There is strong evidence that smoking increases the risk

Transcription

1 Baseline Smoking Status and the Long-Term Risk of Death or Nonfatal Vascular Event in People with Stroke A 10-Year Survival Analysis Joosup Kim, BBiomedSci (Hons); Seana L. Gall, PhD; Helen M. Dewey, PhD; Richard A. L. Macdonell, MD; Jonathan W. Sturm, PhD; Amanda G. Thrift, PhD Background and Purpose Smoking may exacerbate the risk of death or further vascular events in those with stroke, but data are limited. Methods 1589 cases of first-ever and recurrent stroke were recruited between 1996 and 1999 from a defined geographical region in North East Melbourne. Both hospital and nonhospital cases were included. Over a 10-year period, all deaths, recurrent stroke events, and acute myocardial infarctions that were reported at follow-up interviews were validated using medical records. Cox proportional hazards regression was used to assess the association between baseline smoking status (never, ex, and current) and outcome (death, acute myocardial infarction, or recurrent stroke). Results Patients who were current smokers (Hazard Ratio [HR], 1.30; 95% Confidence Interval [CI], ; P=0.012) at the time of their stroke had poorer outcome when compared with those who had never smoked. Among those who survived the first 28 days of stroke, current smokers (HR, 1.42; 95% CI, ; P<0.003) and ex-smokers (HR, 1.18; 95% CI, ; P=0.039) at baseline had poorer outcome than those who had never smoked. Current smokers also had a greater risk of recurrent events than past smokers (HR, 1.23; 95% CI, ; P=0.050). Conclusions Patients who smoked at the time of their stroke or had smoked before their stroke had greater risk of death or recurrent vascular events when compared with patients who were never smokers. There are benefits of smoking cessation, with ex-smokers appearing to have a lesser risk of recurrent vascular events than current smokers. (Stroke. 2012;43: ) Key Words: mortality smoking stroke survival There is strong evidence that smoking increases the risk of stroke 1 and acute myocardial infarction (AMI). 2 Chemicals in cigarettes have been implicated in the development of atherosclerosis 3 and other pathologies leading to stroke and heart disease. 4,5 It is therefore likely that smoking exacerbates the increased risk of death and cardiovascular events in patients with stroke, as has been seen in previous studies of AMI. 6 8 However, community-based stroke studies provide inconsistent results on the role of baseline smoking on long-term outcome after stroke Investigators of the Perth Community Stroke Study (PCSS) found that past smoking was associated with poor outcome, whereas current smoking was not. 11 Others have found no association between smoking at the time of stroke and long-term outcomes. 9,10 We aimed to assess the contribution of smoking status at the time of stroke to the risk of recurrent vascular events (recurrent strokes and myocardial infarctions) and death in a community-based incidence study; the North East Melbourne Stroke Incidence Study (NEMESIS). This was a unique opportunity to examine this potential association as we carefully assessed deaths and subsequent vascular events over 10 years of follow-up after stroke. Methods Patients were recruited from a defined geographical area. Both hospital and nonhospital records were used to identify all cases of stroke occurring between May 1, 1996 and April 30, 1997 (an 8 postcode region in North East Melbourne with a population of 133,816) and between May 1, 1997 and April 30, 1999 (a 22 post code region, population ). Strokes were defined according to the World Health Organization (WHO) clinical definition. 12 All potential cases of stroke were assessed by a panel of stroke experts who decided on whether the signs and symptoms complied with the definition of stroke and so determined ultimate inclusion in the study. NEMESIS adheres to the criteria for an ideal stroke incidence study as described by Sudlow & Warlow. 13 A baseline assessment was conducted after patients gave consent to participate in the study. Sociodemographic and medical details of the patients were obtained by research nurses using medical records Received June 26, 2012; final revision received August 16, 2012; Accepted August 29, From the Department of Medicine, Monash Medical Centre, Southern Clinical School, Clayton, Victoria, Australia (J.K., A.G.T.); Menzies Research Institute Tasmania, University of Tasmania, Hobart, Tasmania, Australia (S.L.G.); Florey Neuroscience Institutes, Heidelberg, Victoria, Australia (H.M.D., A.G.T.); Department of Medicine, University of Melbourne, Heidelberg, Victoria, Australia (H.M.D., R.A.L.M.); Neurology Department, Austin Health, Heidelberg, Victoria, Australia (H.M.D., R.A.L.M.); and the Department of Neurology, Gosford Hospital, Gosford, NSW, Australia (J.W.S.). The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Amanda Thrift, PhD, Stroke and Ageing Research Centre, Southern Clinical School (Monash Medical Centre), Department of Medicine, Monash University, Melbourne, Australia. amanda.thrift@monash.edu 2012 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 3174 Stroke December 2012 at recruitment and during questioning of participants at follow-up interview. Patients who suffered a subarachnoid hemorrhage were not followed-up. In cases where patients were reluctant to be interviewed at their place of residence, research nurses conducted a telephone interview and collected as much information as possible. When a patient was unable to respond because of severe physical or cognitive disability, the majority of information was obtained from a proxy respondent. Definitions Using information in medical records seen at baseline, we classified patients as never smokers, ex-smokers, or current smokers. Never smokers were defined as people who had never smoked in their lifetime, ex-smokers were defined as people who had previously smoked but were nonsmokers at the time of their stroke, and current smokers were defined as people who were smokers at the time of their stroke. A past history of stroke, cardiovascular disease, atrial fibrillation, diabetes mellitus, and history of hypertension was determined by perusal of medical records. Other sociodemographic details of the patients that could potentially impact risk factor management were ascertained. Such factors included age, sex, place of birth, living arrangements, and place of residence. Patients socioeconomic status was estimated using the Australian Bureau of Statistics Index of Relative Socioeconomic Disadvantage (IRSD) score corresponding to their place of residence. The IRSD, which was developed by the Australian Bureau of Statistics, ranks different geographic areas of Australia according to economic and social resources of households within the areas. We used the IRSD available for Census Collection Districts, which is the smallest area for which the IRSD is calculated, consisting of approximately 250 households each. IRSD was then categorized into 3 equally sized tertiles: most disadvantage, moderate disadvantage, and least disadvantage. 14 Follow-Up of Recurrent Strokes When patients reported that they had suffered a recurrent stroke or transient ischemic attack before a follow-up interview, a data recording sheet outlining details of these events was completed by research nurses. Information on each possible event was gathered from several sources, including hospital and nursing home records, death certificates, and coroner s reports. The date and time of onset, history of onset, and main symptoms of the events were recorded. The findings from computed tomography scans or magnetic resonance imaging were recorded. When an electrocardiograph, echocardiogram, carotid Doppler ultrasound scan, cerebral angiogram, or magnetic resonance angiogram was performed on the patient, the findings of these tests were recorded. This information was presented to a panel of neurologists who determined whether the event was a stroke or not. Follow-Up of Acute Myocardial Infarctions When patients reported that they had suffered an AMI at the followup interview, the potential event was followed up by the research nurse using all available sources. The date of the event, history of onset, and main symptoms of the events were recorded. AMIs were categorized as a definite AMI, probable AMI, or definitely not an AMI using an AMI classification algorithm incorporating the presence of ischemic pain, the levels of cardiac enzymes, troponin levels, and the presence of Q-waves in electrocardiogram, which is a similar classification method to that described by Thygesen and colleagues. 15 Follow-Up of Death Participants in the study were followed up using the National Death Index to determine whether any patients lost to follow-up had died. Covariates Statistics Cox proportional hazards regression was conducted to assess the association between smoking and poor outcome, which included death and nonfatal vascular events (AMI or recurrent stroke). The model was adjusted for confounding factors (age, sex, and socioeconomic status) and all other statistically significant baseline variables (variables with a 2-sided P value 0.05). Variables with a P value >0.1 were removed using backward stepwise method. Cumulative hazard functions were generated based on the results of the Cox proportional hazards regression and were stratified by baseline smoking status. Ethics NEMESIS was approved by the ethics committee at each participating institution. Informed consent was obtained before any interview was conducted. When the participant was cognitively impaired dysphasic, or had altered consciousness, consent was obtained from the next-of-kin. Ethics committee approval was also obtained from the Australian Institute of Health and Welfare for access to the National Death Index data. Results Patient Characteristics There were some differences in patient characteristics between never, past, and current smokers (Supplement Table 1). Compared with patients who had never smoked, past and current smokers were younger, were more often male, had greater socioeconomic disadvantage, and were less often residing in an institution. Compared with those who were never smokers, ex-smokers had more often suffered a previous stroke (P=0.026), a previous transient ischemic attack (P=0.041), and a previous AMI (P<0.001). When compared with patients who had never smoked, current smokers less often had a history of hypertension (P<0.001) and less often had atrial fibrillation (P<0.001). Twenty-eight day case-fatality was 23% (359 of 1589). In the group of patients for whom smoking status was unknown, there was a 50% 28-day fatality rate, which was greater than that of never smokers (P<0.001; unadjusted for other variables). Current smokers also appeared to have a lesser 28-day fatality rate than never smokers (unadjusted P=0.061; unadjusted for other variables). Table 1 shows the comparisons between smoking groups when those who did not survive beyond 28 days of their stroke were excluded from analysis. Differences in patient characteristics between each smoking group were similar to those described above. Baseline Factors Associated With Outcome at 10 Years Poststroke Using multivariable Cox proportional hazards regression, we found that greater age, history of stroke, stroke subtype, atrial fibrillation, being born in Australia, previous acute myocardial infarction, diabetes mellitus, and unconsciousness at the time of stroke were associated with death or recurrent vascular event at 10 years poststroke (Supplemental Table 2). Smokers at the time of stroke had poorer outcome (hazard ratio [HR], 1.30; 95% confidence interval [CI], ; P=0.012) when compared with those who were never smokers at the time of stroke. Patients who were current smokers appeared to be at greater risk of death and nonfatal vascular events than ex-smokers when age was incorporated into the model as a continuous variable (HR, 1.22; 95% CI, ; P=0.055).

3 Kim et al Smoking Status and Risk of Death or Vascular Event 3175 Table 1. Baseline Characteristics of Never Smokers, Ex-Smokers, Current Smokers, and Unknown Smokers Who Survived the First 28 Days After Their Index Stroke Variable Never Smoker (n=502) (%) Ex-Smoker (n=422) (%) When patients who did not survive the first 28 days were excluded from the analysis, stroke subtype was no longer associated with death and nonfatal vascular events (Table 2). Amongst those who survived the first 28 days after their stroke, current smoking was associated with poor outcome (HR, 1.42; 95% CI, ; P=0.003), as was past smoking (HR, 1.18; 95% CI, ; P=0.039). When age was incorporated into the model as a continuous variable, current smokers appeared to be at greater risk of death and nonfatal vascular events than exsmokers (HR, 1.23; 95% CI, ; P=0.050). The cumulative hazard functions of recurrent vascular event or death in patients of different smoking categories are shown in Figure 1 (all strokes) and Figure 2 (28-day survivors). After adjusting for all significant and potentially confounding factors in the Cox proportional hazards regression, patients who were current smokers at baseline had the greatest hazard, followed by ex-smokers and then nonsmokers. P Value* Current Smoker (n=182) (%) P Value Unknown Smoker (n=124) (%) P Value Age 0.01 < <65 70 (14) 68 (16) 69 (38) 13 (10) (21) 118 (28) 67 (37) 19 (15) (40) 159 (38) 39 (21) 41 (33) > (25) 77 (18) 7 (4) 51 (41) Male 161 (32) 281 (67) < (59) < (35) IRSD (socioeconomic status) < Most disadvantage 156 (31) 143 (34) 94 (52) 44 (35) Moderate disadvantage 172 (34) 137 (32) 56 (31) 39 (31) Least disadvantage 174 (35) 141 (33) 32 (18) 41 (33) Born in Australia 332 (66) 266 (63) (64) (63) Residing in institution 80 (16) 40 (9) (9) (36) <0.001 First ever stroke at baseline 415 (83) 324 (77) (84) (73) Stroke subtype <0.001 Ischemic stroke 415 (83) 346 (82) 157 (86) 79 (64) Intracerebral Hemorrhage 50 (10) 55 (13) 17 (9) 14 (11) Undetermined stroke 37 (7) 21 (5) 8 (4) 31 (25) Previous TIA 74 (15) 79 (19) (13) (15) Diabetes mellitus 93 (19) 101 (24) (14) (18) Previous AMI 59 (12) 83 (17) (13) (13) Atrial fibrillation 114 (23) 99 (23) (11) (22) History of hypertension 336 (67) 243 (57) (38) < (57) 0.23 Unconscious at time of stroke 68 (14) 62 (15) (11) (25) AMI indicates acute myocardial infarction; IRSD, Index of Relative Socioeconomic Disadvantage TIA, transient ischemic attack. P values < 0.1 in bold. *Never smokers vs ex-smokers. Never smokers vs current smokers. Never smokers vs unknown smokers. Percentage of column total (n). Unknown or missing data: place of birth (4 never smokers, 1 ex-smoker, 6 unknown smokers), residence (2 never smokers, 1 ex-smoker, 10 unknown smokers), previous TIA (1 never smoker, 1 current smoker, 4 unknown smokers), diabetes mellitus (1 never smoker, 1 ex-smoker, 5 unknown smokers), previous AMI (1 ex-smoker, 6 unknown smokers), atrial fibrillation (3 never smokers, 5 ex-smokers, 3 current smokers, 9 unknown smokers), and history of hypertension (1 never smoker, 2 ex-smokers, 8 unknown smokers). Smoking was not independently associated with 28-day fatality after stroke onset (Supplemental Table 3). Fatality at 28 days after stroke onset was independently associated with age, socioeconomic status, atrial fibrillation, unconsciousness at the time of stroke, and stroke subtype. Smoking was associated with outcome at 10 years poststroke in 28-day survivors with ischemic stroke at baseline (Supplemental Table 4 and Figure 1a) but not in those who suffered an intracerebral hemorrhage at baseline (Supplemental Table 5 and Figure 1b). Discussion The major finding from this study is that people who were current or ex-smokers at the time of their stroke had poorer long-term outcomes than those who had never smoked. This association only became apparent after adjusting for age, as smokers had their vascular events (or died) at a much younger age than those who had either never smoked or had ceased smoking. It is unclear why this relationship is present in

4 3176 Stroke December 2012 Table 2. Adjusted Cox Regression Analysis for First Vascular Event or Death in Patients Who Survived the First 28 Days After Their Index Stroke Age* Univariable Multivariable Variable HR 95% CI P Value HR 95% CI P Value < < < < <0.001 > < <0.001 Male* IRSD (socioeconomic status)* Most disadvantage Moderate disadvantage Least disadvantage Born in Australia < Recurrent stroke at baseline < <0.001 Atrial fibrillation < Previous AMI < Diabetes mellitus Unconscious at time of stroke < <0.001 Smoking status Never smoker Ex-smoker Current smoker Institutionalized < History of hypertension Previous TIA Stroke subtype Ischemic stroke 1.00 Intracerebral Hemorrhage Undetermined stroke <0.001 n=1084 in multivariable analysis: patients with unknown smoking status were excluded, patients with recurrent vascular event or death within 28 days were excluded. 802 experienced a nonfatal vascular event or death within 10 years poststroke. Vascular event: stroke or myocardial infarction. P Values <0.1 in bold. AMI indicates acute myocardial infarction; CI, confidence interval; HR, hazard ratio; IRSD, Index of Relative Socioeconomic Disadvantage; TIA, transient ischemic attack. *Potential confounding factors. vthose with ischemic stroke and not in those with intracerebral hemorrhage, but this may be attributable to a small sample size in those with intracerebral hemorrhage. Our findings may be different from those seen in other studies because our sample size was larger, therefore providing greater power to detect a difference between groups. In addition, we used a composite outcome measure incorporating recurrent strokes, myocardial infarctions, and death. Our finding is important as it provides further evidence of the poor outcome among those who smoke. It is particularly important because we demonstrated that smoking is having its greatest impact among those who are younger. Because of the potential years of healthy life that is lost, smoking will indirectly be making a major contribution to the costs of stroke. Smokers in our study, being younger than never smokers, were less likely to have atrial fibrillation and hypertension than never smokers Other investigators have shown that after adjustment for age, smokers have a greater risk of hypertension, 16 abdominal obesity, 17 dyslipidaemia, 18 atrial fibrillation, 19 and diabetes mellitus. 20 When these risk factors coexist they appear to potentiate the risk of vascular events such as heart attack and stroke. 21 We found that atrial fibrillation and diabetes were independently associated with poorer long-term outcomes after stroke. It would appear that targeting education towards younger people, males, and those from disadvantaged backgrounds about the risks associated with smoking would be an effective way to reduce the burden of stroke. Study Limitations This study has some limitations. First, a large proportion of participants residing in aged care institutions were not interviewed at baseline. This meant that there was missing information in many of the variables, including smoking status. If these institutionalized stroke survivors differ in some way from noninstitutionalized stroke survivors, then the fact

5 Kim et al Smoking Status and Risk of Death or Vascular Event 3177 Figure 1. Cumulative hazard of vascular event or death after stroke. Adjusted for age, sex, socioeconomic status, place of birth, stroke subtype, history of stroke, history of acute myocardial infarction, history of atrial fibrillation, history of diabetes mellitus, unconsciousness at time of stroke, and place of residence Figure 2. Cumulative hazard of vascular event or death (in 28-day survivors of stroke). Adjusted for age, sex, socioeconomic status, place of birth, history of stroke, history of acute myocardial infarction, history of atrial fibrillation, history of diabetes, unconsciousness at time of stroke, and place of residence that they are underrepresented may have affected our findings. Second, inaccuracies in patient classification are likely to exist as risk factors that were not mentioned in the medical history were presumed to be absent, such as for smoking status. This may mean that this variable is associated with some misclassification bias. Third, because smoking status was ascertained from inspection of medical records, there was limited information on duration and number of cigarettes smoked by a patient. Therefore, we were unable to assess dose response of smoking on outcome. The limited information on smoking status also meant that we were unable to distinguish between those who had ceased smoking soon before their stroke and those who had quit smoking much earlier. Another limitation is that we did not assess the cessation of smoking in all participants who were current smokers at the time of their stroke. From a previous substudy we know that 37% of patients who smoked at baseline had quit at 5 years poststroke and that most of them had quit immediately after their stroke. 22 These limitations in smoking status classification mean that the differences seen between smoking groups is likely to have been attenuated. Finally, our multivariable model does not include all factors potentially associated with outcome after stroke. For example, data on obesity, serum cholesterol levels, alcohol consumption, and family history of stroke were unavailable at baseline. In future stroke studies, the timing and dose of smoking should be assessed, similar to the assessments conducted in studies of the effect of smoking on outcome after AMI. 6 8 In these studies, differences in survival between pre-ami and post-ami quitters have been investigated, as has the effect of cigarette reductions on mortality in persistent smokers. Similarly, it would be useful to assess the association between mortality and smoking cessation pre- and poststroke. Conclusions We found that smoking was associated with recurrent stroke, myocardial infarction and death after stroke. Those who quit smoking before their stroke had a lesser risk of death, recurrent stroke and myocardial infarction at 10 years post stroke than those who were smokers at the time of stroke. People with stroke who are younger, or male, or living in more disadvantaged areas are more likely to be smokers and these groups should be targeted for smoking cessation for primary prevention of stroke. Acknowledgments The contribution of the following research nurses is acknowledged: Sue Mosley, Mary Staios, and Dennis Young. Li Chun Quang provided assistance with data management. The National Death Registry was used to follow up deaths in the cohort. Sources of Funding This work was supported by grants from the National Health and Medical Research Council (154600, , ), the Victorian Health Promotion Foundation, the Foundation for High Blood Pressure Research, and the National Stroke Foundation. Dr Kim was supported by a Heart Foundation of Australia (co-funded with National Stroke Foundation) postgraduate scholarship (PP 10M 5505). Dr Thrift was supported by a NHMRC Fellowship (438700). None. Disclosures References 1. O Donnell MJ, Xavier D, Liu L, Zhang H, Chin SL, Rao-Melacini P, Rangarajan S, Islam S, Pais P, McQueen MJ, Mondo C, Damasceno A, Lopez-Jaramillo P, Hankey GJ, Dans AL, Yusoff K, Truelsen T, Diener HC, Sacco RL, Ryglewicz D, Czlonkowska A, Weimar C, Wang X, Yusuf S; INTERSTROKE investigators. Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study): a case-control study. Lancet. 2010;376: Teo KK, Ounpuu S, Hawken S, Pandey MR, Valentin V, Hunt D, Diaz R, Rashed W, Freeman R, Jiang L, Zhang X, Yusuf S; INTERHEART Study Investigators. Tobacco use and risk of myocardial infarction in 52 countries in the INTERHEART study: a case-control study. Lancet. 2006;368: Howard G, Wagenknecht LE, Burke GL, Diez-Roux A, Evans GW, McGovern P, Nieto FJ, Tell GS. Cigarette smoking and progression of atherosclerosis: the Atherosclerosis Risk in Communities (ARIC) Study. JAMA. 1998;279: Ambrose JA, Barua RS. The pathophysiology of cigarette smoking and cardiovascular disease: an update. J Am Coll Cardiol. 2004;43:

6 3178 Stroke December Benowitz NL. Cigarette smoking and cardiovascular disease: pathophysiology and implications for treatment. Prog Cardiovasc Dis. 2003;46: Gerber Y, Rosen LJ, Goldbourt U, Benyamini Y, Drory Y; Israel Study Group on First Acute Myocardial Infarction. Smoking status and longterm survival after first acute myocardial infarction a population-based cohort study. J Am Coll Cardiol. 2009;54: Kinjo K, Sato H, Sakata Y, Nakatani D, Mizuno H, Shimizu M, Sasaki T, Kijima Y, Nishino M, Uematsu M, Tanouchi J, Nanto S, Otsu K, Hori M; Osaka Acute Coronary Insufficiency Study (OACIS) Group. Impact of smoking status on long-term mortality in patients with acute myocardial infarction. Circ J. 2005;69: Rea TD, Heckbert SR, Kaplan RC, Smith NL, Lemaitre RN, Psaty BM. Smoking status and risk for recurrent coronary events after myocardial infarction. Ann Intern Med. 2002;137: Dhamoon MS, Tai W, Boden-Albala B, Rundek T, Paik MC, Sacco RL, Elkind MS. Risk of myocardial infarction or vascular death after first ischemic stroke: the Northern Manhattan Study. Stroke. 2007;38: Mohan KM, Crichton SL, Grieve AP, Rudd AG, Wolfe CD, Heuschmann PU. Frequency and predictors for the risk of stroke recurrence up to 10 years after stroke: the South London Stroke Register. J Neurol Neurosurg Psychiatr. 2009;80: Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, Stewart-Wynne EG. Five-year survival after first-ever stroke and related prognostic factors in the Perth Community Stroke Study. Stroke. 2000;31: Hatano S. Experience from a multicentre stroke register: a preliminary report. Bull World Health Organ. 1976;54: Sudlow CL, Warlow CP. Comparing stroke incidence worldwide: what makes studies comparable? Stroke. 1996;27: Pink B. An introduction to Socio-Economic Indexes For Areas (SEIFA). Australian Bureau of Statistics (ABS), Canberra Thygesen K, Alpert JS, White HD; Joint ESC/ACCF/AHA/WHF Task Force for the Redefinition of Myocardial Infarction. Universal definition of myocardial infarction. Eur Heart J. 2007;28: Bowman TS, Gaziano JM, Buring JE, Sesso HD. A prospective study of cigarette smoking and risk of incident hypertension in women. J Am Coll Cardiol. 2007;50: Canoy D, Wareham N, Luben R, Welch A, Bingham S, Day N, Khaw KT. Cigarette smoking and fat distribution in 21,828 British men and women: a population-based study. Obes Res. 2005;13: Masulli M, Riccardi G, Galasso R, Vaccaro O. Relationship between smoking habits and the features of the metabolic syndrome in a non-diabetic population. Nutr Metab Cardiovasc Dis. 2006;16: Heeringa J, Kors JA, Hofman A, van Rooij FJ, Witteman JC. Cigarette smoking and risk of atrial fibrillation: the Rotterdam Study. Am Heart J. 2008;156: Willi C, Bodenmann P, Ghali WA, Faris PD, Cornuz J. Active smoking and the risk of type 2 diabetes: a systematic review and meta-analysis. JAMA. 2007;298: Whisnant JP. Modeling of risk factors for ischemic stroke. The Willis Lecture. Stroke. 1997;28: Gall SL, Dewey HM, Thrift AG. Smoking cessation at 5 years after stroke in the North East Melbourne Stroke Incidence Study. Neuroepidemiology. 2009;32: