Dementia rates have been predicted to increase at an alarming rate in developing countries1 in the near future. Cerebrovascular disease (CVD) is currently the second most common cause of acquired cognitive decline and dementia, and contributes to neurodegenerative dementia.2 The prevalence of vascular cognitive impairment (VCI), involving all cognitive domains and causes of vascular injury, is likely to be greater than that of vascular dementia (VaD).3 Assuming the cerebrovascular mechanisms involved,4 VCI can be deemed as the most common pattern of cognitive decline.5
VCI has been defined as cognitive impairment evoked by or associated with vascular causes.3,6 Despite the heterogeneity of the VCI construct, it has been classified into three subtypes: VaD, Alzheimer's disease (AD) with a vascular component, and VCI that does not fulfill dementia criteria (referred to as vascular cognitive impairment, no dementia (VCIND)).7 Nevertheless, common clinical criteria for VCI diagnosis are lacking. Currently, the most widely used criteria for VaD include the International Classification of Diseases, Tenth Edition (ICD-10);8 the Diagnostic and Statistical Manual for Mental Disorders, Fourth Edition (DSM-IV);9 the State of California Alzheimer Disease Diagnostic and Treatment Centers (ADDTC) criteria,10 and the National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherche et l'Enseignement en Neurosciences (NINDS-AIREN) criteria.11 Apart from the ADDTC, all of these criteria recognize mixed dementia, but to a varied extent. In addition, VCIND is typically defined according to the criteria applied in the Canadian Study of Health and Aging.12 Clinically, VCI is generally under-diagnosis.
The recently developed VCI harmonization standards3 are a breakthrough in the development of diagnostic criteria of VCI. Consequently, a set of empirically derived criteria for VCI is now emerging with a high degree of reliability.13 Because VCI may be preventable,4 diagnosis at the earlier stages of the condition (e.g., VCIND) is particularly important. We therefore aimed to devise a diagnostic algorithm for VCI based on a survey of previous literature and a Delphi consensus method. This consensus method included the proposed diagnostic frameworks for probable, possible and definite VCI. We developed a set of key points to define VCIND, VaD, and mixed VCI/AD subtypes, and produced a flowchart for VCI diagnosis. In preliminary tests, we examined the reliability and validity of our proposed criteria.
Our study applied a combination of qualitative and quantitative methods (Figure 1), as described elsewhere.14 The overall methodology was as follows.
|view in a new window |
Figure 1．Formed new algorithm for VCI diagnosis.
A search strategy was employed for identifying references published in English listed in the PubMed database between January 1990 and December 2007, with the term “criteria/standard”, “clinical path/flow”, “vascular cognitive impairment”, “vascular cognitive deficit/disorder”, “poststroke dementia”, “reliability”, and “validity”. References selected were required to meet the following selection criteria: (1) they were original articles or reviews; (2) they involved clinical observations and case reports; (3) all assessments of cognitive function were measured using standardized neuropsychological tests; and (4) they were published in journals with relatively high Impact Factors. A total of 549 articles were initially obtained. However, 258 references were excluded because they failed to meet the selection criteria. Additional references were selected from “Vascular Cognitive Impairment”.15 We systematically reviewed all references between March 2006 and February 2008. An initial algorithm for VCI diagnosis was produced on the basis of the established clinical path,16 adjusted properly. Diagnostic frameworks for probable, possible and definite VCI, and the key points of definitions for three VCI subtypes were also developed.
We developed and distributed a questionnaire of semi-open and open questions, which included various items and an overview of the rationale of the study. We selected 10 experts in the field with specific knowledge on the specific research area, based on their publication history. Seven experts working specializing in dementia, senile disease, geriopsychosis, or neurology composed an expert panel, other 3 rejected. The questionnaire was sent either by email or conventional post. The experts were asked to make independent judgments on each index based on their professional training and experiences, giving weighted value for their results (score range, 0–10). Meanwhile, each expert was asked to provide a different answer.
The questionnaire distributed in the first round contained 69 items. Good consensus between 7 experts (Cronbach's α=0.96) was obtained. In the first questionnaire, significant moderate to good items-total correlations were found for 50 items (r range, 0.40–0.97). Based on these results, 19 items (r <0.40) were deleted, and an additional 17 new items were added. In the second round, a 67-item questionnaire was distributed. Similarly to the first round, good consensus was arrived between 6 experts (Cronbach's α=0.96) and significant moderate to good items-total correlations were obtained for 50 items (r range, 0.41–0.99) in total. Consequently, another 17 items (r <0.40) were deleted. The criteria were then revised accordingly. For statistic requirement, r values were interpreted as follow: <0.40, poor correlation; 0.40–0.70, moderate correlation; 0.71–1.00, good to excellent correlation. Data were analyzed using SPSS version 11.5 (SPSS Inc, USA).
Totally 100 patients with first or recurrent stroke (including a minority with transient ischemic stroke) participated in the study, aged between 43 to 84 years (76 male and 24 female). All patients were admitted to the Department of Neurology in Beijing Tiantan Hospital between April and June 2008. Stroke was defined according to the World Health Organization definition.17 Other eligibility requirements were (1) Chinese ethnicity, with good spoken Chinese language; (2) an age of ≥18 years; (3) mild dysphasia (a score ≤1) according to the language component of the National Institute of Health Stroke Scale (NIHSS);18 and (4) adequate vision and hearing. The study was approved by the medical ethics committee of the Beijing Tiantan Hospital. All patients or their legal proxy provided informed consent. Exclusion criteria were (1) the presence of a disturbance of consciousness, delirium, major depression, sensorimotor or aphasic disturbances severe enough to compromise cognitive evaluation and the administration of detailed scales; (2) the presence of unrelated neurological diseases other than AD; (3) severe psychiatric disorders, and (4) refusal to participate.
All patients underwent an abbreviated clinical evaluation,3 conducted by trained physicians, including a careful examination of case history. This assessment involved interviews with patients or a reliable informant, physical examinations, and neurological examinations (including the NIHSS). Laboratory evaluations included carotid artery ultrasound, echocardiography, electrocardiogram, computed tomography (CT) or magnetic resonance imaging (MRI) of the head, and an additional CT, MR or digital subtraction angiography. In addition, blood samples were collected to measure glucose, lipids, C-reactive protein, homocysteine, glycosylated hemoglobin, clotting factors, and fibrinogen. Although not part of the harmonization standards, acute onset, progression, occurrence of urinary incontinence and gait disturbance were also gathered through detailed history taking19 and clinical observations. Functioning level prior to the onset of the condition was measured using a modified Rankin Scale.20 Patients showing any change in the course of illness were observed and recorded carefully during their hospital stay.
Based on the recommendations of harmonization standards,3 a battery of neuropsychological tests was performed by a trained neurologist. A Chinese version of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE)21 was completed through an informant within the patients' first week of admission. The IQCODE examines changes experienced by the patients over the past ten years in the performance of daily activities requiring memory and other intellectual abilities. The scale items were scored on a five-point scale as 1 indicating “much better”, 3 “not much change”, and 5 “much worse”. The final score was computed as the average grade of the rated items. If a patient's condition was stable, the following additional neuropsychological tests were administered.
A Chinese version of Folstein's Mini-Mental State Examination was conducted to provide a global assessment of cognitive function, and to evaluate patients with dementia (scores of ≤17, 20, 22 and 23, indicate dementia at different levels of education).22,23 An “Animal Naming” semantic fluency task24 was chosen to assess executive functioning. In this task, all participants were asked to name as many different animals as they could within 60 seconds. For a measurement of information processing speed, the Reitan Trail Making Test Part A25 was administered to all subjects. Because a spatial task requires both organizational and visuoperceptual skills, primary visuospatial ability was further evaluated using the Clock Design Test (CDT).26 CDT is also suitable for examining language comprehension, short-term memory, and executive functioning.
Mental and behavioral status was examined using Jeffrey Cummings Neuropsychiatric Inventory (NPI),27 which concerns 10 aspects of behavioral and psychological symptoms. In most cases, the NPI was completed by a caregiver without the need for an interviewer. To more thoroughly probe poststroke depressive (PSD) condition, the Hamilton Depression Scale28 was administered using an interview and observation. A patient was considered to suffer from PSD if the clinical presentation fulfilled the DSM-IV criteria for depression, and that the mood disorder was due to stroke instead of other causes. In addition, the Activities of Daily Living (ADL) Scale29 was administered to measure patients' everyday functioning. All questions regarded the patient's self-maintenance abilities. If the patient was dependent on others for one of these activities, it was recorded as a disability on the corresponding items. Finally, the severity of cognitive deficits was rated from 0.5 (doubtful) to 3 (severe) on the basis of the Clinical Dementia Rating Scale.30
Diagnosis for VCI and its subtypes
All case vignettes were reviewed by a neurologist (Q.L.Z.) in accordance with the new criteria, and a preliminary judgment was made. If additional questions arose regarding any case vignettes, they were further discussed with a senior neurologist (S.R.G.) to verify the diagnoses. Dementia was distinguished using the DSM revised third edition (DSM-III-R) criteria.31 The cognition of patients affected by cognitive deficits was compared with their cognitive level prior to onset. Totally 66 of the 100 patients met the proposed criteria for probable or possible VCI. Of this sample, 33 patients also fulfilled DSM-III-R criteria for dementia. The 25 case vignettes with probable (n=21) and possible VCI (n=4) were then prepared in a standardized clinical format, and categorized into VCIND (n=9), VaD (n=10) or mixed VCI/AD (n=6) subtypes.
A panel of raters comprised 7 physicians from the affiliated hospital of a separate university. The panel included 2 geriatric neurologists, 1 psychiatrist, and 4 neurologists. Attentively, there were 2 raters participating in above expert's questionnaires. To avoid selection bias, 3 of 7 raters were inexperienced in the diagnosis of dementia. In assessing the 25 case vignettes, each rater was asked to make an independent yes/no judgment regarding the fulfillment of the new criteria for probable or possible VCI according to a provided instruction manual. Judgments were made for each VCI subtype in the same way. Finally, inter-raters reliability was assessed on the basis of diagnostic agreement between them.
Because of the lack of a currently accepted “gold standard” for VCI, we used convergent validity to assess the validity of the proposed criteria. Each rater made a yes/no judgment regarding the patients' fulfillment of currently accepted diagnostic criteria, including the criteria of diagnosis for VCIND,12 the NINDS-AIREN criteria for VaD,11 and the definitional diagnosis for AD/CVD.11,32 We treated this combination of measures as a so-called “gold standard”. Furthermore, each rater was required to provide a differential diagnosis. Finally, convergent validity was examined using correlations between the proposed key points of the definitions for VCIND, VaD or mixed VCI/AD subtypes and their currently accepted diagnostic criteria.
We selected 45 patients with probable VCI to test the diagnostic accuracy (including sensitivity and specificity) of our measures. Accuracy was calculated according to whether the patients met the proposed key points of definitions for the VCIND, VaD, or mixed VCI/AD subtypes. This measure was compared with the fulfillment of the corresponding currently used criteria (mentioned above). Concordance between our new criteria and the current diagnostic criteria was also examined.
Statistical analysis and sample calculation
The number of case vignettes and raters used for the validity of the study were calculated according that 66 of the 100 patients met our new criteria. Inter-rater reliability was analyzed using the diagnostic categories and Fleiss' kappa33 (SAS 9.1). To assess the strength of agreement, a previous study suggested the following guidelines for interpreting k scores: 0.00 to 0.20, slight agreement; 0.21 to 0.40, fair agreement; 0.41 to 0.60, moderate agreement; 0.61 to 0.80, substantial agreement, and 0.81 to 1.00, almost perfect agreement.34 Convergent validity was assessed using the Spearman's r between the compared criteria. In addition, diagnostic accuracy was determined by sensitivity and specificity, which were calculated using a χ2 test. Concordance between the proposed key points of definition for each subtype and the currently accepted criteria was computed using kappa statistics comparing the observed concordance to that predicted by chance. Data were analyzed using SPSS version 11.5 (SPSS Inc, USA). A P value <0.05 was considered statistically significant.
Algorithm of VCI diagnosis
Tables 1 and 2 show how the probable, possible and definite VCI, three VCI subtypes, and vascular causes were defined. Figure 2 shows how we identified each VCI subtype, and specified its causes. In the absence of specific biomarkers, the clinical diagnosis for VCI can only be probabilistic.
|view in a new window |
Table 1. Proposed criteria of diagnosis for probable, possible and definite VCI
|view in a new window |
Table 2. Proposed key points of definitions for three VCI subtypes
|view in a new window |
Figure 2. Flow-chart for VCI diagnosis and workup. *Suspected history but not obvious on neuropsychological tests. †The clinician performing the diagnostic assessment will fully consider the potential bias created by the informant who may minimize or exaggerate the condition.
Table 3 summarizes the judgments made for the 25 cases vignettes with our proposed criteria. There was a significant slight agreement in diagnosis between 7 raters for probable and possible VCI (k=0.13, P=0.001). Moderate agreement of diagnosis was found for three VCI subtypes (k=0.45, P <0.001).
|view in a new window |
Table 3. A final judgment made for the 25 case vignettes with different criteria
The judgments made for the 25 case vignettes with corresponding current criteria are shown in Table 3. From Table 4, it can be seen that we found good convergent validity. Significant correlations ranging from good (4-r range, 0.75–0.92) to perfect (3-r=1.00) were revealed for the VCIND subtype. Furthermore, significant moderate to good correlations were observed for the VaD subtype (1-r=0.46; 5-r range, 0.76–0.92) and the mixed VCI/AD subtype (r=0.92 and 1.00; 4-r range, 0.47–0.70), respectively.
|view in a new window |
Table 4. Convergent validity for the proposed criteria
ROC curves for the VCIND (Figure 3), VaD (Figure 4), and mixed VCI/AD subtypes (Figure 5) indicated a high level of diagnostic accuracy. A sensitivity and specificity analysis distinguishing each subtype from probable VCI produced the following results: (1) VCIND subtype: 75.90 % and 94.40 %; (2) VaD subtype: 34.40% and 100%, and (3) mixed VCI/AD subtype: 87.50% and 98.90%. Importantly, concordance between the new and old diagnostic criteria for the VCIND subtype and VCIND was good (k=0.72). There was a similarly good concordance between the mixed VCI/AD subtype and AD/CVD (k=0.86), but only moderate concordance between the VaD subtype and VaD (k=0.42), all P <0.001.
|view in a new window |
Figure 3. ROC curve predicted the diagnostic accuracy for the VCIND subtype, area under curve as 0.85 (95% CI: 0.75–0.95, P <0.001). Diagonal segments are produced by ties.
Figure 4. ROC curve predicted the diagnostic accuracy for the VaD subtype, area under curve as 0.67 (95% CI: 0.55–0.80, P=0.006). Diagonal segments are produced by ties.
Figure 5. ROC curve predicted the diagnostic accuracy for the mixed VCI/AD subtype, area under curve as 0.93 (95% CI: 0.80–1.07, P <0.001). Diagonal segments are produced by ties.
Major features of the proposed algorithm
The new diagnostic algorithm we developed VCI is similar to the currently accepted algorithm for the diagnosis and evaluation of dementia.35 However, our proposed criteria are focused on cognitive deficits that are likely to be either of vascular origin alone or in combination with AD.
Our study aimed to generate suitable criteria for diagnosing probable, possible and definite VCI, implementing new elements in accord with recent VCI harmonization guidelines. First, definitions of cognitive deficits from at least involvement in any one domain to an impairment of memory or executive function plus additional one (two cognitive deficits defined as lower for dementia). The unimpaired social abilities or ADL (do not due to physical effects of CVD alone) are defined as threshold, distinguishing patients with VCIND from dementia. Second, subcortical cognitive deficits are a prominent feature of VCI diagnosis, because small vessel disease is likely to be the most common cause of VCI.36,37 Third, our definition of CVD is broader than the current definition to take into account any findings emerging from neurological examination that might indicate CVD. Furthermore, our criteria incorporate any neuroimaging findings related to various vascular causes and take them into account for VCI diagnosis. In our algorithm, however, a temporal relationship between CVD and cognitive deficits is not required. A recognized stroke is likewise not required for a VCI diagnosis under our criteria, but this relationship may be inferred from the following features: sudden onset, clinical course (e.g., fluctuation or stepwise deterioration), outcomes and prognosis. Specifically, underlying the proposed criteria is any vascular-related cognitive impairment that might capture the heterogeneity of VCI.
In contrast, we also proposed three clinical subtypes of VCI, based solely on clinical characteristics without considering the effects of vascular injury distribution on neuroimaging. This classification is in accord with the viewpoint that the diagnosis of VCI subtypes is more strongly associated with clinical features than with radiographic features.38 The mixed VCI/AD subtype designation in our proposed criteria should be interpreted with some caution, because patients with subtypes of VCIND or VaD are more likely to concurrently suffer from AD. However, there may be difficulties in distinguishing these combinations clinically. The three VCI subtypes we proposed should be further explored in the future.
Some major elements required in the current criteria are also implemented into our proposed criteria: “patchy” distribution of the deficits (in ICD-10); the course of abrupt onset; fluctuating, or stepwise deterioration (in NINDS-AIREN); optional behavioral and psychological symptoms of dementia (in DSM-III, DSM-III-R and ICD-10), cognitive deficits related to vascular lesions (in DSM-III-R, DSM-IV, ICD-10 and NINDS-AIREN), and features supporting the diagnosis for probable VCI and different levels of certainty of the diagnosis (in ADDTC and NINDS-AIREN). Finally, we also incorporated definitions of dementia focused on impaired social functioning (in DSM-III, DSM-III-R, DSM-IV, and ICD-10).
Rating reliability and validity
Between raters, there was a lower agreement of diagnosis for probable and possible VCI than for three VCI subtypes. It is likely that our proposed key points for each subtype are more concise than for VCI in general, helping the raters to understand them more easily, particularly if they were inexperienced. Nevertheless, our diagnostic agreement for VCI and three VCI subtypes was lower than that found by Wentzel et al with other criteria.13 This discrepancy may be due to levels of rater experience differing between the two studies.
Our new criteria showed good convergent validity. The data shown in Table 4 must be interpreted cautiously because we applied three current criteria as so-called “gold standards”. In psychometric research, the term accuracy can be used interchangeably with validity. In fact, our proposed key points distinguished the VCIND, VaD, or mixed VCI/AD subtypes with accuracy greater than 65%. Apart from the VaD subtype, other subtypes seemed to be highly sensitive (>70%) clinical measures. Moreover, the proposed criteria discriminated each VCI subtype with high specificity (>90%). Additionally, concordance between our new criteria for three VCI subtypes and the corresponding current criteria was significant, ranging between moderate and good levels of concordance.
Unfortunately all experts participating in our study were from the same country, and the sample of experts was relatively small. The sample of patients was likewise very small, and mainly included patients with stroke. However, VCI often develops in patients without stroke, and can coexist with other neurodegenerative diseases, such as frontotemporal lobe degeneration, dementia with Lewy bodies, Parkinson's disease dementia, progressive supranuclear palsy, and Huntington's disease, among others. Because data from neuropathology were not available for all patients, we could not make definite VCI diagnoses for them. A false positive rate could not be calculated and we lacked follow up data. In addition, we did not test the external validity and the discriminative validity for the proposed criteria. Critically, we were unable to evaluate the criterion-related validity because of a lack of an accepted VCI gold standard, particularly in pathology.
Taken together, our findings suggest that the proposed criteria might have good reliability and validity in the clinical assessment of stroke. The algorithm for VCI diagnosis we developed may be a step towards providing an appropriate diagnostic approach for VCI clinically, particularly in the earlier stage of the condition. A prospective multicenter study will be needed in future to assess the reliability and validity of our new diagnostic algorithm in different settings (e.g., unselected heterogeneous community samples). Ultimately, the sensitivity and specificity for the neuropathological diagnosis of VCI will also need to be evaluated.
Acknowledgements: We thank the following experts for participating in a Delphi consensus study: CHEN Kang-ning, GAO Su-rong, PENG Dan-tao, WANG Lu-ning and YU Xin. We also thank the panel of raters: CHEN Kang-ning, GAO Su-rong, LI Tao, NIU Song-tao, SHI Wei-xiong, YANG Xiao-na and ZHOU Heng. We thank RU Xiao-juan for statistical support, LI Hao for Fleiss Kappa, and ZHENG Hua-guang for the ensuring diagnostic accuracy (with SPSS version 13.0), respectively. We gratefully acknowledge GAO Su-rong for helpful comments on this manuscript.
1. Kalaria RN, Maestre GE, Arizaga R, Friedland RP, Galasko D, Hall K, et al. Alzheimer's disease and vascular dementia in developing countries: prevalence, management, and risk factors. Lancet Neurol 2008; 7: 812-826.
2. O'Brien JT, Erkinjuntti T, Reisberg B, Roman G, Sawada T, Pantoni L, et al. Vascular cognitive impairment. Lancet Neurol 2003; 2: 89-98.
3. Hachinski V, Iadecola C, Petersen RC, Breteler MM, Nyenhuis DL, Black SE, et al. National Institute of Neurological Disorders and Stroke-Canadian Stroke Network vascular cognitive impairment harmonization standards. Stroke 2006; 37: 2220-2241.
4. Moorhouse P, Rockwood K. Vascular cognitive impairment: current concepts and clinical developments. Lancet Neurol 2008; 7: 246-255.
5. Román G, Sachdev P, Royall D, Bullock R, Orgogozo J, López-Pousa S, et al. Vascular cognitive disorder: a new diagnostic category updating vascular cognitive impairment and vascular dementia. J Neurol Sci 2004; 226: 81-87.
6. Hachinski VC, Bowler JV. Vascular dementia. Neurology 1993; 43: 2159-2160; author reply 2160-2151.
7. Wentzel C, Rockwood K, MacKnight C, Hachinski V, Hogan DB, Feldman H, et al. Progression of impairment in patients with vascular cognitive impairment without dementia. Neurology 2001; 57: 714-716.
8. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva, Switzerland: World Health Organization; 1992: 50-51.
9. American Psychiatric Association Committee on Nomenclature and Statistics. Diagnostic and statistical manual of mental disorders (DSM-IV), 4th Ed. Washington, D.C.: American Psychiatric Association; 1994.
10. Chui HC, Victoroff JI, Margolin D, Jagust W, Shankle R, Katzman R. Criteria for the diagnosis of ischemic vascular dementia proposed by the State of California Alzheimer's disease diagnostic and treatment centres. Neurology 1992; 42: 473-480.
11. Roman GC, Tatemichi TK, Erkinjuntti T, Cummings JL, Masdeu JC, Garcia JH, et al. Vascular dementia: diagnostic criteria for research studies: report of the NINDS-AIREN International Workshop. Neurology 1993; 43: 250-260.
12. Graham JE, Rockwood K, Beattie BL, Eastwood R, Gauthier S, Tuokko H, et al. Prevalence and severity of cognitive impairment with and without dementia in an elderly population. Lancet 1997; 349: 1793-1796.
13. Wentzel C, Darvesh S, MacKnight C, Shea C, Rockwooda K. Inter-rater reliability of the diagnosis of vascular cognitive impairment at a memory clinic. Neuroepidemiology 2000; 19: 186-193.
14. Gibson G, Timlin A, Curran S, Wattis J. The scope for qualitative methods in research and clinical trials in dementia. Age Ageing 2004; 33: 422-426.
15. Erkinjuntti T, Roman G, Chui H. Diagnostic criteria. Erkinjuntti T, Gauthier HS, Montreal, eds. Vascular cognitive impairment. London: Martin Dunitz Ltd Publication; 2002: 43-55.
16. Panella M, Marchisio S, Di Stanislao F. Reducing clinical variations with clinical pathways: do pathways work? Int J Qual Health Care 2003; 6: 509-521.
17. World Health Organization, MONICA Project Principal Investigators. The World Health Organization MONICA Project (monitoring trends and determinants in cardiovascular disease): a major international collaboration. WHO MONICA Project Principal Investigators. J Clin Epidemiol 1988; 41: 105-114.
18. Brott T, Adams HP Jr, Olinger CP, Marler JR, Barsan WG, Biller J, et al. Measurement of acute cerebral infarctions: a clinical examination scale. Stroke 1989; 20: 864-870.
19. Stephens S, Kenny RA, Rowan E, Kalaria RN, Bradbury M, Pearce R, et al. Association between mild vascular cognitive impairment and impaired activeities of daily living in older stroke survivors without dementia. J Am Geriatr Soc 2005; 53: 103-107.
20. van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke 1988; 19: 604-607.
21. Fuh JL, Teng EL, Lin KN, Larson EB, Wang SJ, Liu CY, et al. The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) as a screening tool for dementia for a predominantly illiterate Chinese population. Neurology 1995; 45: 92-96.
22. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189-198.
23. Wang ZY, Zhang MY, Qu GY, Chen JX, Zhao J. Application of Chinese version of the Mini-Mental State Examination (MMSE). Shanghai Arch Psychiatry (Chin) 1989; 7: 108-111.
24. Chan AS, Poon MW. Performance of 7- to 95-year-old individuals in a Chinese version of the category fluency test. J Int Neuropsychol Soc 1999; 5: 525-533.
25. Reitan RM, Wolfson D. The Halstead-Reitan Neuropsychological Test Battery. Tucson: Neuropsychological Press; 1985.
26. Tuokko K, Hadjistavropoulos T, Miller JA, Beattie BL. The clock test: a sensitive measure to differentiate normal elderly from those with Alzheimer's disease. J Am Geriatr Soc 1992; 40: 579-584.
27. Cummings JL, Mega M, Gray K, Rosenberg-Thompson S, Carusi DA, Gornbein J. The neuropsychiatric inventory: comprehensive assessment of psychopathology in dementia. Neurology 1994; 44: 2308-2314.
28. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23: 56-62.
29. Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969; 9: 179-186.
30. Morris JC. The clinical dementia rating (CDR): current version and scoring rules. Neurology 1993; 43: 2412-2414.
31. Diagnostic and statistical manual of mental disorders, 3rd ed. rev.: DSM-III-R. Washington, D.C.: American Psychiatric Association; 1987.
32. McKhann G, Drachmann D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology 1984; 34: 939-944.
33. Fleiss JL. Measuring nominal scale agreement among many raters. Psychological Bulletin 1971; 76: 378-382.
34. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159-174.
35. Corey-Bloom J, Thal LJ, Galasko D, Folstein M, Drachman D, Raskind M, et al. Diagnosis and evaluation of dementia. Neurology 1995; 45: 211-218.
36. Esiri MM, Wilcock GK, Morris JH. Neuropathological assessment of the lesions of significant in vascular dementia. J Neurosurg Psyhiatry 1997; 63: 749-753.
37. Roman GC, Erkinjuntti T, Wallin A, Pantoni L, Chui HC. Subcortical ischaemic vascular dementia. Lancet Neurol 2002; 1: 426-436.
38. Rockwood K, Black SE, Song XW, Hogan DB, Gauthier S, MacKnight C, et al. Clinical and radiographic subtypes of vascular cognitive impairment in a clinic-based cohort study. J Neurol Sci 2006; 240: 7-14.