Validity of the Montreal Cognitive Assessment as a Screen for Mild Cognitive Impairment and Dementia in African Americans
Journal of Geriatric Psychiatry and Neurology 2014, Vol. 27(3) 199-203 ª The Author(s) 2014 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0891988714524630 jgpn.sagepub.com
Felicia C. Goldstein, PhD1, Angela V. Ashley, MD1, Eric Miller, BS1, Olga Alexeeva, BS1, Lavezza Zanders, BS1, and Veronique King, BS1
Abstract The validity of the Montreal Cognitive Assessment (MoCA) as a screen for mild cognitive impairment (MCI) and dementia was evaluated in African Americans attending an urban outpatient memory disorders clinic. Eighty one patients 50 years old were administered the MoCA and neuropsychological tests. Clinicians, blinded to the MoCA scores, reviewed the neuropsychological findings and reports of instrumental activities of daily living and they assigned a diagnosis of normal cognition (NC; N ¼ 16), MCI (N ¼ 38), or dementia (N ¼ 27). The MoCA scores of the 3 groups were significantly different (NC > MCI > dementia). Using cutoff scores of 24 points for MCI and 22 points for dementia, the MoCA had .95 sensitivity and .63 specificity for MCI and .96 sensitivity and .88 specificity for dementia. The MoCA is a valid and cost-effective screen for cognitive impairment in African Americans but with a higher likelihood of falsely classifying persons with NC as having MCI. Keywords mild cognitive impairment, dementia, Montreal Cognitive Assessment, MoCA, African Americans
The early and accurate detection of cognitive impairment in older adults is of paramount importance, given the evidence that pathological changes in neurodegenerative disorders such as Alzheimer disease (AD) occur decades before individuals meet criteria for dementia1 as well as findings of subtle cognitive compromise associated with medical conditions such as vascular disease that can produce silent strokes.2 Consequently, efforts have increasingly shifted toward the identification of persons with mild cognitive impairment (MCI), a stage in which a person is functioning relatively independently in their environment but has subjective complaints and/or objective cognitive deficits.3 Mild cognitive impairment is a known risk factor for the transition to dementia, especially AD, with an average conversion rate of 4% to 15% per year in community-based studies.4,5 Using available data from the Aging, Demographics, and Memory Study,6 Plassman and colleagues estimated that 5.4 million Americans who were 71 years of age met the criteria for cognitive impairment—no dementia in 2002. At follow-up, an average of 17 months later, 11.7% of these individuals converted to a diagnosis of dementia, with 83% diagnosed with AD and almost 17% having possible vascular dementia. Screening efforts for MCI have typically relied on the MiniMental State Examination (MMSE).7 However, this measure omits subtests sensitive to executive dysfunction and thus may
miss the detection of causes of MCI such as vascular disease or frontotemporal dementia. In addition, the MMSE is now proprietary,8 costing slightly more than US$1 per copy, thus making it costly to use in clinical settings in light of rising health care costs and lower reimbursement rates. The Montreal Cognitive Assessment (MoCA),9 developed by Nasreddine and colleagues, has become increasingly adopted as an alternative screening tool. The MoCA consists of items evaluating memory, language, visuomotor ability, attention, and executive functioning. The MoCA has been found in several studies to be a more sensitive alternative to the MMSE for detecting MCI and dementia.10-14 In the current study, we investigated the validity of the MoCA as a screening tool for cognitive impairment in older adult African American patients attending a memory assessment clinic in the Southeast. We know of no study to date that has separately examined the sensitivity of cutoff scores on the 1
Department of Neurology, Emory University School of Medicine, Atlanta, GA, USA
Received 10/1/2013. Received revised 1/7/2014. Accepted 1/10/2014. Corresponding Author: Felicia C. Goldstein, Wesley Woods Health Center, 1841 Clifton Rd, NE, Atlanta, GA 30329, USA. Email: [email protected]
Downloaded from jgp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 18, 2015
Journal of Geriatric Psychiatry and Neurology 27(3)
MoCA in African Americans despite evidence for the high incidence of MCI and dementia in this ethnic group.5,15 Compared to Caucasians, African Americans are faced with a disproportionate number of vascular comorbidities such as hypertension that could differentially affect their scores on the MoCA.16 In addition, differences in access to and quality of education in African Americans versus Caucasians, especially those growing up in the segregation era in the south, could influence performance.17 A study conducted in Florida by Luis and colleagues12 found that the MoCA had excellent sensitivity and specificity for detecting MCI and dementia. However, their sample was 96% Caucasian and 70 years and older. In the present study, African American patients received the MoCA as well as a neuropsychological battery of cognitive measures. Clinicians, blinded to the MoCA scores, independently used results from the ‘‘gold standard’’ tests and information regarding independence in performing instrumental activities of daily living (IADLs) to make a diagnosis of normal cognition (NC), MCI, or dementia. Sensitivities, specificities, and areas under the curve (AUCs) were calculated to determine the optimal cutoff points to detect cognitive impairment in the sample.
Method Participants The sample consisted of a consecutive series of African American patients 50 years old who had received cognitive assessments as part of their standard workup in the Memory Disorders Clinic at Grady Memorial Hospital. Grady Memorial Hospital is one of the largest public hospitals in the southeast and it is a satellite memory assessment clinic for the National Institutes of Health—National Institute on Aging -supported Emory Alzheimer’s Disease Research Center. Over 90% of the patients are African Americans. This facility is the primary hospital for inner city and medically indigent populations in the 2 largest counties in the Atlanta metro area. We excluded patients from this study with preexisting conditions such as intellectual disabilities, drug and/or substance abuse, and severe psychiatric illness that could affect their performance on the cognitive measures apart from a primary neurodegenerative etiology. The study was approved by the Emory University and Grady Oversight Committee Institutional Review Boards.
Measures The cognitive assessment battery at the Grady Memory Disorders Clinic includes tests that are administered on the same day and in the same order for all individuals. The participants in this study were first evaluated with the MoCA, a brief screening measure that assesses visuoconstructional ability (copying a cube), memory (learning and delayed recall of 5 words), executive functioning (drawing a clock, untimed abbreviated version of Trails B, letter fluency, and similarities), sustained attention and working memory (detecting a target letter
embedded in a string of distractors, serial 7’s, repeating 4 digits, and reversing 3 digits), language (naming 3 pictures and repeating 2 sentences), and orientation. Scores range from 0 to 30, with higher scores signifying better performance. A 30- to 45-minute neuropsychological battery was then administered. The Consortium to Establish a Registry for Alzheimer’s Disease18 assesses multiple cognitive domains encompassing semantic memory (timed animal fluency), language (naming 15 pictures), visuoconstructional ability (copying 4 designs), and verbal and visual memory (immediate and delayed recall and recognition of 10 words, delayed recall of the designs). Measures of sustained attention and working memory (ability to repeat up to 8 digits and reverse up to 7 digits); Digit Span,19 and sequence numbers 1 to 25 (Trails A),20 and executive functioning (a timed extended version of Trails B)20 were also administered.
Procedure Raw scores on the neuropsychological tests (excluding the MoCA) were converted to standard scores using available norms.19,21,22 A printout of each patient’s raw and standard scores, without the score on the MoCA or any identifying information such as the patient’s name, was given to 2 clinicians (FG and AA). The clinicians also received a written summary of the patient’s performance, abstracted from that day’s clinical note, of their ability to perform IADLs such as managing medications, finances, cooking, shopping, and driving. Information concerning IADLs was obtained via patient’s self-report and often corroborated by an interview with an informant. The clinicians independently reviewed this information and then each determined whether the patient met the criteria for NC, MCI, or dementia. Impairment in the cognitive tests was defined as scores more than 1.5 standard deviations (SDs) below the demographically adjusted norms on one or more of the neuropsychological measures. The diagnosis of MCI followed the guidelines set forth by an expert panel including evidence of cognitive impairment on one or more of the objective neuropsychological tests and minimal impairment in IADLs.23 The classification of MCI was not limited to those with memory impairment and included individuals who demonstrated impairments in other single domains or multiple domains. The diagnosis of dementia required that 2 or more cognitive domains be impaired. Memory impairment, while not required for diagnosis, was impaired in >90% of the patients. For those demonstrating difficulties in multiple cognitive domains, the designation of ‘‘dementia’’ was determined based on the individual’s significant loss of functional independence judged to be severe enough to interfere with their social and/or occupational functioning. This typically involved impairments in higher order daily activities that could no longer be performed independently although some patients may have been able to complete basic activities such as dressing, grooming, and bathing. A diagnosis of ‘‘NC’’ required that the demographically corrected z scores were not more than 1.5 SDs below the norms and that IADLs were described as intact.
Downloaded from jgp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 18, 2015
Goldstein et al
Table 1. Demographic Features and MoCA Scores.a
Age, mean (SD); range Number of years of completed education, mean (SD); range Gender, N (%) female MoCA scores,b mean (SD); range MoCA scores,c mean (SD); range
Normal Cognition (N ¼ 16)
MCI (N ¼ 38)
65.8 (7.7); 50-78 13.4 (3.1); 8-20a 11 (68.8) 24.6 (3.1); 16-28a 25.1 (2.9); 17-28a
71.9 (8.9); 55-87 10.9 (2.4); 6-16a 27 (71.1) 19.1 (4.2); 7-25a 19.8 (4.2); 8-26a
Dementia (N ¼ 27) P Value 70.5 (11.4); 51-89 11.7 (2.2); 6-16 19 (70.4) 15.2 (4.1); 9-24a 15.9 (4.0); 10-25a
.10 .01 .93 .001 .001
Abbreviations: MCI, mild cognitive impairment; SD, standard deviation; MoCA, Montreal Cognitive Assessment. a A common superscript indicates a significant difference between the groups. b Uncorrected for education. c Corrected for education by adding 1 point if 12 years.9
Results The final sample included 81 patients. Of these, initial agreement in diagnosis was obtained for 71 (88%) of the patients. For the remaining 10 patients, a consensus diagnosis was reached through discussion. In all, 16 patients met diagnostic criteria for NC, 38 for MCI, and 27 for dementia. Table 1 shows their demographic features. Analysis of variance (ANOVA) indicated a significant group difference in education, F2, 78 ¼ 5.90, P < .01. Post hoc Sheffe analysis revealed that those with NC were more highly educated than those classified as having MCI (P < .01). An ANOVA also revealed a trend for age differences among the groups, F2, 78 ¼ 2.33, P ¼ .10, with normal controls being younger than those with MCI and dementia. The distribution of gender among the groups was comparable, w2(2) ¼ .04, P ¼ .93. To examine group differences, an analysis of covariance, controlling for age and education, was performed on the raw, noneducation-corrected MoCA scores. There was a significant main effect of group, F2,76 ¼ 24.79, P < .001. Post hoc Sheffe analysis revealed significant differences in performance (see Table 1), with normal controls receiving the highest mean score, followed by a decline in those with MCI, and the worse performance by those with dementia. The analyses were again repeated without controlling for demographic differences among the groups and the results were replicated, F2, 78 ¼ 28.91, P < .001. The scores of the 3 groups significantly differed from each other and were in the same direction as described earlier. The boxplots in Figure 1 graphically depict the distributions of MoCA scores and are consistent with the above-mentioned findings. Although there was some overlap among the groups, the general pattern suggested a decline in performance in the expected direction.
Sensitivity and Specificity Analyses Sensitivity, specificity, and AUCs were calculated to assess the diagnostic accuracy of the MoCA at various cutoff scores. Table 2 shows these values, with NC as the comparison group, for those with MCI alone or dementia alone. Based on the recommended cutoff score for impairment of