Article Text
PDF
Abstract
Objective Identification of patients with aneurysmal subarachnoid haemorrhage (aSAH) with cognitive impairment is important for patient management (medical treatment, cognitive rehabilitation and social arrangements). The Montreal Cognitive Assessment (MoCA) is currently recommended over the Mini-Mental State Examination (MMSE) by the US National Institute of Neurological Disorder, in the chronic post-stroke setting. We hypothesised that the MoCA has a better correlation with functional outcome at 3 months than the MMSE.
Methods We carried out a prospective observational study in Hong Kong over a 2 year period, recruiting patients aged 21–75 years with aSAH admitted within 96 h of ictus. The assessments included the modified Rankin Scale, Lawton Instrumental Activity of Daily Living (IADL), Short Form-36, MoCA and MMSE at 3 months after ictus. Analyses were carried out to compare MoCA with MMSE.
Results 90 patients completed the 3 month assessments. Cognitive impairment (MoCA <26) was determined in 73% of patients at 3 months. Delayed cerebral infarction explained the 31–38% variance in cognitive outcomes (MMSE and MoCA) at 3 months. MoCA demonstrated good discrimination of favourable neurological and IADL outcomes similar to the MMSE in receiver operating characteristics curve analyses.
Conclusions MoCA defined cognitive impairment was common at 3 months after aSAH and MoCA correlated with functional outcomes similar, but not superior, to the MMSE.
The study is registered at ClinicalTrials.gov of the US National Institutes of Health (NCT01038193).
Statistics from Altmetric.com
Introduction
Although aneurysmal subarachnoid haemorrhage (aSAH) accounts for only 3% of strokes, its profound consequences and unique window for intervention have justified its classification as a separate entity.1 Estimated independence varies between 36% and 60% after aSAH.2 ,3 Our local study suggested that 27–44% of patients who returned to the community had cognitive dysfunction.4 ,5 Factors that cause brain injury and cognitive impairment, following aSAH, include delayed cerebral ischaemia (DCI), direct injury from cerebral haematoma, raised intracranial pressure and chronic hydrocephalus. Identification of aSAH patients with cognitive impairment is important for patient management (medical treatment, cognitive rehabilitation and social arrangements). However, administration of a battery of detailed neuropsychological assessments is not always practical at the bedside.
The most common cognitive screening assessment tool is the Mini-Mental State Examination (MMSE) which is quick (5–10 min) and easy to administer in a clinic setting. However, the MMSE was originally designed for screening Alzheimer's disease and does not encompass all of the cognitive deficits that might occur following a stroke. It is particularly weak in its ability to measure executive functions, such as abstract thinking, judgement, problem solving and perception, which are cognitive deficits associated with vascular diseases and functional impairment.6
In recognition of MMSE's deficits, the Montreal Cognitive Assessment (MoCA) is currently recommended over the MMSE by the US National Institute of Neurological Disorder, in the chronic post-stroke setting.7 By comparison, the MoCA places more emphasis on tasks of frontal executive function and attention than the MMSE, and it may be more sensitive for the detection of non-Alzheimer's disease dementia. It contains demanding tasks to assess higher level language abilities, memory and complex visuospatial processing,8 and therefore has fewer pronounced ceiling effects than the MMSE. All of these features make the MoCA a theoretically ideal screening tool for aSAH patients with suspected cognitive impairment. Godefroy et al however showed that in a cohort of consecutive stroke patients, at adjusted cut-off scores to optimise sensitivity and specificity, the MoCA was not superior to the MMSE in diagnosing poststroke cognitive impairment.9
Schweizer et al subsequently found that in a 32 aSAH patient study, the MoCA was more sensitive to cognitive impairment than the MMSE using a battery of neurocognitive tests.10 Superior performance on the animal naming and abstraction subtests of the MoCA scores were associated with return to work following aSAH. It would be of interest to see how the MoCA, compared with the MMSE, performs in relation to functional outcomes such as neurological and instrumental activity of daily living.
We therefore investigated the prevalence and risk factors of post-SAH cognitive impairment defined by the MMSE and MoCA, and how performances in each test correlate with patient's functional outcome 3 months after ictus.
Methods
This prospective observational four centre study was carried out in Hong Kong over a 2 year period. The current study is registered at ClinicalTrials.gov of the US National Institutes of Health (NCT01038193) and was approved by the hospital ethics committees. The study conforms to the Declaration of Helsinki and written informed consent was obtained from all participants or their next of kin.
Patient inclusion criteria were: (1) spontaneous SAH with an angiography confirmed aetiology of intracranial aneurysms; (2) hospital admission within 96 h after ictus; (3) age between 21 and 75 years; (4) speaker of Chinese (Mandarin or Cantonese); and (5) informed consent from patients or their next of kin. Patient exclusion criteria were: (a) history of previous cerebrovascular or neurological disease other than unruptured intracranial aneurysm; or (b) history of neurosurgery prior to ictus; or (c) known dementia or cognitive impairment prior to ictus; or (d) unable to cooperate for cognitive assessments (not obeying command or significant dysphasia). A total of 280 consecutive patients were screened and 90 patients completed all 3 month assessments, including the MoCA and MMSE (figure 1). Patients were excluded from the current analysis if they could not complete all of the 3 month assessments, including the MoCA and MMSE.
Flowchart of patient recruitment. MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment.
Cerebral infarction due to DCI is defined as a new cerebral infarction identified on CT after exclusion of procedure related infarctions. Procedure related infarction was defined as new hypodensity appearing on the post-treatment CT at around 12–24 h after aneurysm treatment. All recruited patients had delayed CT of the brain at 2–3 weeks after presentation available for assessment. The diagnoses of cerebral infarction due to DCI were made by site investigators. Clinical deterioration due to DCI is defined as clinical deterioration, presumably caused by cerebral ischaemia after exclusion of other potential causes.11
Assessments
Assessments were carried out 3 months after ictus by one of the two research assistants (psychology graduates) trained by a postdoctoral research psychologist.
Montreal Cognitive Assessment
The MoCA12 is a 1 page, 30 point test that is usually administered within 15 min, and evaluates the following seven cognitive domains: visuospatial/executive functions, naming, verbal memory registration and learning, attention, abstraction, 5 min delayed verbal recall and orientation.7 One point is added for education less than 12 years. The Hong Kong version has previously been validated in Chinese patients with cerebral small vessel disease.12 We also recently reported the application of the MoCA in neurosurgical patients after traumatic and spontaneous intracerebral haemorrhage.13 Cognitive impairment on the MoCA was defined as a score of <26.7
MMSE Chinese (Cantonese) version
The MMSE14 ,15 comprises five sections (orientation, registration, attention and calculation, recall and language). The maximum total score is 30 and the test can usually be completed within 10 min. The Cantonese version was previously validated in a population of demented patients.15 Cognitive impairment on the MMSE was defined as a score of <27.14
Modified Rankin Scale
The Modified Rankin Scale (mRS)16–18 is a valid and clinically relevant disability scale to assess recovery and is commonly used in stroke trials.19 mRS identifies activity limitation and does not identify deficits. It ranges from 0 (no symptom) to 6 (death).
Chinese Lawton Instrumental Activity of Daily Living Scale
The Lawton Instrumental Activity of Daily Living (IADL) Scale20 is an appropriate instrument to assess independent living skills. Items assessed include ability to use the telephone, go shopping, prepare food, do the housekeeping, do the laundry, use of transportation, responsibility for own medications and ability to handle finances. The Chinese version has been validated and used previously.21
Geriatric Depressive Scale
Post-stroke depressive symptoms were assessed using the Chinese version of the 15 item version of the Geriatric Depressive Scale GDS.22–24 The Chinese version of the GDS was previously validated against the Structural Clinical Interview for DSA-IV diagnosis of depression and was found to have high negative predictive value but low positive predictive value.24 The Chinese version of the GDS was previously employed to detect poststroke depression among Chinese in the community.25
Short form-36
The Short form-36 (SF-36) is a 36 item generic general health questionnaire that yields scores on eight health subscales relating to physical health and social and mental well being.26 ,27 The Chinese version has been validated previously, and population based norms are available.28 ,29 The health related quality of life (HRQOL) assessment was omitted when the patient was unable to follow commands or to comprehend and answer the questions correctly. Scores of the various scales are combined to form physical health and mental health component scores. The procedures for scoring and the computation of the scale and component scores of SF-36 have recently been described for aSAH patients using norms for the Hong Kong population.30
Statistical analysis
The trial data were collected on printed forms and entered into a computer using Access 2003 software (Microsoft Inc, Redmond, Washington, USA). Statistical analyses were generated using SPSS for Windows V.15.0 (SPSS Inc) and MedCalc V.12.2.1.0. Categorical data are given as numbers (percentages), unless otherwise specified; numerical data are given as means and SDs; and ordinal data are given as medians and IQRs. A difference with a p value <0.05 was regarded as statistically significant (two tailed test). Categorical data were analysed using the Fisher's exact test or χ2 test, with ORs and 95% CI as appropriate. Correlations between numerical or ordinal data were assessed using Kendall's rank correlation (Kendall's tau b coefficient).
Multiple regression analysis was performed to determine the factors associated with the MMSE and MoCA scores using the enter method, with the F probability of entry set at 0.05 and that of removal set at 0.10. The tolerance value is an indicator of the extent to which the variance of the specified independent variable is not explained by the other independent variables in the model and is calculated using the formula 1−R2 for each variable. The independent variables considered included age, sex, years of education, GDS score, World Federation of Neurosurgical Societies (WFNS) grade, Fisher CT grade, mode of treatment of the aneurysm and (dichotomised with absence as the baseline) intraventricular haemorrhage, intracerebral haemorrhage, cerebral infarction due to DCI and clinical deterioration due to DCI.
Receiver operating characteristic (ROC) curves were constructed to examine the ability of the MoCA and MMSE to differentiate favourable neurological outcome (mRS 0–2) and high IADL score (≥15) at 3 months. The area under the curve (AUC) was calculated for each ROC curve. AUC represents the probability that, when one sample is drawn from a truly normal population and another sample from a truly abnormal population, the score of the normal sample will be higher than that of the abnormal sample. A larger AUC denotes better correlation. AUCs were presented with 95% CI. To differentiate favourable neurological outcome (mRS 0–2) and high IADL score (≥15) at 3 months, cut-off values were then derived at ROC coordinate points where both sensitivity and specificity were optimised using the Youden Index. Sensitivity, specificity, positive and negative predictive values (PPV and NPV) and diagnostic accuracy at the optimal cut-offs were calculated for both the MoCA and MMSE. Statistical significances of the differences between AUCs were assessed with the non-parametric approach of Delong et al.31
Results
Patient cohort
Ninety patients completed the 3 month assessments. The patient profiles of the included and excluded patients (who consented to data collection during acute admission) are shown in table 1. Comparing the included patients with the excluded patients, there was a higher proportion of men, more patients with grade I–II WFNS, more patients undergoing ventricular drainage and a higher proportion with delayed ischaemic neurological deficits.
Clinical data comparisons among included and excluded patients (who consented to data collection during acute admission)
Prevalence of cognitive impairment
A summary of the results of the assessments is shown in table 2. MMSE scores were <27 in 40% (36/90) of patients. MoCA scores were <26 in 73% (66/90) of patients. Of the patients with impaired MoCA scores (<26), 30 (45%) patients had normal MMSE scores (≥27) and 36 (55%) had impaired MMSE scores (<27), whereas all 24 (27%) patients with normal MoCA scores (≥26) had normal MMSE scores (≥27) (Fisher's exact test, p<0.001).
Summary of assessment results
Risk factors
MMSE (B coefficient −0.165, 95% CI −0.272 to −0.059), MoCA (B coefficient −0.269; 95% CI −0.390 to −0.148) and mRS (B coefficient 0.040; 95% CI 0.013 to 0.066) scores were independently associated with age. MMSE (B coefficient −3.832; 95% CI −7.305 to −0.358), MoCA (B coefficient −4.189; 95% CI −8.296 to −0.083) and mRS (B coefficient 0.882; 95% CI 0.081 to 1.684) scores were independently associated with cerebral infarction due to DCI and mRS scores were independently associated with WFNS grade on admission (B coefficient 0.490; 95% CI 0.281 to 0.699).
The R2 values for cerebral infarction due to DCI were 0.314, 0.384 and 0.457 for the MMSE, MoCA and mRS, respectively, indicating that the percentages of the variance were explained by the abovementioned factors.
Functional outcome at 3 months
The 3 month outcomes were assessed in 80% of recruited patients. At 3 months, the mRS was 0–2 in 57% (median (IQR) 2 (2–4.5)) and IADL scores were ≥15 in 70% (median (IQR) 27 (24–29)) of patients. Median GDS score was 7 (IQR 4–12). The mean SF-36 physical health component and mental health component scores were 42±14 and 51±13, respectively.
Sensitivity, specificity and predictive value
With regard to favourable neurological outcome at 3 months (mRS 0–2), the AUC was 0.910 (95% CI 0.985 to 0.835; p<0.001) for the MoCA and 0.873 (95% CI 0.964 to 0.8782; p<0.001) for the MMSE (figure 2A). There were no statistically significant differences between the two ROC curves (z=1.303, p=0.193). At the optimal cut-off of MoCA scores ≥20, sensitivity was 0.857, specificity was 0.778, PPV was 0.938 (95% CI 0.852 to 0.976), NPV was 0.731 (95% CI 0.539 to 0.863) and diagnostic accuracy was 0.879. At the optimal cut-off of MMSE scores ≥25, sensitivity was 0.857, specificity was 0.778, PPV was 0.922 (95% CI 0.830 to 0.966), NPV was 0.679 (95% CI 0.493 to 0.821) and diagnostic accuracy was 0.848.
(A) Receiver operating characteristic (ROC) curves of modified Rankin Scale score of 0–2 with Montreal Cognitive Assessment and Mini-Mental State Examination. (B) ROC curves of Lawton Instrumental Activity of Daily Living ≥15 with Montreal Cognitive Assessment and Mini-Mental State Examination.
For high IADL scores (≥15), the AUC was 0.832 (95% CI 0.944 to 0.719; p<0.001) for the MoCA and 0.806 (95% CI 0.926 to 0.686; p<0.001) for the MMSE (figure 2B). There were no statistically significant differences between the two ROC curves (z=1.031, p=0.303). At the optimal cut-off of the MoCA scores ≥20, sensitivity was 0.866, specificity was 0.696, PPV was 0.892 (95% CI 0.794 to 0.947), NPV was 0.654 (95% CI 0.462 to 0.806) and diagnostic accuracy was 0.824. At the optimal cut-off of the MMSE scores ≥26, sensitivity was 0.821, specificity was 0.739, PPV was 0.929 (95% CI 0.843 to 0.969), NPV was 0.613 (95% CI 0.438 to 0.763) and diagnostic accuracy was 0.832.
Correlations between the 3 month outcomes
mRS scores significantly correlated with both the MoCA scores (Kendall's tau b coefficient −0.413; p<0.001) and the MMSE scores (Kendall's tau b coefficient, −0.362; p<0.001).
IADL scores significantly correlated with both MoCA scores (Kendall's tau b coefficient 0.493; p<0.001) and MMSE scores (Kendall's tau b coefficient 0.485; p<0.001). MoCA and MMSE scores were significantly intercorrelated (Kendall's tau b coefficient 0.657; p<0.001). GDS were significantly correlated with MoCA (Kendall's tau b coefficient −0.191; p=0.027) and MMSE scores (Kendall's tau b coefficient −0.198; p=0.024) but MMSE and MoCA were not correlated with SF-36 physical health and mental health component scores.
Discussion
Cognitive impairment is common at the 3 month assessment
Seventy-three per cent of survivors demonstrated MoCA defined cognitive impairment at 3 months. In patients with aSAH, MoCA defined cognitive impairment was more prevalent than MMSE defined cognitive impairment. Cognitive impairment after aSAH is due to a combination of focal and diffuse brain injury, occurring during the initial haemorrhage or subsequently. The latter could be due to delayed cerebral infarction, raised intracranial pressure from associated haematoma, hydrocephalus, toxic effects of blood products, procedure related brain injuries and infection. These factors are especially relevant in our cohort which included poor grade aSAH. Our reported high rate of cognitive dysfunction is in accordance with previous reports.2 ,7 These data are important and can assist future study design and sample size calculation.
MMSE and MoCA demonstrated good discrimination of favourable neurological and IADL outcomes at 3 months
Our study suggests that the MoCA demonstrated good discrimination of favourable neurological and IADL outcomes similar to the MMSE in ROC curve analyses. Cognitive impairment, as assessed by the MoCA, has been found to be related to activities of daily life in patients with surgically evacuated spontaneous and traumatic intracerebral haematoma13 although the cut-off MoCA scores (<20) for functional outcome were lower than our proposed cut-off MoCA scores (<26) to screen for cognitive impairment.
In the original MoCA study, 75% of patients with mild cognitive impairment on neuropsychological testing had a normal MMSE but abnormal MoCA.7 In the Oxford Vascular Study of transient ischaemic attack and stroke, MoCA detected deficits in multiple domains that are not shown by MMSE, including executive function and attention (not tested by the MMSE) and recall and repetition (MMSE items less demanding).32 However, Godefroy et al highlighted that sensitivity and specificity are cut-off sensitive, and cautioned that the use of published cut-offs could be problematic and questioned the advantage of the MoCA over the MMSE in acute stroke patients.9 In acute stroke patients, the MoCA had high sensitivity and low specificity, which might be due to the high frequency of instrumental deficits, such as language and visuoconstructive domains.9 The MMSE is relatively predictive of these instrumental deficits.
Delayed cerebral infarction is an independent strong predictor for poor MoCA and MMSE performance
In multiple regression analyses, cerebral infarction due to DCI explained 31–38% of the variance in cognitive outcomes (MMSE and MoCA). Published case series have confirmed that DCI (clinical deterioration and/or cerebral infarction due to DCI) is associated with cognitive impairment and poor neurological outcome at 3 months and 1 year.33 ,34 A recent consensus paper11 suggests changing the surrogate marker for aSAH trials from DCI to cerebral infarction due to DCI and our finding that cerebral infarction due to DCI independently explains 48% of the variance in neurological outcome (mRS) and predicts MoCA and MMSE outcomes supports this.
Cognitive outcomes are not predictive of general HRQOL outcome at 3 months
Neither MoCA nor MMSE scores were correlated with SF-36 physical and mental component scores. Our findings are in accordance with previous studies using generic HRQOL assessments.30 ,35 Stroke specific HRQOL assessments, such as the Stroke Impact Scale, include parameters such as memory and communication domains in the scale construct and have been validated.36 We would recommend that future neurovascular studies use stroke specific questionnaires for a better evaluation of the disease burden.
Limitations of the current study
There are limitations to our study. First, there was no gold standard assessment of cognitive function in the current study and the cut-off MoCA scores for cognitive impairment were defined from previous validation studies in non-SAH patients.5 ,14 Godefroy et al commented that high sensitivity of MoCA cut-off scores were associated with low specificity in acute stroke patients.9 We thus were unable to assess whether the MoCA overestimated the prevalence of cognitive impairment in the current study; second, published stroke studies of the MoCA focused on patient populations with less neurological deficits, where the MoCA might perform differently and not have the same correlations with the functional measures; third, test–retest and inter-rater reliability was not assessed in the study although it has been reported previously in an ischaemic stroke population12; fourth, only 90 of 158 patients could be assessed cognitively in the current study—it is probable that patients with severe cognitive impairment had associated inability to complete the cognitive assessments and mortality, and hence the cognitive impairment rate was underestimated; fifth, the diagnoses of cerebral infarction due to DCI were made by site investigators and in future studies, blinded radiologists may provide more accurate assessment of cerebral infarction due to DCI; and finally, the MoCA and MMSE were administered among a battery of assessments, and fatigue may have increased the likelihood of error, although we offered a 5−10 min rest in the middle of the assessment sessions.
Conclusions
We were able to show that 73% of patients with aSAH had MoCA defined cognitive impairment, and that cerebral infarction due to DCI was independently associated with poor functional and cognitive outcomes at 3 months. The MoCA and MMSE showed good accuracy in differentiating functional outcomes (mRS and IADL) in patients at 3 months.
References
Footnotes
Cognitive Dysfunction after Aneurysmal Subarachnoid Haemorrhage Site Investigators: (1) Prince of Wales Hospital: George Wong, Wai Sang Poon, Vincent Mok. (2) Kwong Wah Hospital: John Kwok, Kwong Yau Chan, Peter Woo, Calvin Mak, Peter Pang. (3) Princess Margaret Hospital: Yin Chung Po, Tony Chan, Wai Kei Wong, Simon Lee. (4) Pamela Youde Nethersole Eastern Hospital: Chi Keung Wong, Michael Lee, Rebecca Ng, Alain Wong, Vincent Pang.
Funding This study was supported by the Neurosurgery Research and Training Fund, the Chinese University of Hong Kong.
Competing interests None.
Ethics approval Ethics approval was provided by the Joint NTEC-CUHK Clinical Research Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.