (r = 0.414, p < 0.0001). Sex did not affect total NUCOG score (Table 2). 3.2.2. MMSE As with the NUCOG, age in non-demented neurological (r = 0.239, p = 0.031) and control groups (r = 0.421, p < 0.0001) negatively correlated with MMSE score. However, with respect to education, the performance on MMSE was only positively correlated with the control group. Sex also did not affect MMSE scores (Table 2). 3.3. Comparison of NUCOG and MMSE scores between subject groups Both NUCOG and MMSE scores differed significantly between the four study groups (F3, 528 = 275.389, p < 0.0001 and F3, 528 = 171.529, p < 0.0001 respectively), with the highest scores in control group and the lowest in dementia group. However, when controlled for both age and level of education, the multivariate analysis of variance (MANOVA) did not indicate any difference for total NUCOG and MMSE scores in differentiating the patient groups (p = 0.428 and p = 0.846, respectively). Particularly, both NUCOG and MMSE failed to discriminate between epilepsy and neurological groups, while both tools were able to differentiate the dementia and neurological/epilepsy, dementia and control, and control and neurological/epilepsy groups (p < 0.0001). It was also worth noting that age (p = 0.002) and education level (p = 0.001) were shown to influence the NUCOG scores but not MMSE scores (p = 0.182 and p = 0.109, respectively) according to the MANOVA results. Age and education level adjusted estimations are shown in Table 3. 3.4. NUCOG profile across main patient groups RMANOVA was utilized to investigate how the cognitive profile (five domains of NUCOG) differed across diagnostic groups, with age and educational level as covariates. NUCOG subscale scores differed significantly between different groups (p < 0.0001). Additionally, interactions between subscales and age (p < 0.0001)/
Table 1 Comparisons of the Chinese version of the Neuropsychiatry Unit Cognitive Assessment Tool and Mini-Mental State Examination across different patient groups
Age Sex, male Education, years MMSE Total NUCOG Total NUCOG Attention NUCOG Visuoconstructional NUCOG Memory NUCOG Executive NUCOG Language
Epilepsy n = 144
Neurological n = 81
Dementia n = 44
Control n = 260
33.39 ± 13.060 52.1% 10.56 ± 2.961 26.26 ± 3.430 81.74 ± 11.814 16.90 ± 2.348 17.21 ± 2.511 14.65 ± 3.433 14.89 ± 3.623 17.92 ± 2.426
55.70 ± 10.213 60.5% 11.16 ± 2.754 25.94 ± 2.467 81.38 ± 7.586 17.26 ± 1.439 17.29 ± 1.662 14.71 ± 2.767 14.43 ± 3.044 17.65 ± 1.562
45.50 ± 18.575 63.6% 8.91 ± 2.321 17.91 ± 6.305 52.41 ± 15.361 11.39 ± 5.491 12.07 ± 3.115 8.41 ± 3.294 7.97 ± 3.670 12.58 ± 3.593
34.52 ± 15.662 38.8% 13.16 ± 2.871 28.62 ± 1.571 93.23 ± 5.483 18.82 ± 1.088 18.76 ± 1.285 17.91 ± 2.084 18.19 ± 2.076 19.57 ± 0.703
The significances regarding the cognitive data were
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Validation of the Chinese version of the NUCOG cognitive screening tool in patients with epilepsy, dementia and other neurological disorders Lan Gao a, Shu-Chuen Li a,⇑, Li Xia b, Songqing Pan b, Dennis Velakoulis c,d, Mark Walterfang c,d a
School of Biomedical Sciences and Pharmacy, The University of Newcastle, Callaghan, NSW, Australia Neurology Department, Renmin Hospital of Wuhan University, Wuhan, Hubei, China c Neuropsychiatry Unit, Royal Melbourne Hospital, Melbourne, VIC, Australia d Melbourne Neuropsychiatry Centre, Melbourne Health and University of Melbourne, VIC, Australia b
a r t i c l e
i n f o
Article history: Received 15 April 2013 Accepted 24 September 2013
Keywords: Cognition Dementia Epilepsy Mini-Mental State Examination Neuropsychiatry Unit Cognitive Assessment Tool Screening
a b s t r a c t We aimed to develop and validate a Chinese version of the Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG) for use in Chinese-speaking subjects internationally. Patients and healthy controls were recruited from two hospitals between July and October 2012. Receiver operating characteristic (ROC) curves were utilized to test criterion validity. Convergent validity was assessed via correlations between NUCOG and the Mini-Mental State Examination (MMSE). Reliability was measured by internal consistency (Cronbach’s a). Patients with epilepsy (n = 144), neurological diseases (n = 81), dementia (n = 44), and controls (n = 260) completed the NUCOG and the MMSE. Overall, both NUCOG and MMSE scores differed significantly across the four groups with the highest scores in the control group and the lowest in the dementia group (p < 0.0001). The NUCOG scores could differentiate between patients with certain seizure types, stroke and transient ischemic attack. Compared to the MMSE, the NUCOG exhibited a higher area under the ROC curve. The convergent validity was substantially correlated, and internal consistency was very high (0.922). The Chinese version of NUCOG was demonstrated to be a sensitive and reliable screening tool for cognitive impairment in a Chinese-speaking population not only in China, but also in countries where there is a sizeable population of ethnic Chinese. Additionally, our study also showed the NUCOG could better differentiate cognitive function in patients with certain seizure types, stroke and transient ischemic attack than the MMSE. This potentially expands the clinical usefulness of NUCOG, enabling clinicians to measure the cognitive profile of patients with epilepsy and ischemic cerebrovascular diseases. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction The identification of cognitive impairment in at-risk individuals, including patients with neurological disorders, relies on the synthesis of information obtained from history-taking, physical and mental state examination, and investigative testing of cognition. Screening patients for cognitive assessment in busy medical settings is most commonly undertaken through the use of standardized cognitive screening tools (CST), which combine a number of individual test items to test a broad range of cognitive functions. Most of these tools were initially developed as English-language tools, with some being translated and validated in a limited subset of non-English languages. A small number of CST are available to test patients whose chief language is Chinese, including the MiniMental State Examination (MMSE) [1], the Beijing version of the
⇑ Corresponding author. Tel.: +61 2 4921 5921; fax: +61 2 4921 2044. E-mail address: [email protected] (S.-C. Li). 0967-5868/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2013.09.020
Montreal Cognitive Assessment (BJ-MoCA) [2], and the Saint Louis University Mental Status Examination (SLUMS) [3]. Balancing the need for sensitivity, breadth of assessment and brevity of administration in the construction of CST is difficult, and many available tools have significant limitations. For instance, the MMSE may be insensitive to early cognitive impairment, and lacks spatial recall and executive function testing. It also has a floor effect in patients with severe cognitive deficits [4]. While the MoCA was specifically developed to screen patients with mild cognitive impairment and mild dementia, the Chinese-language version may less successfully discriminate between these two disorders [5,6] and may be less successful in detecting cognitive impairment in neurological disorders [2]. To our knowledge the validity of the Chinese version of the SLUMS is yet to be reported [3]. Given the limitations of these instruments, the availability of another valid and portable Chinese-language screening tool for cognitive impairment may provide clinicians with an additional option to reliably detect cognitive impairment in Chinese-speaking patients. Such patients can be found in mainland Chinese settings, in addition
L. Gao et al. / Journal of Clinical Neuroscience 21 (2014) 980–987
to populations of ethnic Chinese living in many Southeast Asian countries and the increasing number of Chinese immigrants who have relocated to non-Chinese-speaking countries or regions (such as the USA, Canada, Australia, Europe, and Singapore) in recent years [7–9]. The availability of such a tool, allowing cognitive assessment to be carried out in the mother tongue of the respondent, minimizes comprehension difficulties as a confounding variable in the interpretation of test results. The Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG) was developed with the intention to address some of the limitations of existing screening tools. It includes extensive executive and spatial function testing, and produces a multidimensional cognitive score. The NUCOG comprises five domains: attention, visuoconstructional function, memory, executive function, and language, with a maximum score of 20 points in each domain [10]. The utility and validity of NUCOG has been demonstrated in a variety of neurological disorders including Alzheimer’s disease, frontotemporal and subcortical dementia, major mood and psychotic illness, epilepsy, stroke, Parkinson’s disease and head injury [10–12]. According to these validation studies, the NUCOG is capable of differentiating between patients with neurological and psychiatric illness associated with comorbid cognitive impairment and controls, and reliably discriminates between demented and non-demented individuals. Like a number of other cognitive tools, the NUCOG appears to have a greater specificity and sensitivity than the MMSE in the detection of significant cognitive impairment [10,11]. Additionally, a Persian translation of the NUCOG essentially demonstrated near-identical validity and reliability in a large mixed neurological, psychiatric and dementia clinical population [13]. This study aimed to translate the NUCOG for clinical use in Chinese patients with epilepsy, non-dementia neurological disease, and dementia, and to test its validity and reliability in these clinical disorders compared to a group of healthy controls. We also aimed to compare the performance of the Chinese-language versions of the NUCOG and MMSE to determine if the NUCOG demonstrated a superior capacity to detect cognitive impairment in these populations.
2. Methods The study was approved by the Ethics Committees of the two hospital study sites in China. All the participants provided informed consent prior to their inclusion in the study.
2.1. Subjects Adult patients from both inpatient and outpatient clinics from the Renmin Hospital of Wuhan University, and the Fifth Hospital of Wuhan (Wuhan, Hubei, China) were randomly selected by attending physicians or consultant neurologists/epileptologists between July and October 2012. The patients to be recruited included patients diagnosed with epilepsy, non-dementia neurological diseases, and those diagnosed with dementia according to the Diagnostic and Statistical Manual revision IV (DSM-IV) criteria. The epilepsy group was divided into four distinct subgroups of patients with predominantly simple partial, complex partial, secondary generalized seizures and primary generalized seizures. Adults without a history of illness associated with cognitive deficit, or any other neurological or psychiatric disease, were deemed as meeting the inclusion criteria for the healthy control group. These subjects were primarily recruited from medical school students, relatives and caregivers of aforementioned patients, and hospital general staff.
981
All the interviews were administered by L.G. (who was trained by the tool’s developers M.W. and D.V. at the Royal Melbourne Hospital, Australia) and L.X. 2.2. Measures 2.2.1. MMSE The MMSE is a well-established and widely-used screening tool for cognitive impairment. The Chinese version of the MMSE has been extensively deployed for a variety of disorders, including those disorders examined in this study [3,14–17]. 2.2.2. NUCOG The NUCOG is a 21 item tool which examines a comprehensive range of cognitive function and requires no instrumentation. Administration of NUCOG results in a total score for a respondent out of a possible 100, comprising scores out of 20 on its five cognitive domains. Performance on the English language version of the NUCOG has been demonstrated to correlate strongly with MMSE scores and formal neuropsychological testing [11]. 2.3. Translation process 2.3.1. Forward and backward translation Two Chinese physicians (L.G. and L.X.) independently translated the English version of NUCOG into simplified Chinese. The two Chinese versions of NUCOG were synthesized into one via thorough discussion between the two translators and using the input of the third consultant neurologist (S.Q.P.). Then another two bi-lingual physicians who were blinded to the original NUCOG as well as the study design performed the back-translation process. Lastly, the two back translations were submitted for review by the NUCOG developers for consistency with the original item content of the tool (M.W. and D.V.). 2.3.2. Cultural and linguistic adaptations The initial Chinese version of NUCOG was sent to a consultant neurologist, an epileptologist, and a neuropsychiatrist to evaluate the equivalence with the original version. Due to linguistic differences between English and Chinese languages, two key changes were proposed by these experts to achieve the equivalent difficulty and depth of the test. 1. Repetition: ‘‘banana, artillery, constitutional’’ were replaced by ‘‘WuDangShan, TianXiaWuDi, XiYangWuXianHao’’ for the equivalent pronounced length of these words. 2. Writing: ‘‘The boy’s aunt made a large pie out of steak and dough’’ was replaced by a Chinese language version of ‘‘Mother rode a horse to buy mangos’’ to match the same syntactical difficulty. 2.4. Pilot testing Forty subjects, including hospital general staff (n = 20), interns (n = 12), nurses (n = 4), and outpatients of a Neurology Clinic (n = 4) were recruited to complete the draft Chinese NUCOG. From their qualitative input, wording and phrasing were changed accordingly to avoid confusion in understanding. 2.5. Statistical analysis Demographic data (age and education) were compared using the Kruskal-Wallis test, while the chi-squared test was used to analyze sex proportion. One-way analysis of variance was employed to analyze cognitive data. Correlations between
982
L. Gao et al. / Journal of Clinical Neuroscience 21 (2014) 980–987
demographic variables and cognitive data were undertaken using Spearman’s correlation coefficient. Receiver operator characteristic (ROC) curves were constructed to examine criterion validity. Spearman’s correlations between NUCOG and MMSE were assessed to evaluate the convergent validity. Reliability was tested by internal consistency using Cronbach’s a. A comparison of NUCOG subscales across diagnostic groups was conducted using repeated-measures analysis of variance (RMANOVA), with within-subject effects tested using the Greenhouse-Geisser test, controlling for age and years of education. Furthermore, the performance of the NUCOG and MMSE across diagnostic groups was compared via multivariate ANOVA, using age and years of education as covariates. Statistical analyses were performed using the Statistical Package for the Social Sciences version 20.0 software (SPSS, Chicago, IL, USA). 3. Results In total 529 subjects were enrolled in this study, including patients diagnosed with epilepsy (n = 144), non-dementia neurological diseases (n = 81), dementia (n = 44) and healthy controls (n = 260) (Table 1). Among the patients with non-dementia neurological diseases, the diagnoses included stroke (n = 26), dizziness/vertigo (n = 26), transient ischemic attack (TIA; n = 13), headache (n = 2), meningitis (n = 7), myasthenia gravis (n = 1), Parkinson’s disease (n = 1), peripheral facial neuritis (n = 4), and Hunt syndrome (n = 1). All neurological patients were enrolled from an inpatient unit within the first 7 days of admission. The dementia group included patients with senile dementia of the Alzheimer type (SDAT, diagnosed according to USA National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association, [NINCDS-ADRDA] criteria), vascular dementia (VD, diagnosed according to the USA National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences [NINDS-AIREN] criteria) and other forms of dementia (OD, diagnosed according to DSM-IV criteria for dementia and clinician consensus). 3.1. Demographic data The demographic variables differed across the subject groups. The patient groups differed significantly with respect to age, with the neurological patient group the oldest, followed by the dementia group, and then the control and epilepsy groups whose mean age did not differ (p > 0.05). For level of education, the control group had the highest mean years of education, followed by the neurological and epilepsy groups (which did not differ), with the least-educated patients belonging to the dementia group
(p < 0.0001). Sex also varied among groups, with the control group having the lowest proportion of male participants with the dementia group including the most (Table 1). 3.2. Relationship between demographic data and cognitive data 3.2.1. NUCOG Age negatively correlated with NUCOG scores in the epilepsy (r = 0.234, p = 0.005), non-dementia neurological (r = 0.266, p = 0.017) and control groups (r = 0.401, p < 0.0001). In contrast, educational level positively correlated with the total NUCOG score in the non-dementia neurological (r = 0.328, p = 0.003) and control groups (r = 0.414, p < 0.0001). Sex did not affect total NUCOG score (Table 2). 3.2.2. MMSE As with the NUCOG, age in non-demented neurological (r = 0.239, p = 0.031) and control groups (r = 0.421, p < 0.0001) negatively correlated with MMSE score. However, with respect to education, the performance on MMSE was only positively correlated with the control group. Sex also did not affect MMSE scores (Table 2). 3.3. Comparison of NUCOG and MMSE scores between subject groups Both NUCOG and MMSE scores differed significantly between the four study groups (F3, 528 = 275.389, p < 0.0001 and F3, 528 = 171.529, p < 0.0001 respectively), with the highest scores in control group and the lowest in dementia group. However, when controlled for both age and level of education, the multivariate analysis of variance (MANOVA) did not indicate any difference for total NUCOG and MMSE scores in differentiating the patient groups (p = 0.428 and p = 0.846, respectively). Particularly, both NUCOG and MMSE failed to discriminate between epilepsy and neurological groups, while both tools were able to differentiate the dementia and neurological/epilepsy, dementia and control, and control and neurological/epilepsy groups (p < 0.0001). It was also worth noting that age (p = 0.002) and education level (p = 0.001) were shown to influence the NUCOG scores but not MMSE scores (p = 0.182 and p = 0.109, respectively) according to the MANOVA results. Age and education level adjusted estimations are shown in Table 3. 3.4. NUCOG profile across main patient groups RMANOVA was utilized to investigate how the cognitive profile (five domains of NUCOG) differed across diagnostic groups, with age and educational level as covariates. NUCOG subscale scores differed significantly between different groups (p < 0.0001). Additionally, interactions between subscales and age (p < 0.0001)/
Table 1 Comparisons of the Chinese version of the Neuropsychiatry Unit Cognitive Assessment Tool and Mini-Mental State Examination across different patient groups
Age Sex, male Education, years MMSE Total NUCOG Total NUCOG Attention NUCOG Visuoconstructional NUCOG Memory NUCOG Executive NUCOG Language
Epilepsy n = 144
Neurological n = 81
Dementia n = 44
Control n = 260
33.39 ± 13.060 52.1% 10.56 ± 2.961 26.26 ± 3.430 81.74 ± 11.814 16.90 ± 2.348 17.21 ± 2.511 14.65 ± 3.433 14.89 ± 3.623 17.92 ± 2.426
55.70 ± 10.213 60.5% 11.16 ± 2.754 25.94 ± 2.467 81.38 ± 7.586 17.26 ± 1.439 17.29 ± 1.662 14.71 ± 2.767 14.43 ± 3.044 17.65 ± 1.562
45.50 ± 18.575 63.6% 8.91 ± 2.321 17.91 ± 6.305 52.41 ± 15.361 11.39 ± 5.491 12.07 ± 3.115 8.41 ± 3.294 7.97 ± 3.670 12.58 ± 3.593
34.52 ± 15.662 38.8% 13.16 ± 2.871 28.62 ± 1.571 93.23 ± 5.483 18.82 ± 1.088 18.76 ± 1.285 17.91 ± 2.084 18.19 ± 2.076 19.57 ± 0.703
The significances regarding the cognitive data were
