You are here

Article Text

Article menu

Download PDFPDF

Cerebrovascular disease
Research paper
Impact of stroke-associated infection on long-term survival: a cohort study
  1. Joseph Kwan1,
  2. Ruth Mary Pickering2,
  3. Dorit Kunkel3,
  4. Carolyn Fitton3,
  5. Damian Jenkinson1,
  6. V Hugh Perry4,
  7. Ann M Ashburn3,
  8. on behalf of the Stroke Association Rehabilitation Research Centre
  1. 1The Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust, Bournemouth, UK
  2. 2Public Health Sciences and Medical Statistics, University of Southampton, Southampton, UK
  3. 3University of Southampton, Southampton, UK
  4. 4Centre for Biological Sciences, University of Southampton, Southampton, UK
  1. Correspondence to Professor Ann M Ashburn, University of Southampton, Southampton General Hospital, Mailpoint 886, Southampton, Hampshire SO16 6YD, UK; ann{at}soton.ac.uk

Abstract

Background and objective The effects of stroke-associated infection (SAI) on long-term survival are unclear. We performed a prospective evaluation to explore risk factors of SAI, and compared survival status over the 3 years following stroke onset between those who experienced SAI and those who did not.

Methods Consecutive patients with acute stroke admitted to a stroke unit between April 2005 and December 2006 were invited to participate. We prospectively collected data on demographics, pathological and clinical stroke subtype, stroke severity, and neurological and functional consequences, and abstracted additional data on occurrence and timing of SAI in hospital from medical notes. Survival status 3 years after stroke onset was obtained.

Results We recruited 413 acute stroke patients, 161 (39%) experienced SAI. After excluding patients with infection at onset, patients with intracerebral haemorrhage (p=0.014), dysphagia (p=0.003) and urinary incontinence/catheterisation (p=0.000) were at higher risk of infection after controlling for case mix. The risk of death in hospital was greater following an SAI (HR 3.56; 95% CI 1.94 to 6.53; p=0.000), as was risk of death calculated over the whole 3-year follow-up period among those acquiring SAI within 2 weeks of onset (HR 1.66; 95% CI 1.14 to 2.40; p=0.031).

Conclusions SAIs have long-lasting effects on patient survival. This serves to emphasise the importance of immediate access to organised stroke unit care for people with acute stroke, with active physiological monitoring and protocols for early detection and treatment of SAIs.

  • Stroke
  • Infectious Diseases
  • Epidemiology

https://doi.org/10.1136/jnnp-2012-302552

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

    Infections are the most common complication during the first few weeks after a stroke, occurring in an estimated 30% of acute stroke patients included in a recent systematic review and meta-analysis.1 Pneumonia alone occurs in 7–22% of strokes,2 and in one study, 27% of patients with post-stroke bacterial pneumonia died within 30 days compared to only 4% of non-infected patients.3 In another study,4 patients who experienced post-stroke infections had approaching three times the odds of dying within the first 5 days compared to those who did not experience infection. Importantly, this association with poor short-term outcome was independent of case mix, stroke severity, and adverse physiological disturbances including fever, hypoxia and hyperglycaemia, which also independently affected outcome.4

    Although it is recognised that post-stroke infections can lead to poorer short-term survival and outcome, at least during the first 3 months after stroke,2 ,5–8 very few studies have explored their influence on longer-term survival and outcome.2 ,9 In one longitudinal study, post-stroke infection was an independent predictor of functional outcome (measured using the Barthel index and modified Rankin score) at final follow-up on average 15 months post-stroke.9 Pre-stroke bacterial infections may also influence stroke severity and outcome but the precise relationship remains unclear8; some animal studies have even demonstrated that pre-stroke endotoxin exposure may be neuroprotective.10 ,11 In this large-scale longitudinal study, we sought to determine risk factors for stroke-associated infection (SAI) and the impact of SAI on survival during the first 3 years post-stroke.

    Methods

    Ethics approval was granted through the Dorset NHS Local Research Ethics Committee (ethics approval no. 04/Q2201/91). Patients diagnosed with stroke and consecutively admitted to the Royal Bournemouth and Christchurch NHS Foundation Trust acute stroke and stroke rehabilitation units were eligible for recruitment between April 2005 and December 2006. The stroke unit is situated within a district general hospital, with 650 new stroke admissions per year. Potential participants were recruited as soon as possible after admission to the unit and outlying wards. We introduced a two-stage procedure of consent and assent. Explicit written informed consent was sought as soon as a potential participant was well enough, and capable of receiving and understanding the study information which had been left with them for at least one day. If the patient had been discharged when the researcher returned to get consent, the study information detailed how to contact the research team to arrange a home visit. Study information was left by the bedside of those deemed too ill to consent, asking next of kin for assent to retain information from medical records for research purposes. Patients for whom assent was attained were approached later for consent if they recovered sufficiently while still in hospital. We were thus able to use information from the medical records for both those who consented, and those for whom assent was obtained. Patients were under the care of specialists in the stroke unit and managed according to unit guidelines which were based on the then current versions of the Royal College of Physicians guidelines12 and National Institute for Health and Clinical Excellence (NICE) quality standards for stroke.13 The research team had no direct involvement in the daily clinical care of study participants. Details of recruitment to the study have been previously described; we enrolled 87% of eligible patients to the study.14

    Demographic information, stroke severity (Oxford Community Stroke Project (OCSP) classification of ischaemic stroke,15 Glasgow Coma Score (GCS),16 pre-stroke and admission functional status (Functional Ambulation Score, FAS),17 and neurological consequences on admission including dysphagia and continence status, were abstracted from medical notes prospectively for most patients, and checked prior to analysis in some cases. Survival status 3 years post-stroke onset was established (from case notes, electronic databases and attempting to make contact with consented participants) for all but 10 participants who were treated as censored at the last date they were known to be alive (discharge for three participants, and up to 27 months post-onset for the remainder).

    Infection data

    SAI is an infection associated with the initial stroke event. In the literature, its definition has varied across studies, from infections occurring up to 3 days,18 up to 7 days,19 ,20 and up to 14 days21 after stroke onset. We defined SAI to include pre-stroke infection occurring up to 7 days before onset, as well as infections occurring any time during the initial hospital admission. Pre-stroke infections shortly before onset, and later in-hospital infections were included because they could impact on the initial recovery process and thus clinical outcome during follow-up. Moreover, any infections found to be present on the day of stroke onset might in fact have begun several days before the stroke onset. SAIs were categorised as those occurring: (a) up to 7 days before and including the day of onset; (b) day 1 to the end of the first 2 weeks post-onset; and (c) after the first 2 weeks post-onset and up to discharge. The definitions of specific infections were similar to those used in previously published studies.4 Urinary tract infection (UTI) was defined as the presence of relevant clinical symptoms and/or signs such as dysuria, urinary frequency with positive microbiological cultures, or negative cultures with leucocytosis (>11×109/l), fever (temperature ≥37.5°C) or both. Pneumonia was defined as the presence of relevant clinical symptoms and/or signs (such as purulent cough, unilateral inspiratory crackles, bronchial breath sounds) with at least one of: leucocytosis, fever, or a positive chest radiograph. Lower respiratory tract infection (LRTI) differed from pneumonia by an absence of abnormality on the chest radiograph. Cellulitis was defined as a spreading bacterial infection just below the skin.

    Data on SAIs were abstracted from case notes by a consultant stroke specialist. Every patient was examined for signs and symptoms of SAIs daily as part of the routine ward round by the stroke team, hence all infections were prospectively identified, investigated, treated and recorded systematically and according to unit guidelines. Data on infections occurring prior to stroke were abstracted from the clinical history, information provided by the general practitioner (or hospital computer system), and drug history, particularly the use of antibiotics (type, dosage, timing).

    Statistical analysis

    Admission characteristics of the groups according to timing of infection were summarised using descriptive statistics. HRs of hospital acquired infection among those without infection at onset were examined in Cox regression models, with participants with no infection treated as censored at death or discharge (whichever came first). The cumulative incidence of hospital acquired infection estimated separately for each OCSP subtype and primary intracerebral haemorrhage (PICH) in the presence of discharge or death in hospital, was obtained post-estimating a competing risks model using the command ‘stcrreg’ in Stata.22 All HRs are presented with 95% CIs and likelihood ratio tests were performed for each term uncontrolled and controlled for age, gender, history of stroke, pre-stroke FAS, living status, OCSP subtype, dysphagia, incontinence, inability to mobilise or lift both arms, and abnormal GCS verbal component at admission.23 The relationship between infection status and survival up to 3 years post-onset was examined in proportional hazards Cox models controlling for the same factors, with withdrawals and participants alive at last contact treated as censored. The proportionality assumption was examined by fitting separate models for each of the 3 years of follow-up. Kaplan–Meier survival curves for groups defined by infection status were drawn for the whole 3-year follow-up period and separately for each year, and log rank tests were carried out. Risk of death associated with infection on or before onset was examined in a Cox model censored at the time of discharge, and the model was expanded with a time dependent infection covariate (which took the value 0 before infection but switched and remained equal to 1 after an infection was recorded). Because of the smaller number of in-hospital deaths, these latter models were controlled for a reduced number of factors.

    Results

    Of the 416 participants recruited to the study, the medical records of 413 were reviewed for infection status and were included in the analysis (case notes were unavailable for three people). Participants stayed on average 44 days in hospital (median 33, min–max 1–197 days). Among the 347 who survived to discharge, average length of stay was longer (mean 45, median 33, min–max 1–70 days) than for the 66 who died in hospital, but the longest lengths of stay were in this latter group (mean 41, median 31, min–max 3–197 days) (mean 41, median 31, min-max 3-197 days). Date of discharge from (or death in) hospital was known for all participants. After discharge 115 participants died during the 3-year follow-up period; 47 participants withdrew or were alive at last contact; and 185 were alive at 3 years. The 47 who were not followed up until either death or 3 years withdrew for the following reasons: medically unwell (n=3, median follow-up 31 months); moved away from the area (n=14, median follow-up 15 months); patient decision with no other reason recorded (n=20, median follow-up 6 months); and patients for whom we obtained assent to review medical records from a relative but who did not consent to be contacted post-discharge (n=10, median follow-up 10 months).

    A total of 161 (39%) participants experienced at least one infection while in hospital—36 of first infections (22%) occurred on or up to 7 days before stroke onset, 66 (41%) during the first 2 weeks following stroke onset, and 59 (37%) after the first 2 weeks. Table 1 summarises the demographic characteristics and clinical characteristics on admission for the groups defined according to timing of first infection. We found that those who survived to discharge without infection were younger (mean age 76 vs 81 years), more likely to be independent in ambulation pre-stroke (80% vs 64%), and less likely to have a severe OCSP total anterior circulation infarct (TACI) subtype stroke (8% vs 22%) compared to patients acquiring infection during the first 2 weeks post-stroke onset (table 1). The groups experiencing infection in hospital had high rates of dysphagia (71% and 73%), urinary incontinence (56% and 54%) and low Glasgow Coma scores on admission, and thus represent a more severe group than those surviving to discharge. Among patients with ischaemic stroke, pneumonia was the commonest infection type, occurring on or up to 7 days before stroke onset, whereas UTI was the commonest infection type after the first day of stroke onset (table 2). Among patients with PICH, pneumonia and UTI were the commonest infection types but numbers were small. In total, 20 of the 36 (56%) infections that occurred on or before onset were pneumonia, compared to 39 of the 125 (31%) that occurred during hospitalisation (OR 2.76; 95% CI 1.20 to 6.33; p=0.0144).

    View this table:
    Table 1

    Demographic and clinical characteristics (on admission) of the different groups defined by timing of first infection (n=413)

    View this table:
    Table 2

    Type of first infection according to timing of occurrence, for groups with infarct and PICH separately (n=161 infections)

    Predictors of post-stroke infection

    Table 3 shows the predictors of acquiring a first post-stroke infection in hospital among those without infection at stroke onset (n=377). In uncontrolled analysis, patients with dysphagia and urinary incontinence on admission were at highest risk (p=0.000 for both factors); a higher risk of infection was also experienced by older patients, those more dependent in ambulation pre-stroke, and those with abnormal GCS verbal score. The OCSP subtype TACI, and patients with PICH (uncontrolled HR 1.65 compared to all ischaemic strokes combined, p=0.031) were also at significantly increased risk of infection. After controlling for the other factors, only PICH, dysphagia (p=0.003) and urinary incontinence on admission (p=0.000) remained independent predictors. The PICH group had greater risk of infection than the TACI subtype, and significantly higher risk compared to the infarct subtypes combined (controlled HR=1.81, p=0.014). The influence of OCSP subtype and PICH on the cumulative incidence of infection in the first 8 weeks post-stroke onset can be seen in figure 1. The higher cumulative incidence of infection in the TACI subtype and PICH group reflect the higher risk of infection in these groups, but also their longer in-hospital stays.

    View this table:
    Table 3

    HRs for first infection from the Cox regression model

    Figure 1
    Figure 1

    Cumulative incidence of infection in hospital during the first 8 weeks post-stroke onset by Oxford Community Stroke Project (OCSP) subtype and primary intracerebral haemorrhage (PICH) (n=375). LACI, lacunar infarct; PACI, partial anterior circulation infarct; POCI, posterior circulation infarct; TACI, total anterior circulation infarct.

    SAI and in-hospital mortality

    Infection occurring on or up to 7 days before onset was associated with a slightly higher risk of mortality before discharge (uncontrolled HR=1.46, p=0.272) but this was not statistically significant (see table 4), and after controlling for age, pre-stroke FAS or OCSP subtype TACI there was no association with mortality. The results from the time dependent analysis, which allows the risk of in-hospital mortality to change as a study participant's infection status changed from no infection to infection, are also shown in table 4. The risk following infection was significantly higher in both the uncontrolled and controlled models (controlled HR=3.56; 95% CI 1.94 to 6.53; p=0.000).

    View this table:
    Table 4

    HRs for in-hospital mortality associated with infection

    SAI and long-term mortality

    Risk of death over the 3-year follow-up period associated with infection status and factors known at stroke onset were estimated in a model assuming constant risk over follow-up (table 5). A gradient in risk was found with the highest risk in those with infection before, or up to 2 weeks post-onset (uncontrolled HR 2.93; 95% CI 2.12 to 4.05, p=0.000) compared to those without infection. The impact of these higher risks on survival throughout the 3-year follow-up period is shown in the Kaplan–Meier plot in figure 2A. Since the groups with infection were shown to be more severe on admission than the majority of patients who survived to discharge (table 1), we also examined their risk after controlling for the demographic characteristics and measures of severity in table 5. The groups with infection within 2 weeks of onset remained at significantly higher risk (controlled HR 1.66: 95% CI 1.44 to 2.40; p=0.031). Older age, male gender, pre-stroke FAS, OCSP subtype TACI and dysphagia on admission were also found to be independent risk factors. Although PICH was shown to be an independent risk factor for infection (table 3), this group was not at particularly high risk in either uncontrolled or controlled analysis of long-term mortality. The assumption of constant risk associated with timing of infection over the entire follow-up period underlying the analyses of table 5 was examined by estimating the model separately for each year of follow-up (table 6). The risk of death associated with infection acquired up to 2 weeks post-stroke was high during the first year post-onset, but during the second and third years, the risk was lower and not statistically significant. This pattern can be seen in the Kaplan–Meier curves specific to each year of follow-up (figure 2B–D).

    View this table:
    Table 5

    HRs from Cox models fitted to mortality over 3 years

    View this table:
    Table 6

    HRs for mortality from Cox models including timing of infection, fitted to each year of follow-up separately

    Figure 2
    Figure 2

    Kaplan–Meier survival curves over 3 years post-onset for groups defined by time of infection (n=413), and separately for each year of follow-up.

    Discussion

    This is the first prospective study of the impact of SAI on the long-term survival of people with stroke over a 3-year period, with very few participants lost to causes other than death. Patients with hospital-acquired infection were at significantly higher risk of mortality while in hospital and during the first year of follow-up post-onset, and this risk was independent of their profile with respect to age, pre-stroke walking ability and Oxford subtype of stroke, or dysphagia, incontinence, ability to mobilise or talk on admission. The impact on the proportion surviving could still be seen at the end of the 3-year follow-up period, though risks experienced during the second year among those surviving the first year, or during the third year among those surviving the second, were not higher amongst those with SAI. Approaching 70% of patients who remained infection-free in hospital were still alive 3 years post-stroke onset, compared to about 30% of those who developed infection either prior to their stroke or up to 2 weeks later. The group acquiring infection 2 weeks or more after onset had intermediate survival of about 50% at 3 years. Like others, we found urinary incontinence and dysphagia to be independent predictors of SAI. Unlike other researchers we found those suffering a haemorrhage type stroke to be more likely to acquire SAI than those with ischaemic stroke.

    To a large extent our sample of stroke patients and the care they received, is representative of the population of patients in stroke units in the UK. Care was provided according to the Royal College of Physicians guidelines12 and NICE quality standards.13 Data submitted by the unit to the latest National Sentinel Stroke Audit 2010,24 compared with national averages, show the mean age of female patients in the unit to be 80 years (national mean 79 years), and that of male patients to be 77 years (national mean 73 years). Moreover, 68% of patients in the unit were independent pre-stroke (national rate 76%), hence patients admitted to the unit appear to be slightly older and more dependent than those in other stroke units in the UK, and this may in part account for the longer mean length of stay (44 days) in the unit compared to the national average (24 days). Other factors that influence length of stay include clinical and administrative factors, social services provision, and primary healthcare structures and processes.

    There are some limitations to our study. We only recorded information on infections acquired during (or shortly before) the acute hospital stay. Information on infection was obtained by retrospective review of medical records rather than prospectively, but this was carried out by a senior consultant stroke physician. We did not collect a specific measure of stroke severity such as the National Institutes of Health (NIH) stroke scale,25 but did have symptoms at admission (age, the verbal component of the GCS, ability to lift both arms and mobilise unaided) recommended by Counsell et al 23 as predictors of outcome 30 days and 6 months after stroke, along with two further symptoms linked to severity, dysphagia and incontinence. It is possible that some of the factors showing independent risk of infection (table 3), or death (tables 4 ⇑–6) in controlled models, are reflecting poor outcome associated with other aspects of severity not captured by the controlling factors included in our models.

    We have demonstrated a relationship between SAI and survival that was detectable 3 years after stroke onset, and was independent of demographic characteristics of participants and symptoms at admission. Like Vargas et al 18 we found those with lacunar infarcts—by definition, small, deep, cerebral infarction caused by small vessel occlusion—were at lowest risk of acquiring infection in hospital among those without infection on admission, although the reduction in risk in our study was not as extreme as that reported by those authors. An unanticipated finding was that the risk of developing infection was greater in participants with intracerebral haemorrhage than in any of the Oxford classification subtypes of infarct. Vargas et al also examined infection in the group of patients with haemorrhagic stroke but did not find them to be at substantially higher risk of infection. Importantly, the risk of infection we found remained high and statistically significant after controlling for dysphagia, incontinence, age, and the verbal component of the GCS. Speculating as to possible explanations for this finding, it could be that intracerebral haemorrhage is associated with inherently different inflammatory processes from ischaemic stroke.26 ,27 A second possibility is that the difference in treatment as stated in published stroke guidelines12 ,13—such as avoidance of antiplatelet and cholesterol-lowering medications—predisposes people with haemorrhagic stroke to higher risk of infection; another possibility is that the causal route is in the opposite direction and that people with a predisposition to infection are more likely to have haemorrhagic stroke. We have found no published evidence in relation to any of these conjectures. Finally, there is the possibility of a type I error, that is, that there is no underlying difference in infection rate between the two types of stroke (uncontrolled p=0.031, controlled p=0.014), and the finding remains to be replicated in data from different settings.

    In conclusion, we have demonstrated that SAIs have a long-lasting impact on patient survival. This serves to emphasise the importance of immediate access to acute stroke units for people with acute stroke: organised stroke unit care with protocols for active physiological monitoring and early detection and treatment of SAIs has been shown to improve the delivery of care and reduce rates of complications including chest and other infections.28 In England and Wales, the National Sentinel Audit of Stroke24 reveals that risk of pneumonia has decreased from 16% in 2008 to 13% in 2010, and risk of UTI has similarly decreased from 9% to 6%. This suggests that the stroke unit environment plays a role in moderating risk of hospital acquired infection.

    Current national guidelines do not recommend the use of prophylactic antibiotics in acute stroke, although the PANTHERIS trial provides some evidence to support the idea that prophylactic antibiotic therapy may reduce post-stroke infections and improve outcome.29 Our findings support further evaluation, in particular relating to the possibility of differences in the value of antibiotic prophylaxis between ischaemic and haemorrhagic stroke.

    References

    1. ↵
      1. Westendorp WF,
      2. Nederkoorn PJ,
      3. Vermeij J-D,
      4. et al
      . Post-stroke infection: a systematic review and meta-analysis. BMC Neurology 2011;11:110.
      OpenUrlCrossRefPubMed
    2. ↵
      1. Kwan J,
      2. Jenkinson D,
      3. Vassallo M,
      4. et al
      . System infections after acute stroke. Aging Health 2008;491:85–94.
      OpenUrl
    3. ↵
      1. Katzan IL,
      2. Cebul RD,
      3. Husak SH,
      4. et al
      . The effect of pneumonia on mortality among patients hospitalized for acute stroke. Neurology 2003;60:620–5.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Kwan J,
      2. Hand P
      . Infection after acute stroke is associated with poor short-term outcome. Acta Neurologica Scandinavica 2007;115:331–8.
      OpenUrlPubMed
    5. ↵
      1. Aslanyan S,
      2. Weir CJ,
      3. Diener HC,
      4. et al
      . Pneumonia and urinary tract infection after acute ischaemic stroke: a tertiary analysis of the GAIN International trial. Eur J Neurol 2004;11:49–53.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Ovbiagele B,
      2. Hills NK,
      3. Saver JL,
      4. et al
      . Frequency and determinants of pneumonia and urinary tract infection during stroke hospitalization. J Stroke Cerebrovasc Dis 2006;15:209–13.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Hong KS,
      2. Kang DW,
      3. Koo JS,
      4. et al
      . Impact of neurological and medical complications on 3-month outcomes in acute ischaemic stroke. Eur J Neurol 2008;15:1324–31.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Emsley HC,
      2. Hopkins SJ
      . Acute ischaemic stroke and infection: recent and emerging concepts. Lancet Neurol 2008;7:341–53.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Hilker R,
      2. Poetter C,
      3. Findeisen N,
      4. et al
      . Nosocomial pneumonia after acute stroke: implications for neurological intensive care medicine. Stroke 2003;34:975–81.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Tasaki K,
      2. Ruetzler CA,
      3. Ohtsuki T,
      4. et al
      . Lipopolysaccharide pre-treatment induces resistance against subsequent focal cerebral ischemic damage in spontaneously hypertensive rats. Brain Res 1997;748:267–70.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Rosenzweig HL,
      2. Lessov NS,
      3. Henshall DC,
      4. et al
      . Endotoxin preconditioning prevents cellular inflammatory response during ischemic neuroprotection in mice. Stroke 2004;35:2576–81.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      Intercollegiate Stroke Working Part. National clinical guideline for stroke. 3rd edn. London: Royal College of Physicians, 2008.
    13. ↵
      National Institute for health and Clinical Excellence. Stroke: diagnosis and initial management of acute stroke and transient ischaemic attack (TIA), 2008.
    14. ↵
      1. Pickering RM,
      2. Kunkel D,
      3. Fitton C,
      4. et al
      . In-hospital recruitment to observational studies of stroke. Int J Rehabil Res 2010;33:56–63.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Bamford J,
      2. Sandercock P,
      3. Dennis M,
      4. et al
      . Classification and natural history of clinically identifiable subtypes of cerebral infarction. Lancet 1991;337:1521–6.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Teasdale G,
      2. Murray G,
      3. Parket L,
      4. et al
      . Adding up the Glasgow Coma Scale. Acta Neurochirorgica Supplementum (Wien) 1979;28:13–16.
      OpenUrl
    17. ↵
      1. Holden MK,
      2. Gill KM,
      3. Magliozzi MR,
      4. et al
      . Clinical gait assessment in the neurologically impaired. Reliability and meaningfulness. Phys Ther 1984;64:35–40.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Vargas M,
      2. Horcajada JP,
      3. Obach V,
      4. et al
      . Clinical consequences of infection in patients with acute stroke: is it prime time for further antibiotic trials? Stroke 2006;37:461–5.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Chammorro Á,
      2. Urra X,
      3. Planas AM
      . Infection after acute ischemic stroke, a manifestation of brain-induces imunodepression. Stroke 2007;38:1097–103.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Vermeij FH,
      2. Scholte op Reimer WJM,
      3. de Man P,
      4. et al
      . Stroke-associated infection is an independent risk factor for poor outcome after acute ischemic stroke: data from the Metherlands stroke survey. Cerebrovasc Dis 2009;27:465–71.
      OpenUrlCrossRefPubMed
    21. ↵
      1. Zhang DP,
      2. Yan FL,
      3. Xu HQ,
      4. et al
      . A decrease of human leucocyte antigen-DR expression on monocytes in peripheral blood predicts stroke-associated infection in critically-ill patients with acute stroke. Eur J Neurol 2009;16:498–505.
      OpenUrlCrossRefPubMed
    22. ↵
      StataCorp. Stata release 11. Statistical software. College Station, TX: StataCorp LP, 2009.
    23. ↵
      1. Counsell C,
      2. Dennis M,
      3. McDowall M
      . Predicting functional outcome in acute stroke: comparison of a simple six variable model with other predictive systems and informal clinical prediction. J Neurol Neurosurg Psychiatry 2004;75:401–5.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      Royal College of Physicians. National Sentinel Audit of Stroke. 2010.
    25. ↵
      1. Brott T,
      2. Adams HP,
      3. Olinger CP,
      4. et al
      . Measurements of acute cerebral infarction: a clinical examination scale. Stroke 1989;20:864–70.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Grau AJ,
      2. Buggle F,
      3. Schnitzler P,
      4. et al
      . Fever and infection early after ischemic stroke. J Neurol Sci 1999;171:115–20.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Becker KJ
      . Targeting the central nervous system inflammatory response in ischemic stroke. Curr Opin Neurol 2001;14:349–53.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Evans A,
      2. Perez I,
      3. Harraf F,
      4. et al
      . Can differences in management processes explain different outcomes between stroke unit and stroke-team care? Lancet 2002;358:1586–92.
      OpenUrl
    29. ↵
      1. Harms H,
      2. Prass K,
      3. Meisel C,
      4. et al
      . Preventive antibacterial therapy in acute ischemic stroke: a randomized controlled trial. PLoS One 2008;3:e2158.
      OpenUrlCrossRefPubMed

    Footnotes

    • Contributors JK was involved in developing the research question, collected data on infection and was responsible for drafting the paper. RMP was a grant applicant, involved in developing the research question, responsible for the statistical analysis, and contributed to drafting the paper. DK was a grant applicant, involved in developing the research question, oversaw the conduct of the study and the collection of the data, and contributed to drafting the paper. CF was involved in collecting the data and contributed to drafting the paper. DJ was a grant applicant, involved in developing the research question, and contributed to drafting the paper. VHP was a grant applicant, involved in developing the research question, and contributed to drafting the paper. AM was the principal investigator, involved in developing the research question, and contributed to drafting the paper.

    • Funding The study was funded by the Stroke Association.

    • Competing interests None.

    • Ethics approval Dorset NHS Local Research Ethics Committee (ethics no. 04/Q2201/91).

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

    Linked Articles