You are on page 1of 5

Journal of the Neurological Sciences 335 (2013) 164–168

Contents lists available at ScienceDirect

Journal of the Neurological Sciences


journal homepage: www.elsevier.com/locate/jns

Effects of smoking on outcomes after acute atherothrombotic stroke in


Japanese men
Naoko Kumagai a,b,⁎, Yoshiyasu Okuhara a, Tatsuo Iiyama b, Yasunori Fujimoto c, Hidehiro Takekawa d,
Hideki Origasa e, Yu Kawanishi c, Takenori Yamaguchi f
a
Center of Medical Information Science, Kochi Medical School, Kochi University, Kohasu, Oko-cho, Nankoku City, Kochi 783-8505, Japan
b
Integrated Center for Advanced Medical Technologies, Kochi Medical School, Kochi University, Kohasu, Oko-cho, Nankoku City, Kochi 783-8505, Japan
c
Department of Neurosurgery, Kochi Medical School, Kochi University, Kohasu, Oko-cho, Nankoku City, Kochi 783-8505, Japan
d
Department of Neurology, Dokkyo University School of Medicine, Kitakobayashi 880, Mibu, Shimotsuga, Tochigi 321-0293, Japan
e
Division of Biostatistics and Clinical Epidemiology, University of Toyama, Toyama, 2630 Sugitani, Toyama City, Toyama 930-0194, Japan
f
National Cerebral and Cardiovascular Center, 5-7-1 Fujishiro-dai, Suita, Osaka 565-8565, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Background: The effects of smoking on clinical outcomes following acute stroke remain controversial.
Received 16 June 2013 Methods: We evaluated the influence of smoking on 90-day outcomes after acute atherothrombotic stroke in 292
Received in revised form 27 August 2013 Japanese men extracted from the database of the Edaravone and Argatroban Stroke Therapy for Acute Ischemic
Accepted 16 September 2013 Stroke randomized parallel-group trial that tested the safety and efficacy of edaravone and argatroban therapy
Available online 25 September 2013
in 814 patients in 2004–2008. Smokers were matched with non-smokers of the same age for identical age
distribution in the smoker and non-smoker groups. Poor 90-day outcomes (defined as death, Barthel index b 60,
Keywords:
Acute atherothrombotic stroke
or modified Rankin score N 3) were evaluated using a logistic regression model. Significant variables (P b 0.05)
Age disparity between smokers and in univariate analysis were further evaluated by multivariate logistic regression analysis using a forward-
non-smokers selection method.
Gender difference Results: Body temperature, age, National Institute of Health Stroke Scale score at admission, systolic blood pres-
Japanese men sure, and smoking status were selected in the final model. Smokers had significantly increased odds of poor 90-
Poor outcome day functional outcomes independent of other statistically significant predictor variables (adjusted odds ratio,
Prognostic factor 2.28; 95% confidence interval, 1.15–4.55; P = 0.019).
Smoking
Conclusions: In Japanese men, smoking leads to poor functional outcomes at 3 months after acute atherothrombotic
stroke.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction and diabetes mellitus, have been demonstrated to be prognostic factors


[2,3]. Although smoking is a well-established risk factor for all forms of
Stroke is a common neurological disorder leading to late-life disabil- stroke [4–6], the prognostic influence of smoking—an individual behav-
ity. More than one-third of patients who experience a stroke return to ior and a modifiable factor—following acute ischemic stroke remains a
their home with some level of permanent disability [1]. The sustained matter of debate [7–9]. The prevalence of tobacco use has remained
impact of stroke has a devastating impact not only on the individual largely unchanged over the last quarter of a century [4]. Therefore, clar-
and their caregivers but also on the wider family, particularly children ifying the role of smoking as a prognostic factor after acute stroke is im-
[1]. It is critically important to improve stroke outcomes in order to re- portant not only for improving stroke outcomes but also for appropriate
duce the individual and social burden of care. The degree of functional promotion of quitting the smoking habit.
recovery after acute stroke is influenced by many factors; thus far, cer- We considered that the conflicting results available for smoking sta-
tain factors, including the location of lesions, severity of the ischemic ep- tus and the prognosis for stroke are results of inadequate adjustment for
isode, age, and the presence of other disorders, such as hypertension confounding factors. Fewer women are smokers as compared to men,
and age disparities have been noted in smoking and non-smoking
groups in available researches on smoking and its prognostic effects
[7–9]. Therefore, in the present study, we targeted the stroke subtype
and gender and included age-matched subjects in order to eliminate
⁎ Corresponding author at: Integrated Center for Advanced Medical Technologies, Kochi
Medical School, Kochi University, Kohasu, Oko-cho, Nankoku City, Kochi 783-8505, Japan.
confounding factors. This study aimed to clarify the role of the smoking
Tel.: +81 88 880 2719; fax: +81 88 880 2543. habit as a prognostic factor for 90-day functional outcomes following
E-mail address: kumagain@kochi-u.ac.jp (N. Kumagai). acute atherothrombotic stroke.

0022-510X/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jns.2013.09.023
N. Kumagai et al. / Journal of the Neurological Sciences 335 (2013) 164–168 165

2. Patients and methods in Table 1. We analyzed whether smoking status was associated with
poor outcomes at 90 days following acute atherothrombotic stroke
2.1. Patients using this dataset.

Data from the Edaravone and Argatroban Stroke Therapy for Acute
2.2. Ethics statement
Ischemic Stroke study (EAST) database were used for analysis. The
EAST study was a randomized parallel-group trial conducted between
The EAST study protocol was approved by the ethical committees of
August 2004 and May 2008 that tested the safety and efficacy of combi-
the participating institutions and all participants provided written in-
nation therapy with argatroban and edaravone in 814 patients with
formed consent.
acute noncardioembolic and nonlacunar stroke in an unblended fashion
in Japan (ClinicalTrials.gov identifier: NCT00153946) [10]. Edaravone is
a free radical scavenger and a neuroprotective agent and argatroban is a 2.3. Baseline factors and outcome
selective thrombin inhibitor. The detailed study design is available at
ClinicalTrials.gov. The EAST study showed no significance differences Age, smoking status, previous history of infarction, previous history of
between combination therapy with argatroban and edaravone and transient ischemic attack (TIA), ischemic heart disease, hypertension, di-
argatroban alone with respect to safety and efficacy. Although previous abetes, hyperlipidemia, atrial fibrillation, systolic blood pressure (SBP),
reports have suggested that thrombolytic therapy may be more effec- diastolic blood pressure (DBP), body temperature, administration time
tive in smokers than in non-smokers and that current smoking does from stroke onset, size of infarction, number of ischemic lesions, use of
not affect the outcome of tissue plasminogen activator (t-PA) therapy statin, use of aspirin, randomized treatment of the EAST study (mono-
after adjusting for age [11,12], none of the patients in our study received therapy with argatroban vs. combination therapy with argatroban and
t-PA therapy. edaravone), and the National Institutes of Health Stroke Scale (NIHSS)
First, data for 486 Japanese male patients with atherothrombotic score at admission were selected as baseline covariate predictors. The
stroke from the EAST database were extracted for this study since the variable of the NIHSS score was divided into 3 categories, i.e., 1–4,
majority of smokers in the EAST study were men and few women re- 5–15, or ≥16. All patients were treated within 24 h from stroke onset
ported the smoking habit. Second, because the mean age of the smoking and were examined with both computed tomography and magnetic res-
group was considerably lower than that of the non-smoking group in onance imaging at baseline.
the extracted database, we matched the age distribution between the
smoking and non-smoking groups in order to minimize the confound-
ing factors related to age. For such matching, 1 patient from the smoking Table 1
group was paired with 1 patient of the same age in the non-smoking Characteristics of smokers vs. non-smokers.

group. If there were no patients of the same age in both the smoking Characteristic Non- Smoker P
and non-smoking groups, such non-matched patients were excluded smoker (n = 146) value
from this study. Consequently, 146 pairs of non-smokers and smokers, (n = 146)

i.e., a total of 292 patients, from the EAST database were included in n % n %
the present study, as shown in Fig. 1. Thus, we ensured that the age dis- Agea (years) ≤50 2 (1.4) 2 (1.4)
tribution in the smoker and non-smoker groups was identical, as shown 51–60 16 (11.0) 16 (11.0)
61–70 60 (41.1) 60 (40.1)
71–80 56 (38.4) 56 (38.4)
N80 12 (8.1) 12 (8.1)
Enrolled subjects Administration time from stroke Mean ± SD 11.5 ± 6.2 12.5 ± 6.6 0.18
(n = 814) onset (h)
Systolic blood pressure (mm Hg) Mean ± SD 161.6 ± 158.6 ± 0.28
25.3 23.6
Diastolic blood pressure (mm Hg) Mean ± SD 87.9 ± 16.2 84.0 ± 14.7 0.034
Body temperature (°C) Mean ± SD 36.4 ± 0.53 36.4 ± 0.50 0.46
Japanese patients with acute NIHSS score at admission Median 4 [2,7] 5 [3,8] 0.10
atherothrombotic stroke (n = 701) [25%,75%]
1–4 88 (60.3) 70 (47.9) 0.10
4–15 51 (34.9) 68 (46.6)
≥16 7 (4.8) 8 (5.5)
Ischemic heart disease Yes 22 (15.1) 10 (6.8) 0.025
Previous infarction: Yes 21 (14.4) 25 (17.1) 0.52
Male subjects (n = 486) Female subjects (n = 215) Previous TIA Yes 5 (3.4) 3 (2.1) 0.72
Hypertension Yes 94 (64.4) 89 (61.0) 0.55
Smokers Smokers
Diabetes Yes 53 (36.3) 42 (28.8) 0.17
(n = 254; 52.3%) (n = 28; 9.9%) Hyperlipidemia Yes 41 (28.1) 42 (28.8) 0.90
Atrial fibrillation Yes 7 (4.8) 3 (2.1) 0.20
Number of ischemic lesions None 22 (15.1) 15 (10.3) 0.16
Single 107 (73.3) 104 (71.2)
Multiple 17 (11.6) 27 (18.5)
One patient from the smoking group was Size of infarction None/ 55 (37.7) 33 (22.6) 0.005
paired with one patient of the same age from ≤15 mm
the non-smoking group N15 mm 91 (62.3) 113 (77.4)
Argatroban/argatroban and
edaravone
Argatroban and edaravone 73 (50.0) 81 (55.5) 0.35
Use of statin Yes 23 (15.8) 22 (15.1) 0.87
Non-smokers (n = 146) and smokers (n = 146) Use of aspirin Yes 111 (76.0) 117 (80.1) 0.40
Total = 292 (Outcome n = 54) For analysis of continuous variables, t-test or Wilcoxon's rank-sum test was used; chi-
square test or Fisher's exact test was used to analyze categorical variables.
TIA, transient ischemic attack; NIHSS, National Institutes of Health Stroke Scale.
a
Fig. 1. Patient flowchart. Age was used as the matching variable.
166 N. Kumagai et al. / Journal of the Neurological Sciences 335 (2013) 164–168

Table 2
Associations between a poor outcome and a baseline factor.

Baseline factor Good outcome Poor outcome OR (95% CI) P value

n (%) n = 238 n (%) n = 54

Smoking Smoker 111 (46.6) 35 (64.8) 2.11 (1.14–3.89) 0.017⁎


Age (1 year) 69.0 ± 7.58 71.6 ± 8.53 1.05 (1.01–1.09) 0.025⁎
Previous infarction Yes 34 (14.3) 12 (22.2) 1.71 (0.82–3.58) 0.16
Previous TIA Yes 7 (2.9) 1 (1.9) 0.62 (0.08–5.17) 0.66
Hypertension Yes 149 (62.6) 34 (63.0) 1.02 (0.55–1.87) 0.96
Diabetes Yes 73 (30.7) 22 (40.7) 1.55 (0.85–2.86) 0.16
Hyperlipidemia Yes 68 (28.6) 15 (27.8) 0.96 (0.50–1.86) 0.91
Ischemic heart disease Yes 30 (12.6) 2 (3.7) 0.27 (0.06–1.15) 0.077
Atrial fibrillation Yes 9 (3.8) 1 (1.9) 0.48 (0.06–3.87) 0.49
Body temperature (°C) Mean ± SD 36.4 ± 0.51 36.6 ± 0.50 3.21 (1.69–6.13) b0.001⁎
Systolic blood pressure (mm Hg) 158.3 ± 23.7 168.0 ± 26.6 1.02 (1.00–1.03) 0.009⁎
Diastolic blood pressure (mm Hg) 85.5 ± 15.5 87.8 ± 15.9 1.01 (0.99–1.03) 0.34
NIHSS score at admission b0.001⁎
1–4 145 (60.9) 13 (24.0) 1
5–15 85 (35.7) 34 (63.0) 4.46 (2.23–8.92) b0.001⁎
≥16 8 (3.4) 7 (13.0) 9.76 (3.05–31.2) b0.001⁎
Size of infarction N15 mm 158 (66.4) 46 (85.2) 2.91 (1.31–6.46) 0.009⁎
Number of ischemic lesions 0.22
None 34 (14.3) 3 (5.6) 1
Single 170 (71.4) 41 (75.9) 2.73 (0.80–9.34) 0.11
Multiple 34 (14.3) 10 (18.5) 3.33 (0.84–13.2) 0.086
Argatroban/argatroban and edaravone
Argatroban and edaravone 123 (51.7) 31 (57.4) 1.26 (0.69–2.29) 0.45
Administration time from stroke onset (h) 12.1 ± 6.43 11.6 ± 6.32 0.99 (0.94–1.03) 0.54
Use of statin Yes 38 (16.0) 7 (13.0) 0.78 (0.33–1.87) 0.58
Use of aspirin Yes 187 (78.6) 41 (75.9) 0.86 (0.43–1.73) 0.67

TIA, transient ischemic attack; NIHSS, National Institutes of Health Stroke Scale.
An association between each baseline variable and stroke outcome was evaluated using univariate logistic regression analysis and its odds ratio (OR) and 95% confidence interval (CI) were
estimated.
⁎ Indicates P b 0.05.

The outcome at 90 days after the onset of acute atherothrombotic 3. Results


stroke was evaluated as the study endpoint. The Barthel index (BI)
and the modified Rankin scale (mRS), which are the most common out- 3.1. Characteristics of the entire EAST study dataset
come measures used in stroke research and focus on stroke-related dis-
ability and motor function recovery, were used in this study. We defined The original EAST database of acute atherothrombotic stroke pa-
poor outcome as the presence of any of the following: death, mRS N 3, tients shows that the male subpopulation includes 52.3% smokers,
or BI b 60 on 90 days [9,13]. while the female subpopulation only includes 9.9% smokers (Fig. 1).
The mean age of the male non-smoking group was 72.0 years, while
2.4. Statistical analysis that of the male smoking group was 65.8 years; further, the smoking
group also included extremely young male patients (P b 0.001).
The characteristics between non-smokers and smokers were com-
pared using the chi-square test or Fisher's exact test for categorical var-
3.2. Baseline characteristics between non-smokers and smokers
iables, and the two-sample t-test or Wilcoxon's rank-sum tests for
continuous variables.
The smokers had a lower value of DBP, a lower frequency of ischemic
The association between a baseline factor and outcome was investi-
heart diseases, and larger infarction sizes (P b 0.05) at baseline (Table 1).
gated by using univariate and multivariate logistic regression analyses.
A variable was considered potentially associated with the outcome
when the p value was less than 0.05 in univariate logistic analysis. Var- 3.3. Association between baseline factors and poor outcome
iables that reached this significance threshold were further evaluated by
multivariate logistic regression analysis, using a forward-selection As shown in Table 2, age (OR = 2.11, P = 0.017), SBP (OR = 1.02,
method with a significance level of entry of 0.05. The interactions of fac- P = 0.009), body temperature (OR = 3.21, P b 0.001), NIHSS score at
tors were also evaluated through a forward-selection method. The the time of admission (score: 5–15 and ≥16; OR = 4.46, OR = 9.76,
goodness of fit was evaluated using the Hosmer–Lemeshow test, and respectively, P b 0.001), and size of infarction (OR = 2.91, P = 0.009)
predictive ability was confirmed by c-statistics based on receiver oper- were selected as candidate covariates, in addition to smoking status
ating characteristic analysis. External validation was performed using (OR = 2.11, P = 0.017).
a cross-validation method. In the cross-validation analysis, 233 patients, Multivariate logistic regression model with forward-selection includ-
i.e., 80% of the total patients were randomly selected from the entire ed smoking status (OR, 2.28; 95% CI, 1.15–4.55; P = 0.019), age (OR,
dataset then, the c-statistics and odds ratio were estimated with the 1.06; 95% CI, 1.01–1.11; P = 0.014), SBP (OR, 1.02; 95% CI, 1.00–1.03;
multivariate logistic model using this selected dataset. This process was P = 0.0096), body temperature (OR, 3.27; 95% CI, 1.59–6.72; P =
repeated 1000 times. Therefore, 1000 c-statistic values and 1000 odds 0.001), and NIHSS score at admission (score: 5–15 and ≥16; OR, 4.05;
ratio for smoking were generated. The averages, minimum, and maxi- 95% CI, 1.93–8.51; P b 0.001; and OR, 9.45; 95% CI, 2.64–33.8; P b 0.001,
mum values have been shown in this study. respectively) as shown in Table 3. No interaction terms were selected.
All statistical tests were two-tailed, and the level of statistical signif- The final model for c-statistic was 0.80, and the P value for the Hosmer–
icance was set at 0.05. All analyses were performed using SAS statistical Lemeshow goodness-of-fit test was 0.61. The model demonstrated ex-
software, version 9.1 (SAS Institute Inc., Cary, NC). cellent discrimination and acceptable calibration.
N. Kumagai et al. / Journal of the Neurological Sciences 335 (2013) 164–168 167

Table 3 patients than did the non-smoking groups [5–9,11,12]. A large age
Associations between poor outcomes and smoking, with controlling confounding variables. difference cannot be controlled by multivariate regression analysis and
Risk factor Adjusted OR (95% CI) P value can yield confusing results. Although the latter study used age-stratified
analyses, the number of outcomes was small and could not adequately
Smoking
Smoker 2.28 (1.15–4.55) 0.019 control the confounding factors.
Age (years) 1.06 (1.01–1.11) 0.014 Second, an increasing body of evidence has shown gender-specific
Systolic arterial pressure 1.02 (1.00–1.03) 0.0096 differences in the etiology of stroke as well as in stroke symptoms and
Body temperature 3.27 (1.59–6.72) 0.0013
prognosis after ischemic stroke [15–18]. Some of these differences may
NIHSS score at admission
have important implications because of the gender-specific differences
5–15 4.05(1.93–8.51) b0.001 in responses to nitric oxide, which is a component of cigarette smoke
≥16 9.45 (2.64–33.8) b0.001 [19,20]. Typically, fewer women report a smoking habit than men
NIHSS, National Institutes of Health Stroke Scale. [5–8,11,12]. Therefore, analyzing these factors together may cause fur-
The model included smoking and candidate confounding variables, i.e., age, systolic ther bias while estimating the effects of smoking.
arterial pressure, body temperature, NIHSS score and size of infarction, with P b 0.05 in Third, although the prognostic effect of smoking status was differen-
univariate analysis. The forward selection approach in multivariate logistic regression
tially related to the mechanism of stroke [21], previous studies have not
was performed; then, smoking age, systolic arterial pressure, body temperature, and
NIHSS were selected. The size of infarction was not selected; it was confirmed that there well accounted for the differences in ischemic stroke subtypes. Specifi-
were no interactions between any of the 2 variables when the selection criterion was set cally, they did not perform stratified analysis by subtype, or if they did
as P b 0.05. so, the methods failed to adjust for other confounding factors.
Hosmer–Lemeshow test (chi-square = 6.38, degree of freedom = 8, P = 0.61). These oversights together could have affected the contradictory re-
C-statistics = 0.80.
sults of these previous studies. All the atherothrombotic patients includ-
ed in the EAST study also demonstrated an age disparity related to the
smoking status along with few female smokers, similar to that in
other studies. Therefore, in the present study, we included only male pa-
3.4. Validation tients with atherothrombotic stroke and matched the age distribution
for the smoker and non-smoker groups to be identical, as shown in
The final model, including the variables of smoking status, age, body Table 1. Six factors, i.e., smoking status, age, SBP, body temperature,
temperature, systolic body pressure, and NIHSS scores were used and NIHSS score at admission, and size of infarction were selected as poten-
odds ratio and c-statistics were estimated in the model with the tial candidate prognosis factors in univariate analysis. Subsequently, the
resampled dataset. The distribution of 1000 different c-statistics ranged multivariable model revealed smoking status, age, SBP, body tempera-
from a minimum value to a maximum value of 0.75 to 0.85, with a mean ture, and NIHSS score at admission to be significant prognosis factors,
value of 0.80 for the resampling population, which was the same as the with no interaction terms among them. Only a single factor, i.e., size of
c-statistic (0.80) derived from the entire dataset. The distribution of infarction, was excluded from the final model through the forward-
1000 different ORs of smoking ranged from 1.27 to 5.04, with a mean selection step, as shown in Table 3. The reason for exclusion of size of
of 2.36 for the resampling population, which was significantly more infarction may be the lack of statistical power supporting this factor;
than 1.00. the size of infarction was weakly correlated with age and smoking status,
and such correlations are known to decrease statistical power [22,23].
4. Discussion Thus, smokers showed a significantly increased likelihood of poor
functional outcomes at 90 days in the final multivalent model (Table 3).
Our findings demonstrate that Japanese male smokers in the EAST The final multivariate model demonstrated excellent discrimination
study experienced poorer outcomes than did non-smokers following and acceptable calibration based on results of c-statistics and Hosmer–
acute atherothrombotic stroke; smoking was thus indicated to be an Lemeshow test. Our cross-validation analysis also supported the reli-
independent, poor prognostic factor for Japanese men after acute ability of our results.
atherothrombotic stroke. Two previous studies for this issue have At baseline, the smokers had larger infarct sizes as compared to non-
shown contradictory results, wherein one study suggested that current smokers, as shown in Table 1. With respect to the infarct size, although
or recent smokers with acute ischemic stroke experienced poorer func- size of infarction was not included in the final multivariate model, pos-
tional outcomes at 3 months as compared to non-smokers [8], whereas sibly due to the lack of statistical power, this was significantly associated
the other observed that current smokers with acute ischemic stroke with poor outcomes at 90 days in the univariate logistic model
experienced better long-term outcomes over a 6-month period than did (Table 2). The size of infarction has already been demonstrated to be
non-smokers [9]. These conflicting results have 3 potential explanations, a prognostic factor [24–27]. In addition, nicotine has previously been
as clarified below. shown to increase the size of infarction in animal models [28,29]. Nicotine
First, although both results were obtained after adjusting the co- may thus be directly or indirectly associated with factors causing larger
variates by using multivariate regression analysis, there was an age infarcts, thereby leading to poor functional outcomes in smokers.
disparity between smokers and non-smokers, with smokers being A limitation of the present study is that this was a secondary analysis
significantly younger than non-smokers. The age differences between of a previously completed randomized trial rather than a prospective
non-smokers and smokers in both studies were considerable; therefore, study specifically designed to address the present question. Therefore,
the confounding factors could not be sufficiently adjusted in these we were unable to collect detailed information on the number of ciga-
studies. Age difference is a critical problem that has also affected inves- rettes smoked daily, the duration of smoking, or changes in the smoking
tigations examining the association between smoking status and prog- habit of the investigated patients at 90 days after stroke. We were also
nosis after myocardial infarction. Because younger people tend to have unable to provide a continuous scale of infarct size. Moreover, associa-
lesser comorbidities and are more physically fit, the results of better tions between Japanese male smokers and low education level, consid-
outcomes or decreased mortality rates after an acute myocardial infarc- ered to be related to low income, little physical activity, and irregular
tion were considered to be due to the younger age of the smoking group breakfast habits, have been suggested [30]; while it is possible that
(as compared to that of the non-smoking group) rather than any these factors are potentially related to outcomes, this information was
benefits from smoking [14]. We reviewed several papers concerning not collected in the present study. Further investigations are needed
smoking status and stroke and found that in these studies, the smoking to clarify the relationship between smoking status and clinical outcomes
groups included significantly younger and a higher number of male following acute atherothrombotic stroke, including the association
168 N. Kumagai et al. / Journal of the Neurological Sciences 335 (2013) 164–168

between smoking and infarct size. However, it should be noted that all NCT00153946). http://clinicaltrials.gov/ct2/show/study/NCT00153946. [Accessed on
19 August 2011].
the patients were examined with both computed tomography and mag- [11] Ovbiagele B, Saver JL. The smoking-thrombolysis paradox and acute ischemic stroke.
netic resonance imaging (the infarct sizes in all patients were assessed). Neurology 2005;65:293–5.
Further, although certain clinical factors such as serum cholesterol [12] Kufner A, Nolte CH, Galinovic I, Brunecker P, Kufner GM, Endres M, et al.
Smoking-thrombolysis paradox: recanalization and reperfusion rates after in-
levels were not examined, clinically many important factors were in- travenous tissue plasminogen activator in smokers with ischemic stroke. Stroke
cluded in our analysis, and our model was proven to have a good fit by 2013;44(2):407–13.
c-statistics and the Hosmer–Lemeshow test. Moreover, our validation [13] Sulter G, Steen C, De Keyser J. Use of the Barthel index and modified Rankin scale in
acute stroke trials. Stroke 1999;30:1538–41.
testing supported the reliability of our results. [14] Aune E, Røislien J, Mathisen M, Thelle DS, Otterstad JE. The “smoker's paradox” in
To the best of our knowledge, this is the first report to address the in- patients with acute coronary syndrome: a systematic review. BMC Med 2011;9:97.
fluence of smoking on functional outcomes after acute atherothrombotic [15] Petrea RE, Beiser AS, Seshadri S, Kelly-Hayes M, Kase CS, Wolf PA. Gender differences
in stroke incidence and poststroke disability in the Framingham Heart Study. Stroke
stroke by analyzing an age-balanced dataset that specified both gender
2009;40:1032–7.
and the stroke subtype. Our findings indicate that smoking increases [16] Palm F, Urbanek C, Wolf J, Buggle F, Kleemann T, Hennerici MG, et al. Etiology, risk
the likelihood of poor functional outcomes among male Japanese patients factors and sex differences in ischemic stroke in the Ludwigshafen stroke study, a
who experience acute atherothrombotic stroke. population-based stroke registry. Cerebrovasc Dis 2012;33(1):69–75.
[17] Towfighi A, Markovic D, Ovbiagele B. Persistent sex disparity in midlife stroke prev-
alence in the United States. Cerebrovasc Dis 2011;31(4):322–8.
Conflicts of interest [18] Palnum KH, Mehnert F, Andersen G, Ingeman A, Krog BR, Bartels PD, et al. Medical
prophylaxis following hospitalization for ischemic stroke: Age- and sex-related dif-
ferences and relation to mortality. Cerebrovasc Dis 2010;30(6):556–66.
The authors declare that there are no conflicts of interest. [19] Oksala NK, Oksala A, Erkinjuntti T, Pohjasvaara T, Kunnas T, Vataja R, et al. Long-term
survival after ischemic stroke in postmenopausal women is affected by an interac-
tion between smoking and genetic variation in nitric oxide synthases. Cerebrovasc
References Dis 2008;26(3):250–8.
[20] Park EM, Cho S, Frys KA, Glickstein SB, Zhou P, Anrather J, et al. Inducible nitric oxide
[1] Wolfe C, Rudd A. The Burden of Stroke White Paper: raising awareness of the global synthase contributes to gender differences in ischemic brain injury. J Cereb Blood
toll of stroke-related disability and death. http://www.safestroke.org/Portals/10/ Flow Metab 2006;26(3):392–401.
FINAL%20Burden%20of%2020Stroke.pdf; 2007. [Accessed March 2011. Stroke.pdf. [21] Murat M, Erturk O. Ischemic stroke subtypes: risk factors, functional outcome and
Accessed March 2009]. recurrence. Neurol Sci 2002;22(6):449–54.
[2] Nakamura K, Nakagawa H, Sakurai M, Murakami Y, Irie F, Fujiyoshi A, et al. Influence [22] Hsieh FY. Sample size tables for logistic regression. Stat Med 1989;8(7):795–802.
of smoking combined with another risk factor on the risk of mortality from coronary [23] Courvoisier DS, Combescure C, Agoritsas T, Gayet-Ageron A, Perneger TV. Performance
heart disease and stroke: pooled analysis of 10 Japanese cohort studies. Cerebrovasc of logistic regression modeling: beyond the number of events per variable, the role of
Dis 2012;33:480–91. data structure. J Clin Epidemiol 2011;64(9):993–1000.
[3] Kuwashiro T, Kamouchi M, Ago T, Hata J, Sugimori H, Kitazono T. The factors as- [24] Pan SL, Wu SC, Wu TH, Lee TK, Chen TH. Location and size of infarct on functional
sociated with a functional outcome after ischemic stroke in diabetic patients: the outcome of noncardioembolic ischemic stroke. Disabil Rehabil 2006;28:977–83.
Fukuoka Stroke Registry. J Neurol Sci 2012;313(1–2):110–4. [25] Saunders DE, Clifton AG, Brown MM. Measurement of infarct size using MRI predicts
[4] Shah RS, Cole JW. Smoking and stroke: the more you smoke the more you stroke. prognosis in middle cerebral artery infarction. Stroke 1995;26(12):2272–6.
Expert Rev Cardiovasc Ther 2010;8(7):917–32. [26] Beloosesky Y, Streifler JY, Burstin A, Grinblat J. The importance of brain infarct size
[5] Wolf PA, D'Agostino RB, Kannel WB, Bonita R, Belanger AJ. Cigarette smoking as a and location in predicting outcome after stroke. Age Ageing 1995;24(6):515–8.
risk factor for stroke: the Framingham Study. JAMA 1988;259:1025–9. [27] Reith J, Jørgensen HS, Pedersen PM, Nakayama H, Raaschou HO, Jeppesen LL, et al.
[6] Kim DE, Lee KB, Jang IM, Roh H, Ahn MY, Lee J, et al. Associations of cigarette Body temperature in acute stroke: relation to stroke severity, infarct size, mortality,
smoking with intracranial atherosclerosis in the patients with acute ischemic stroke. and outcome. Lancet 1996;347(8999):422–5.
Clin Neurol Neurosurg 2012;114(9):1243–7. [28] Bradford ST, Stamatovic SM, Dondeti RS, Keep RF, Andjelkovic AV. Nicotine aggra-
[7] Weng WC, Huang WY, Chien YY, Wu CL, Su FC, Hsu HJ, et al. The impact of smoking vates the brain postischemic inflammatory response. Am J Physiol Heart Circ Physiol
on the severity of acute ischemic stroke. J Neurol Sci 2011;308:94–7. 2011;300:H1518–29.
[8] Ovbiagele B, Weir CJ, Saver JL, Muir KW, Lees KR. IMAGES investigators: Effect of smoking [29] Paulson JR, Yang T, Selvaraj PK, Mdzinarishvili A, Van der Schyf CJ, Klein J, et al. Nicotine
status on outcome after acute ischemic stroke. Cerebrovasc Dis 2006;21:260–5. exacerbates brain edema during in vitro and in vivo focal ischemic conditions.
[9] Bang OY, Park HY, Lee PH, Kim GM, Chung CS, Lee KH, et al. Improved outcome after J Pharmacol Exp Ther 2010;332:371–9.
atherosclerotic stroke in male smoker. J Neurol Sci 2007;260:43–8. [30] Hu L, Sekine M, Gaina A, Nasermoaddeli A, Kagamimori S. Association of smoking
[10] National Institute of Health. ClinicalTrials.gov. (Edaravone and Argatroban Stroke behavior and socio-demographic factors, work, lifestyle and mental health of
Therapy Study for Acute Ischemic Stroke (EAST); ClinicalTrials.gov Identifier: Japanese civil servants. J Occup Health 2007;49(6):443–52.

You might also like