You are here

Article Text

Article menu
Cerebrovascular disease
Research paper
The cognitive burden of stroke emerges even with an intact NIH Stroke Scale Score: a cohort study
  1. Tatu Kauranen1,2,
  2. Siiri Laari1,3,
  3. Katri Turunen1,3,
  4. Satu Mustanoja3,
  5. Peter Baumann2,
  6. Erja Poutiainen1,3
  1. 1University of Helsinki, Institute of Behavioral Sciences, Helsinki, Finland
  2. 2Department of Neurology and Clinical Neurophysiology, Lapland Central Hospital, Rovaniemi, Finland
  3. 3Helsinki University Central Hospital, Helsinki, Finland
  1. Correspondence to Tatu Kauranen Lapland Central Hospital, Department of Neurology and Clinical Neurophysiology, P.O. Box 8041, FIN-96200 Rovaniemi, Finland; tatu.kauranen{at}lshp.fi

Abstract

Background We aim to facilitate recognition of the cognitive burden of stroke by describing the parallels between cognitive deficits and the National Institutes of Health Stroke Scale (NIHSS), a widely used measure of stroke severity.

Methods A consecutive cohort of 223 working-age patients with an acute first-ever ischaemic stroke was assessed neuropsychologically within the first weeks after stroke and at a 6-months follow-up visit and compared with 50 healthy demographic controls. The NIHSS was administered at the time of hospital admittance and upon discharge from the acute care unit. The associations between total NIHSS scores and domain-specific cognitive deficits were analysed correlatively and with a binary logistic regression.

Results Of the NIHSS measurements (admittance median=3, range 0–24; discharge median=1, range 0–13), the total score at the time of discharge had systematically stronger correlations with cognitive impairment. Adjusted for demographics, the NIHSS discharge score stably predicted every cognitive deficit with ORs ranging from 1.4 (95% CI 1.2 to 1.6) for episodic memory to 1.9 (95% CI 1.5 to 2.3) for motor skills. The specificities of the models ranged from 89.5–97.7%, but the sensitivities were as low as 11.6–47.9%. Cognitive deficits were found in 41% of patients with intact NIHSS scores and in all patients with NIHSS scores ≥4, a finding that could not be accounted for by confounding factors.

Conclusions Cognitive deficits were common even in patients with the lowest NIHSS scores. Thus, low NIHSS scores are not effective indicators of good cognitive outcomes after stroke.

  • Cerebrovascular Disease
  • Cognition
  • Neuropsychology
  • Stroke

http://dx.doi.org/10.1136/jnnp-2013-305585

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Cognitive impairments predict nearly every stroke outcome measure.1–4 As a consequence, the efficient allocation of limited support resources requires the ability to identify patients with and without poststroke cognitive deficits. However, the resources required for thorough neuropsychological assessments to evaluate cognitive impairments vary between and within hospitals. Without effective neuropsychological assessments, assumptions of a stroke patient's cognitive state rely on known risk factors for poststroke cognitive impairments,5 and other measures of stroke severity.

    The NIH Stroke Scale (NIHSS),6 is a widely used stroke severity measure. Although this scale was not designed to measure cognition, several studies have investigated the relationship between the NIHSS and cognition. The interest has focused primarily on the focal deficits to which the NIHSS has been proven insensitive, specifically neglect,7 and aphasia.8 Whether the NIHSS is also insensitive to other domain-specific cognitive deficits remains unexplored. In the prediction of severe global poststroke cognitive impairments in the elderly, the NIHSS has shown some promise.9 For prioritising purposes, information regarding the probability of a good recovery is also needed.

    The aim of this cohort study was to describe the relationship between total NIHSS scores and the emergence of cognitive impairments following stroke. Based on previous studies of neglect and aphasia, we hypothesised that the NIHSS would also be insensitive to other domain-specific cognitive deficits. However, we expected the lowest total NIHSS scores to predict good cognitive outcomes on a global level of cognitive dysfunction.

    Methods

    Setting and participants

    Data were collected consecutively during 2007–2009 in two Finnish hospitals: Helsinki University Central Hospital and Lapland Central Hospital. The inclusion criteria were a first-ever diagnosed supratentorial ischaemic stroke in native Finnish-speaking patients between the ages of 18 years and 65 years. Patients were excluded if they had a neurological or psychiatric history or comorbidity based on their clinical history and patient reports. All patients in this cohort were treated according to the institutional guidelines for stroke care. A control group, which consisted of 50 healthy subjects with similar demographic characteristics who were eligible based on all of the inclusion/exclusion criteria except for the stroke, was also assessed. The controls were recruited from the spouses, siblings and friends of the patients. This cohort has been partially described in previous reports.3 ,10 The Ethics Committee of Helsinki University Central Hospital approved the study and consent procedure, and all participants signed an informed consent form.

    Neurological and medical data

    The NIHSS was completed first upon admittance to the hospital and again upon discharge from the acute care unit, and the total scores were used. A stroke neurologist evaluated the lesion size and the presence of potential leukoaraiosis and silent infarcts using CT or MRI and defined the pathophysiological aetiology of the infarct.11 In addition, the functional status,12 was assessed as a part of the neuropsychological assessment.

    Patient medical reports were collected and checked for the six principal vascular risk factors: atrial fibrillation, diabetes, hypercholesterolaemia, hypertension, overweight (defined as a body mass index >25) and smoking, which were encoded by pooling them to create an accumulation sum score that provides the number of principal vascular risk factors.

    Cognitive and mood state data

    A neuropsychologist completed a neuropsychological assessment according to a written study protocol twice for each patient; first, after a neurologist had determined that the patient's acute condition had been stabilised and that the patient was ready for discharge from the acute care unit, and again at the 6-months follow-up to validate the initial findings. The control participants were also assessed twice, with a 3-months interval between testing sessions, according to an identically written study protocol.

    Neuropsychological assessments included measures of seven cognitive domains. Each cognitive domain was measured with three separate tests as follows: executive functions with the Trail Making Test Part B,13 ,14 phonemic fluency15 and Go/no-go task16; psychomotor speed with time to complete the Trail Making Test Part A,13 ,14 copying tasks (time per correct response)16 and the modified Token Test (time per correct response)17; episodic memory with a sum score of immediate and delayed recall in Wechsler Memory Scale Revised Logical Memory I and II,18 learning a series of 10 unrelated words with an added delayed recall score16 and a shortened Benton Visual Retention Test with added delayed recall and delayed recognition scores19; working memory with the Digit Span subtest of the Wechsler Adult Intelligence Scale, Third Edition,20 a homogeneous interference task16 and a heterogeneous interference task16; language with the Modified Token Test,17 a shortened Visual Naming task from the Boston Diagnostic Aphasia Examination21 and repetition of a long sentence16; visual spatial and construction skills with the copying of four geometric figures,15 ,22 clock arms with 10 clocks15 and a visuospatial searching task23; and motor skills with the bimanual hand movements task,16 fist-palm-edge task16 and finger tapping test.15 The inner consistency of within-domain tests ranged from 0.74 to 0.93 (Cronbach's α).

    Mood state was evaluated as part of both of the neuropsychological assessments using a modified Profile of Mood States (POMS)24 questionnaire that included questions about mood, apathy and fatigue. At the initial evaluation, a 10-item version was used; at follow-up, a 38-item version was used. Total symptom sum scores were used in the analyses to control for the effect of altered mood state, apathy and fatigue on cognitive performance.

    Statistical analyses

    The statistical analyses were computed using IBM SPSS Statistics software, V.20.0 (International Business Machines Corporation, Armonk, New York, USA). A total of 10 cells with missing data in the data matrix, found within six different neuropsychological test scores of the patients, were imputed using the means of the variables. POMS data was missing for 11 patients. Comparisons between the patients and controls and the dropout analysis were calculated using t tests, U tests or χ2 statistics, depending on the variable distributions. All neuropsychological tests were administered to all participants, and test failures (eg, because of severe aphasia or neglect) were scored as zero points or the maximum time. The patients’ follow-up neuropsychological assessments were compared with the controls’ follow-up performance to control for the effect of learning on test performance. A patient's neuropsychological test performance was considered defective if it was below the 10th percentile of the control group's performance. Spearman's correlations were calculated for the numbers of defective within-domain test performances for each of the seven cognitive domains in the initial neuropsychological assessment and the total NIHSS scores upon admittance and discharge. To characterise the sample and perform the subsequent logistic regression models, domain-specific cognitive deficits were binary-encoded for cases that had at least two defective within-domain test scores. This binary method of defining cognitive deficits on the basis of accumulating defective performance was chosen to enhance the clinical relevance of the cognitive deficits, and it has also been used previously.3 ,25 Subsequently, enter-method binary logistic regression models adjusted for gender, age and education (years) were calculated for those cognitive domains that had a significant correlation with NIHSS scores. Logistic regression models were evaluated with regard to the significance of predictors as measured by Wald statistics, ORs with CIs and the sensitivities and specificities of the models. In addition, the total number of cognitive deficits (0, 1, 2 or at least 3) and the presence of at least one cognitive deficit were also calculated to characterise global, cross-domain cognitive dysfunction.

    Results

    Sample characteristics

    During data collection, 38 patients refused to participate, 19 were lost due to logistic reasons such as quick discharges or hospital transfers and 6 severely injured patients were excluded because they could not complete the neuropsychological assessments. Originally, 230 patients with ischaemic stroke were included in the study. However, after the retrospective exclusion of seven patients with specific pervasive developmental disorders (F80–89) according to the 10th revision of the International Statistical Classification of Diseases and Related Health Problems, the final number of patients included in the current study was 223 (63% male). The mean age of the patients was 54 years (SD=9.8), and they had a mean of 12 years of education (SD=2.7). The average duration of their hospital stay was 10 days (SD=5.2 days). Online supplementary table S1 presents the characteristics of the cohort. The distributions of gender, age and education did not differ between the patient and control groups; however, the patients reported more mood symptoms and had more vascular risk factors than the controls did. Large artery atherosclerosis was the most common pathophysiological cause of stroke. The cohort recovered well clinically (NIHSS at discharge median 1, range 0–13) and functionally (Barthel Index median 100, range 25–100). A total of 136 patients (61.0%) exhibited at least one cognitive deficit during the initial neuropsychological assessments, which were conducted an average of 8 days after the stroke (SD=4.5).

    Of the initial cohort, N=190 participated in the 6-months follow-up. The follow-up neuropsychological assessments were conducted an average of 178 days (SD=14.2) after the stroke. During the follow-up, at least one cognitive deficit was found in 88 patients (46.3%). All cognitive deficits, with the exception of executive dysfunction and motor skills, showed a significant trend of recovery from the initial assessment to the follow-up (p<0.05). The 33 dropouts from the 6-months follow-up were less educated and were more likely to have leukoaraiosis and domain-specific cognitive deficits than the followed-up patients (p<0.05), with the exception of visual construction and spatial deficits. The dropouts also had a higher total number of cognitive deficits.

    NIHSS and cognition

    The total NIHSS scores at admittance and discharge were correlated (p<0.01) with each of the seven cognitive impairment scores, that is, the number of defective within-domain test performances during the initial assessment. The correlations between the NIHSS admittance score and cognitive impairment scores ranged from r=0.18 (episodic memory) to r=0.40 (motor skills). The NIHSS discharge score associations were systematically stronger, and the correlations with the cognitive impairment scores ranged from r=0.34 (working memory) to r=0.44 (motor skills).

    Separate logistic regression models adjusted for gender, age and education were computed to predict domain-specific cognitive deficits within the first weeks after the stroke using NIHSS scores upon admittance and discharge. Figure 1 illustrates the odds of having each of the seven cognitive deficits when the NIHSS score increased by 1 unit, starting from 0. Adjusted for demographics, the NIHSS discharge score stably predicted every cognitive deficit with ORs ranging from 1.37 (95% CI 1.18 to 1.60) for episodic memory to 1.87 (95% CI 1.52 to 2.29) for motor skills. The NIHSS score at discharge was systematically more effective in predicting deficits than the admittance score. The predictive models had rather high specificities, which ranged from 89.5–97.7%, but the sensitivities were as low as 11.6–47.9%. Figure 2 further illustrates this insensitivity by describing the emergence of cognitive deficits across different total NIHSS scores and their persistence at the 6-months follow-up. Approximately two patients out of five with a NIHSS score of 0 initially had at least one cognitive deficit, as did half of the patients with a score of 1. Among patients with NIHSS scores of 2 or 3, the prevalence of initial cognitive impairment increased linearly, and all of the patients in this functionally well-recovered working-age cohort who had NIHSS scores of 4 or higher exhibited at least one cognitive deficit. The prevalence of cognitive deficits specific to every domain except motor skills was significantly reduced between the initial assessment and the follow-up (p<0.05). The common relationship between cognitive deficits and the lowest NIHSS scores, however, persisted at the 6-months follow-up assessment.

    Figure 1
    Figure 1

    Risk of domain-specific cognitive deficits increases with every increase in the NIH Stroke Scale (NIHSS) score. ORs are from binary logistic regression models adjusted for gender, age and education. “A deficit exists” is the predicted category. All ORs relate to Wald tests statistically significant at p<0.01. Error bars represent the 95% CI.

    Figure 2
    Figure 2

    Emergence of at least one cognitive deficit within the first weeks after a first-ever ischaemic stroke and at 6-months follow-up across total NIH Stroke Scale (NIHSS) scores upon discharge from the acute care unit.

    To control for confounding factors, we compared those patients with at least one initial cognitive deficit (N=136) with the cognitively intact patients (N=87) in terms of their accumulation of six principal vascular risk factors, silent infarcts, leukoaraiosis and mood state. The occurrence of at least one cognitive deficit did not account for the accumulation of the six principal risk factors (cognitively impaired/intact patients, respectively; median=3, range=0–6/median=3, range=0–5; U=4919.500, p=0.071), silent infarcts (37 (27%)/18 (21%); χ2(1)=1.113, p=0.291), leukoaraiosis (37 (27%)/20 (23%); χ2(1)=0.730, P=0.393) or mood state (median=13, range 1–31/median 11, range 2–32; U=4547.500, p=0.052). To further test the role of these possibly confounding factors—especially mood state, which only narrowly avoided reaching significance in the analysis of the total sample—we repeated the comparisons for the N=135 patients who had the lowest NIHSS scores (0–1) upon discharge from the acute care unit. This repeat analysis did not alter the results, as none of the factors were significantly saturated in the groups of 64 cognitively impaired/71 intact patients, respectively (vascular risk factors: median=3, range=0–6/median=3, range=0–5, U=1871.000, p=0.123; silent infarcts: 20 (31%)/16 (23%), χ2(1)=1.671, p=0.196; leukoaraiosis: 21 (33%)/17 (25%), χ2(1)=1.196 p=0.274; mood state: median=13, range=1–31/median 11, range=2–32, U=1863.000, p=0.156).

    Discussion

    As hypothesised, the insensitivity of the NIHSS in detecting cognitive deficits was not limited to acute neglect and aphasia,5 ,8 ,9 ,26 but was also found in a broader range of cognitive functions. Previously, Cumming et al9 had shown in elderly patients that a cognitively oriented subscale of the NIHSS could predict severe global poststroke cognitive impairments, as indicated by a reference diagnosis. We had a significantly younger, working-aged cohort and detailed neuropsychological assessments and used a strict method of defining cognitive deficits rather than a diagnostic category. Consequently, we expected to be able to describe the parallels of good clinical and cognitive recovery. However, in contrast to our hypothesis, we could not provide even a tentative NIHSS cut-off score that would predict intact cognition even after controlling for the effects of vascular risk factors, silent infarcts, leukoaraiosis and mood state. This finding is alarming, considering the impact of the NIHSS on our evaluations of stroke severity and the power of cognitive impairments within the first weeks after stroke to predict various other stroke outcome measures, such as return to work,3 quality of life and mood.2 The observed prevalence rates of poststroke cognitive deficits among the patients with the mildest strokes emphasise the importance of neuropsychological assessments in understanding the sequelae of stroke and for appropriate treatment planning.

    The validity of our results depends on the clinical relevance of the definitions of cognitive deficits and success in the sampling of the study. In defining cognitive deficits, we used a rather conservative method, which has been proven clinically relevant in predicting the ability to return to work after stroke in the same population.3 The clinical relevance was accomplished through the use of a thorough neuropsychological assessment, with three separate tests for each cognitive domain. To denote a deficit, we required at least two out of the three within-domain test scores to be defective. This method of chasing a repeating pattern of defective test performance instead of single deviant test scores mimics clinical neuropsychological decision-making and significantly narrows the criteria for a defective performance. As a consequence of this conservative definition, prevalence of having at least one cognitive deficit within the first weeks after stroke in our study was somewhat lower (61%) than previously reported (approximately 70%).5 The initial neuropsychological assessments were completed after the neurologist had evaluated a patient to be ready for discharge. This timing was chosen to minimise the effect of fluctuating acute conditions such as delirium and fatigue on the neuropsychological test results. The effects of fatigue, apathy and mood state were also evaluated with the POMS questionnaire. By scoring an administered but totally failed test as poor performance, we took the risk of incurring type I error in encoding other cognitive domains than the truly impaired ones as defective. For example, if a patient also failed a test other than language-related tests because of his or her aphasia, the patient was more likely to also have other domains scored as defective, irrespective of whether the patient truly had any neuropsychological impairment other than aphasia. Therefore, the prevalence rates of the domain-specific cognitive deficits we report might be marginally inflated, although they are similar to those previously observed.1 This, however, does not distort our main argument concerning the prevalence of having at least one conservatively defined cognitive deficit associated with the lowest NIHSS scores.

    The sample of the study was restricted to working-age (18–65 years) patients with a first-ever stroke. As the patients were receiving modern acute stroke care, the favourable clinical and functional outcome of the cohort was not unexpected. The small amount of more severely injured patients in the sample, however, raises concerns regarding the generalisability of the results. We excluded six patients because they were too severely injured to complete the neuropsychological assessment. These exclusions certainly cause a slight underestimation of the median NIHSS score and Barthel Index of our data for the purpose of generalising the results to the stroke population as a whole. However, as we are arguing for the relatively high prevalence of cognitive deficits among patients with less severe strokes, the good clinical and functional recoveries of our sample are not an argument against the generalisability of our results. In fact, due to the good clinical and functional outcomes, this working-age cohort arguably presents an essential stroke subpopulation at risk of suffering from non-detected cognitive burden. In addition, focusing on these relatively young stroke survivors supported our aim to minimise the bias caused by potential pre-existing cognitive impairment, about which it is generally impossible to gather explicit information in a clinical setting. Other factors that may have caused inaccuracy in the results we presented are the relatively modest sample size, particularly in the control group, the different sizes of the patient and control groups and the use of only the imaging techniques that were available in standard clinical practice.

    To conclude, our results demonstrate how the cognitive burden of stroke begins to accumulate in the absence of other markers of stroke severity, even in a working-age cohort that has achieved good functional recovery. Thus, there appears to be no safe limit for assuming intact cognition in stroke patients based on low NIHSS scores. Given the established possibilities for treating cognitive deficits after stroke,27 our results encourage the early and systematic assessment of stroke patients’ cognition.

    Acknowledgments

    University Instructor Jari Lipsanen was consulted regarding statistical methods.

    References

    1. ↵
      1. Nys GMS,
      2. van Zandvoort MJE,
      3. de Kort PLM,
      4. et al
      . The prognostic value of domain-specific cognitive abilities in acute first-ever stroke. Neurology 2005;64:821–7.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Nys GMS,
      2. van Zandvoort MJE,
      3. van der Worp HB,
      4. et al
      . Early cognitive impairment predicts long-term depressive symptoms and quality of life after stroke. J Neurol Sci 2006;247:149–56.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Kauranen T,
      2. Turunen K,
      3. Laari S,
      4. et al
      . The severity of cognitive deficits predicts return to work after a first-ever ischaemic stroke. J Neurol Neurosurg Psychiatry 2013;84:316–21.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Wiberg B,
      2. Kilander R,
      3. Sundström J,
      4. et al
      . The relationship between executive dysfunction and post-stroke mortality: a population-based cohort study. BMJ Open 2012;2:e000458.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Gottesman RF,
      2. Hillis AE
      . Predictors and assessment of cognitive dysfunction resulting from ischaemic stroke. Lancet Neurol 2010;9:895–905.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Goldstein LB,
      2. Bertels C,
      3. Davis JN
      . Interrater reliability of the NIH Stroke Scale. Arch Neurol 1989;46:660–2.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Hillis AE,
      2. Wityk RJ,
      3. Barker PB,
      4. et al
      . Change in perfusion in acute nondominant hemisphere stroke may be better estimated by tests of hemispatial neglect than by the National Institutes of Health Stroke Scale. Stroke 2003;34:2392–6.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Croquelois A,
      2. Wintermark M,
      3. Reichhart M,
      4. et al
      . Aphasia in hyperacute stroke: language follows brain penumbra dynamics. Ann Neurol 2003;54:321–9.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Cumming TB,
      2. Blomstrand C,
      3. Bernhardt J,
      4. et al
      . The NIH stroke scale can establish cognitive function after stroke. Cerebrovasc Dis 2010;30:7–14.
      OpenUrlCrossRefPubMed
    10. ↵
      1. Turunen K,
      2. Kauranen T,
      3. Laari S,
      4. et al
      . Cognitive deficits after subcortical infarction are comparable with deficits after cortical infarction. Eur J Neurol 2013;20:286–92.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Adams HP Jr.,
      2. Bendixen BH,
      3. Kappelle LJ,
      4. et al
      . Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35–41.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Mahoney FI,
      2. Barthel D
      . Functional evaluation: the Barthel Index. Md State Med J 1965;14:56–61.
      OpenUrlPubMed
    13. ↵
      1. Reitan RM
      . Validity of the trail making test as an indication of organic brain damage. Percept Mot Skills 1958;8:271–6.
      OpenUrlCrossRef
    14. ↵
      1. Poutiainen E,
      2. Kalska H,
      3. Laasonen M,
      4. et al
      ., eds. Trail—making—testi. Käsikirja (The trail-making test. A finnish manual). Helsinki: Psykologien Kustannus Oy, 2010.
    15. ↵
      1. Lezak MD,
      2. Howieson DB,
      3. Loring DW
      . Neuropsychological assessment. 4th edn. New York: Oxford University Press, 2004.
    16. ↵
      1. Christensen A-L
      . Luria's neuropsychological investigation text. 2nd edn. Copenhagen: Munksgaard, 1979.
    17. ↵
      1. De Renzi E,
      2. Faglioni P
      . Normative data and screening power of a shortened version of the Token Test. Cortex 1978:14;41–9.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Wechsler D
      . WMS-R Käsikirja (The WMS-R. A finnish manual). Helsinki: Psykologien Kustannus Oy, 1996.
    19. ↵
      1. Benton AL
      . Revised visual retention test. 4th edn. New York: Psychological Corporation, 1974.
    20. ↵
      1. Wechsler D
      . WAIS-III Käsikirja (The WAIS-III. A finnish manual). Helsinki: Psykologien Kustannus Oy, 2005.
    21. ↵
      1. Laine M,
      2. Niemi J,
      3. Koivuselkä-Sallinen P,
      4. et al
      . BDAT- Bostonin Diagnostinen Afasiatesti (The standardized Finnish version of the Boston Diagnostic Aphasia Examination). Helsinki: Psykologien Kustannus Oy, 1997.
    22. ↵
      1. Carlesimo GA,
      2. Fadda L,
      3. Caltagirone C
      . Basic mechanisms of constructional apraxia in unilateral brain-damaged patients: role of visuo-perceptual and executive disorders. J Clin Exp Neuropsychol 1993;15:342–58.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Vilkki J
      . Hemi-inattention in visual search for parallel lines after focal cerebral lesions. J Clin Exp Neuropsychol 1989;11:319–31.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. McNair DM,
      2. Lorr M
      . An analysis of mood in neurotics. J Abnorm Soc Psychol 1964;69:620–7.
      OpenUrlCrossRefWeb of Science
    25. ↵
      1. Ganguli M,
      2. Ratcliff G,
      3. Dekosky ST
      . Cognitive test scores in community-based older adults with and without dementia. Aging Ment Health 1997;1:176–80.
      OpenUrlCrossRef
    26. ↵
      1. Gottesman RF,
      2. Kleinman JT,
      3. Davis C,
      4. et al
      . The NIHSS-plus: improving cognitive assessment with the NIHSS. Behav Neurol 2010;22:11–15.
      OpenUrlCrossRefPubMed
    27. ↵
      1. Cicerone KD,
      2. Langenbahn DM,
      3. Braden C,
      4. et al
      . Evidence-based cognitive rehabilitation: updated review of the literature from 2003 through 2008. Arch Phys Med Rehabil 2011;92:519–30.
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors All of the authors included in the manuscript fulfil the criteria for authorship. There is no one else who fulfils these criteria but has not been included as an author. TK conceptualised the study, undertook data collection, designed the statistical analysis and analysed the data and drafted and revised the draft paper. He is the guarantor. SL and KT undertook data collection and drafted and revised the manuscript. SM undertook data collection and revised the manuscript. PB undertook data collection and revised the manuscript. EP conceptualised the study, administrated the funding and drafted and revised the manuscript.

    • Funding This study was funded by The Finnish Social Insurance Institution's (KELA) Research Department, Lapland Hospital District's EVO fund and the Finnish Cultural Foundation's Lapland Regional fund.

    • Competing interests None.

    • Ethics approval The Ethics Committee of Helsinki University Central Hospital.

    • Provenance and peer review Not commissioned; externally peer reviewed.