Relationship between Poststroke Cognition, Baseline Factors, and Functional Outcome: Data from ‘‘Efficacy of Nitric Oxide in Stroke’’ Trial Sandeep Ankolekar, MRCP,*† Cheryl Renton, MSc,* Gillian Sare, MRCP, PhD,*† Sharon Ellender, RGN,* Nikola Sprigg, MD, MRCP,*‡ Joanna M. Wardlaw, FRCR, FMedSci,x and Philip M. W. Bath, FRCPath, FRCP,*‡ for the ENOS Trial Investigators
Background: Poststroke cognitive impairment is common and identification of prognostic factors associated with it and its relationship with other functional outcomes may help in developing preventative strategies. Methods: Previously independent patients with acute stroke, enrolled into the ongoing ‘‘Efficacy of Nitric Oxide in Stroke’’ trial, were assessed by telephone on day 90 for cognitive impairment using modified versions of ‘‘Mini Mental State Examination’’ (MMSE-M) and ‘‘Telephone Instrument for Cognitive Status’’ (TICS-M) scales and category fluency. The relationship of cognitive impairment with baseline prognostic factors and other functional outcomes at day 90 were studied. Results: The analysis included 1572 patients, mean age 69 years (standard deviation, 12), and female 40%. By 90 days, 246 patients had died, and cognitive impairment was present in 38%. Increasing age, stroke severity, heart rate, and presence of cerebral atrophy on baseline neuroimaging were associated with cognitive impairment (all P , .001). Hypertension and atrial fibrillation were also associated with category fluency and MMSE-M, respectively. Cognition was significantly related to other functional outcomes, TICS-M with dependency (modified Rankin Scale, rs 5 2.562, P , .001); disability (Barthel Index, rs 5 .577, P , .001); mood (Zung Depression Score, rs 5 2.542, P , .001); and quality of life (Euro Quality of life-5 Descriptor, rs 5 .519, P , .001). Conclusions: In previously independent individuals, cognitive impairment was common 3 months after stroke and related to increasing age, stroke severity, hypertension, atrial fibrillation, and cerebral atrophy on brain scanning. Cognition was related to dependency, disability, low mood, and quality of life. Hence, treatment directed toward reducing dependency might also reduce cognitive impairment. Key Words: Poststroke cognitive impairment—stroke—functional outcome—dementia— ENOS—randomized controlled trial. Ó 2014 by National Stroke Association
From the *Stroke Trials Unit, Division of Clinical Neurosciences, University of Nottingham, Nottingham; †Leeds General Infirmary, Leeds; ‡Nottingham University Hospitals NHS Trust, Nottingham; and xDivision of Clinical Neurosciences, University of Edinburgh, Edinburgh, United Kingdom. Received November 2, 2013; revision received March 12, 2014; accepted April 14, 2014. ENOS is funded by the Medical Research Council (grant G0501797) and was previously funded by BUPA Foundation and The Hypertension Trust (501100000355). P.M.W.B. is The Stroke Association Profes-
sor of Stroke Medicine. S.A. and C.R. are funded by the Medical Research Council. The funding body for ENOS and the Sponsor did not have any input into the study. Address correspondence to Philip M.W. Bath, FRCPath, FRCP, Division of Clinical Neurosciences, University of Nottingham, City Hospital campus, Hucknall Road, Nottingham NG5 1PB, United Kingdom. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.04.022
Journal of Stroke and Cerebrovascular Diseases, Vol. 23, No. 7 (August), 2014: pp 1821-1829
1821
S. ANKOLEKAR ET AL.
1822
Poststroke dementia is common and can occur in up to 40% of patients over the subsequent 5 years following a stroke.1 When cognitive impairment is diagnosed in the first few months after stroke, it may progress to dementia, remain stable, or improve over the following months to years.2,3 Hence, recognizing risk factors in the acute phase of stroke that predict cognitive impairment might help to develop preventative strategies. Similarly, identifying the association between cognitive impairment and other functional outcomes after stroke, such as disability, dependence, depression, and reduced quality of life, may further help in improving outcome, for example, treating preventable or reversible conditions such as dependency and depression might improve cognitive function.4,5 However, poststroke cognition is rarely studied in acute stroke trials6; only 3 of the 190 trials published in English language literature between 1976-2003 had specific cognitive outcome measures.7 Although several cross-sectional and observational studies have assessed the relationship of poststroke cognition with other functional outcomes and baseline factors, these are limited by the small number of patients and, hence, the lack of external validity beyond the populations studied. We report here the relationship between cognitive impairment and baseline factors and functional outcome in 1572 patients enrolled in the ongoing international ‘‘Efficacy of Nitric Oxide in Stroke’’ (ENOS) trial with data from 18 countries across 5 continents.8
Methods ENOS Trial ENOS is an ongoing, international, multicentre, prospective, randomized, single-blind, controlled trial investigating the management of blood pressure in acute stroke.8 Briefly, patients maybe enrolled if they were previously independent, have an acute ischemic or hemorrhagic stroke, are within 48 hours of symptom onset, have high systolic blood pressure (140-220 mm Hg), and have limb weakness. As patients presenting with acute stroke were included in the study, assessment of prestroke cognitive impairment was not possible. The requirement that patients should be independent before their index stroke means that none should have had significant cognitive impairment before their index stroke. Patients are randomized to transdermal glyceryl trinitrate or control (with patients blinded to treatment), and those taking blood pressure lowering drugs immediately before their stroke are also randomized to continue or stop these; interventions are given for 7 days. The trial adheres to Good Clinical Practice, is approved by the UK Medicines and Healthcare Regulatory Agency and national Research Ethics Committee, and by local research ethics committees at each participating site, and is registered as ISRCTN99414122. Functional assessments, including cognition, are assessed in each country centrally over the telephone at 90 days, blinded to treatment assign-
ment. Basic demographic data were checked over the phone to ensure validity of the patient. The patients reported here were enrolled from 18 countries from 7 geographic regions (Africa, America, South Asia, Southeast Asia, Australasia, Europe, and United Kingdom).
Baseline Assessments Information on baseline factors prognostic for functional outcome is collected before randomization.
Brain Imaging Local investigators were asked to answer the following questions with information based on their (neuro-) radiology departmental report: (1) Is there evidence of mass effect? (2) Is there evidence of cerebral atrophy? (3) Is there evidence of leukoaraiosis or periventricular lucency? and (4) Is there evidence of any previous strokes? The location and extent of the imaging abnormalities were not collected from the local radiology report.
Cognitive Assessments Three cognitive assessments were, ‘‘Mini Mental State Examination-Modified’’ (MMSE-M), ‘‘Telephone Instrument for Cognitive Status-Modified’’ (TICS-M), and category fluency (animal naming), each being assessed by a trained observer over the telephone in each patients’ national language 90 days after randomization (Table S1 in Appendix). MMSE-M and TICS-M were chosen as they are global measures of cognition and the category fluency because it assesses executive function, which is more likely to be affected in vascular cognitive impairment. MMSE-M was collected during the period October 2001-January 2008 and July 2009-July 2011 (a database error meant that data were not collected in between), and TICS-M and category fluency (animal naming) during January 2008-July 2011. MMSE-M is an abbreviated form of the standard validated MMSE suitable for telephone use (Table S1 in Appendix). Patients with an MMSE-M score ,14 were considered a priori to have cognitive impairment (equivalent to ,24 on the complete MMSE).9 TICS-M is a global cognition measure that has been validated for use in poststroke patients.10 Patients with a TICS-M score ,20 were considered a priori to have cognitive impairment (equivalent to ,21 on the TICS-M score with 39 points).11 Animal naming is a test of category fluency and involves patients recalling as many animals as possible in 1 minute. Performance was computed as the number of appropriate items without repetitions. Patients naming ,10 animals were considered a priori to have cognitive impairment.12 (Animal naming was chosen instead of the verbal fluency assessment to make it language-script independent.) Patients who died before 90 days were included in the analyses for 3 reasons: (1) inclusion of death is standard in other functional assessments in stroke, for example,
ENOS POST STROKE COGNITION
modified Rankin Scale (mRS, death 5 6), Barthel Index (BI, death 5 25), and EuroQOL-5 Descriptor (EQ-5D, death 5 0); (2) inclusion of people who die prevents bias from exclusion of patients with severe stroke (who are at the highest risk of death); and (3) death is the final outcome in severe dementia. Patients who died were assigned a score of 21 for each cognition measure. Analyses were also performed without death included.
Functional Outcome Assessments Dependency (mRS),13 disability (BI),13 mood (Zung Depression Scale [ZDS]14), and quality of life (EuroQOL- 5 Descriptors [EQ-5D] and EuroQOL-Visual Analog Scale [EQ-VAS]15) were assessed by telephone at day 90 by a trained central assessor in each country blinded to treatment. mRS assesses dependency and ranges from 0 (no symptoms) to 5 (completely dependent and bed ridden). BI measures disability and ranges from 0 (completely disabled and dependent) to 100 (no disability). ZDScale is a measure of mood and ranges from 0 (normal mood) to 100 (severe depression). EuroQOL is a standardized instrument assessing quality of life and includes 5 dimensions (EQ-5D; mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) along with a summary visual analog scale (EQ-VAS). The EQ-5D descriptors also allow the derivation of a single index value. Death was included in each scale: mRS 5 6, BI 5 25, ZDS 5 25, EQ-5D 5 0, and EQ VAS 5 21.
Statistical Methods Comparisons between groups were assessed using the t test (continuous data), Mann–Whitney U test (ordinal data), and c2 test (binary data). Spearman rank correlation coefficient (rs, for continuous or ordinal variables) or rank biserial coefficient (rrb, Somer D, for binary variables) were used to assess univariate associations between measures of cognition (MMSE-M, TICS-M, and animal naming) and baseline factors (age, sex, hypertension, hyperlipidemia, diabetes, atrial fibrillation, previous stroke, heart rate, systolic blood pressure, temperature, stroke severity, ischemic or hemorrhagic stroke type, side of stroke, and baseline imaging findings). Multiple variable analyses were performed using stepwise linear regression with P ..10 as the criteria for exclusion of covariates. The relationship and agreement between MMSE-M, TICS-M, and animal naming were assessed using Spearman rank correlation and kappa statistic. Because the analyses are exploratory and require confirmation, no adjustment for multiple comparisons is made. Statistical analysis was performed using the Statistical Package for Social Sciences, version 19 (SPSS-19) for Mac (IBM Corporation). Significance was set at P ,.05 and 95% confidence intervals are shown.
Results Between July 2001 and July 2011, 2450 patients were enrolled into ENOS from 18 countries. Of 2141 patients
1823
who had completed day 90 follow-up, cognition data were available for 1572 patients, including 246 who died. A total of 482 patients did not have any cognition data whereas 87 patients had partial data on the TICS-M and MMSE-M. Missing data was because of: (1) inability to answer some or all of the questions over the telephone, for example, due to dysphasia, fatigue, or illness and (2) readmission to hospital making direct telephone follow-up of the participant difficult logistically. The characteristics of patients alive at day 90 with and without cognition data are shown in Table 1; patients with missing TICS-M or MMSE-M were more likely to be older, female, hypertensive, have a history of atrial fibrillation (AF) or previous stroke, or have a severe stroke or dysphasia (all P , .001). Such patients were also more likely to be dependent, disabled, and have a poorer quality of life at day 90 (P , .01, data not shown). Patients with incomplete TICS-M or MMSE-M data (n 5 85 of 569) performed poorly on the animal naming test as compared with those with complete data (median [interquartile range], 1 [5] vs. 10 [8]; Mann–Whitney U test, P , .001). There was no difference in baseline characteristics between patients with partial or completely missing TICSM and MMSE-M data (P . .10, data not shown).
Cognition and Other Functional Measures at Day 90 Figure 1 shows the distribution of scores at 90 days for the 3 cognitive scales, and Table 2 gives the summary statistics for the same scales. At 90 days just under 50% of patients were classified as cognitively impaired or dead; cognitive impairment was present in just under 40% of ENOS enrollees when excluding those who had died by day 90. Cognitive scores did not differ between patients with ischemic stroke and primary intracerebral hemorrhage (Mann–Whitney U test: MMSE-M, 14 [7, 17] vs. 13.5 [6.3, 17], P 5 .56; TICSM, 20 [10, 24] vs. 20 [6, 24], P 5 .93; and category fluency, 9 [3, 14] vs. 8 [.3, 12], P 5 .10. Functional outcome measures at day 90 are given in Table 2.
Predictors of Cognition at Day 90 The relationships between baseline prognostic factors and subsequent cognitive impairment were assessed in univariate analyses, these including patients who died (Table 3). Female sex, increasing age, heart rate and stroke severity (Scandinavian Stroke Scale), left-sided strokes, total anterior circulation syndrome (TACS) and a history of AF, previous stroke, or hypertension were each significantly associated with cognitive impairment at day 90. Presentation with a lacunar syndrome and history of hypercholesterolemia were more likely to be associated with normal or very mild cognitive impairment. On multiple variable analysis, increased age, heart rate and stroke severity, and TACS remained significant for all
S. ANKOLEKAR ET AL.
1824
Table 1. Baseline characteristics of participants enrolled into the ENOS trial with outcome assessed at day 90. Number (%), median (interquartile range), or mean (standard deviation); comparison by chi-square test, Mann–Whitney U test, or t test. Participants with cognition data include 246 patients who died Variables
Cognition data present
Cognition data absent
2p
Number Age, y, mean (SD) Sex, female, n (%) Recruited from, n (%) Africa America, North Asia, South Asia, Southeast Australasia Europe United Kingdom Risk factors, n (%) High blood pressure Hyperlipidemia Diabetes mellitus IHD Previous stroke/TIA Atrial fibrillation Premorbid mRS, median (IQR) Blood pressure, mean (SD) Systolic Diastolic Heart rate, bpm, mean (SD) Scandinavian Stroke Scale (SSS), mean (SD) Dysphasia, n (%) Stroke type, n (%) Ischemic stroke Primary intracerebral hemorrhage Stroke syndrome, n (%) (16) TACS PACS LACS POCS Left hemisphere stroke, n (%) Brain scan, N 5 1564, n (%) Atrophy Leukoaraiosis Mass effect from stroke Previous stroke(s) Temperature, oC Blood glucose, mmol/L
1572 69.2 (12.3) 631 (40.1)
569 72.6 (11.1) 264 (46.4)
,.001 .010 ,.001
14 (.9) 30 (1.9) 157 (10.0) 182 (11.6) 49 (3.1) 167 (10.6) 973 (61.9)
1 (.2) 3 (.5) 3 (.5) 104 (18.3) 6 (1.1) 65 (11.4) 387 (68.0)
969 (61.6) 423 (26.9) 265 (16.9) 252 (16.0) 339 (21.6) 163 (10.4) 0 (0, 0)
405 (71.2) 147 (25.8) 92 (16.2) 93 (16.3) 142 (25.0) 94 (16.5) 0 (0, 0)
,.001 .62 .71 .86 .097 ,.001 .55
167.7 (19.2) 90.2 (13.4) 76.9 (14.9) 35.7 (12.8)
169.7 (18.8) 90.1 (12.7) 76.7 (15.3) 30.3 (13.2)
.031 .92 .84 ,.001
488 (31.0)
285 (50.5)
,.001 .92
1304 (83.0) 268 (17.0)
471 (82.8) 98 (17.2)
386 (24.6) 446 (28.4) 671 (42.7) 69 (4.4) 873 (55.5)
214 (37.6) 183 (32.2) 165 (29.0) 7 (1.2) 253 (44.5)
,.001
363 (23.2) 313 (20.0) 208 (13.2) 392 (25.1) 36.6 (.57) 7.3 (2.9)
174 (30.6) 129 (22.7) 75 (13.2) 138 (24.3) 36.6 (.52) 7.2 (2.7)
,.001 .18 .94 .70 .36 .073
,.001
Abbreviations: bpm, beats per minute; IHD, ischemic heart disease; IQR, interquartile range; LACS, lacunar syndrome; mRS, modified Rankin Scale; PACS, partial anterior circulation syndrome; POCS, posterior circulation syndrome; SD, standard deviation; TACS, total anterior circulation syndrome; TIA, transient ischemic attack.
3 cognition scales. The adjusted R square values for stepwise linear regression models were MMSE-M, r2 5 .351; TICS-M, r2 5 .349; and category fluency, r2 5 .230. The results of brain scanning, as reported by each investigators’ (neuro-) radiology department were available for 2132 patients (Computerized Tomography (CT)
scan 5 1992, Magnetic Resonance Imaging (MRI) scan 5 140). The presence of cerebral atrophy was significantly related to cognitive impairment as assessed by the 3 cognition scales (Table 3). Other imaging factors—mass effect from the qualifying stroke lesion, previous stroke, and leukoaraiosis—were associated in univariate but not in
ENOS POST STROKE COGNITION
1825
Figure 1. Distribution of scores for the modified Mini Mental State Examination, modified Telephone Instrument for Cognitive Status and category fluency (animal naming)
multiple variable analysis with 1 or more of the cognition scales. The results of analyses did not differ qualitatively after excluding patients who died; however, adjusted R square was reduced for each of MMSE-M (.35-.13), TICS-M (.35-.16), and category fluency (.23-.08; Table S2 in Appendix).
Relation between Cognition and Other Functional Outcomes Measures of cognitive impairment were each related to dependency (mRS), disability (BI), depression (ZDS), and reduced quality of life (EQ-5D and EQ-VAS; Table 3). There was a strong agreement between the 3 cognition scales; MMSE-M and TICS-M (rs 5 .83, P , .001), MMSEM and category fluency (rs 5 .72, P , .001), and TICS-M and category fluency (rs 5 .74, P , .001). Moderate agreement was noted between the 3 cognition scales in identifying cognitive impairment using predetermined cut-off scores; MMSE-M and TICS-M (k 5 .61, P , .001), MMSEM and category fluency (k 5 .50, P , .001), and TICS-M and category fluency (k 5 .48, P , .001). The agreement between the cognition scales was less robust when death was excluded from the analysis (Table S2 in Appendix).
Discussion The aims of the present study were first, to identify factors measured during the acute phase of stroke that are related to subsequent cognitive impairment and second, to assess the relationship between cognition at 90 days poststroke with other measures of functional outcome.
The strengths of the study are that it is one of the largest studies of cognition after acute stroke to date, includes multiple race ethnicity groups from 18 countries across 5 continents, uses unified methodology, involves patients largely representative of hospitalized stroke patients, is based on prospectively collected data, and uses data collected during the current and last decade. Poststroke cognitive impairment (PSCI) was related with increasing age, stroke severity, heart rate, TACS, and the presence of atrophy on brain scanning, in both unadjusted and adjusted analyses. Several studies (Table 4), including a systematic review,1 have reported associations between subacute PSCI and age,16-21 and stroke severity,16,17,20,22 and the present study confirms these findings. This suggests that the severity of the index stroke in an aging brain maybe sufficient to cause PSCI in the short-term. The relationship of cognition with AF, left hemisphere stroke, and hypertension was less consistent across the 3 cognition measures. Several studies, albeit smaller, have reported associations of subacute cognition with AF17-19 and left-sided strokes19,20 but none with hypertension. Others studies have shown an association with female sex,18,23,24 previous stroke,20,22,25 diabetes,16,22 and alcohol intake19; the first 3 were not found in this study and the latter factor is not measured in ENOS. Although high blood pressure (BP) during the acute phase of stroke is independently associated with poor functional outcome,26,27 an independent association between baseline BP and subsequent cognitive impairment was not seen in this study. The combined effect of these risk factors may become more important with time, as suggested by their association with the long-term
S. ANKOLEKAR ET AL.
1826
Table 2. Cognition and other functional outcome measures at day 90 using central telephone assessment in 1572 patients with cognition data, including 246 who died. Number (%) or median (interquartile range) Functional outcome Died MMSE-M, N 5 1218, died 192 Median (IQR) Dead or cognitive impairment, MMSE-M ,14, n (%) TICS-M, N 5 1041, died 178 Median (IQR) Dead or cognitive impairment, TICS-M ,20, n (%) Category fluency, N 5 1044, died 178 Median (IQR) Dead or cognitive impairment, Fluency ,10, n (%) mRS, N 5 1572, died 246 Median (IQR) Dead or dependent, mRS .2, n (%) BI, N 5 1572, died 246 Median (IQR) Dead or disabled, BI #90, n (%) ZDS, N 5 1526, died 246 Median (IQR) Dead or depressed, ZDS $70, n (%) Depressed at 90 days, n (%) EuroQOL, N 5 1571 for EQ-5D and 1522 for VAS EQ-5D Index, median (IQR) EQ-VAS, median (IQR)
Number 246 14.0 (7, 17) 582 (47.8)
21 (10, 24) 546 (48.4)
9 (3, 14) 546 (52.3)
3 (1, 4) 874 (55.6)
90 (46.3, 100) 902 (57.4) 52.5 (37.5, 72.5) 422 (27.7) 176 (13.8)
.64 (.14, .81) 65 (40, 80)
Abbreviations: BI, Barthel Index; EQ-5D, EuroQOL-5 Descriptor; EQ-VAS, EuroQOL-Visual Analog Scale; IQR, interquartile range; MMSE-M, Mini Mental State Examination; mRS, Modified Rankin Scale; TICS-M, Telephone Instrument for Cognitive Status-Modified; ZDS, Zung Depression Scale.
development of dementia. Interestingly, hyperlipidemia was inversely associated with cognitive impairment. This apparent paradox is difficult to explain but earlier observational studies have reported that hyperlipidemia is a risk factor for cognitive impairment in younger patients28,29 but appears to be protective in older individuals.30 Silent cerebrovascular disease detected by neuroimaging such as leukoaraiosis and silent infarcts are 10-20 times more prevalent than manifest strokes.31 Such lesions are independently associated with cognitive impairment in the elderly and may lower the threshold for dementia cumulatively, as seen with Alzheimer disease.32,33 Although, several studies have found an association between PSCI and silent infarcts,34 cerebral atrophy35 and white matter lesions,24,35 only cerebral atrophy remained significant on multiple variable analysis in the present study.
When assessing the inter-relationships between the 4 measures of functional outcome at 3 months, PSCI was associated with each of functional dependency (mRS), disability (BI), depression (ZDS), and reduced quality of life (EQ-5D, EQ-VAS). This is consistent with findings from previous but smaller studies.18,20,36-38 The strong inter-relationships between these measures suggest that interventions that attenuate dependency, the usual primary outcome in acute stroke trials, may also reduce subsequent cognitive decline. The present study has several limitations. First, the data come from a randomized controlled trial with multiple exclusion criteria that will have limited the range of included patients. Of relevance, ENOS excludes patients without motor impairment, and those who are normotensive or have premorbid dependency. These constraints may have attenuated some relationships, for example, between cognition and stroke severity, and cognition and BP. Counteracting this limitation is that the size of the data is an order of magnitude larger than previous studies, as summarized in a recent review article.1 Second, 27% of our patients were unable to answer telephone-based cognition questions because of factors such as dysphasia, fatigue, or being in hospital. This is somewhat higher than 18% seen in another acute stroke trial assessing cognition although it included only 294 patients39 Missing data are a recognized problem in studies assessing cognition, especially in elderly patients, and is likely to be higher in larger studies.40 These patients had more severe strokes (and other differences as summarized in Table 1), again biasing the sample. However, this limitation maybe less relevant in this study because patients who died (who will have had the most severe strokes) were included. As a result, the study will have tended to include mildmoderate and very severe stroke, but missed those with severe stroke. Third, there was no evaluation of prestroke cognitive impairment and it is difficult to say if factors other than the index stroke contributed to the day 90 cognition. However, exclusion of patients with mRS .2 at baseline means it is unlikely that many, if any, patients assessed in this study had overt dementia at baseline. Fourth, we used the interpretation of brain scan provided at the local center, this introducing variation in interpretation. Local reporting will also have focused on more overt than subtle imaging lesions. In due course, central blinded adjudication will be available for the whole trial. Fifth, as the study is international, we had to choose questions that were language and culturally independent, for example, using animal naming for category fluency rather than verbal fluency using words beginning with F, A, or S (which would have no relevance in Chinese patients). Cultural differences may affect some responses to the study questions and bias relationships, and international variations are known to exist for mRS, MMSE, and EQ-5D.41-43 Nevertheless, including patients from Africa, South Asia, Southeast Asia, Australasia–Oceania, Europe, and
ENOS POST STROKE COGNITION
1827
Table 3. Univariate and multiple variable relationships between cognitive measures at day 90 and (i) baseline factors and (ii) other functional outcome measures at day 90. Shown are correlation coefficients (r) with P values from the univariate analyses (Spearman rank correlation coefficient (rs) for continuous variables and rank biserial coefficient (rrb, Somer D) for binary variables). Patients who died were included in the analysis. MMSE-M Variable Day 0 (baseline) Age Sex, female Atrial fibrillation Diabetes mellitus Hyperlipidemia Hypertension Previous stroke/TIA Systolic blood pressure Heart rate Temperature Stroke severity (SSS) Stroke syndrome (16) Lacunar vs. rest TACS vs. rest Stroke type (IS vs. PICH) Stroke side (left vs. right) Brain scan Atrophy Leukoaraiosis matter disease Mass effect from stroke Previous stroke(s) Serum glucose Day 90 Modified Rankin Scale Barthel Index Zung Depression Score EQ-5D Index EQ Visual Analog Scale
TICS-M
Category fluency
r
2p
r
2p
r
2p
2.343 2.140 2.366 2.010 .056 2.102 2.136 2.021 2.117 .021 .378
,.001y ,.001 ,.001y .808 .132 .002 .001 .468 ,.001y .477 ,.001y
2.366 2.066 2.297 2.033 .046 2.130 2.111 .005 2.125 .035 .377
,.001y .074 ,.001 .502 .262 ,.001 .015 .883 ,.001y .261 ,.001y
2.304 2.093 2.294 2.031 .093 2.131 2.118 2.020 2.109 2.050 .378
,.001y .009 ,.001 .536 .026y ,.001y .008 .516 ,.001y .106 ,.001y
.283 2.472 2.027 2.120
,.001 ,.001y .562 ,.001y
.236 2.393 .004 2.118
,.001 ,.001y .934 .001
.242 2.408 2.075 2.091
,.001 ,.001y .091 .010
2.258 2.109
,.001y .009
2.246 2.072
,.001y .117
2.209 2.073
,.001y .110
2.190 2.109 2.115
,.001 .008 .002
2.102 2.042 2.051
.105 .346 .101
2.211 2.050 .036
,.001 .250 .245
2.551 .571 2.532 .507 .532
,.001 ,.001 ,.001 ,.001 ,.001
2.562 .577 2.542 .519 .528
,.001 ,.001 ,.001 ,.001 ,.001
2.558 .590 2.589 .518 .560
,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: EQ, Euro Quality of life; EQ-5D, Euro Quality of life-5 Descriptor; IS, ischemic stroke; PICH, primary intracerebral hemorrhage; SSS, Scandinavian Stroke Scale; TACS, total anterior circulation syndrome; TIA, transient ischemic attack. yAssociations those remain significant on multiple variable analyses.
North America will have increased substantially the external validity of the findings. Sixth, telephone assessment of cognition will be affected by extraneous factors such as hearing loss in an older population and whether the participant or a caregiver delivered the responses. Finally, only basic measures of cognition were used in this study; MMSE-M is biased toward attention and orientation and suffers from a ceiling effect. However, TICS-M includes assessment of multiple cognitive domains covering memory, orientation, attention, and language. Importantly, as the prevalence of cognitive impairment found in this study is comparable with other similar studies, the limitations of these scales maybe more theoretic than real in this population. Previous studies have ignored death as a cognitive outcome but we included it because death is part of other
functional assessments used in stroke such as mRS, BI, and EQ-5D and is the final outcome in end stage dementia. In addition, there is the statistical advantage that patients who die are not discarded from analysis, important because this group comprises a significant minority of the cohort, and including them removes the bias of excluding patients who are older and/or have more severe stroke. The additional sample size also adds statistical power to analyses. In summary, PSCI is common at day 90 and is associated with increasing age, stroke severity, history of AF and hypertension, and cerebral atrophy on neuroimaging. The presence of these factors maybe useful clinically in identifying people at high risk of developing PSCI. Furthermore, PSCI is associated with other functional outcomes at day 90 including dependency, disability, poor mood, and
S. ANKOLEKAR ET AL.
1828
Table 4. Baseline factors in patients with acute stroke that are related to subsequent cognitive impairment (2 weeks to 6 months) in other studies. Baseline Predictor Baseline demographics Age Education Sex, female Atrial fibrillation Previous stroke Diabetes Hypertension Hyperlipidemia Stroke severity Left vs. right OCSP stroke presentation Heart rate Baseline blood pressure Temperature Glucose Baseline CT scan Cerebral Atrophy Mass effect from stroke Previous Stroke Leukoaraiosis
Univariate analysis
Multiple variable analysis
(18-26)y (20, 22, 24-28) (19, 24, 29)y (17-20)y (19-23, 25, 26)y (17, 20, 23, 26) -y -y (17-19, 21, 23, 27, 29)y (20, 21, 25)y (17, 22)y -y (28) (18)
(18-23, 26)y (20, 22, 25, 26) (18-20)y (20, 21, 23, 25, 26) (17, 23) -y -y (17, 18, 21, 23, 29)y (20, 21)y (17)y -y -
(22, 29, 30)y y y (27, 29, 30)y
(30)y -y (29, 30)
-: denotes no association in other studies; y: this study.
reduced quality of life; hence, beneficial treatment for one may have useful effects on the others. Acknowledgment: We thank patients and investigators who are taking part in ENOS, and the UK Stroke Research Network for facilitating recruitment. Trial Steering Committee: Pierre Amarenco (Independent expert, France), Philip Bath (Chief Investigator, UK), Kennedy Lees (UK), Keith Muir (Independent expert, UK); Stuart Pocock (Statistical consultant, UK), Angela Shone (Sponsor representative, UK), Jane Sinclair (Funder representative, UK), Graham Venables (Independent chair, UK), Joanna Wardlaw (Neuroradiology lead, UK), David Whynes (Health Economics lead, UK). International Advisory Committee: Chris Bladin (Australia), Valeria Caso (Italy), Hui Meng Chang (Singapore), Ronan Collins (Eire), Anna Czlonkowska (Poland), Anwar El Etribi (Egypt), Abdul Raman Ghani (Malaysia), John Gommans (New Zealand), Stephen Phillips (Canada), Kameshwar Prasad (India), Asita de Silva (Sri Lanka), Szabolcs Szatmari (Romania), Exuperio de Tejedor (Spain), Yong-Jun Wang (China), Lawrence Wong (Hong Kong) competing interests. Competing Interests: P.M.W.B. is Chief Investigator and J.W. is neuroimaging lead for ENOS.
Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10. 1016/j.jstrokecerebrovasdis.2014.04.022
References 1. Pendlebury ST, Rothwell PM. Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: a systematic review and meta-analysis. Lancet Neurol 2009;8:1006-1018. 2. Ballard C, Rowan E, Stephens S, et al. Prospective follow-up study between 3 and 15 months after stroke: improvements and decline in cognitive function among dementia-free stroke survivors .75 years of age. Stroke 2003;34:2440-2444. 3. Rasquin SM, Lodder J, Verhey FR. Predictors of reversible mild cognitive impairment after stroke: a 2-year followup study. J Neurol Sci 2005;229-230:21-25. 4. Jorge RE, Acion L, Moser D, et al. Escitalopram and enhancement of cognitive recovery following stroke. Arch Gen Psychiatry 2010;67:187-196. 5. Kimura M, Robinson RG, Kosier JT. Treatment of cognitive impairment after poststroke depression : a doubleblind treatment trial. Stroke 2000;31:1482-1486. 6. Ankolekar S, Geeganage C, Anderton P, et al. Clinical trials for preventing post stroke cognitive impairment. J Neurol Sci 2010;299:168-174. 7. Anderson CA, Arciniegas DB, Filley CM. Treatment of acute ischemic stroke: does it impact neuropsychiatric outcome? J Neuropsychiatry Clin Neurosci 2005;17:486-488. 8. The ENOS Trial Investigators. Glyceryl trinitrate vs. control, and continuing vs. stopping temporarily prior antihypertensive therapy, in acute stroke: rationale and design of the Efficacy of Nitric Oxide in Stroke (ENOS) trial (ISRCTN99414122). Int J Stroke 2006;1:245-249. 9. Folstein M, Folstein S, McHugh P. ‘‘Mini-mental state:’’ a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198. 10. Barber M, Stott DJ. Validity of the Telephone Interview for Cognitive Status (TICS) in post-stroke subjects. Int J Geriatr Psychiatry 2004;19:75-79.
ENOS POST STROKE COGNITION 11. de Jager CA, Budge MM, Clarke R. Utility of TICS-M for the assessment of cognitive function in older adults. Int J Geriatr Psychiatry 2003;18:318-324. 12. Tombaugh TN, Kozak J, Rees L. Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Arch Clin Neuropsychol 1999; 14:167-177. 13. Sulter G, Steen C, Jacques De K. Use of the Barthel Index and Modified Rankin Scale in acute stroke trials. Stroke 1999;30:1538-1541. 14. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry 1965;12:63-70. 15. Euroqol Group. EuroQol–a new facility for the measurement of health-related quality of life. Health Policy 1990; 16:199-208. 16. Censori B, Manara O, Agostinis C, et al. Dementia after first stroke. Stroke 1996;27:1205-1210. 17. Barba R, Martinez-Espinosa S, Rodriguez-Garcia E, et al. Poststroke dementia: clinical features and risk factors. Stroke 2000;31:1494-1501. 18. Tamam B, Tasdemir N, Tamam Y. [The prevalence of dementia three months after stroke and its risk factors]. Turk Psikiyatri Derg 2008;19:46-56. 19. Zhou DH, Wang JY, Li J, et al. Study on frequency and predictors of dementia after ischemic stroke: the Chongqing stroke study. J Neurol 2004;251:421-427. 20. Lin JH, Lin RT, Tai CT, et al. Prediction of poststroke dementia. Neurology 2003;61:343-348. 21. Rasquin SM, Verhey FR, van Oostenbrugge RJ, et al. Demographic and CT scan features related to cognitive impairment in the first year after stroke. J Neurol Neurosurg Psychiatry 2004;75:1562-1567. 22. Henon H, Durieu I, Guerouaou D, et al. Poststroke dementia: incidence and relationship to prestroke cognitive decline. Neurology 2001;57:1216-1222. 23. Madureira S, Guerreiro M, Ferro JM. Dementia and cognitive impairment three months after stroke. Eur J Neurol 2001;8:621-627. 24. Tang WK, Chan SS, Chiu HF, et al. Frequency and determinants of poststroke dementia in Chinese. Stroke 2004; 35:930-935. 25. Pohjasvaara T, Erkinjuntti T, Ylikoski R, et al. Clinical determinants of poststroke dementia. Stroke 1998;29: 75-81. 26. Leonardi-Bee J, Bath PMW, Phillips SJ, et al. Blood pressure and clinical outcomes in the International Stroke Trial. Stroke 2002;33:1315-1320. 27. Willmot M, Leonardi-Bee J, Bath PM. High blood pressure in acute stroke and subsequent outcome: a systematic review. Hypertension 2004;43:18-24. 28. Whitmer RA, Sidney S, Selby J, et al. Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 2005;64:277-281.
1829 29. Dufouil C, Richard F, Fievet N, et al. APOE genotype, cholesterol level, lipid-lowering treatment, and dementia: the Three-City Study. Neurology 2005;64:1531-1538. 30. Solfrizzi V, Panza F, Colacicco AM, et al. Vascular risk factors, incidence of MCI, and rates of progression to dementia. Neurology 2004;63:1882-1891. 31. Longstreth WT Jr. Brain abnormalities in the elderly: frequency and predictors in the United States (the Cardiovascular Health Study). Cardiovascular Health Study Collaborative Research Group. J Neural Transm Suppl 1998;53:9-16. 32. Snowdon DA, Greiner LH, Mortimer JA, et al. Brain infarction and the clinical expression of Alzheimer disease. The Nun Study. JAMA 1997;277:813-817. 33. Schneider JA, Wilson RS, Bienias JL, et al. Cerebral infarctions and the likelihood of dementia from Alzheimer disease pathology. Neurology 2004;62:1148-1155. 34. Mok V, Chang C, Wong A, et al. Neuroimaging determinants of cognitive performances in stroke associated with small vessel disease. J Neuroimaging 2005; 15:129-137. 35. Jokinen H, Kalska H, Mantyla R, et al. White matter hyperintensities as a predictor of neuropsychological deficits post-stroke. J Neurol Neurosurg Psychiatry 2005; 76:1229-1233. 36. Kauhanen M, Korpelainen JT, Hiltunen P, et al. Poststroke depression correlates with cognitive impairment and neurological deficits. Stroke 1999;30:1875-1880. 37. Tatemichi TK, Desmond DW, Stern Y, et al. Cognitive impairment after stroke: frequency, patterns, and relationship to functional abilities. J Neurol Neurosurg Psychiatry 1994;57:202-207. 38. Mok VCT, Lam WWM, Fan YH, et al. Cognitive impairment and functional outcome after stroke associated with small vessel disease. J Neurol Neurosurg Psychiatry 2004;75:560-566. 39. Cumming TB, Bernhardt J, Linden T. The Montreal cognitive assessment: short cognitive evaluation in a large stroke trial. Stroke 2011;42:2642-2644. 40. Hardy SE, Allore H, Studenski SA. Missing data: a special challenge in aging research. J Am Geriatr Soc 2009; 57:722-729. 41. Quinn TJ, Dawson J, Walters MW, et al. Variability in modified Rankin scoring across a large cohort of international observers. Stroke 2008;39:2975-2979. 42. Scazufca M, Almeida OP, Vallada HP, et al. Limitations of the Mini-Mental State Examination for screening dementia in a community with low socioeconomic status: results from the Sao Paulo Ageing & Health Study. Eur Arch Psychiatry Clin Neurosci 2009;259:8-15. 43. Sainsbury A, Seebass G, Bansal A, et al. Reliability of the Barthel Index when used with older people. Age Ageing 2005;34:228-232.
Background: Poststroke cognitive impairment is common and identification of prognostic factors associated with it and its relationship with other functional outcomes may help in developing preventative strategies. Methods: Previously independent patients with acute stroke, enrolled into the ongoing ‘‘Efficacy of Nitric Oxide in Stroke’’ trial, were assessed by telephone on day 90 for cognitive impairment using modified versions of ‘‘Mini Mental State Examination’’ (MMSE-M) and ‘‘Telephone Instrument for Cognitive Status’’ (TICS-M) scales and category fluency. The relationship of cognitive impairment with baseline prognostic factors and other functional outcomes at day 90 were studied. Results: The analysis included 1572 patients, mean age 69 years (standard deviation, 12), and female 40%. By 90 days, 246 patients had died, and cognitive impairment was present in 38%. Increasing age, stroke severity, heart rate, and presence of cerebral atrophy on baseline neuroimaging were associated with cognitive impairment (all P , .001). Hypertension and atrial fibrillation were also associated with category fluency and MMSE-M, respectively. Cognition was significantly related to other functional outcomes, TICS-M with dependency (modified Rankin Scale, rs 5 2.562, P , .001); disability (Barthel Index, rs 5 .577, P , .001); mood (Zung Depression Score, rs 5 2.542, P , .001); and quality of life (Euro Quality of life-5 Descriptor, rs 5 .519, P , .001). Conclusions: In previously independent individuals, cognitive impairment was common 3 months after stroke and related to increasing age, stroke severity, hypertension, atrial fibrillation, and cerebral atrophy on brain scanning. Cognition was related to dependency, disability, low mood, and quality of life. Hence, treatment directed toward reducing dependency might also reduce cognitive impairment. Key Words: Poststroke cognitive impairment—stroke—functional outcome—dementia— ENOS—randomized controlled trial. Ó 2014 by National Stroke Association
From the *Stroke Trials Unit, Division of Clinical Neurosciences, University of Nottingham, Nottingham; †Leeds General Infirmary, Leeds; ‡Nottingham University Hospitals NHS Trust, Nottingham; and xDivision of Clinical Neurosciences, University of Edinburgh, Edinburgh, United Kingdom. Received November 2, 2013; revision received March 12, 2014; accepted April 14, 2014. ENOS is funded by the Medical Research Council (grant G0501797) and was previously funded by BUPA Foundation and The Hypertension Trust (501100000355). P.M.W.B. is The Stroke Association Profes-
sor of Stroke Medicine. S.A. and C.R. are funded by the Medical Research Council. The funding body for ENOS and the Sponsor did not have any input into the study. Address correspondence to Philip M.W. Bath, FRCPath, FRCP, Division of Clinical Neurosciences, University of Nottingham, City Hospital campus, Hucknall Road, Nottingham NG5 1PB, United Kingdom. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.04.022
Journal of Stroke and Cerebrovascular Diseases, Vol. 23, No. 7 (August), 2014: pp 1821-1829
1821
S. ANKOLEKAR ET AL.
1822
Poststroke dementia is common and can occur in up to 40% of patients over the subsequent 5 years following a stroke.1 When cognitive impairment is diagnosed in the first few months after stroke, it may progress to dementia, remain stable, or improve over the following months to years.2,3 Hence, recognizing risk factors in the acute phase of stroke that predict cognitive impairment might help to develop preventative strategies. Similarly, identifying the association between cognitive impairment and other functional outcomes after stroke, such as disability, dependence, depression, and reduced quality of life, may further help in improving outcome, for example, treating preventable or reversible conditions such as dependency and depression might improve cognitive function.4,5 However, poststroke cognition is rarely studied in acute stroke trials6; only 3 of the 190 trials published in English language literature between 1976-2003 had specific cognitive outcome measures.7 Although several cross-sectional and observational studies have assessed the relationship of poststroke cognition with other functional outcomes and baseline factors, these are limited by the small number of patients and, hence, the lack of external validity beyond the populations studied. We report here the relationship between cognitive impairment and baseline factors and functional outcome in 1572 patients enrolled in the ongoing international ‘‘Efficacy of Nitric Oxide in Stroke’’ (ENOS) trial with data from 18 countries across 5 continents.8
Methods ENOS Trial ENOS is an ongoing, international, multicentre, prospective, randomized, single-blind, controlled trial investigating the management of blood pressure in acute stroke.8 Briefly, patients maybe enrolled if they were previously independent, have an acute ischemic or hemorrhagic stroke, are within 48 hours of symptom onset, have high systolic blood pressure (140-220 mm Hg), and have limb weakness. As patients presenting with acute stroke were included in the study, assessment of prestroke cognitive impairment was not possible. The requirement that patients should be independent before their index stroke means that none should have had significant cognitive impairment before their index stroke. Patients are randomized to transdermal glyceryl trinitrate or control (with patients blinded to treatment), and those taking blood pressure lowering drugs immediately before their stroke are also randomized to continue or stop these; interventions are given for 7 days. The trial adheres to Good Clinical Practice, is approved by the UK Medicines and Healthcare Regulatory Agency and national Research Ethics Committee, and by local research ethics committees at each participating site, and is registered as ISRCTN99414122. Functional assessments, including cognition, are assessed in each country centrally over the telephone at 90 days, blinded to treatment assign-
ment. Basic demographic data were checked over the phone to ensure validity of the patient. The patients reported here were enrolled from 18 countries from 7 geographic regions (Africa, America, South Asia, Southeast Asia, Australasia, Europe, and United Kingdom).
Baseline Assessments Information on baseline factors prognostic for functional outcome is collected before randomization.
Brain Imaging Local investigators were asked to answer the following questions with information based on their (neuro-) radiology departmental report: (1) Is there evidence of mass effect? (2) Is there evidence of cerebral atrophy? (3) Is there evidence of leukoaraiosis or periventricular lucency? and (4) Is there evidence of any previous strokes? The location and extent of the imaging abnormalities were not collected from the local radiology report.
Cognitive Assessments Three cognitive assessments were, ‘‘Mini Mental State Examination-Modified’’ (MMSE-M), ‘‘Telephone Instrument for Cognitive Status-Modified’’ (TICS-M), and category fluency (animal naming), each being assessed by a trained observer over the telephone in each patients’ national language 90 days after randomization (Table S1 in Appendix). MMSE-M and TICS-M were chosen as they are global measures of cognition and the category fluency because it assesses executive function, which is more likely to be affected in vascular cognitive impairment. MMSE-M was collected during the period October 2001-January 2008 and July 2009-July 2011 (a database error meant that data were not collected in between), and TICS-M and category fluency (animal naming) during January 2008-July 2011. MMSE-M is an abbreviated form of the standard validated MMSE suitable for telephone use (Table S1 in Appendix). Patients with an MMSE-M score ,14 were considered a priori to have cognitive impairment (equivalent to ,24 on the complete MMSE).9 TICS-M is a global cognition measure that has been validated for use in poststroke patients.10 Patients with a TICS-M score ,20 were considered a priori to have cognitive impairment (equivalent to ,21 on the TICS-M score with 39 points).11 Animal naming is a test of category fluency and involves patients recalling as many animals as possible in 1 minute. Performance was computed as the number of appropriate items without repetitions. Patients naming ,10 animals were considered a priori to have cognitive impairment.12 (Animal naming was chosen instead of the verbal fluency assessment to make it language-script independent.) Patients who died before 90 days were included in the analyses for 3 reasons: (1) inclusion of death is standard in other functional assessments in stroke, for example,
ENOS POST STROKE COGNITION
modified Rankin Scale (mRS, death 5 6), Barthel Index (BI, death 5 25), and EuroQOL-5 Descriptor (EQ-5D, death 5 0); (2) inclusion of people who die prevents bias from exclusion of patients with severe stroke (who are at the highest risk of death); and (3) death is the final outcome in severe dementia. Patients who died were assigned a score of 21 for each cognition measure. Analyses were also performed without death included.
Functional Outcome Assessments Dependency (mRS),13 disability (BI),13 mood (Zung Depression Scale [ZDS]14), and quality of life (EuroQOL- 5 Descriptors [EQ-5D] and EuroQOL-Visual Analog Scale [EQ-VAS]15) were assessed by telephone at day 90 by a trained central assessor in each country blinded to treatment. mRS assesses dependency and ranges from 0 (no symptoms) to 5 (completely dependent and bed ridden). BI measures disability and ranges from 0 (completely disabled and dependent) to 100 (no disability). ZDScale is a measure of mood and ranges from 0 (normal mood) to 100 (severe depression). EuroQOL is a standardized instrument assessing quality of life and includes 5 dimensions (EQ-5D; mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) along with a summary visual analog scale (EQ-VAS). The EQ-5D descriptors also allow the derivation of a single index value. Death was included in each scale: mRS 5 6, BI 5 25, ZDS 5 25, EQ-5D 5 0, and EQ VAS 5 21.
Statistical Methods Comparisons between groups were assessed using the t test (continuous data), Mann–Whitney U test (ordinal data), and c2 test (binary data). Spearman rank correlation coefficient (rs, for continuous or ordinal variables) or rank biserial coefficient (rrb, Somer D, for binary variables) were used to assess univariate associations between measures of cognition (MMSE-M, TICS-M, and animal naming) and baseline factors (age, sex, hypertension, hyperlipidemia, diabetes, atrial fibrillation, previous stroke, heart rate, systolic blood pressure, temperature, stroke severity, ischemic or hemorrhagic stroke type, side of stroke, and baseline imaging findings). Multiple variable analyses were performed using stepwise linear regression with P ..10 as the criteria for exclusion of covariates. The relationship and agreement between MMSE-M, TICS-M, and animal naming were assessed using Spearman rank correlation and kappa statistic. Because the analyses are exploratory and require confirmation, no adjustment for multiple comparisons is made. Statistical analysis was performed using the Statistical Package for Social Sciences, version 19 (SPSS-19) for Mac (IBM Corporation). Significance was set at P ,.05 and 95% confidence intervals are shown.
Results Between July 2001 and July 2011, 2450 patients were enrolled into ENOS from 18 countries. Of 2141 patients
1823
who had completed day 90 follow-up, cognition data were available for 1572 patients, including 246 who died. A total of 482 patients did not have any cognition data whereas 87 patients had partial data on the TICS-M and MMSE-M. Missing data was because of: (1) inability to answer some or all of the questions over the telephone, for example, due to dysphasia, fatigue, or illness and (2) readmission to hospital making direct telephone follow-up of the participant difficult logistically. The characteristics of patients alive at day 90 with and without cognition data are shown in Table 1; patients with missing TICS-M or MMSE-M were more likely to be older, female, hypertensive, have a history of atrial fibrillation (AF) or previous stroke, or have a severe stroke or dysphasia (all P , .001). Such patients were also more likely to be dependent, disabled, and have a poorer quality of life at day 90 (P , .01, data not shown). Patients with incomplete TICS-M or MMSE-M data (n 5 85 of 569) performed poorly on the animal naming test as compared with those with complete data (median [interquartile range], 1 [5] vs. 10 [8]; Mann–Whitney U test, P , .001). There was no difference in baseline characteristics between patients with partial or completely missing TICSM and MMSE-M data (P . .10, data not shown).
Cognition and Other Functional Measures at Day 90 Figure 1 shows the distribution of scores at 90 days for the 3 cognitive scales, and Table 2 gives the summary statistics for the same scales. At 90 days just under 50% of patients were classified as cognitively impaired or dead; cognitive impairment was present in just under 40% of ENOS enrollees when excluding those who had died by day 90. Cognitive scores did not differ between patients with ischemic stroke and primary intracerebral hemorrhage (Mann–Whitney U test: MMSE-M, 14 [7, 17] vs. 13.5 [6.3, 17], P 5 .56; TICSM, 20 [10, 24] vs. 20 [6, 24], P 5 .93; and category fluency, 9 [3, 14] vs. 8 [.3, 12], P 5 .10. Functional outcome measures at day 90 are given in Table 2.
Predictors of Cognition at Day 90 The relationships between baseline prognostic factors and subsequent cognitive impairment were assessed in univariate analyses, these including patients who died (Table 3). Female sex, increasing age, heart rate and stroke severity (Scandinavian Stroke Scale), left-sided strokes, total anterior circulation syndrome (TACS) and a history of AF, previous stroke, or hypertension were each significantly associated with cognitive impairment at day 90. Presentation with a lacunar syndrome and history of hypercholesterolemia were more likely to be associated with normal or very mild cognitive impairment. On multiple variable analysis, increased age, heart rate and stroke severity, and TACS remained significant for all
S. ANKOLEKAR ET AL.
1824
Table 1. Baseline characteristics of participants enrolled into the ENOS trial with outcome assessed at day 90. Number (%), median (interquartile range), or mean (standard deviation); comparison by chi-square test, Mann–Whitney U test, or t test. Participants with cognition data include 246 patients who died Variables
Cognition data present
Cognition data absent
2p
Number Age, y, mean (SD) Sex, female, n (%) Recruited from, n (%) Africa America, North Asia, South Asia, Southeast Australasia Europe United Kingdom Risk factors, n (%) High blood pressure Hyperlipidemia Diabetes mellitus IHD Previous stroke/TIA Atrial fibrillation Premorbid mRS, median (IQR) Blood pressure, mean (SD) Systolic Diastolic Heart rate, bpm, mean (SD) Scandinavian Stroke Scale (SSS), mean (SD) Dysphasia, n (%) Stroke type, n (%) Ischemic stroke Primary intracerebral hemorrhage Stroke syndrome, n (%) (16) TACS PACS LACS POCS Left hemisphere stroke, n (%) Brain scan, N 5 1564, n (%) Atrophy Leukoaraiosis Mass effect from stroke Previous stroke(s) Temperature, oC Blood glucose, mmol/L
1572 69.2 (12.3) 631 (40.1)
569 72.6 (11.1) 264 (46.4)
,.001 .010 ,.001
14 (.9) 30 (1.9) 157 (10.0) 182 (11.6) 49 (3.1) 167 (10.6) 973 (61.9)
1 (.2) 3 (.5) 3 (.5) 104 (18.3) 6 (1.1) 65 (11.4) 387 (68.0)
969 (61.6) 423 (26.9) 265 (16.9) 252 (16.0) 339 (21.6) 163 (10.4) 0 (0, 0)
405 (71.2) 147 (25.8) 92 (16.2) 93 (16.3) 142 (25.0) 94 (16.5) 0 (0, 0)
,.001 .62 .71 .86 .097 ,.001 .55
167.7 (19.2) 90.2 (13.4) 76.9 (14.9) 35.7 (12.8)
169.7 (18.8) 90.1 (12.7) 76.7 (15.3) 30.3 (13.2)
.031 .92 .84 ,.001
488 (31.0)
285 (50.5)
,.001 .92
1304 (83.0) 268 (17.0)
471 (82.8) 98 (17.2)
386 (24.6) 446 (28.4) 671 (42.7) 69 (4.4) 873 (55.5)
214 (37.6) 183 (32.2) 165 (29.0) 7 (1.2) 253 (44.5)
,.001
363 (23.2) 313 (20.0) 208 (13.2) 392 (25.1) 36.6 (.57) 7.3 (2.9)
174 (30.6) 129 (22.7) 75 (13.2) 138 (24.3) 36.6 (.52) 7.2 (2.7)
,.001 .18 .94 .70 .36 .073
,.001
Abbreviations: bpm, beats per minute; IHD, ischemic heart disease; IQR, interquartile range; LACS, lacunar syndrome; mRS, modified Rankin Scale; PACS, partial anterior circulation syndrome; POCS, posterior circulation syndrome; SD, standard deviation; TACS, total anterior circulation syndrome; TIA, transient ischemic attack.
3 cognition scales. The adjusted R square values for stepwise linear regression models were MMSE-M, r2 5 .351; TICS-M, r2 5 .349; and category fluency, r2 5 .230. The results of brain scanning, as reported by each investigators’ (neuro-) radiology department were available for 2132 patients (Computerized Tomography (CT)
scan 5 1992, Magnetic Resonance Imaging (MRI) scan 5 140). The presence of cerebral atrophy was significantly related to cognitive impairment as assessed by the 3 cognition scales (Table 3). Other imaging factors—mass effect from the qualifying stroke lesion, previous stroke, and leukoaraiosis—were associated in univariate but not in
ENOS POST STROKE COGNITION
1825
Figure 1. Distribution of scores for the modified Mini Mental State Examination, modified Telephone Instrument for Cognitive Status and category fluency (animal naming)
multiple variable analysis with 1 or more of the cognition scales. The results of analyses did not differ qualitatively after excluding patients who died; however, adjusted R square was reduced for each of MMSE-M (.35-.13), TICS-M (.35-.16), and category fluency (.23-.08; Table S2 in Appendix).
Relation between Cognition and Other Functional Outcomes Measures of cognitive impairment were each related to dependency (mRS), disability (BI), depression (ZDS), and reduced quality of life (EQ-5D and EQ-VAS; Table 3). There was a strong agreement between the 3 cognition scales; MMSE-M and TICS-M (rs 5 .83, P , .001), MMSEM and category fluency (rs 5 .72, P , .001), and TICS-M and category fluency (rs 5 .74, P , .001). Moderate agreement was noted between the 3 cognition scales in identifying cognitive impairment using predetermined cut-off scores; MMSE-M and TICS-M (k 5 .61, P , .001), MMSEM and category fluency (k 5 .50, P , .001), and TICS-M and category fluency (k 5 .48, P , .001). The agreement between the cognition scales was less robust when death was excluded from the analysis (Table S2 in Appendix).
Discussion The aims of the present study were first, to identify factors measured during the acute phase of stroke that are related to subsequent cognitive impairment and second, to assess the relationship between cognition at 90 days poststroke with other measures of functional outcome.
The strengths of the study are that it is one of the largest studies of cognition after acute stroke to date, includes multiple race ethnicity groups from 18 countries across 5 continents, uses unified methodology, involves patients largely representative of hospitalized stroke patients, is based on prospectively collected data, and uses data collected during the current and last decade. Poststroke cognitive impairment (PSCI) was related with increasing age, stroke severity, heart rate, TACS, and the presence of atrophy on brain scanning, in both unadjusted and adjusted analyses. Several studies (Table 4), including a systematic review,1 have reported associations between subacute PSCI and age,16-21 and stroke severity,16,17,20,22 and the present study confirms these findings. This suggests that the severity of the index stroke in an aging brain maybe sufficient to cause PSCI in the short-term. The relationship of cognition with AF, left hemisphere stroke, and hypertension was less consistent across the 3 cognition measures. Several studies, albeit smaller, have reported associations of subacute cognition with AF17-19 and left-sided strokes19,20 but none with hypertension. Others studies have shown an association with female sex,18,23,24 previous stroke,20,22,25 diabetes,16,22 and alcohol intake19; the first 3 were not found in this study and the latter factor is not measured in ENOS. Although high blood pressure (BP) during the acute phase of stroke is independently associated with poor functional outcome,26,27 an independent association between baseline BP and subsequent cognitive impairment was not seen in this study. The combined effect of these risk factors may become more important with time, as suggested by their association with the long-term
S. ANKOLEKAR ET AL.
1826
Table 2. Cognition and other functional outcome measures at day 90 using central telephone assessment in 1572 patients with cognition data, including 246 who died. Number (%) or median (interquartile range) Functional outcome Died MMSE-M, N 5 1218, died 192 Median (IQR) Dead or cognitive impairment, MMSE-M ,14, n (%) TICS-M, N 5 1041, died 178 Median (IQR) Dead or cognitive impairment, TICS-M ,20, n (%) Category fluency, N 5 1044, died 178 Median (IQR) Dead or cognitive impairment, Fluency ,10, n (%) mRS, N 5 1572, died 246 Median (IQR) Dead or dependent, mRS .2, n (%) BI, N 5 1572, died 246 Median (IQR) Dead or disabled, BI #90, n (%) ZDS, N 5 1526, died 246 Median (IQR) Dead or depressed, ZDS $70, n (%) Depressed at 90 days, n (%) EuroQOL, N 5 1571 for EQ-5D and 1522 for VAS EQ-5D Index, median (IQR) EQ-VAS, median (IQR)
Number 246 14.0 (7, 17) 582 (47.8)
21 (10, 24) 546 (48.4)
9 (3, 14) 546 (52.3)
3 (1, 4) 874 (55.6)
90 (46.3, 100) 902 (57.4) 52.5 (37.5, 72.5) 422 (27.7) 176 (13.8)
.64 (.14, .81) 65 (40, 80)
Abbreviations: BI, Barthel Index; EQ-5D, EuroQOL-5 Descriptor; EQ-VAS, EuroQOL-Visual Analog Scale; IQR, interquartile range; MMSE-M, Mini Mental State Examination; mRS, Modified Rankin Scale; TICS-M, Telephone Instrument for Cognitive Status-Modified; ZDS, Zung Depression Scale.
development of dementia. Interestingly, hyperlipidemia was inversely associated with cognitive impairment. This apparent paradox is difficult to explain but earlier observational studies have reported that hyperlipidemia is a risk factor for cognitive impairment in younger patients28,29 but appears to be protective in older individuals.30 Silent cerebrovascular disease detected by neuroimaging such as leukoaraiosis and silent infarcts are 10-20 times more prevalent than manifest strokes.31 Such lesions are independently associated with cognitive impairment in the elderly and may lower the threshold for dementia cumulatively, as seen with Alzheimer disease.32,33 Although, several studies have found an association between PSCI and silent infarcts,34 cerebral atrophy35 and white matter lesions,24,35 only cerebral atrophy remained significant on multiple variable analysis in the present study.
When assessing the inter-relationships between the 4 measures of functional outcome at 3 months, PSCI was associated with each of functional dependency (mRS), disability (BI), depression (ZDS), and reduced quality of life (EQ-5D, EQ-VAS). This is consistent with findings from previous but smaller studies.18,20,36-38 The strong inter-relationships between these measures suggest that interventions that attenuate dependency, the usual primary outcome in acute stroke trials, may also reduce subsequent cognitive decline. The present study has several limitations. First, the data come from a randomized controlled trial with multiple exclusion criteria that will have limited the range of included patients. Of relevance, ENOS excludes patients without motor impairment, and those who are normotensive or have premorbid dependency. These constraints may have attenuated some relationships, for example, between cognition and stroke severity, and cognition and BP. Counteracting this limitation is that the size of the data is an order of magnitude larger than previous studies, as summarized in a recent review article.1 Second, 27% of our patients were unable to answer telephone-based cognition questions because of factors such as dysphasia, fatigue, or being in hospital. This is somewhat higher than 18% seen in another acute stroke trial assessing cognition although it included only 294 patients39 Missing data are a recognized problem in studies assessing cognition, especially in elderly patients, and is likely to be higher in larger studies.40 These patients had more severe strokes (and other differences as summarized in Table 1), again biasing the sample. However, this limitation maybe less relevant in this study because patients who died (who will have had the most severe strokes) were included. As a result, the study will have tended to include mildmoderate and very severe stroke, but missed those with severe stroke. Third, there was no evaluation of prestroke cognitive impairment and it is difficult to say if factors other than the index stroke contributed to the day 90 cognition. However, exclusion of patients with mRS .2 at baseline means it is unlikely that many, if any, patients assessed in this study had overt dementia at baseline. Fourth, we used the interpretation of brain scan provided at the local center, this introducing variation in interpretation. Local reporting will also have focused on more overt than subtle imaging lesions. In due course, central blinded adjudication will be available for the whole trial. Fifth, as the study is international, we had to choose questions that were language and culturally independent, for example, using animal naming for category fluency rather than verbal fluency using words beginning with F, A, or S (which would have no relevance in Chinese patients). Cultural differences may affect some responses to the study questions and bias relationships, and international variations are known to exist for mRS, MMSE, and EQ-5D.41-43 Nevertheless, including patients from Africa, South Asia, Southeast Asia, Australasia–Oceania, Europe, and
ENOS POST STROKE COGNITION
1827
Table 3. Univariate and multiple variable relationships between cognitive measures at day 90 and (i) baseline factors and (ii) other functional outcome measures at day 90. Shown are correlation coefficients (r) with P values from the univariate analyses (Spearman rank correlation coefficient (rs) for continuous variables and rank biserial coefficient (rrb, Somer D) for binary variables). Patients who died were included in the analysis. MMSE-M Variable Day 0 (baseline) Age Sex, female Atrial fibrillation Diabetes mellitus Hyperlipidemia Hypertension Previous stroke/TIA Systolic blood pressure Heart rate Temperature Stroke severity (SSS) Stroke syndrome (16) Lacunar vs. rest TACS vs. rest Stroke type (IS vs. PICH) Stroke side (left vs. right) Brain scan Atrophy Leukoaraiosis matter disease Mass effect from stroke Previous stroke(s) Serum glucose Day 90 Modified Rankin Scale Barthel Index Zung Depression Score EQ-5D Index EQ Visual Analog Scale
TICS-M
Category fluency
r
2p
r
2p
r
2p
2.343 2.140 2.366 2.010 .056 2.102 2.136 2.021 2.117 .021 .378
,.001y ,.001 ,.001y .808 .132 .002 .001 .468 ,.001y .477 ,.001y
2.366 2.066 2.297 2.033 .046 2.130 2.111 .005 2.125 .035 .377
,.001y .074 ,.001 .502 .262 ,.001 .015 .883 ,.001y .261 ,.001y
2.304 2.093 2.294 2.031 .093 2.131 2.118 2.020 2.109 2.050 .378
,.001y .009 ,.001 .536 .026y ,.001y .008 .516 ,.001y .106 ,.001y
.283 2.472 2.027 2.120
,.001 ,.001y .562 ,.001y
.236 2.393 .004 2.118
,.001 ,.001y .934 .001
.242 2.408 2.075 2.091
,.001 ,.001y .091 .010
2.258 2.109
,.001y .009
2.246 2.072
,.001y .117
2.209 2.073
,.001y .110
2.190 2.109 2.115
,.001 .008 .002
2.102 2.042 2.051
.105 .346 .101
2.211 2.050 .036
,.001 .250 .245
2.551 .571 2.532 .507 .532
,.001 ,.001 ,.001 ,.001 ,.001
2.562 .577 2.542 .519 .528
,.001 ,.001 ,.001 ,.001 ,.001
2.558 .590 2.589 .518 .560
,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: EQ, Euro Quality of life; EQ-5D, Euro Quality of life-5 Descriptor; IS, ischemic stroke; PICH, primary intracerebral hemorrhage; SSS, Scandinavian Stroke Scale; TACS, total anterior circulation syndrome; TIA, transient ischemic attack. yAssociations those remain significant on multiple variable analyses.
North America will have increased substantially the external validity of the findings. Sixth, telephone assessment of cognition will be affected by extraneous factors such as hearing loss in an older population and whether the participant or a caregiver delivered the responses. Finally, only basic measures of cognition were used in this study; MMSE-M is biased toward attention and orientation and suffers from a ceiling effect. However, TICS-M includes assessment of multiple cognitive domains covering memory, orientation, attention, and language. Importantly, as the prevalence of cognitive impairment found in this study is comparable with other similar studies, the limitations of these scales maybe more theoretic than real in this population. Previous studies have ignored death as a cognitive outcome but we included it because death is part of other
functional assessments used in stroke such as mRS, BI, and EQ-5D and is the final outcome in end stage dementia. In addition, there is the statistical advantage that patients who die are not discarded from analysis, important because this group comprises a significant minority of the cohort, and including them removes the bias of excluding patients who are older and/or have more severe stroke. The additional sample size also adds statistical power to analyses. In summary, PSCI is common at day 90 and is associated with increasing age, stroke severity, history of AF and hypertension, and cerebral atrophy on neuroimaging. The presence of these factors maybe useful clinically in identifying people at high risk of developing PSCI. Furthermore, PSCI is associated with other functional outcomes at day 90 including dependency, disability, poor mood, and
S. ANKOLEKAR ET AL.
1828
Table 4. Baseline factors in patients with acute stroke that are related to subsequent cognitive impairment (2 weeks to 6 months) in other studies. Baseline Predictor Baseline demographics Age Education Sex, female Atrial fibrillation Previous stroke Diabetes Hypertension Hyperlipidemia Stroke severity Left vs. right OCSP stroke presentation Heart rate Baseline blood pressure Temperature Glucose Baseline CT scan Cerebral Atrophy Mass effect from stroke Previous Stroke Leukoaraiosis
Univariate analysis
Multiple variable analysis
(18-26)y (20, 22, 24-28) (19, 24, 29)y (17-20)y (19-23, 25, 26)y (17, 20, 23, 26) -y -y (17-19, 21, 23, 27, 29)y (20, 21, 25)y (17, 22)y -y (28) (18)
(18-23, 26)y (20, 22, 25, 26) (18-20)y (20, 21, 23, 25, 26) (17, 23) -y -y (17, 18, 21, 23, 29)y (20, 21)y (17)y -y -
(22, 29, 30)y y y (27, 29, 30)y
(30)y -y (29, 30)
-: denotes no association in other studies; y: this study.
reduced quality of life; hence, beneficial treatment for one may have useful effects on the others. Acknowledgment: We thank patients and investigators who are taking part in ENOS, and the UK Stroke Research Network for facilitating recruitment. Trial Steering Committee: Pierre Amarenco (Independent expert, France), Philip Bath (Chief Investigator, UK), Kennedy Lees (UK), Keith Muir (Independent expert, UK); Stuart Pocock (Statistical consultant, UK), Angela Shone (Sponsor representative, UK), Jane Sinclair (Funder representative, UK), Graham Venables (Independent chair, UK), Joanna Wardlaw (Neuroradiology lead, UK), David Whynes (Health Economics lead, UK). International Advisory Committee: Chris Bladin (Australia), Valeria Caso (Italy), Hui Meng Chang (Singapore), Ronan Collins (Eire), Anna Czlonkowska (Poland), Anwar El Etribi (Egypt), Abdul Raman Ghani (Malaysia), John Gommans (New Zealand), Stephen Phillips (Canada), Kameshwar Prasad (India), Asita de Silva (Sri Lanka), Szabolcs Szatmari (Romania), Exuperio de Tejedor (Spain), Yong-Jun Wang (China), Lawrence Wong (Hong Kong) competing interests. Competing Interests: P.M.W.B. is Chief Investigator and J.W. is neuroimaging lead for ENOS.
Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10. 1016/j.jstrokecerebrovasdis.2014.04.022
References 1. Pendlebury ST, Rothwell PM. Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: a systematic review and meta-analysis. Lancet Neurol 2009;8:1006-1018. 2. Ballard C, Rowan E, Stephens S, et al. Prospective follow-up study between 3 and 15 months after stroke: improvements and decline in cognitive function among dementia-free stroke survivors .75 years of age. Stroke 2003;34:2440-2444. 3. Rasquin SM, Lodder J, Verhey FR. Predictors of reversible mild cognitive impairment after stroke: a 2-year followup study. J Neurol Sci 2005;229-230:21-25. 4. Jorge RE, Acion L, Moser D, et al. Escitalopram and enhancement of cognitive recovery following stroke. Arch Gen Psychiatry 2010;67:187-196. 5. Kimura M, Robinson RG, Kosier JT. Treatment of cognitive impairment after poststroke depression : a doubleblind treatment trial. Stroke 2000;31:1482-1486. 6. Ankolekar S, Geeganage C, Anderton P, et al. Clinical trials for preventing post stroke cognitive impairment. J Neurol Sci 2010;299:168-174. 7. Anderson CA, Arciniegas DB, Filley CM. Treatment of acute ischemic stroke: does it impact neuropsychiatric outcome? J Neuropsychiatry Clin Neurosci 2005;17:486-488. 8. The ENOS Trial Investigators. Glyceryl trinitrate vs. control, and continuing vs. stopping temporarily prior antihypertensive therapy, in acute stroke: rationale and design of the Efficacy of Nitric Oxide in Stroke (ENOS) trial (ISRCTN99414122). Int J Stroke 2006;1:245-249. 9. Folstein M, Folstein S, McHugh P. ‘‘Mini-mental state:’’ a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198. 10. Barber M, Stott DJ. Validity of the Telephone Interview for Cognitive Status (TICS) in post-stroke subjects. Int J Geriatr Psychiatry 2004;19:75-79.
ENOS POST STROKE COGNITION 11. de Jager CA, Budge MM, Clarke R. Utility of TICS-M for the assessment of cognitive function in older adults. Int J Geriatr Psychiatry 2003;18:318-324. 12. Tombaugh TN, Kozak J, Rees L. Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Arch Clin Neuropsychol 1999; 14:167-177. 13. Sulter G, Steen C, Jacques De K. Use of the Barthel Index and Modified Rankin Scale in acute stroke trials. Stroke 1999;30:1538-1541. 14. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry 1965;12:63-70. 15. Euroqol Group. EuroQol–a new facility for the measurement of health-related quality of life. Health Policy 1990; 16:199-208. 16. Censori B, Manara O, Agostinis C, et al. Dementia after first stroke. Stroke 1996;27:1205-1210. 17. Barba R, Martinez-Espinosa S, Rodriguez-Garcia E, et al. Poststroke dementia: clinical features and risk factors. Stroke 2000;31:1494-1501. 18. Tamam B, Tasdemir N, Tamam Y. [The prevalence of dementia three months after stroke and its risk factors]. Turk Psikiyatri Derg 2008;19:46-56. 19. Zhou DH, Wang JY, Li J, et al. Study on frequency and predictors of dementia after ischemic stroke: the Chongqing stroke study. J Neurol 2004;251:421-427. 20. Lin JH, Lin RT, Tai CT, et al. Prediction of poststroke dementia. Neurology 2003;61:343-348. 21. Rasquin SM, Verhey FR, van Oostenbrugge RJ, et al. Demographic and CT scan features related to cognitive impairment in the first year after stroke. J Neurol Neurosurg Psychiatry 2004;75:1562-1567. 22. Henon H, Durieu I, Guerouaou D, et al. Poststroke dementia: incidence and relationship to prestroke cognitive decline. Neurology 2001;57:1216-1222. 23. Madureira S, Guerreiro M, Ferro JM. Dementia and cognitive impairment three months after stroke. Eur J Neurol 2001;8:621-627. 24. Tang WK, Chan SS, Chiu HF, et al. Frequency and determinants of poststroke dementia in Chinese. Stroke 2004; 35:930-935. 25. Pohjasvaara T, Erkinjuntti T, Ylikoski R, et al. Clinical determinants of poststroke dementia. Stroke 1998;29: 75-81. 26. Leonardi-Bee J, Bath PMW, Phillips SJ, et al. Blood pressure and clinical outcomes in the International Stroke Trial. Stroke 2002;33:1315-1320. 27. Willmot M, Leonardi-Bee J, Bath PM. High blood pressure in acute stroke and subsequent outcome: a systematic review. Hypertension 2004;43:18-24. 28. Whitmer RA, Sidney S, Selby J, et al. Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 2005;64:277-281.
1829 29. Dufouil C, Richard F, Fievet N, et al. APOE genotype, cholesterol level, lipid-lowering treatment, and dementia: the Three-City Study. Neurology 2005;64:1531-1538. 30. Solfrizzi V, Panza F, Colacicco AM, et al. Vascular risk factors, incidence of MCI, and rates of progression to dementia. Neurology 2004;63:1882-1891. 31. Longstreth WT Jr. Brain abnormalities in the elderly: frequency and predictors in the United States (the Cardiovascular Health Study). Cardiovascular Health Study Collaborative Research Group. J Neural Transm Suppl 1998;53:9-16. 32. Snowdon DA, Greiner LH, Mortimer JA, et al. Brain infarction and the clinical expression of Alzheimer disease. The Nun Study. JAMA 1997;277:813-817. 33. Schneider JA, Wilson RS, Bienias JL, et al. Cerebral infarctions and the likelihood of dementia from Alzheimer disease pathology. Neurology 2004;62:1148-1155. 34. Mok V, Chang C, Wong A, et al. Neuroimaging determinants of cognitive performances in stroke associated with small vessel disease. J Neuroimaging 2005; 15:129-137. 35. Jokinen H, Kalska H, Mantyla R, et al. White matter hyperintensities as a predictor of neuropsychological deficits post-stroke. J Neurol Neurosurg Psychiatry 2005; 76:1229-1233. 36. Kauhanen M, Korpelainen JT, Hiltunen P, et al. Poststroke depression correlates with cognitive impairment and neurological deficits. Stroke 1999;30:1875-1880. 37. Tatemichi TK, Desmond DW, Stern Y, et al. Cognitive impairment after stroke: frequency, patterns, and relationship to functional abilities. J Neurol Neurosurg Psychiatry 1994;57:202-207. 38. Mok VCT, Lam WWM, Fan YH, et al. Cognitive impairment and functional outcome after stroke associated with small vessel disease. J Neurol Neurosurg Psychiatry 2004;75:560-566. 39. Cumming TB, Bernhardt J, Linden T. The Montreal cognitive assessment: short cognitive evaluation in a large stroke trial. Stroke 2011;42:2642-2644. 40. Hardy SE, Allore H, Studenski SA. Missing data: a special challenge in aging research. J Am Geriatr Soc 2009; 57:722-729. 41. Quinn TJ, Dawson J, Walters MW, et al. Variability in modified Rankin scoring across a large cohort of international observers. Stroke 2008;39:2975-2979. 42. Scazufca M, Almeida OP, Vallada HP, et al. Limitations of the Mini-Mental State Examination for screening dementia in a community with low socioeconomic status: results from the Sao Paulo Ageing & Health Study. Eur Arch Psychiatry Clin Neurosci 2009;259:8-15. 43. Sainsbury A, Seebass G, Bansal A, et al. Reliability of the Barthel Index when used with older people. Age Ageing 2005;34:228-232.
