Secondary prophylactic treatment and long‐term prognosis after TIA and different subtypes of stroke. A 25‐year follow‐up hospital‐based observational study

Abstract Objectives To assess long‐term prognosis after transient ischemic attack (TIA)/subtypes of stroke relative to secondary prophylactic treatment(s) given. Materials and Methods Retro/prospective follow‐up of patients hospitalized in the Stroke Unit or in the Department of Neurology, Linköping, in 1986 and followed up to Feb. 2011. Results A total of 288 men were followed up for 2254 years (mean 7.8 years) and 261 women for 1984 years (mean 7.6 years). In men, the distribution to anticoagulants (AC) (warfarin treatment) was 18%, antiplatelet therapy (APT) usually ASA 75 mg/day 54%, untreated 27%, unknown 2%. In women, the distribution to AC was 15%, APT 60%, untreated 23%, unknown 2%, respectively. Mortality rates at 1 year, 10 years, and 25 years for men were 21%, 67%, and 93%, respectively, versus the rates in women of 24%, 71%, and 90%, respectively. Survival curves showed markedly increased risk of death compared to the normal population. AC treatment was more favorable for men regarding the annual risk of stroke, compared with APT (9.4% vs. 9.8%), as well as the risks of MI, (5.6% vs. 6.7%), and death (8.1% vs. 10.3%), compared to women for stroke (11.6% vs. 8.8%) and MI (5.3% vs. 3.7%) but not for death (8.3% vs. 8.4%). The risk of fatal bleeding was 0.86% annually on AC compared to 0.17% on APT. According to Cox regression analysis included patients with TIA/ischemic stroke, first‐line treatment had beneficial effects on survival: AC OR 0.67 (0.5–0.9), APT 0.67 (0.52–0.88) versus untreated. Conclusions Patients with a history of TIA/stroke had a higher mortality rate versus controls, providing support for both primary and secondary prophylaxis regarding vascular risk factors for death. This study also provided support for secondary prophylactic treatment with either AC or ASA (75 mg once daily) to reduce the vascular risk of death unless there are contraindications.

Atrial fibrillation (AF) is an important risk factor for cardiac embolism, especially in combination with other risk factors, with a high risk of severe stroke and/or recurrence (Friberg, Benson, Rosenqvist, & Lip, 2012;Goto et al., 2008;Kim et al., 2011). In primary or secondary prophylactic treatment of patients with AF, AC and APT have each been shown to be a better alternative than a placebo, with warfarin better than APT but with an increased risk of bleeding (Alberts, Eikelboom, & Hankey, 2012;Blackshear et al., 1996;Connolly et al., 2008;Fuster et al., 2006;Hart, Pearce, & Aguilar, 2007;Hylek et al., 2003;Laupacis et al., 1994). Novel oral anticoagulants (NOACs) have been reported to have the same effects or to perform even better than warfarin with an equal or lower risk of major bleeding (Alberts et al., 2012;Hankey, 2014;Hori et al., 2013;Lopes et al., 2012), and NOACs have shown large differences in their favor regarding the risk of stroke or systemic embolism, compared with ASA (Alberts et al., 2012;Diener et al., 2012).
Several randomized trials using secondary prophylactic treatment, either with AC or APT or conducting comparisons between AC and APT after TIA/stroke due to arterial thromboembolism have found APT to have the same effects as AC, or APT has been deemed to be a better option because of a lower risk of major bleeding or other factors (Antithrombotic Trialists' Collaboration, 2009;Campbell, Smyth, Montalescot, & Steinhubl, 2007;De Schryver, Algra, Kappelle, van Gijn, & Koudstaal, 2012;Gouya et al., 2014;Hankey, 2014;Lemmens, Chen, Ni, Fieuws, & Thijs, 2009;Maasland et al., 2009;Sandercock, Counsell, Tseng, & Cecconi, 2014). Urgent secondary prophylactic treatment(s) decrease the risk of (recurrent) stroke considerably after TIA or minor stroke due to arterial thromboembolism (Rothwell et al., 2007). However, the risk of recurrent stroke did not decrease over time with APT in patients with arterial embolisms (Lemmens et al., 2009). A higher dose of aspirin than 75-100 mg once daily explained the increased risk of side effects, but it did not provide any better protection regarding the outcomes of cardiovascular events (Campbell et al., 2007).
Patients with either TIA and/or an ischemic or hemorrhagic stroke have besides an increased risk of suffering (recurrent) stroke, increased risks of experiencing myocardial infarction (MI) or a vascular cause of death over long-term observation (Appelros, Gunnarsson, & Terent, 2011;Brønnum-Hansen, Davidsen, & Thorvaldsen, 2001;Burns et al., 2011;Dhamoon, Sciacca, Rundek, Sacco, & Elkind, 2006;Eriksson & Olsson, 2001; Hardie, Hankey, Jamrozik, Broadhurst, Touze et al., 2005). Warfarin alone or in combination with aspirin was superior to aspirin alone regarding endpoints, but it increased the risk of major, nonfatal bleeding after MI (Hurlen, Abdelnoor, Smith, Erikssen, & Arnesen, 2002). In fact, the long-term effects of administering AC/APT after TIA/stroke due to arteriosclerotic disease for the remainder of the patient′s life remains unknown (Cleland, 2006). The present observational trial included patients with TIA/stroke hospitalized in the Stroke Unit or in the Department of Neurology in 1986. The purpose of this trial was to report the long-term prognosis after TIA/subtypes of stroke relative to secondary prophylactic treatment(s) administered or not over a long-term observation period, and to determine predictors of stroke, MI and death.

| MATERIALS AND METHODS
This study enrolled 549 patients, of whom 362 patients (66%) (M 171, W 191)  The criteria for other diagnoses and classifications were provided in a previous study (Eriksson & Olsson, 2001). Cause of death was based on the underlying cause defined by the World Health Organization (WHO) as the disorder that started a chain of events leading to death.
Previous study had reported predictors of stroke, and death during long term observation time, but the using of secondary prophylactic treatment only in the acute phase (Eriksson & Olsson, 2001). The aim of this study was to report the long-term prognosis for TIA and different subgroups of stroke with a focus on documenting events of strokes, myocardial infarctions, deaths, depending on the diagnosis, sex and type of secondary prophylactic treatment(s) (AC, APT) administered or not during the observation time. Other aims of the study was to report other factors that could be of importance for the long-term course: such as the occurrence of cancers in this cohort of patients, occurrence of atrial fibrillation on ECG (not known before), prescription of lipid-lowering drugs/ACE inhibitor or angiotensin-IIreceptor blockers.
The study began in 2003 after approval by the Ethics Committee of the University Hospital in Linköping, and a completed application was filed in 2008. The results reported in this trial were based on data reported previously (Eriksson & Olsson, 2001), as well as new data supplied in medical records from different clinics, general practitioners, cancer registries, death certificates, and autopsy reports, which were all scrutinized, and from complete telephone interviews and/or (in a few cases letters to) patients or their relatives/caregivers six times since 2003. All of the included patients were followed up to Feb. 2011

| Anticoagulant/antiplatelet therapy
Patients were treated after admission either with AC (usually in combination with heparin for the first few days) or with APT, which usually consisted of half a tablet of Albyl ® Minor (Recip) 250 mg given once daily, except patients with reduced consciousness/severe disability and/or e.g. history of a bleeding disorder. For each patient, the benefit/risk profile assessment was made before the choice of treatment, or no treatment after having ruled out another cause of the patient's illness. The results from the analysis of International Normalized Ratio (INR) values among the AC-treated patients have not been reported, but the recommendation was that the INR values should be 2-3. After 1990, half a tablet of Albyl ® Minor was replaced by Trombyl ® 75 mg once daily. Events of strokes, myocardial infarctions, and deaths were classified according to the treatment(s) given. Short interruption(s) of treatment due to surgery or other reasons inside or outside the hospital did not change this classification of treatment. Procedure-related events of strokes/MIs were classified by the secondary prophylactic treatment that was given before the surgery. Fatal index strokes were reported at the latest treatment, whereas fatal recurrent strokes were classified as either AC or APT given or both or untreated at the time of the event of stroke even if >1 month had elapsed from ictus to death. This classification was also applied in patients with other causes of death if secondary prophylactic treatment was stopped due to a serious disease. Of patients treated with AC plus APT included only AC in the observation period unless otherwise stated.
The study reports events of strokes, MIs, deaths per 100-years of observation time. Intention to treat analysis included all of the patients with TIA/ischemic stroke. year. Systolic/diastolic blood pressure was a continuous variable with a linear increase in risk per mm Hg increase in blood pressure.

| RESULTS
The baseline characteristics of the 549 included patients, of whom 261 patients (48%) were women, are shown in Table 1. Among patients admitted from the catchment area of Linköping 53% of all ischemic strokes occurred in women, of which 29% were cardioembolic (CE) cerebral infarctions, compared to 19% in men (p = .056).
Among women with CE, 72% were older than 74 years old vs. 59% in men (ns). The corresponding distributions of intracerebral hemorrhage (ICH) were 55%, and 45%, respectively. Among women, 81% were older than 74 years old vs. 31% in men (ns). Approximately, one-fourth of the included men admitted from the catchment area of Linköping with stroke had a history of a previous stroke versus one fifth of the women. Among the patients admitted due to TIA, 38% of the men had remaining symptoms and signs at admission, compared with 18% of the women. In total, 15% of the men and 13% of the women had TIAs <3 months before the latest symptoms of TIA/stroke (before stroke 10% and 11%, respectively). There were no statistically significant differences between sexes from onset of acute neurological symptoms and signs and admission (men, mean 3.1 days, std. 11.8, median 0, women 2.8; 11.5; 0). Among men and women, 11% and 9%, respectively, had been hospitalized for more than one day at another hospital/clinic before admission. CT -scans were performed in 77% of men and 67% of women. In patients with TIA/atherosclerotic cerebral infarction (ACI) with symptoms within the carotid artery supplying area, doppler/angiography was performed in 81 men (51%) versus 65 women (48%). Greater than 50% stenosis/occlusion of the symptomatic internal carotid artery was found in 33% of men versus 23% in women (ns). 11 (2) 95 (47) 88 (37) 19 (8) 17 (4) 26 (9) 12 (1) 21 (  .01

| Outcomes during the first month
.044 Missing value,% No CT-scan/autopsy in 44 cases. T A B L E 1 (Continued) for men on APT, and 54% in untreated men versus 62%, 38%, and 76%, respectively in women on AC, women on APT and untreated women, given total annual risks on "treatment" of 45%, 41% and 65%, respectively.

| Events of strokes and myocardial infarctions
Kaplan-Meier estimates of the probabilities of having a stroke for each diagnosis during the whole observation time in men and women are shown in Table 2. In total, 130 men (45%) suffered 199 strokes (55 fatal), compared with 123 women (47%) who had 187 strokes (66 fatal) (Figure 1) Eighty-three (42%) unspecified strokes occurred in men versus 90 (48%) in women (Table 3).
In men, the recurrence of TIA was reported in 47% compared to 35% in women, whereas the occurrence of TIA among stroke patients was considerably lower: in men, 12%; in women, 8%. TIA + stroke occurred in 12% of men versus 6% of women, whereas new events of either TIA and/or stroke occurred in 50% of men versus 51% of women. Stroke due to carotid surgery occurred in one man with a history of TIA and in 2 women with histories of ACI. In total, eight men had strokes <6 months after MI (<30 days, 4; >3 months, 2), and 5 women had had 6 strokes (<30 days in 5 cases). During the observation period, 53 men had had paroxysmal or permanent AF on ECG (not known before), and in total, 33 of these patients had 53 strokes, of whom 19 patients had 24 strokes after registered AF, whereas 20 of 30 women had 27 strokes, of whom 16 women had 19 strokes after recorded AF.

| Prescription of lipid-lowering drugs/ACE inhibitors or angiotensin-II-receptor blockers and occurrence of malignant cancers
During the entire observation period, a total of 29 men (10%) and 19 women (7%) were treated with statins, whereas treatment either with ACE inhibitors or with angiotensin-II-receptor blockers (ARB) drugs, was administered to 63 men (22%) and 47 women (18%).
Sixty-nine men had 80 malignant cancers (ICD-10, C00-C97), of which 22 had been diagnosed before the admission, compared with 74 malignant cancers in 71 women, of which 40 had been diagnosed before admission.

| Death due to a recurrent stroke
The risk of death due to recurrent stroke was 25% in women versus 19% in men. The differences between sexes existed mainly for the first recurrent stroke: 65% versus 47%. In total, 36 of these deaths occurred within one month after the onset of stroke in men, compared with 44 in women. The causes of death in men were as follows: stroke alone in 19 cases; in combination with pneumonia in 22; sepsis in 3; MI in one; heart failure in 5 and other causes in 5 (status epilepticus 1, generalized arteriosclerosis 3, and chronic bronchitis 1).

| Mortality with regard to diagnosis and sex in the Swedish population
During the observation period, 269 (93%) men died versus 235 (90%) women. Kaplan-Meier estimates were calculated of the probabilities of survival for each diagnosis during the observation period in men and women (  Figure 3, whereas the distribution of deaths over the long-term observation period is shown in Figure 4. Of the included patients with DM at admission, the mean age was 71.4 years, and 2% of the men survived; in women, the mean age was 73.2 years, and 0% survived. In men with increased fasting blood glucose (DM not known), the mean age was 72, compared to; 5%, in women with a mean age of 75.4; 7%. Of men with no diabetes the mean age was 68.4, compared to; 8%, in women with a mean age of 70.1; 15%. The differences between the groups existed mainly due to death from index stroke: known diabetes mellitus occurred in 15% of men and 19% of women; increased fasting blood glucose in 32% of men and 24% of women; and 8% and 9% of men and women, respectively, had no diabetes. Stroke during the observation period was more common than MI among the survivors: men, 63% versus 16% (p = .003); and women, 42% versus 19%, (p = .071). At the last follow-up, the risk of death due to cardiosclerosis, heart failure, or arrhythmias was higher in men than in women (p = .014). Fatal cancer occurred in 26 men (known cancer before admission in 6 cases) (9%) versus 13 women (known cancer before admission in 3 cases) (5%) (p = .069).
At 25

| Outcomes during secondary prophylactic treatment
Total observation time for 288 men was 2254 years (mean 7.8 years) and for 261 women 1984 years (mean 7.6 years). In men, the distribution of AC (warfarin treatment) was 18%, compared to APT in 54%, no treatment in 27%, and unknown in 2%; in women, the percentage on AC was 15%, compared to APT in 60%, no treatment in 23%, and unknown in 2%. Table 6  According to the intention to treat analysis, including all of the patients with TIA/ischemic stroke, patients allocated AC had lower annual risk regarding events of strokes, MIs, and death than with APT in men and women and APT versus no treatment, except for MI in men.

| Secondary prophylactic treatment and side effects, occurrences of venous thrombosis, pulmonary embolism and embolization not in the brain
Fatal bleeding occurred in 6 men, of whom 3 had bleeding on AC treatment alone (duodenal ulcer 1, ICH 1, SAH 1); in combination with APT there was one case of ICH, in 2 cases of APT ICH, as well as four cases in women (ICH on AC 2, ICH on APT 1, SAH untreated 1) ( Table 8). The risk of fatal bleeding was annually 0.86% on AC compared to 0.17% on APT. Venous thrombosis was reported at a rate of 0.4% per year in men on APT versus 0.5% in women.
Pulmonary embolism occurred in 8 men who were untreated (1.4% per year) versus in 12 women (2.6% per year). There were statistically significant differences between AC treatment versus no treatment regarding the risk of pulmonary embolism (p = .000), and between APT and no treatment (p = .005), but not between AC-treated versus APT-treated patients. In total, 86 events of venous thrombosis, pulmonary embolism, embolization (not in the brain) and/or side effects were reported in 63 men, compared to 69 reports in 55 women. The annual risks of complications reported above and side effects during AC treatment were 5.9% in men and 4.4% in women, while the corresponding values on APT in men and women were 3.7% and 3%, and in untreated men and women, 3% and 4.4%, respectively.

| Predictors of stroke, myocardial infarction and death
According to Cox regression analysis, ACI and CE versus ICH were shown to be predictors of an event of stroke (

| DISCUSSION
In this study, Kaplan-Meier analysis showed a higher risk among women than men of having a stroke during the first year −16.8% which can be compared with 22.4% (range 15-28%) after 4-5 years (Norrving, 2003). Log rank tests showed differences between diagnosis with an increased risk in men with CE versus TIA and ICH, as well as in women for CE versus TIA and LI and for; ACI versus LI. The risk of having a stroke in areas other than the index symptoms and signs was high in both sexes and the annual risk of stroke remained high for a very long time in both sexes. Analyses of the annual risk of stroke depending on the diagnosis and sex after 10 and after 25 years of observation, respectively, found that there was a reduction in the annual risk of stroke in men and women at index diagnoses of CE and ACI, but the annual risk of stroke was slightly increased for women with TIA and in men with ICH, while the annual risk of stroke was unchanged in men with TIA, LI, and in women with LI and ICH, respectively.
Of all strokes, 12.1% in men and 10.1% in women occurred among those patients who first had verified AF during the observation period. Few patients (1.4% men, 1.9% women) had a recurrent stroke <3 months after a MI. These figures were slightly higher than in other studies of in-hospital stroke or during the first month (Albaker et al., 2011;Witt et al., 2006).
In this trial Cox regression analysis showed index diagnoses of CE and ACI to be statistically significant predictors of having a recurrent stroke, and other important risk factors were previous stroke, history of diabetes mellitus/fasting blood glucose ≥6.1 mmol/L or hypertension /treatment with antihypertensive drugs. Compared with the previous study (Eriksson & Olsson, 2001), where the variable diagnosis was not included in the analysis: age, severity, previous stroke and systolic blood pressure were each reported as predictors for recurrent stroke. According to intention to treat, the annual risk of stroke in patients with TIA/ischemic stroke for patients allocated to AC was 7.7% in men and 8.4% in women; with allocation to APT, the rates were 10.8% and 11.4%, and in untreated patients, they were 12% and 11.4%, respectively. Lower risks on AC versus APT were reported in men with diagnoses of TIA and ACI but not of LI or for CE patients. In women, lower risks were found on APT versus AC in patients with TIA, ACI, and LI but not in patients with CE among the TIA/ischemic stroke patients. The findings that women's responses to AC treatment were better than those of men among patients with CE were previously reported (Laupacis et al., 1994). It is not likely to assume that differences from therapeutic INR 2-3 will explain these differences (Wieloch et al., 2011). Many recurrences with unchanged treatment occurring of one and the same patient is of great importance for the interpretation of results. Another explanation could be that it was the selection of few patients with high risk among the AC-treated men (Kim et al., 2011). The observation period of the patients with CE cerebral infarction in this trial could not be compared with NOAC trials in which the mean and median The cumulative risk of a first event of MI was slightly higher in men, but the differences from women were small except for during the last 5 years of the observation period: 65.3% (43.6-87%) versus 53.3% (42.5-64.1%). Log rank tests established differences between diagnoses regarding the risk of MI, with increased risks in patients with TIA and ACI versus CE and ICH in men and in women with ACI versus CE and ICH.
The predictors of MI were age, angina pectoris, systolic blood pressure, and diagnosis versus ICH, with almost the same high risk for patients with TIA and ACI, and a lower risk for patients with LI, with the lowest risk in patients with CE stroke. Other predictors were previous MI and heart failure. According to intention to treat analysis, the annual risk of MI among men on AC was 5.3% versus 3.8% in women; on APT the rates were 7.5% and 5.3%, and without treatment they were 4.3% and 7%, respectively. The low annual risk of MI among the untreated men could partly be explained by the low frequency of MIs early during the observation period and partly by these patients later receiving AC or APT. Including all of the patients, AC was a slightly better alternative than APT in men, with annual risks of MI of 5.6% versus 6.7%, and the differences with more favorable effects for AC treatment were found in patients admitted due to TIA, CE, and LI. In women, APT was a better alternative than AC with annual risks of 3.7% versus 5.3%, and differences in favour of APT were found in patients with TIA, ACI, and LI.
Stroke mortality has decreased in many countries, but the incidence has not (Feigin et al., 2009;Vaartjes et al., 2013). Ten treatable risk of MI is quite different than that for stroke (Yusuf et al., 2004), but hypertension is one important risk factor for stroke, MI or vascular cause of death (Carlson & Böttiger, 1985;Conroy et al., 2003;Håheim et al., 1993;Harmsen et al., 1990;Hu et al., 2005;Psaty et al., 2001;Qizilbash, Lewington, Duffy, & Peto, 1995;Seshadri et al., 2006) and antihypertensive treatment(s) reduces these risks (Hackam & Spence, 2007;Hankey, 2014;Mancia et al., 2013;Turnbull, 2003;Turnbull et al., 2005;. Survival after first or recurrent stroke has been reported to be <10% after 20 years, the risk factors for death were age, previous stroke, subtype of stroke, and different stroke features . Lower mortality rates for patients with TIA versus stroke after one year were reported by a study in Japan after hospital discharge (7% for stroke patients and 3.5% for TIA patients) (Kimura et al., 2005). Stroke units have improved the outcomes in patients with large-vessel infarcts but not in those with lacunar syndrome (Evans et al., 2002). In the present trial, there was also a quite different prognosis regarding survival for patients admitted due to TIA versus other ischemic subtypes of stroke after 15 years: TIA, men 38% (22-54.4%) and women 47% (23.4-70.8%); ACI, men 17% (11.1-23.1%) and women 20% (13.6-27%); CE, men 2% (0-6.7%) and women 4% Approximately 75% of all deaths were caused by death due to index stroke, or a recurrent stroke in both sexes during the first year, thereafter percentage of a recurrent stroke as the cause of death was higher in women than in men during the first 10 years. Fatal recurrent stroke had thereafter reduced percentage impact as a cause of death, even more after 20 years, for both sexes. Another vascular cause of death was of increased percentage significance in women between 10 and 20 years compared with men, but not after 20 years.
In total, the frequency of vascular causes of death was slightly lower in men 74.3% versus women 75.9%. Women had a higher risk of death due to index stroke/recurrent stroke 40.2% versus 33.3%; ns but a lower risk of death due to cancer 5% versus 9%. In patients with increased fasting blood glucose levels/diabetes mellitus, increased mortality rates have been reported (Emerging Risk Factors Collaboration, 2011;Eriksson & Olsson, 2001) as in the present trial where the differences existed mainly for index stroke. Men had a higher risk of death due to cardiosclerosis, heart failure, or arrhythmia versus women (p = .014), which could have partly depended on a higher frequency of several MIs. No classification was performed at admission to assess whether the heart failure was systolic or diastolic, but the prevalence and mortality rate were reported to be almost the same (Bhatia et al., 2006;Owan et al., 2006   2.6 (11) 6.4 (7) 0 5.7 (12) 3.7 (2) 5.9 (2) 6.6 (8) sensitivity cardiac troponin with a diagnostic threshold of 16 ng/L and through these means, improving outcomes due to tailored treatment (Shah et al., 2015).
The total risks of death per year, including all ICH patients, were 11.9% in men and 11.8% in women. Compared with the normal population, differences in survival increased after one year, including in all men and women. After more than 12 years in women versus 23 years in men, the differences regarding survival decreased compared with the first year and with the normal population, but there were great differences depending on TIA/subgroup of stroke. Including all patients' survival curves after 15 years showed higher mortality rate in men compared to women.
The previous study (Eriksson & Olsson, 2001) that did not include the variable secondary prophylactic treatment but the patient group with ICH in Cox regression analysis found that age, stroke severity, previous stroke, heart failure, diabetes/fasting blood glucose ≥6.1 mmol/L were each to be important predictors of death. In this study heart failure had been replaced by previous MI and hypertension as predictors but the other predictors were the same: this as the previous study supported adequate treatment for many risk factors for vascular death, both as primary and secondary prevention.
Cox regression analysis and intention to treat analysis, as analyses of events on given treatments provided support for secondary prophylactic treatment, either with AC or ASA 75 mg once daily, over the long-term observation period unless there were contraindications.
In this trial, AC treatment seemed to have been more favorable than APT in men, whereas women had similar or better effects of APT versus AC except for regarding CE cerebral infarction. In contrast, the risk of severe bleedings on AC treatment was higher than with no treatment (p = .023). Gastrointestinal bleeding is a common side effect of APT (Antithrombotic Trialists' Collaboration, 2009) and in the present trial, the annual risks either dyspepsia, gastrointestinal bleeding, or anemia were considerably higher with AC or ASA versus no treatment. Treatment with AC and APT combined increased the risk of severe ICH (Toyoda et al., 2009) and the intensity of anticoagulation (Blackshear et al., 1996;Connolly et al., 2008;Hylek et al., 2003). Åsberg, Henriksson, Farahmand, and Terént (2013) (Alberts et al., 2012;Diener et al., 2012;Hankey, 2014;Hori et al., 2013;Lopes et al., 2012)  2.6 (31)  2008). Many patients with "TIA" at admission were therefore not included in this trial. Another limitation of the study was that many risk factors were not analyzed, especially smoking habits (Carlson & Böttiger, 1985;Conroy et al., 2003;Eriksson, 1985