Factors Affecting Formation and Growth of Intracranial Aneurysms
A Long-Term Follow-Up Study
Background and Purpose—We sought to investigate factors determining growth rate of unruptured intracranial aneurysms as well as formation of new (de novo) aneurysms in patients from a time period when unruptured aneurysms were not treated surgically.
Methods—Eighty-seven patients (79 had ruptured aneurysms clipped at start of follow-up) with 111 unruptured aneurysms as well as an additional 7 patients (2 with and 5 without unruptured aneurysms) who developed new aneurysms were followed from the 1950s to the 1970s until death or subarachnoid hemorrhage or until the last contact. Patients’ cerebral arteries were examined later either with conventional (control) angiography (n=38) and/or, for those alive during 1996–1998, with 3-dimensional CT angiography (n=57). In addition, 10 patients were studied at neuropathological autopsy.
Results—Mean±SD duration of follow-up was 18.9±9.4 years (range, 1.2 to 38.9 years). Unruptured aneurysms increased in size ≥1 mm in 39 of the 87 patients (45%) and ≥3 mm in 31 (36%). New aneurysms were found in 15 of the 89 patients and in 5 without an unruptured aneurysm at the beginning of follow-up. Aneurysm rupture was associated very significantly (P<0.001) with aneurysm growth during follow-up. Of several potential risk factors tested, only cigarette smoking (odds ratio [OR], 3.92; 95% CI, 1.29 to 11.93) and female sex (OR, 3.36; 95% CI, 1.11 to 10.22) were, after adjustment for age, significant (P<0.05) independent risk factors for occurrence of aneurysm growth of ≥1 mm. Only cigarette smoking (OR, 3.48; 95% CI, 1.14 to 10.64; P<0.05) was associated with growth of ≥3 mm. Age- and hypertension-adjusted risk factors for aneurysm formation were female sex (OR, 4.73; 95% CI, 1.16 to 19.38) and cigarette smoking (OR, 4.07; 95% CI, 1.09 to 15.15).
Conclusions—Women and cigarette smokers are at increased risk for intracranial aneurysm formation and growth. Cigarette smoking in particular hastens aneurysm growth. Cessation of smoking is important for patients with unruptured aneurysms and possibly also for those with a prior subarachnoid hemorrhage.
The prognosis for aneurysmal subarachnoid hemorrhage (SAH), despite improvements in medical and neurosurgical treatment, has remained largely unaffected. Overall case-fatality still reaches 40% to 50%.1 2 An additional 10% to 20% of patients will survive but be severely disabled, with only approximately 40% of SAH patients recovering to reach independent status. SAH is also the most common form of stroke and cause of stroke mortality among young adults.3
Aneurysms develop during adulthood, with risk for SAH increasing linearly with age.1 3 4 5 During the last few decades, several studies have been published concerning risk factors for SAH (ie, for ruptured intracranial aneurysm)4 5 6 7 8 and for future rupture of a verified unruptured aneurysm4 5 9 10 11 12 but not for formation and growth rate of intracranial aneurysms. Because fewer than half of all aneurysms ever rupture, it is essential to determine separately the risk factors for both aneurysm formation and its growth.4 5 12 This is also important for detection of mechanisms by which different risk factors (eg, cigarette smoking) increase the risk for SAH. Previously, growth rate of an aneurysm has also been associated with its rupture.10 13
In this long-term cohort study, patients with unruptured aneurysms diagnosed before 1979 were followed by conventional angiography and/or by 3-dimensional (3D) CT angiography or examined at autopsy to define risk factors for aneurysm growth and formation.
Subjects and Methods
The study comprised 94 patients (mean age, 38.7 years; range, 14.6 to 60.6 years; 56 [60%] women; 89 with unruptured aneurysms) who were at the Department of Neurosurgery, Helsinki University Central Hospital, during 1956–1978, for aneurysm evaluation and whose aneurysms were later investigated. Before 1979, unruptured aneurysms were not operated on in our clinic.10 12 Only 4 of all 142 patients with unruptured aneurysms diagnosed before 1979 who were alive and had not been studied with control angiography never came to the outpatient department. They were elderly (>80 years) and/or had a severe incapacitating disease and were excluded from this study, as were those 49 patients with unruptured aneurysms who had already died without control angiography. These excluded patients (53 of 142) differed from the patients of the present cohort only by their higher age (mean, 46.9 years; range, 24.7 to 60.7 years) at diagnosis of unruptured aneurysm.
Of the 89 patients with unruptured intracranial aneurysms, 79 (89%) had SAH at the beginning of follow-up. Only the ruptured aneurysm was operated on, and the occlusion of the aneurysm was confirmed by postoperative angiography. Unruptured aneurysms of these patients were later measured either with conventional and/or 3D CT angiography (n=72) or at autopsy (n=7). Aneurysms of 8 patients with only unruptured aneurysms (4 symptomatic and 4 incidental) were monitored later either with angiography (n=7) or at autopsy (n=1). Two additional patients died of SAH caused by a verified de novo aneurysm. These patients are included in de novo analyses but not in aneurysm growth analyses because the unruptured aneurysms were not measured at autopsy. The study included not only patients with angiographic follow-up but also those whose aneurysms were measured at autopsy, since aneurysm growth or formation may also be associated with case fatality of SAH.
In addition, 5 patients had no unruptured aneurysm (4 with a single clipped ruptured aneurysm and 1 with normal angiography) before 1979 but later were treated for verified ruptured de novo aneurysms. These 5 patients were included only in risk factor studies of de novo aneurysms but not in incidence studies.
If the parent artery was clipped together with the aneurysm, the patient was excluded from the study because of the possibility of formation of a new aneurysm or the chance of enlargement of a preexisting aneurysm due to increase in hemodynamic stress in the remaining arteries.10 12
During 1996–1998, 57 patients were studied in the outpatient department, where they were personally interviewed (S.J.) with the use of a structured questionnaire, and patients’ cerebral vessels were studied with 3D CT angiography (K.P. and M.P.). The questionnaire elicited data on patients’ height and weight; previous diseases and hospital visits; regular drug use, including analgesics, stimulants, and narcotics; approximate intake of alcohol; current and previous smoking status; and family history of verified intracranial aneurysm cases in first-degree relatives (parents, siblings, and offspring).
Smoking was categorized as follows: never a smoker, formerly a regular cigarette smoker, or currently a cigarette smoker, with 10 and 20 cigarettes per day as cutoff points. Starting point and duration of smoking as well as time of cessation of smoking also were recorded. Alcohol consumption was recorded as approximate grams of absolute ethanol consumed within 1 week during the total follow-up period (standard drink=12 g of alcohol). Problem drinking was assessed by means of the CAGE questionnaire, described in detail elsewhere.7 8 12 Patients with ≥2 positive answers to the 4 questions were considered CAGE-positive, a sensitive indicator of previous and current excessive drinking (sensitivity and specificity 80% to 90%).7 8 12
In addition to the 57 subjects, an additional 37 patients who were already deceased had previously been followed for their aneurysms by conventional angiography (n=27) or at autopsy (n=10). During 1997–1998, questionnaires were completed during telephone interviews with relatives of these patients. Data on these patients were also available from medical records and from our previous follow-up studies and interviews in the 1970s and 1980s.10 12
In the outpatient department, blood samples were obtained for measurement of markers of alcohol intake: erythrocyte mean cell volume (MCV) (normal range, 80 to 96 fL) and γ-glutamyltransferase (GGT) (normal range, 0 to 44 U/L). Besides heavy drinking, cigarette smoking has also been shown to elevate MCV values.7 8
Recording of Blood Pressure Values
Blood pressure (BP) values were recorded before diagnosis and during follow-up. In patients with multiple BP measurements, values in the first quarter and last quarter of the follow-up were averaged. In the outpatient department, BP was measured twice with an interval of 30 minutes with the subject in a supine position after a rest of 15 minutes, with the average of these values recorded for statistical comparisons. Definite hypertension was defined as systolic pressure repeatedly >160 mm Hg, diastolic pressure >95 mm Hg, or use of antihypertensive medication.
All angiographies performed at the start and during follow-up as well as 3D CT (helical) angiographies in the outpatient department were examined by 2 neuroradiologists (K.P. and M.P.) who had no knowledge of the patients’ case histories. The size of each aneurysm was based on its greatest diameter measured in standard conventional angiographic projections and with 1.1-fold magnification of images taken into account; also recorded were location, shape (round, oval, or multilobed), and direction (14 different directions grouped into 3 main categories for statistical analyses: upward, downward, and horizontal). Conventional follow-up angiographies had been done for 38 patients (20 because of SAH from unruptured aneurysm, 5 because of SAH from a de novo aneurysm, and 13 to check unruptured aneurysms). Neither digital subtraction nor magnetic resonance angiographies were used in this study.
Those 57 patients admitted to the outpatient department were studied by 3D CT angiography to determine change in aneurysm size and shape and to determine whether de novo aneurysms had appeared. Conventional follow-up angiographies had been done for 27 patients (18 SAH cases), a 3D CT angiography alone for 46 patients (1 SAH case), and both of these for 11 patients (7 SAH cases).
Helical CT angiography was performed with a HiSpeed Advantage scanner (General Electric Medical Systems). Three-dimensional CT angiograms were made with a computer workstation (Advantage Windows, General Electric). A helical CT angiogram has been shown to be a safe, accurate, rapid, relatively noninvasive, and reliable method (sensitivity and specificity >85%) to show aneurysms as small as 2 mm in diameter.14 15 16 In 50 patients, only 4 of 62 aneurysms (6%) seen in previous angiography could not be seen in 3D CT angiography. These were aneurysms 2 to 4 mm in diameter located in the internal carotid artery close to bone structures of the skull base and were recorded to be of size similar to that in the primary angiography, ie, these had not grown. Similarly, one incidental aneurysm with conventional angiographic follow-up was recorded that had decreased in size from 26 to 8 mm, likely because of intra-aneurysmal thrombus formation.
Aneurysm growth rates were measured as the difference between maximum aneurysm diameter between the initial and the follow-up angiography/autopsy and as difference in diameter per follow-up year. The third growth rate variable was relative change in diameter per follow-up year (difference/initial diameter/follow-up year).
Data were analyzed with the Biomedical Data Package statistical programs (BMDP Statistical Software Inc, version 1993, release 7.0, University of California at Los Angeles) and with Statistical Product and Service Solutions (SPSS for Windows, release 188.8.131.529, SPSS Inc). Fisher’s exact 2-tailed test, Pearson χ2 test, test for linear trend, Mann-Whitney U test, Student’s t test, Wilcoxon signed rank test, paired t test, and Spearman rank correlation coefficients (rs) were used when appropriate.
Odds ratios (ORs) and 95% CIs of independent factors associated with occurrence of growth of an unruptured aneurysm and formation of a de novo aneurysm were analyzed by unconditional multiple logistic regression. Those variables tested were as follows: age at diagnosis; sex; body mass index (BMI) (calculated as weight/height2 [kg/m2]); size, location, and direction of the largest unruptured aneurysm; diameter of ruptured aneurysm in cases with prior SAH; presence of multiple unruptured aneurysms; aneurysm group (symptomatic or incidental compared with those with prior SAH); cigarette smoking status, current smoking, number of cigarettes smoked daily, and duration of smoking; alcohol consumption and CAGE status; source of information on the patient’s health habits; family history of intracranial aneurysms; definite hypertension; antihypertensive medication; and BP values.
A maximum-likelihood stepwise forward elimination procedure was used, with selection of variables to be added on the basis of the magnitude of their probability values (<0.05). Multiple linear regression analyses were performed to establish independent association between risk factors and aneurysm growth rate variables. A 2-tailed P-value <0.05 was statistically significant.
Aneurysm Growth Rate and Rupture
The largest aneurysm in each patient (n=87) grew significantly (P<0.01) during the follow-up (mean±SD, 2.5±3.7 mm; range, 0 to 17 mm; 0.31±0.86 mm/y; 12.1±42.5%/y). Corresponding growth rates in patients with angiographic follow-up (n=79) were 2.3±3.8 mm, 0.31±0.89 mm/y, and 11.4±43.7%/y. For patients with multiple unruptured aneurysms, the growth rate of the second largest aneurysm was correlated with that of the largest aneurysm (rs=0.387, n=27, P=0.046). In patients with a later SAH, the largest aneurysm always ruptured, except in one case in which the second largest aneurysm grew (13 mm) more than the aneurysm that was originally largest (1 mm) and then ruptured.
The average annual aneurysm rupture rate was 1.6% for the 87 patients (26 SAH cases/1648 person-years of follow-up) and 1.4% for the 79 patients with angiographic follow-up (21 SAH cases/1529 person-years of follow-up). Aneurysms that later ruptured had increased significantly (P<0.0001) more often in size (≥1 mm) than had the largest aneurysms in those without any rupture (26 of 26 [100%] versus 13 of 61 [21%]). Those aneurysms that ruptured had also grown significantly (P<0.0001) more than had the largest aneurysms in those without any rupture (6.3±4.2 versus 0.8±1.9 mm; 0.95±1.37 versus 0.04±0.09 mm/y; 38.3±72.0 versus 0.9±2.3%/y).
The mean±SD diameter of aneurysms at the time of rupture was 11.2±6.3 mm (11.4±6.8 mm when measured by angiography), which was significantly (P<0.0001) greater than the largest aneurysm (6.0±4.2 mm) of those with no aneurysm rupture by the end of follow-up. The diameter of the ruptured aneurysm in those with a fatal rupture (13.2±8.1 mm) was nonsignificantly higher than that of those with a nonfatal rupture (10.0±4.8 mm). However, if those ruptured aneurysms measured only at autopsy were estimated to be at least 40% greater before death, then mortality was significantly (P<0.05) associated with aneurysm size (15.3±8.0 versus 10.0±4.8 mm).
Aneurysm Growth Rate and Baseline Characteristics
Baseline characteristics by occurrence of growth (≥1 mm) of the largest unruptured aneurysm are shown in Table 1⇓. Unruptured aneurysms increased in size (≥1 mm) in 39 of 87 patients (45%) and in 33 of 79 patients (42%) who underwent angiographic follow-up. Aneurysm growth was ≥2 mm in 35 of 87 patients (40%) and ≥3 mm in 31 patients (36%).
Current smoking at the time of control measurement of aneurysm size was significantly associated with aneurysm growth (P<0.05) (Table 1⇑). Among those whose aneurysms had grown, 11 (32%) smoked >20 cigarettes daily compared with 8 (17%) of those without aneurysm growth. Current smoking also was more common (P=0.050) in men (56%) than in women (34%).
Growth rate variables for the largest aneurysm are shown according to sex and smoking status in Table 2⇓. Current smoking but not sex was associated significantly with growth rate variables. Growth rate did not differ between nonsmokers and those who quit during follow-up (1.38±2.52 versus 1.33±2.69 mm).
Spearman rank correlation coefficients between aneurysm growth rate and other continuous variables are shown in Table 3⇓. Since growth of an aneurysm was frequently associated with its rupture, follow-up time and age at end of follow-up were inversely correlated with aneurysm growth. BP values measured in the outpatient department in connection with 3D CT angiography were not associated with aneurysm growth rate (rs range, −0.148 to 0.179). Among current smokers and quitters, number of cigarettes smoked daily tended to correlate with aneurysm growth (rs range, 0.252 to 0.278), but duration of smoking as well as age at starting of smoking did not correlate (rs range, −0.023 to −0.203). Although MCV and GGT did not correlate with aneurysm growth (rs range, 0.068 to 0.160), MCV values were higher in current smokers (P=0.080) and GGT values higher in heavy alcohol drinkers (P<0.05).
Risk Factors for Aneurysm Growth
Multiple stepwise logistic regression showed that current cigarette smoking at end of follow-up and female sex were significant independent risk factors for aneurysm growth of ≥1 mm, but only current smoking was a risk factor for growth of ≥3 mm. ORs after adjustment for age at diagnosis are shown separately for all patients with unruptured aneurysms as well as for those with angiographic follow-up in Table 4⇓. After additional adjustment for hypertension or family history of aneurysms, only current smoking was a significant risk factor for aneurysm growth of ≥1 mm (adjusted OR, 4.20; 95% CI, 1.27 to 13.92). Cigarette smoking even at the beginning of follow-up significantly predicted later aneurysm growth (age- and sex-adjusted OR, 3.86; 95% CI, 1.37 to 10.89).
By stepwise linear regression, aneurysm growth rate variables both in all patients and in those with angiographic follow-up were associated independently only with current smoking at the end of follow-up. Standardized regression coefficients ranged from 0.220 to 0.293 (P=0.008 to P=0.054).
De Novo Aneurysms
During 1789 patient-years of follow-up in 89 patients with unruptured aneurysms, 15 developed a total of 19 de novo aneurysms in locations at which the initial angiography showed no aneurysm. Probability of de novo aneurysm formation cases was 0.84%/y (95% CI, 0.47% to 1.37%), and probability of aneurysms was 1.06%/y (95% CI, 0.64% to 1.65%). Patients with only 3D CT or conventional angiographic follow-up included 12 de novo cases during 1637 angiographic follow-up years, yielding a probability of 0.73% (95% CI, 0.38% to 1.28%) for average annual aneurysm formation cases.
Women had a significantly (P=0.027) higher risk for de novo aneurysm formation than did men (1.15%/y; 95% CI, 0.62% to 1.96% versus 0.30%/y; 95% CI, 0.04% to 1.09%, respectively), which could not be explained by difference in duration of follow-up (21.3±8.7 years for women and 18.3±10.2 years for men) or by the proportion of 3D CT angiographies by sex. The proportion of de novo aneurysms found in 3D CT angiographies was similar to that in conventional angiographies (15% versus 17%, respectively).
New aneurysm cases were found in current smokers (1.08%/y; 95% CI, 0.43% to 2.22%) nonsignificantly more often than in nonsmokers (0.46%/y; 95% CI, 0.15% to 1.08%). However, current smokers were followed for a shorter time than others (17.1±9.4 versus 24.0±7.7 years, respectively; P<0.001) since their follow-up was more often terminated by SAH from a formerly unruptured aneurysm. Presence of de novo aneurysms was also not significantly associated with age at diagnosis or at end of follow-up, diameter or growth rate of unruptured aneurysm, BMI, aneurysm group, later SAH from unruptured aneurysm, duration of follow-up, alcohol variables, BP values, or hypertension. Among those who used antihypertensive medication, presence of de novo aneurysms was less frequent than among those who did not (1 of 26 versus 14 of 63; P=0.058). There were too few patients (n=8) with a family history of aneurysms to find any association with de novo aneurysm formation.
De novo aneurysms developed during a mean follow-up time of 18.8±7.7 years, a period not differing significantly from that of those without de novo aneurysms. In cases in which a de novo aneurysm caused SAH, the follow-up time was somewhat shorter (14.7±8.1 years; range, 3.4 to 28.4 years). In addition to verified de novo aneurysms, 4 patients had small (3 to 5 mm) aneurysms at the tip of the basilar artery or in the posterior inferior cerebellar artery on 3D CT angiography, the presence of which had not been noted because of a lack of earlier vertebral angiograms. These were not considered de novo aneurysms.
Risk Factors for De Novo Aneurysm Formation
Of a total of 94 patients, 20 had de novo aneurysms (16 women, 4 men; 10 current smokers, 5 non–current smokers). Of these 20 patients, 5 had no unruptured aneurysms at beginning of follow-up. They later developed new aneurysms that caused SAH, but these patients survived to undergo angiography. Stepwise multiple logistic regression showed that female sex and current smoking at end of follow-up were the only significant (P<0.05) independent risk factors for de novo aneurysm formation. A logistic regression model after adjustment for age at diagnosis and definite hypertension is shown in Table 5⇓.
On the basis of these findings, current cigarette smoking and female sex seem to be factors affecting both aneurysm formation and growth. Women in particular were at high risk for aneurysm formation, and cigarette smoking hastened aneurysm growth. The faster the growth, the more likely was the rupture. Aneurysm size may also be associated with mortality after rupture.
Patients with a Prior SAH or Multiple Aneurysms
The results of this study cannot be broadly generalized since this patient population comprised mainly patients with prior SAH and multiple aneurysms, which are the most commonly diagnosed unruptured aneurysms. These patients may have an increased risk for rupture of an unruptured aneurysm.11 They also are younger11 and more likely cigarette smokers17 18 than those with an incidental or a single aneurysm. Of the 142 patients with unruptured aneurysms diagnosed in our clinic before 1979, the present study included 63%. They were relatively young at diagnosis since most older patients without a later SAH were already deceased because of a long follow-up at the time of planned CT angiography. However, the average annual aneurysm rupture rate of 1.4% to 1.6% among our patients did not differ from that of most other studies of unruptured aneurysms,4 5 9 12 suggesting that we have not studied a high-risk subgroup of patients.
Differences between SAH patients either with a single aneurysm or with multiple aneurysms can also be risk factors for aneurysm formation, since the presence of multiple aneurysms does not seem to increase the risk for rupture of an aneurysm.9 10 11 12 Indeed, the risk factors for aneurysm formation in this study were the same (female sex and cigarette smoking) as in previous studies of independent risk factors for multiple intracranial aneurysms.17 18 It is thus quite possible that the present study population differed from the general population or those with an incidental single aneurysm only in prevalence of cigarette smoking and of female sex.
Risk Factors for Aneurysm Formation and Growth
In this study univariate statistics revealed only slightly less than a significant association between female sex and aneurysm growth as well as between current smoking and de novo aneurysm formation. However, because cigarette smoking was more common in men, multivariate ORs became more significant than univariate ratios. Cigarette smoking is more frequent among men than among women, with this habit decreasing with age, in both SAH patients and general population.7 8 12 19
According to our study, women may have an OR as high as 4 for risk for aneurysm formation compared with men. The reason that women do not have a much higher risk for SAH may be due to aneurysm growth rate, which is important in aneurysm rupture. The growth rate itself did not differ by sex. Therefore, a marked difference in aneurysm formation could not have been seen on the basis of sex in autopsy studies.4 20 At autopsy, several aneurysms, especially small ones, can be missed, since an aneurysm is at least 30% to 60% larger before death than after death and before fixation.20
Since female preponderance for SAH is not significant until the fifth decade, the cause of aneurysm formation in women is thought to be secondary to hormonal factors21 ; it has been presumed that estrogen has an inhibitory effect on aneurysm formation, and cigarette smoking also has antiestrogenic properties. The collagen content of cerebral arteries may also diminish after menopause, favoring aneurysm formation.21
Cigarette smoking has been shown, in several studies,6 7 8 12 19 to increase risk for SAH in all age groups, but the mechanism by which smoking increases this risk has remained unknown. The prevalence of smoking in SAH patients ranges from 45% to 75%, whereas in the general adult population it is 20% to 35%.6 7 8 12 19 According to findings of our study, it is likely that smoking causes increased risk for SAH by the formation of an aneurysm and especially by hastening growth of a preexisting aneurysm. Of smoking habits, number of cigarettes smoked daily seems to be more important in terms of aneurysm growth than duration of smoking or age at which one began smoking. In addition, those who had quit smoking had not increased their risk for aneurysm growth. Therefore, cessation of cigarette smoking seems always to be justified for patients with either unruptured aneurysms or prior SAH.
Recent studies suggest that cigarette smokers’ plasma and artery wall elastase/α1-antitrypsin imbalance (ie, increased elastase activity and/or decreased α1-antitrypsin activity) may contribute to either aneurysm formation or SAH.22 23 24 25 Topical application of elastase (but of neither collagenase nor papaverine) experimentally to the artery wall causes saccular aneurysm formation, growth, and even rupture.26 Thus, it is quite possible that regular cigarette smoking increases elastase activity in the artery wall, and this, together with hemodynamic stress, can cause aneurysm formation and even hasten aneurysm growth, leading to rupture.
Risk for SAH From De Novo Aneurysm
Diagnosis of de novo aneurysm is rare, since these are almost always diagnosed after a rupture. If one assumes, on the basis of our previous study,12 that the risk of rupture of an unruptured aneurysm is approximately 1.3%/y and that the rate of formation of an aneurysm according to this present study is 0.84%/y, patients aged <60 years whose aneurysms have been treated successfully seem to have a risk for SAH of 11 per 100 000 per year from a de novo aneurysm. This incidence is even lower than that of the general adult population.1 3 4 The risk of suffering SAH from a potential de novo aneurysm seems to be so low that future angiographic monitoring cannot routinely be recommended to detect such lesions.
This research was supported in part by grants from the Maire Taponen Foundation and the Paavo Nurmi Foundation.
- Received July 13, 2000.
- Revision received October 9, 2000.
- Accepted October 13, 2000.
- Copyright © 2001 by American Heart Association
Fogelholm R, Hernesniemi J, Vapalahti M. Impact of early surgery on outcome after aneurysmal subarachnoid hemorrhage: a population-based study. Stroke. 1993;24:1649–1654.
Broderick JP, Brott TG, Duldner JE, Tomsick T, Leach A. Initial and recurrent bleeding are the major causes of death following subarachnoid hemorrhage. Stroke. 1994;25:1342–1347.
Sarti C, Tuomilehto J, Salomaa V, Sivenius J, Kaarsalo E, Narva E, Salmi K, Torppa J. Epidemiology of subarachnoid hemorrhage in Finland from 1983 to 1985. Stroke. 1991;22:848–853.
Weir B. Aneurysms Affecting the Nervous System. Baltimore, Md: Williams & Wilkins; 1987.
Longstreth WT Jr, Nelson LM, Koepsell TD, van Belle G. Cigarette smoking, alcohol use, and subarachnoid hemorrhage. Stroke. 1992;23:1242–1249.
Juvela S, Hillbom M, Numminen H, Koskinen P. Cigarette smoking and alcohol consumption as risk factors for aneurysmal subarachnoid hemorrhage. Stroke. 1993;24:639–646.
Juvela S. Risk factors for multiple intracranial aneurysms. Stroke. 2000;31:392–397.