Montreal Cognitive Assessment
One Cutoff Never Fits All
Background and Purpose—The objective of this study is to examine the discrepancy between single versus age and education corrected cutoff scores in classifying performance on the Montreal Cognitive Assessment (MoCA) in patients with stroke or transient ischemic attack.
Methods—MoCA norms were collected from 794 functionally independent and stroke- and dementia-free persons aged ≥65 years. magnetic resonance imaging was used to exclude healthy controls with significant brain pathology and medial temporal lobe atrophy. Cutoff scores at 16th, 7th, and 2nd percentiles by age and education were derived for the MoCA and MoCA 5-minute Protocol. MoCA performance in 919 patients with stroke or transient ischemic attack was classified using the single and norm-derived cutoff scores.
Results—The norms for the Hong Kong version of the MoCA total and domain scores and the total score of the MoCA 5-minute protocol are described. Only 65.1% and 25.7% healthy controls and 45.2% and 19.0% patients scored above the conventional cutoff scores of 21/22 and 25/26 on the MoCA. Using classification with norm-derived cutoff scores as reference, locally derived cutoff score of 21/22 yielded a classification discrepancy of ≤42.4%. Discrepancy increased with higher age and lower education level, with the majority being false positives by single cutoffs. With the 25/26 cutoff of the original MoCA, discrepancy further increased to ≤74.3%.
Conclusions—Conventional single cutoff scores are associated with substantially high rates of misclassification especially in older and less-educated patients with stroke. These results caution against the use of one-size-fits-all cutoffs on the MoCA.
The Montreal Cognitive Assessment (MoCA)1 is recommended as a cognitive screen for vascular cognitive impairment by the National Institute of Neurological Disorders and Stroke–Canadian Stroke Network International Harmonization Standards.2 Single cutoff scores are frequently used to classify cognitive impairment in different MoCA versions. The objective of this study is to compare the discrepancy in the classification of MoCA performance in patients with stroke or transient ischemic attack using single versus age and education corrected cutoff scores derived from a group of functionally independent community older adults free of stroke, dementia, and significant brain pathologies, including medial temporal lobe atrophy, a neuroimaging marker of Alzheimer disease.3 The age and education corrected norms for the total scores of the Hong Kong version of the MoCA4 and the MoCA 5-minute protocol5 are also described.
Detailed participant inclusion/exclusion criteria are described in the Methods section in the online-only Data Supplement. Scores on the MoCA and MoCA 5-minute protocol of the healthy controls were stratified according to age on the first level and then education on the second level. Age was divided into 3 strata: 65 to 69, 70 to 79, and ≥80 years. Five education strata of 0 to 3, 4 to 6, 7 to 9, 10 to 12, and >12 years were nested within each age stratum. Education was divided into 0 to 6 and >6 years stratum in the oldest age group (ie, ≥80 years) because of the relatively small sample size in this group. Cutoff scores at 16th, 7th, and 2nd percentile are described for the psychometric classification for Major (2nd percentile) and Mild (16th percentile) Neurocognitive Disorders in the Diagnostic and Statistical Manual of Mental Disorders 5th edition as well as Petersen revised diagnostic criteria of mild cognitive impairment (7th percentile).6
Performance of patients were classified as unimpaired or impaired by single cutoff scores of 21/22 (local version)4 and 25/26 (original English version)1 and age and education corrected percentile cutoffs. Using norm-derived cutoff as reference, classifications by single cutoff scores were described as consistent when the classifications of the single and norm-derived cutoff scores agreed with each other; a miss was defined when a patient classified as impaired by the norm was classified as unimpaired by the single cutoffs; a false positive was determined when a patient classified as unimpaired by the norm was classified as impaired by the single cutoff scores.
Seven hundred ninety-four healthy controls and 919 patients with stroke were recruited. Group comparisons are shown in Table I in the online-only Data Supplement. The Table shows the age and education corrected cutoff scores for total scores of the MoCA and MoCA 5-minute protocol; 69.1% and 29% healthy controls and 48.1% and 20.9% patients with stroke scored above the 21/22 and 25/26 cutoff scores, respectively on the MoCA. Consistency between 21/22 and norm-derived cutoffs were observed in 68.1%, 54.9%, and 57.6% at 16th, 7th, and 2nd percentiles, respectively. In general, consistency decreased with higher age and lower education level. With the 21/22 cutoff, the majority of misclassifications were false positives, with the highest rate ≤55.8% among patients with lowest education (0–3 years). This cutoff score also tended to miss cases identified as impaired by the norms among patients with increasing education level and with decreasing age. With the 25/26 cutoff, consistency further reduced to 57.6%, 46.6%, and 32.7% at 16th, 7th, and 2nd percentiles, respectively. False positives cases were found in all age and education levels and were highest in patients between 70 and 79 years of age (69.9%) and those with 4 to 6 years of education (74.3%). Figure shows the discrepancy in classification by single versus norm-derived cutoff scores across age and education groups.
One-size-fits-all cutoffs are commonly used to define impaired performance on the MoCA. Comparing with age and education corrected cutoffs, we showed a classification discrepancy of ≤74.3% for the 25/26 cutoff in original English MoCA and 55.8% for the local 21/22 cutoff. In general, single cutoff scores tended to classify older and less-educated patients as impaired.
The normative comparison approach is well established in clinical neuropsychology.7 The key advantage of this over other approaches (eg, single cutoffs derived from normal versus specific condition) is that norms can be applied regardless of the clinical condition and diagnostic criteria. Nevertheless, it should be emphasized that cutoff scores serve only as an arbitrary classification for objective test data. In clinical practice, some patients with low education having low test scores (eg, <20) might not be defined as cognitively impaired by the norms. However, as the level of cognition as reflected by the low scores may translate to limitations in daily functioning, these patients should receive appropriate clinical attention regardless of the classification of test performance. Furthermore, it is important to stress that test scores should be interpreted in light of other clinical data, such as clinical history, collateral information, and behavioral observations.
It should be noted that the MoCA was not originally designed for stroke and its performance heavily relies on expressive speech and motor responses. Therefore, language and physical deficits commonly seen after stroke may significantly bias its performance. Like cutoff scores, there is no one-size-fits-all cognitive test. The clinician should be mindful of test limitations and be flexible when choosing tests suitable to the background and abilities of the individual patient.
The strength of our study was the inclusion of a large cohort of elderly persons, in which the norms were derived. Normality in this study was defined in terms of functional level such that a wide spectrum of scores found in functionally independent individuals could be captured. This is also the first normative data study on the MoCA with significant brain pathologies and medial temporal lobe atrophy screened out using magnetic resonance imaging. A major limitation of our study is that norms were only available for people aged ≥65 years. Data from younger samples should be collected in the future.
Sources of Funding
This study was supported by the General Research Fund of Research Grants Council (reference number CUHK471911).
Guest Editor for this article was Terence J. Quinn, MD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011226/-/DC1.
- Received August 17, 2015.
- Revision received September 10, 2015.
- Accepted August 17, 2015.
- © 2015 American Heart Association, Inc.
- Hachinski V,
- Iadecola C,
- Petersen RC,
- Breteler MM,
- Nyenhuis DL,
- Black SE,
- et al
- Wong A,
- Nyenhuis D,
- Black SE,
- Law LS,
- Lo ES,
- Kwan PW,
- et al
- Lezak MD,
- Howieson DB,
- Loring DW.