You are here

Article Text

Article menu
Cerebrovascular disease
Research paper
Poststroke dementia is associated with recurrent ischaemic stroke
  1. Gerli Sibolt1,2,
  2. Sami Curtze1,2,
  3. Susanna Melkas1,2,
  4. Jukka Putaala1,2,
  5. Tarja Pohjasvaara1,2,
  6. Markku Kaste1,2,
  7. Pekka J Karhunen3,4,
  8. Niku K J Oksala3,4,5,
  9. Timo Erkinjuntti1,2
  1. 1Department of Neurology, Helsinki University Central Hospital, Helsinki, Finland
  2. 2Department of Neurological Sciences, University of Helsinki, Helsinki, Finland
  3. 3Forensic Medicine, School of Medicine, University of Tampere, Tampere, Finland
  4. 4The Laboratory Centre Research Unit, Tampere University Hospital, Tampere, Finland
  5. 5Division of Vascular Surgery, Department of Surgery, Tampere University Hospital, Tampere, Finland
  1. Correspondence to Dr Gerli Sibolt,  Department of Neurology, Helsinki University Central Hospital, P.O. Box 340 (Haartmaninkatu 4), Helsinki 00029 HUS, Finland; gerli.sibolt{at}hus.fi

Abstract

Objective To investigate whether poststroke dementia (PSD) diagnosed after ischaemic stroke predicts recurrent ischaemic stroke in long-term follow-up.

Methods We included 486 consecutive patients with ischaemic stroke (388 with first-ever stroke) admitted to Helsinki University Central Hospital who were followed-up for 12 years. Dementia was diagnosed in 115 patients using the Diagnostic and Statistical Manual of Mental Disorders, 3rd edition (DSM-III) criteria. The effects of risk factors and  PSD on survival free of recurrent stroke were estimated using Kaplan–Meier log-rank analyses, and the HRs for stroke recurrence were calculated using Cox proportional hazards models.

Results In the entire cohort, patients with PSD had a shorter mean time to recurrent stroke (7.13 years, 95% CI 6.20 to 8.06) than patients without dementia (9.41 years, 8.89 to 9.92; log rank p<0.001). This finding was replicated in patients with first-ever stroke (6.89 years, 5.85 to 7.93 vs 9.68 years, 9.12 to 10.24; p<0.001). In Cox univariate analysis, PSD was associated with increased risk for recurrent stroke both in the entire cohort (HR 2.02; 95% CI 1.47 to 2.77) and in those with first-ever stroke (2.40; 1.68 to 3.42). After adjustment for the significant covariates of age, atrial fibrillation, peripheral arterial disease and hypertension, PSD was associated with increased risk for recurrent stroke both in the entire cohort (1.84; 1.34 to 2.54) and in those with first-ever stroke (2.16; 1.51 to 3.10).

Conclusions Poststroke dementia predicts recurrence of ischaemic stroke in long-term follow-up and should be considered when estimating prognosis.

  • STROKE
  • DEMENTIA
  • VASCULAR DEMENTIA

https://doi.org/10.1136/jnnp-2012-304084

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

    Ischaemic stroke is a risk factor for cognitive impairment and dementia,1 ,2 and poststroke dementia (PSD) is associated with impaired survival.3 The prevalence of PSD is expected to increase in the future due to ageing of the population and improved poststroke survival.

    The potential association between PSD and recurrence of ischaemic stroke has been infrequently investigated. Only three previous studies have focused on the incidence of recurrent stroke in patients with PSD; the study of Moroney et al included 242 ischaemic stroke patients,4 that of Hénon et al 202 patients having either ischaemic or haemorrhagic stroke,5 and that of Loeb et al6108 patients with lacunar infarction. In the 3-year follow-up study of Hénon et al5 (prospective setting, patients’ mean age of 75 years), 110 of 142 survivors underwent evaluation for dementia at the first follow-up visit, where 29 patients had new-onset PSD. In the prospective study of Loeb et al6 (lacunar infarct patients, patients’ mean age of 65 years), 25 patients developed dementia during the 4-year follow-up.

    In the 5-year follow-up study of Moroney et al4 (prospective study, patients’ mean age of 72 years), 208 patients had at least one follow-up visit, and 62 had PSD. In all these studies, the association between PSD and stroke recurrence remains uncertain. Loeb et al6 and Moroney et al4 found PSD to predict recurrence of stroke. However, in the series of Hénon et al5 the presence of white matter lesions, but not PSD, predicted stroke recurrence.

    We hypothesised that the diagnosis of PSD based on a detailed clinical mental status examination could predict the recurrence of ischaemic stroke. Our hypothesis was tested in a well-characterised large ischaemic poststroke cohort with a 12-year follow-up that was controlled for major potential confounders.

    Methods

    Patients

    The consecutive Helsinki Stroke Ageing Memory (SAM) cohort consisted of 1622 patients with suspected stroke (all Finnish, Caucasian) admitted to Helsinki University Hospital between 1 December 1993 and 30 March 1995. Excluded were patients without ischaemic stroke (n=175), with intracerebral (n=229) or subarachnoid (n=69) haemorrhage. We further excluded those younger than 55 years (n=258) or older than 85 years (n=88), those not living in Helsinki (n=158), and those not speaking the Finnish language (n=3).

    Of the 642 patients initially meeting the inclusion criteria and invited to a follow-up visit 3 months later, 71 (11.1%) had died before the follow-up, 82 (12.8%) declined to participate, and 3 were lost to follow-up due to unknown cause. The final SAM cohort thus consisted of 486 patients, 388 of whom had their first-ever stroke.7 Twelve-year follow-up data were available for 446 of the 451 patients tested for the diagnosis of dementia.

    The study was approved by the ethics committee of the Department of Clinical Neurosciences, Helsinki University Central Hospital, Helsinki, Finland. Informed consent was obtained from all patients.

    Imaging

    All patients underwent brain CT at the acute phase; additionally, 396 patients (81.5%) underwent MRI at 3 months. White matter lesions were rated in accordance with the Leukoaraiosis and Disability in the Elderly (LADIS) rating as none to mild, moderate or severe.8

    Clinical evaluation

    Medical and neurological history was taken as described previously.2 Patients’ educational status was dichotomised as follows: low (0–6 years of formal education) and high (>6 years of formal education). Smoking habits were categorised on admission as non-smoking and smoking (current or former). History of myocardial infarction, cardiac failure, atrial fibrillation, arterial hypertension, peripheral arterial disease (PAD) and diabetes was investigated by reviewing all available hospital charts, in addition to a structured interview of the subject and a knowledgeable informant. History of hypertension was defined at the time of study inclusion as systolic blood pressure ≥160 mm Hg and/or diastolic blood pressure ≥95 mm Hg. Diabetes was defined as previously documented diagnosis, current use of insulin or oral antidiabetic medication, or fasting blood glucose >7.0 mmol/l.

    Clinical mental status examination assessing the main neuropsychological domains (attention; orientation; short-term memory; long-term memory; executive functions, including abstract thinking, judgement and problem solving; aphasia; apraxia; agnosia; motor control; constructional and visuospatial abilities; and personality change) was done as described elsewhere.2 Dementia was diagnosed in 115 patients (25.5%) using the Diagnostic and Statistical Manual of Mental Disorders, 3rd edition (DSM-III) criteria. Cognitive impairment without dementia (CIND) was defined as cognitive impairment in any of the above-mentioned domains after the exclusion of patients with dementia as described previously.9 Assessment of prestroke cognitive decline involved interviews with the patient and a knowledgeable informant. The interviews included structured questions on abnormality in the cognitive domains as well as assessment of social functions before the index stroke. We enquired about the start date of symptoms and their duration, focusing especially on the 1-year period preceding the index stroke. On the basis of all available history from the patient and the knowledgeable informant, a board-certified neurologist independently judged whether the patient had had prestroke cognitive decline or not. Patients with prestroke cognitive decline included those with borderline and definite dementia.

    Patients were followed-up till 21 September 2006 using extensive national registers kept by the National Institute for Health and Welfare. The national ward register contains ICD-9 and ICD-10 diagnosis codes of all hospital treatment periods. ICD-9 codes 433 and 434 and ICD-10 codes I63.0–I63.9 were considered as ischaemic stroke. Survival data and causes of death were obtained from Statistics Finland. Death certificate data were reviewed by a forensic pathology specialist.3 Patients who had died from brain infarction without registered hospital treatment were considered to have had a fatal recurrent stroke.

    Data analysis and statistics

    Pearson χ2 test, or Mantel–Haenszel test, where appropriate (dichotomous variables), one-way analysis of variance (continuous variables), and binary logistic regression function were used to analyse associations between dementia, demographics and risk factors. Levene's test was used to assess the equality of variance of age. Kaplan–Meier log-rank analysis was used to evaluate the impact of PSD on the time to first recurrent stroke. The cumulative recurrence risks and their 95% CIs were calculated by using life-table function. Patients who died from causes other than recurrent ischaemic stroke were censored. The cumulative hazard function was plotted, and we checked that the proportional hazards assumption was met for each parameter included in further models.

    Variable selection for multivariable regression was done by checking the predictive value of confounders used in previous studies (age, gender, education, smoking status, atrial fibrillation,  PAD, hypertension and diabetes) for recurrent stroke using the log-rank test with p<0.2 as a selection criterion. We also analysed white matter lesions dichotomised as none to mild, versus moderate to severe, due to their association with small vessel disease.10 A multivariable model with forced entry was used to check the predictive value of these variables for PSD because variables predicting PSD might cause a collinearity problem towards recurrent stroke. A multivariate Cox regression proportional hazards analysis with forced entry of significant confounders was used. Statistical significance was set at p<0.05.

    All statistical analyses were performed using SPSS Statistics 20 for Linux (IBM Corp., Armonk,  New York, USA).

    Results

    Demographics and risk factors for the entire population are presented in table 1, and for first-ever stroke patients in online  supplementary data 1. During the 3022 person-years of follow-up, 201 patients (41.4%) suffered from at least one non-fatal or fatal recurrent ischaemic stroke. After 12 years of follow-up, 132 patients of the whole cohort (27.2%), but only 15 (13%) in the PSD group were alive, and only 4 (3.5%) had survived a recurrent stroke. Of the 117 non-demented survivors, 26 (22.2%) had survived at least one recurrent stroke. Median follow-up until recurrent stroke or the end of the study was 5.4 years (IQR 2.4–11.0). Among all patients, and in the first-ever stroke group, the mean age of non-demented patients was lower than that of demented patients (70.4 vs 72.9 years; p<0.01 and 70.3 vs73.2 years; p<0.01, respectively). Stroke aetiology and recurrence rates for demented and non-demented patients are provided in online  supplementary data 2.

    View this table:
    Table 1

    Demographics and risk factors of 486 consecutive patients with ischaemic stroke in the Helsinki Stroke Ageing Memory study cohort

    In the whole series, PSD patients had a shorter mean time to recurrent stroke (7.13 years; 95% CI 6.20 to 8.06) than patients without dementia (9.41; 8.89 to 9.92) in Kaplan–Meier log-rank analysis (figure 1).

    Figure 1
    Figure 1

    Cumulative recurrence risk of non-fatal or fatal ischaemic stroke in 446 patients of the Helsinki Stroke Ageing Memory study cohort followed up for 12 years. Stratified for patients diagnosed with poststroke dementia and for patients not fulfilling dementia criteria 3 months after the stroke.

    When the analyses were limited to patients with first-ever stroke, the time to recurrent stroke was 6.89 (5.85 to 7.93) years for patients with PSD and 9.68 (9.12 to 10.24) years for non-demented patients. Recurrent stroke occurred significantly earlier for patients suffering from atrial fibrillation (p=0.02), PAD (p=0.03), and cognitive impairment before stroke (p<0.01).

    In Cox univariate analysis, PSD was associated with increased risk for recurrent stroke both in the entire cohort (HR 2.02; 1.47 to 2.77) and in those with first-ever stroke (2.40; 1.68 to 3.42). CIND, compared with non-demented poststroke patients, did not significantly predict recurrent stroke (1.49; 0.92 to 2.41). Variable selection for multivariable regression was done by checking the predictive value of plausible confounders (age, gender, smoking status, atrial fibrillation, PAD, hypertension, diabetes and white matter lesions) for recurrent stroke using the log-rank test with p<0.2 as a selection criterion. Gender, smoking status and diabetes did not predict recurrent stroke and were therefore excluded from the model. Low education and cognitive impairment before the incident stroke were associated significantly with PSD using Mantel–Haenszel common OR estimate and in a multivariable logistic regression model (table 2), and were therefore excluded from the Cox regression model due to potential collinearity. As white matter lesions are strongly associated with dementia,11 collinearity might be suspected with PSD diagnosis, and therefore, three different Cox models were created, separately for PSD and white matter lesions, and including both PSD and white matter lesions.

    View this table:
    Table 2

    Association between potential covariates and diagnosis of poststroke dementia in the Helsinki Stroke Ageing Memory study cohort (n=486)

    In Cox multivariable regression adjusting for age, atrial fibrillation, PAD and hypertension, only age (1.07 per year; 1.04 to 1.09) and PSD (1.84; 1.34 to 2.54) were associated with increased risk for recurrent stroke in the entire cohort. In first-ever stroke patients, age (1.06 per year; 1.03 to 1.08) and PSD (2.16; 1.51 to 3.10) were associated with recurrent stroke (table 3). CIND as compared with non-demented stroke patients did not predict recurrent stroke (1.42; 0.87 to 2.32) after Cox multivariable adjustment.

    View this table:
    Table 3

    Association between potential covariates and recurrence of fatal or non-fatal ischaemic stroke in the Helsinki Stroke Ageing Memory study cohort (n=486) in the 12-year follow-up

    Discussion

    Our study showed that PSD diagnosed 3 months after index ischaemic stroke is a significant independent predictor of stroke recurrence during a 12-year follow-up, particularly in patients with first-ever stroke.

    This result is in line with Moroney et al,4 who investigated the effect of PSD on the recurrence of stroke, although they reported a higher HR (2.71; 1.36 to 5.42) than we found (1.84; 1.34 to 2.54). A lower HR might be due to a higher percentage of non-stroke death causes with a longer follow-up. The CI in our study is narrower than that of previous studies, reflecting higher precision of the HR due to larger cohort size. Hénon et al,5 by contrast, found that white matter lesions were a better predictor than PSD for stroke recurrence (1.70; 1.23 to 2.36 vs1.25; 0.47 to 3.33). White matter lesions are known to predict cognitive decline and dementia,11 and we have previously shown in the same cohort that severe white matter lesions are related to stroke recurrence for up to 5 years poststroke.12 Poststroke dementia and white matter lesions are both correlated with cerebral small vessel disease,13 and the present study adds to existing knowledge regarding the clinical importance and prognostic value of small vessel disease from the cognitive point of view. As small vessel disease has a grimmer prognosis than large artery occlusion,14 a comparison of small vessel disease and other causes of stroke would be interesting. In our cohort, we performed classification of the stroke causes consistent with those used in the Trial of Org 10172 in Acute Stroke Treatment (TOAST),15 but since over 60% of stroke causes were not clearly due to small vessel, large artery or cardioembolic aetiology, the analyses would not have been of additional value.

    Not only dementia, but also CIND, is known to predict mortality in this poststroke cohort,9 but for stroke recurrence, CIND was not a significant predictor, although there was a clear trend. This is in line with another study investigating the recurrence of vascular events in poststroke patients with CIND.16 The lack of significance might be simply due to a lack of power as a result of missing data for certain neuropsychological domains. Moreover, false positives may exist as stroke deficits might cause impairment in one of the main neuropsychological domains.

    A major strength of our study is its extensive 12-year follow-up, compared with the follow-ups of 3–5 years in previous studies,4–6 and the cohort of 446 patients compared with 202,5 2424 and 1086 patients.

    Relative to the previous studies,4–6 our study has longer person-years of follow-up (3022 vs 647), more recurrent ischaemic strokes (201 vs 45), and more PSD patients (115 vs 62).

    Demographics in our study were similar to previous studies, with a median age of 72 years and equal distribution of genders. Other strengths include clinical evaluation of PSD and reliability of follow-up diagnoses. Stroke diagnoses in Finnish Hospital Discharge Registries have been validated against a population-based stroke registry with fairly good positive predictive values (85–92%).17

    A limitation of our study is that we cannot elaborate the recurrence risk for different stroke aetiologies, but only for ischaemic stroke in general. Additionally, there might be selection bias as cardiogenic strokes tend to be more severe and more often fatal in the first month, so that stroke patients with cardiogenic strokes might have died before the first evaluation at 3-month poststroke, or they were in too bad a condition to participate in the study. Another limitation is that we have only one clinical evaluation at 3 months, and no data regarding whether patients initially without dementia eventually develop dementia during the follow-up. However, a 3-month poststroke assessment is realistic in clinical practice and is standard in major studies on PSD,18 although it has been noted that some patients may still be recovering from cognitive dysfunction.19 Nevertheless, we were able to show that testing for dementia 3 months after the index stroke can be useful for predicting the recurrence of stroke. Thus far, it remains unclear whether the association between PSD and recurrence of stroke is due to a biological link, or to failing secondary prevention of patients with PSD. Suffering from cognitive impairment is known to weaken adherence to medication.20 On the other hand, an increased number of infarcts have been found in patients with Alzheimer disease compared with controls,21 and evidence suggests a link between cerebral small vessel disease and Alzheimer disease.22 We did not distinguish Alzheimer disease from other forms of dementia in this study. For the same cohort, we have been able to show before, that risk factors for Alzheimer disease, as cognitive impairment before stroke, and low education, are associated with poststroke dementia.3 ,23 Thus, it is possible that some patients progressed to Alzheimer disease after the stroke.

    Poststroke dementia generally, and especially after the first-ever stroke, is a strong risk factor for recurrent stroke. Therefore, detailed clinical and even neuropsychological evaluations after stroke should be considered to estimate prognosis and facilitate secondary prevention.

    References

    1. ↵
      1. Leys D,
      2. Hénon H,
      3. Mackowiak-Cordoliani M-A,
      4. et al
      . Poststroke dementia. Lancet Neurol 2005;4:752–9.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Melkas S,
      2. Laurila JV,
      3. Vataja R,
      4. et al
      . Post-stroke delirium in relation to dementia and long-term mortality. Int J Geriatr Psychiatry 2012;27:401–8.
      OpenUrlPubMed
    3. ↵
      1. Melkas S,
      2. Oksala NKJ,
      3. Jokinen H,
      4. et al
      . Poststroke dementia predicts poor survival in long-term follow-up: influence of prestroke cognitive decline and previous stroke. J Neurol Neurosurg Psychiatr 2009;80:865–70.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Moroney JT,
      2. Bagiella E,
      3. Tatemichi TK,
      4. et al
      . Dementia after stroke increases the risk of long-term stroke recurrence. Neurology 1997;48:1317–25.
      OpenUrlCrossRef
    5. ↵
      1. Hénon H,
      2. Vroylandt P,
      3. Durieu I,
      4. et al
      . Leukoaraiosis more than dementia is a predictor of stroke recurrence. Stroke 2003;34:2935–40.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Loeb C,
      2. Gandolfo C,
      3. Croce R,
      4. et al
      . Dementia associated with lacunar infarction. Stroke 1992;23:1225–9.
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Pohjasvaara T,
      2. Erkinjuntti T,
      3. Vataja R,
      4. et al
      . Comparison of stroke features and disability in daily life in patients with ischemic stroke aged 55 to 70 and 71 to 85 years. Stroke 1997;28:729–35.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Pantoni L,
      2. Basile AM,
      3. Pracucci G,
      4. et al
      . Impact of age-related cerebral white matter changes on the transition to disability—the LADIS study: rationale, design and methodology. Neuroepidemiology 2005;24:51–62.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Oksala NK,
      2. Jokinen H,
      3. Melkas S,
      4. et al
      . Cognitive impairment predicts poststroke death in long-term follow-up. J Neurol Neurosurg Psychiatry 2009;80:1230–5.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Oksala NK,
      2. Oksala A,
      3. Pohjasvaara T,
      4. et al
      . Age related white matter changes predict stroke death in long term follow-up. J Neurol Neurosurg Psychiatry 2009;80:762–6.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Jokinen H,
      2. Kalska H,
      3. Ylikoski R,
      4. et al
      . Longitudinal cognitive decline in subcortical ischemic vascular disease—the LADIS Study. Cerebrovasc Dis 2009;27:384–91.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Melkas S,
      2. Sibolt G,
      3. Oksala NKJ,
      4. et al
      . Extensive white matter changes predict stroke recurrence up to five years after first-ever ischemic stroke. Cerebrovasc Dis 2012;34:191–8.
      OpenUrlCrossRefPubMed
    13. ↵
      Anon. 2001–2011: a decade of the LADIS (Leukoaraiosis And DISability) Study: what have we learned about white matter changes and small-vessel disease? Cerebrovasc Dis 2011;32:577–88.
      OpenUrlCrossRefPubMed
    14. ↵
      1. Melkas S,
      2. Putaala J,
      3. Oksala NK,
      4. et al
      . Small-vessel disease relates to poor poststroke survival in a 12-year follow-up. Neurology 2011;76:734–9.
      OpenUrlCrossRef
    15. ↵
      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
    16. ↵
      1. Narasimhalu K,
      2. Ang S,
      3. De Silva DA,
      4. et al
      . The prognostic effects of poststroke cognitive impairment no dementia and domain-specific cognitive impairments in nondisabled ischemic stroke patients. Stroke 2011;42:883–8.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Tolonen H,
      2. Salomaa V,
      3. Torppa J,
      4. et al
      . The validation of the Finnish Hospital Discharge Register and Causes of Death Register data on stroke diagnoses. Eur J Cardiovasc Prev Rehabil 2007;14:380–5.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Tatemichi TK,
      2. Paik M,
      3. Bagiella E,
      4. et al
      . Dementia after stroke is a predictor of long-term survival. Stroke 1994;25:1915–19.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Ivan CS,
      2. Seshadri S,
      3. Beiser A,
      4. et al
      . Dementia after stroke. Stroke 2004;35:1264–8.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Campbell NL,
      2. Boustani MA,
      3. Skopelja EN,
      4. et al
      . Medication adherence in older adults with cognitive impairment: a systematic evidence-based review. Am J Geriatr Pharmacother 2012;10:165–77.
      OpenUrlCrossRefPubMed
    21. ↵
      1. Suter OC,
      2. Sunthorn T,
      3. Kraftsik R,
      4. et al
      . Cerebral hypoperfusion generates cortical watershed microinfarcts in Alzheimer disease. Stroke 2002;33:1986–92.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Utter S,
      2. Tamboli IY,
      3. Walter J,
      4. et al
      . Cerebral small vessel disease-induced apolipoprotein E leakage is associated with Alzheimer disease and the accumulation of amyloid beta-protein in perivascular astrocytes. J Neuropathol Exp Neurol 2008;67:842–56.
      OpenUrlCrossRefPubMed
    23. ↵
      1. Ojala-Oksala J,
      2. Jokinen H,
      3. Kopsi V,
      4. et al
      . Educational history is an independent predictor of cognitive deficits and long-term survival in postacute patients with mild to moderate ischemic stroke. Stroke 2012;43:2931–5.
      OpenUrlAbstract/FREE Full Text

    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 GS and SC conceived the study, drafted the manuscript, and performed the statistical analyses. TE conceived the study and obtained study funding. All authors interpreted the data and edited the manuscript for intellectual content.

    • Funding This study was supported by the Clinical Research Institute, Helsinki University Central Hospital, and the Medical Research Fund of Helsinki University Central Hospital.

    • Competing interests None.

    • Ethics approval Ethics committee of the Department of Clinical Neurosciences, Helsinki, Finland.

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