Modifiable Risk Factors for Subarachnoid Hemorrhage

Abstract

Given the relatively low incidence, high prehospital death rate, substantial geographical differences, and complex disease origin (combination of genetic and environmental risk factors), epidemiologic research on subarachnoid hemorrhage (SAH) and its risk factors is challenging. In practice, we are more or less forced to exploit compromised study designs and nonrepresentative data in such circumstances where it is almost impossible to gather comprehensive data through an optimal design. For example, hospital-based patient cohorts, administrative data repositories, and short-term population-based studies from small geographical regions are often used to research the incidence, case fatality, and risk factors of SAH, regardless of their inherent and self-evident limitations. Since studies on the epidemiology of SAH focus largely on identifying possible risk factors that could aid in disease diagnostics, treatment, and prevention, we aimed to review recent evidence on modifiable risk factors for SAH. In this context, we also try to explain the methodological reasons behind some of the conflicting results and to discuss the primary strengths and limitations of different study designs used in the field of SAH epidemiology. Based on our findings, smoking, high blood pressure, and possibly low physical activity are the only risk factors with high-quality evidence supporting their causal role in SAH. In addition, since all 3 commonly used study designs in SAH epidemiology, namely, hospital-based, population-based, and administrative register-based studies, have their own strengths and limitations, the most robust risk factor estimates and other epidemiologic measures of SAH can likely be established by combining various overlapping and high-quality sources of information in the future.

Introduction

Despite its relatively low and decreasing incidence,¹ subarachnoid hemorrhage (SAH) has remained one of the most common cardiovascular and neurologic causes of death and disability, especially among middle-aged people.² Although the prognosis of patients with SAH has improved over recent decades, up to one-fourth of patients with SAH still die before reaching hospital care.³ Considering the high SAH death rate and the substantial role of modifiable risk factors in SAH mortality and morbidity,⁴ evidence-based primary prevention strategies relying on high-quality epidemiologic estimates are of considerable interest when aiming to decrease the burden of SAH.

The relatively low incidence,¹ high prehospital death rate,³ substantial geographical differences,^1,3,5,6 and complex disease origin of both environmental and genetic components^7,8 make SAH epidemiology a challenging topic to study. In optimal scenarios, high-quality studies on SAH epidemiology, particularly focusing on its incidence and risk factors, should include decade-long follow-ups of hundreds of thousands of people from multiple geographical regions to achieve sufficient and representative sample sizes. Furthermore, prospectively collected data in such studies should consist of both genetic and environmental factors to evaluate the independent and causal effects of potential risk factors. Understandably, conducting such high-quality studies including all those features is often overly cumbersome. This has consequently led to the usage of more practical and convenient yet often suboptimal study designs with a higher risk of systematic biases, which may in turn lead to controversial and misinterpreted findings.

From the clinical point of view, the most important contemporary studies on SAH epidemiology focus on modifiable risk factors because most of these risk factors contribute to or even cause SAH. However, studies on risk factors of noncommunicable diseases, such as SAH, produce often conflicting results. Taking into account the importance of SAH risk factors, the occasional contradiction of the results, and the lack of comprehensive review articles within the past 2 decades, we aimed to review the recent literature on SAH risk factors and their potential pitfalls which could explain part of the controversial findings. Moreover, we discuss the strengths and limitations of different study designs and how their optimization could produce standardized high-quality incidence, case-fatality, and risk factor estimates of SAH in the future. Our structured search strategy in the PubMed database was based on the following terms (without specific language or publication date restrictions): (“subarachnoid hemorrhage” OR “subarachnoid haemorrhage”) AND (incidence OR prevalence OR deaths OR outcome OR mortality OR case-fatality OR burden OR “risk factors”) AND (population OR community OR epidemiology).

What Are the Independent and Modifiable Risk Factors of SAH?

Some previous studies have reported conflicting results on the modifiable risk factors of SAH and their independence. Although there is a consensus on the significance of smoking^7-14 and high blood pressure^7-12,15 as the most important risk factors for SAH, findings concerning other risk factors, such as alcohol consumption,^{8-11,13,16-19} physical activity,^9,12,18-23 adverse lipid profile,^12,24,25 obesity,^{10,12,18,19,26,27} diabetes,^{10,12,16,19,28-30} and the use of hormone replacement therapy (HRT),^16,31,32 are more puzzling. In the following chapter and Table 1, we aim to summarize the best available evidence on the most frequently reported modifiable risk factors for SAH. In addition, we provide clarification on the factors behind the conflicting results and give concrete examples of shortcomings in epidemiologic methodology that contribute to misinterpretations.

Table 1.

Available Evidence on Modifiable Risk Factors of SAH

Risk factor	Associative evidence of conventional observational studies	Causal evidence of MR and twin studies	Overall effect
Smoking	High-quality population-based studies have reported that smoking increases the risk of SAH8-11	MR7,12 and twin13 studies have reported a causal association between smoking and SAH	Causes SAH
High blood pressure	High-quality population-based studies have reported that high systolic and diastolic blood pressure increase the risk of SAH8-11	MR studies have reported a causal association between high blood pressure and SAH7,12,15	Causes SAH
Low physical activity	High-quality population-based cohort studies9,20 have reported that low physical activity associates with an increased risk of SAH	MR studies have reported a suggestive causal association between low physical activity and SAH12,18,19	Probably causes SAH
Adverse lipid profile	High-quality population-based studies have reported that high levels of total cholesterol, high levels of LDL-cholesterol, and low levels of HDL-cholesterol associate with an increased risk of SAH, especially in men24,25	Some MR studies18,33 have reported inverse associations between SAH risk and LDL-cholesterol, HDL-cholesterol and triglyceride levels, but no evidence occur by sex. Other MR studies have not found significant associations12,18	May associate with SAH risk but inconclusive findings of causality
Alcohol consumption	Cohort and case-control studies16,17 have found an association between moderate to high alcohol consumption and SAH risk, but the independent effect from smoking and high blood pressure has remained unclear in high-quality studies8-11	MR18,19 and twin studies13 have reported no independent effect of alcohol use and SAH risk	No high-quality evidence for independent association
Obesity	Population-based studies have reported an inverse association between BMI and SAH risk, but this has been attributed to indirect effects via smoking and high blood pressure26,27,34	MR studies have reported modest causal relationships between obesity and SAH risk12,18,19	No high-quality evidence for independent association
Diabetes	Case-control and small cohort studies have reported an inverse association between diabetes and SAH risk, but high-quality evidence from large population-based cohort studies comprehensively considering possible confounders is lacking10,16,28,35	MR12,19,29 and twin studies30 have found no causal relationship between DM2 and SAH risk	No high-quality evidence for independent association
Female hormonal factors	Both inverse16,31 and positive32 associations have been reported between HRT and SAH risk. Similarly, findings in COC, age at menarche, and age at menopause are inconclusive.16,31 No high-quality studies exist among persons without exposure to smoking and high blood pressure	One MR study found a causal relationship between higher serum levels of SHBG and lower levels of BioT, and SAH risk.37 Two MR studies found no causal relationship between age at menopause and SAH risk37,38	No high-quality evidence for independent association
Air pollution	Ecological studies have reported conflicting results about the association between air pollution and SAH.4,42,43 No high-quality evidence with individual-level data on potential confounders exist	No MR or twin studies available	No high-quality evidence for independent association
Poor diet	Some evidence suggests that decreased fat intake, fruit consumption and following EAT-Lancet diet (sustainable and healthy diet) may associate with the decreased risk of SAH,44,45 but no high-quality studies with comprehensive SAH identification and confounder control are available	MR studies have shown no clear causal evidence between dietary factors and risk of SAH/IA18,19	No high-quality evidence for independent association
Sleep problems	Population-based studies have not found a significant association between poor sleep duration or quality and SAH risk46,47	Two MR studies12,18 found a causal relationship between insomnia and SAH but one46 found no association between long or short sleep duration and SAH risk	No high-quality evidence for independent association
Poor oral health	Hospital-based case-control and small population-based cohort studies have reported that severe periodontitis and gingivitis associate with an increased risk of SAH,48 but large, prospective, and population-based studies with comprehensive confounding control are not available	No MR or twin studies available	No high-quality evidence for independent association

Abbreviations: BioT = bioavailable testosterone; COC = combined oral contraceptives; DM2 = type 2 diabetes mellitus; HDL = high-density lipoprotein; HRT = hormone replacement therapy; IA = intracranial aneurysm; LDL = low-density lipoprotein; MR = Mendelian randomization; SAH = subarachnoid hemorrhage; SHBG = sex hormone-binding globulin; TC = total cholesterol.

Smoking and High Blood Pressure: The Causal Risk Factors of SAH

Although smoking and high blood pressure have consistently been associated with the increased risk of SAH in many high-quality population-based studies over several decades,^8-11 studies providing evidence of the causal relationships were lacking until recently. Although conventional observational studies can provide supportive evidence for cause-and-effect relationships, even high-quality population-based observational studies are prone to reverse causation as well as residual and unmeasured confounding. Randomized trials (gold-standard design to observe cause-and-effect relationships) are not ethically and practically suitable for assessing the causal relationship between risk factors and SAH, since risk-free participants cannot be randomized for widely known health risks (e.g., smoking and high blood pressure). Consequently, it has been challenging to prove the causal effects between different risk factors and SAH. However, during the past 5 years, evidence of the causal factors of SAH has started to emerge. In 2021, a large twin study reported that smoking was a key risk factor in explaining why only one of the twins suffered from fatal SAH.¹³ Since twins are known to share not only many of their genes but also many environmental factors, high-quality twin studies have been considered to be able to provide evidence from cause-and-effect relationships with a low risk of confounding effects. More recently, the causal effects of smoking and high blood pressure on SAH have also been shown by studies using Mendelian randomization (MR).^7,12,14,15 This analytic approach is somewhat analogous to randomized controlled trials and is based on inherited genetic variants that are randomly allocated at conception. By determining the genetic variants that are associated with a risk factor of interest (first core assumption of MR), have no association with known confounders (second core assumption of MR), and do not have a direct effect on the outcome of interest (third core assumption of MR), well-conducted MR studies can overcome the most important challenges of traditional observational studies to assess causality. In other words, since the genetic variants are randomly determined at conception, risk factor-associated variants precede any outcome of interest (minimal risk of reverse causation), while other variants of known and unknown confounders are assumed to distribute randomly (minimal risk of residual and unmeasured confounding). For SAH, previous MR studies^7,12,14,15 have shown 1.5–3 times higher odds of SAH among persons with genetic variants related to smoking initiation, duration, and consumption, as well as to high systolic and diastolic blood pressure values; findings that are also in line with the results of conventional observational studies.^8-11 Given the rather indisputable evidence, conclusions that smoking and high blood pressure not just associate with but actually cause SAH are justified, and can also be applied to clinical practice for health education of patients with SAH and other high-risk persons.

Low Physical Activity: Another Probable Cause of SAH

Low physical activity has also been associated with the increased risk of SAH by large, prospective, and population-based observational studies with comprehensive case identification and control for confounding factors.^9,20-23 The beneficial effect of physical activity has especially been related to regular activity during leisure time and commuting (often voluntary behavior for health-seeking purposes), whereas some evidence suggests that strenuous physical activity at work may not have such a favorable effect on the risk of SAH.²⁰ In addition to observational studies, several MR studies^12,18,19 have provided suggestive causal evidence of the benefits of moderate to vigorous physical activity regarding the risk of SAH. However, these findings are not as consistent as those for smoking and high blood pressure (presented in the previous section). More modest associations may be attributed at least partly to the fact that established genetic variants are associated with all types of physical activities, including physically strenuous work. Nevertheless, the positive effect of physical activities is also pathophysiologically plausible because regular physical activity improves endothelial function, decreases blood pressure, and lowers systemic and vessel wall inflammation, which are all key contributors to SAH development.²⁰ Therefore, the current evidence suggests that low physical activity may be another cause of SAH. Concerning patients with unruptured intracranial aneurysms (UIAs), there is no need to avoid physical activity to decrease the rupture risk of UIAs. In fact, in light of current evidence, patients with UIA and people with an elevated risk of SAH should be encouraged for health-seeking physical activities because this very likely decreases the overall risk of SAH.

Adverse Lipid Profile: A Potential Risk Factor for SAH With Conflicting Evidence of Causality

An adverse lipid profile has been related to both increased and decreased risk of SAH. A systematic review²⁴ reported that there is substantial heterogeneity between existing studies. However, only 2 studies in the review²⁴ fulfilled the criteria for high quality, reporting that elevated total cholesterol is associated with an increased risk of SAH, but only in men. Similarly, a recent high-quality study²⁵ reported that the increased risk of SAH was not only observed among men with elevated total cholesterol levels, but also in men with high levels of low-density lipoprotein (LDL)-cholesterol and ApoB lipoprotein. Moreover, the same study²⁵ reported that the risk of SAH was increased among men and women with decreased levels of high-density lipoprotein cholesterol and increased levels of ApoA1 lipoprotein, even after adjustments for relevant confounders such as smoking and high systolic blood pressure. These lipid profile-related risk factors are associated with general vessel wall inflammation, more specifically with cell death in vascular smooth muscle cells that may contribute to the formation and/or rupture of intracranial aneurysm (IA).²⁵ On the other hand, recently published MR studies have not reported consistent findings—some studies^18,33 have noted that decreased levels of LDL cholesterol and triglycerides may be causative factors for SAH, whereas others^12,18 have not found such causative features. Therefore, the independent effect of adverse lipid profile on SAH is questionable, and the findings may be attributed to uncontrolled known and unknown confounders. In summary, there is no high-quality evidence to support treating dyslipidemia with an indication to reduce the risk of SAH.

Alcohol Consumption: A Commonly Reported but Improbable Risk Factor of SAH

Moderate to heavy alcohol consumption is another modifiable factor that is often associated with an increased risk of SAH. Indeed, 2 systematic reviews and pooled analyses of several case-control studies and cohort studies have reported that persons consuming more alcohol suffer more frequently from SAH.^16,17 The suggested pathophysiologic explanations for this association are alcohol-induced hypertension and endothelial damage. On the other hand, large, prospective, and population-based cohort studies^8-11 with comprehensive case identification and confounder control have not found an independent association between alcohol consumption and the risk of SAH. Studies addressing the independent role of heavy drinking in the risk of SAH have 1 surmountable challenge: the number of heavy drinkers who do not smoke at all or have hypertensive blood pressure values is extremely low. In other words, heavy drinkers tend to smoke and have higher blood pressure values, and these well-established SAH risk factors confound the results of observational association studies that focus on the independent role of alcohol. Supporting the view that confounding explains inconclusive findings, twin¹³ and MR studies^18,19 also suggest that other known or unknown confounders may explain the increased SAH risk among persons with moderate to high consumption of alcohol. Therefore, limiting the use of alcohol to reduce the risk of SAH per se is not scientifically justified.

High Body Mass Index: Obesity Does Not Protect From SAH

Obesity represents one of the most substantially increased health risks worldwide. However, many large, prospective, and population-based cohort studies have paradoxically associated high body mass index (BMI) with a decreased risk of SAH.^10,26,27,34 Of interest, a recent high-quality study provided a probable explanation for this controversial paradox.²¹ The study reported an inverse association between BMI and SAH, but the risk seemed to originate mainly from the indirect effects through smoking.²⁶ In short, persons with low BMI are more commonly (heavy) smokers, and this (heavy) smoking—not low BMI—causes the increased SAH risk. In fact, in a subgroup analysis, the study reported no significant association between BMI and SAH among nonsmoking persons.²¹ Recent MR studies have also reported weak associations between obesity and SAH risk,^12,18,19 suggesting that obesity does not protect against SAH.

Diabetes: Controversial Findings Without High-Quality Evidence

Some retrospective case-control and small cohort studies, as well as their pooled analyses, have reported that diabetic patients may have a decreased risk of SAH.^16,35 Unfortunately, the previous studies have rarely included nonhospitalized SAH deaths, determined the type of diabetes, or assessed possible confounders. Based on prospective follow-up studies with relatively accurate case identification and confounder control, no exceptional SAH risk has been reported in either type 1²⁸ or type 2¹⁰ diabetic patients. Similarly, no significant associations between type 2 diabetes and the risk of SAH have been reported by recent MR^12,19,29 and twin studies.³⁰ Therefore, diabetes is unlikely an independent risk or protective factor for SAH.

Female Hormonal Factors: Do Women Have an Increased Risk of SAH?

Owing to the observations that postmenopausal women have around 1.5 times higher SAH incidence compared with age-matched men,^1,5,10,36 several studies have investigated whether SAH risk depends on factors affecting female hormone levels, such as HRT, combination oral contraceptives, age of menarche and menopause, and number of pregnancies.^16,31,32 However, according to previous systematic reviews on the topic,^16,31 most of the evidence is based on retrospective hospital-based case-control studies with conflicting findings. For example, although one study has related HRT to the increased risk of SAH,³² others have contrarily related it to a decreased risk.^16,31 Moreover, there is a lack of large, prospective, and population-based risk factor studies that include comprehensively nonhospitalized SAH events (i.e., sudden SAH deaths outside hospitals) and that investigate female hormonal factors independently from other SAH risk factors, such as smoking and high blood pressure. Inconsistent findings have also been found in previous MR studies.^7,37,38 Recently, a few high-quality risk factor studies^10,39,40 have reported that postmenopausal women in particular are significantly more prone to the hazardous effects of smoking. These findings suggest that hormonal changes in menopause seem to affect cerebrovascular arteries or the cardiovascular system such way that smoking after menopause becomes even more dangerous and therefore increases the risk of SAH more drastically in postmenopausal women. This may explain at least partially the women's increased SAH incidence and UIA prevalence.^39,41 In short, everyone—particularly women approaching menopause—should stop smoking, also from the epidemiologic viewpoint of SAH.

Air Pollution: Poorly Studied Yet Probable System-Level Risk Factor Requiring Further Person-Level Evidence

Based on the ecological Global Burden of Disease study analyses, ambient particulate matter pollution and indoor household pollution represent risk factors with the largest attribution to SAH-related burden after high blood pressure and smoking.⁴ Although the potential harms of air pollution have been recognized in various cardiovascular diseases including strokes, high-quality evidence on SAH is lacking. Many previous studies^4,42,43 are based on ecological correlations of hospital-based patients with SAH or clusters of all (hemorrhagic) stroke types. Moreover, because there are no high-quality prospective and population-based cohort studies with individual-level data including nonhospitalized patients with SAH and long-term air pollutant concentrations, conclusions regarding the substantial attribution of air pollution to SAH burden may not be fully justified. Nevertheless, given that exposure to air pollution causes many similar pathophysiologic responses to smoking (e.g., increasing inflammation, endothelial dysfunction, and oxidative stress) and that it is a more societal risk factor with potentially higher system-level impact, future studies on the topic are warranted. These studies are especially important in low-income regions, such as Sub-Saharan Africa, where the prevalence and prevention possibilities of other SAH risk factors differ.

Unhealthy Diet, Sleep Problems, and Poor Oral Health: A Versatile Group of Associative Factors With a Limited Amount of High-Quality Evidence

Many other lifestyle-related characteristics such as unhealthy diet,^44,45 sleep duration,^46,47 and poor oral health⁴⁸ have also been associated with the risk of SAH. However, since the findings of these factors rely mostly on scattered hospital-based case-control or small cohort studies, and have a limited amount of data on possible confounders and nonhospitalized SAHs, their consideration as independent risk factors for SAH requires further evidence. For example, although an Australian case-control study of 383 patients with SAH⁴⁴ and a Danish cohort study of 113 hospitalized patients with SAH⁴⁵ reported that adherence to low-fat and plant-based diets may be associated with decreased risk of SAH, previous MR studies have found no clear causal evidence between any of the dietary factors and the risk of SAH or IA.^18,19 On the other hand, 2 MR studies^12,18 have associated insomnia with the increased risk of SAH, but this is not supported by the clinical findings from prospective population-based cohort studies.^46,47

Challenges in Study Designs Addressing the Incidence, Case-Fatality, and Risk Factors of SAH

Prospective, population-based, and long-term follow-up studies are often considered as a gold-standard study design to investigate the epidemiologic measures of different diseases. However, retrospective studies based on hospital records or administrative/insurance registries have been widely used to study the incidence, case-fatality, and risk factors of SAH because they represent more time-efficient and cost-efficient alternatives that arise as part of standard treatment processes. In the following section (summarized in Table 2), we discuss the strengths and limitations of these 3 common study designs (population-based, hospital-based, and administrative register-based) and their overall reliability.

Table 2.

Typical Strengths, Limitations, and Practical Usability of Hospital-Based, Population-Based, and Administrative Register-Based Studies on Epidemiologic Research of SAH

	Hospital-based studies	Population-based studies	Administrative register-based studies
Strengths	Cost-effective Large sample sizes Routinely collected data Extensive clinical data on SAH-specific characteristics such as severity, amount of bleeding, diagnostics, treatments, and in-hospital outcomes	Prospective data collection More representative samples Extensive data on participants' demographics, chronic diseases, lifestyle, and outcomes May include nonhospitalized outcomes and their premorbid risk factors	Cost-effective Large sample sizes for comprehensive subgroup analyses Routinely collected data Representative samples with nationwide coverage over several decades May include nonhospitalized and hospitalized patients with high case accuracy
Limitations	Exclude nonhospitalized SAHs Poor generalizability to other hospitals, regions, and countries Retrospective and often incomplete data on pre-SAH characteristics	Expensive and time-consuming Cover small geographical regions Relatively small sample sizes hindering comprehensive subgroup analyses Baseline data often collected years before SAH Limited data on SAH characteristics, diagnostics, treatments, and in-hospital outcomes	Retrospective design Crude registrations include many false positive SAHs, requiring modifications and external validation Data rely on coding practices which can differ between centers and regions No individual-level data on many potential risk factors or SAH characteristics
Overall usability	Due to the substantial risk of selection/survival bias, not suitable for SAH epidemiology on their own Important for validation and in-hospital information of prospective and register-based studies Can also be used for in-hospital prognostic models and therapeutic trials	A gold-standard study design to investigate the incidence and risk factor estimates of SAH among specific populations Generalization of regional studies to consider national populations requires careful consideration To achieve sufficient sample sizes, standardized international collaboration studies are recommended	Studies with accurate and externally validated case identification provide a cost-effective platform to investigate incidence and case-fatality rates of SAH at a nationwide level over a long period, and within various subgroups and geographical regions If coverage is comprehensive, it can also be linked to identify SAH events among participants of population-based cohort studies

Abbreviation: SAH = subarachnoid hemorrhage.

Hospital-Based Cohort Studies

Hospital-based cohort studies represent an affordable study design for SAH epidemiology with data on admission status, bleeding severity, location of ruptured IA, diagnostic modalities, treatment options, in-hospital outcomes, etc. This study design is particularly useful in assessing hospital-specific SAH incidence rates, mortality/morbidity rates, and risk factors associated with such outcomes.^16,49 The inherent challenge with hospital-based cohorts is that they do not represent all patients with SAH because the most severe SAH cases never reach hospitals and hospital admission criteria vary between different centers and regions. Studies with high autopsy rates of outside-hospital sudden deaths have reported that as many as one-fourth of all SAH events occur before hospitalization.³ Thus, the exclusion of a large proportion of SAH events leads to a notable underestimation of SAH incidence and mortality. Moreover, hospital-based risk factor studies are exposed to selection and survival biases because sudden-death SAHs have more adverse risk factor profiles in comparison with hospitalized patients.⁵⁰ For example, some of the previous hospital-based studies have found that smokers and hypertensive patients with SAH have a better prognosis than nonsmokers and normotensive patients.⁵¹ However, since heavy smokers and hypertensive SAH cases die more likely already before they are included in the hospital-based cohorts, the results that are based on hospitalized patients with SAH are likely distorted.⁵¹ In fact, if nonhospitalized SAH deaths with their known risk factor statuses are included in risk factor analyses, smoking and hypertension are strongly associated with an increased risk of SAH death.⁵¹ Similarly, a protective role of obesity has been reported in several retrospective hospital-based cohorts including patients with SAH regardless of their initial clinical condition on admission.⁵² This observation has in turn been related to a wider theory called the obesity paradox⁵² according to which critically ill patients with high BMI could have improved endurance against a catabolic phase during critical illnesses. However, based on a recent prospective and multicenter cohort study including a homogeneous group of good-grade patients with SAH, patients with high BMI had an increased risk of poor outcomes after SAH.⁵³ Therefore, it is unlikely that obesity per se would protect from poor outcomes after SAH. In other words, because people with the worst risk factor profiles die from SAH before reaching hospitals, the use of hospital-based cohorts in assessing general epidemiologic estimates of SAH cannot be recommended.

Prospective Population-Based Cohort Studies

Prospective population-based cohorts include nonhospitalized SAH events, particularly if the identification of such cases relies on autopsy reports or death certificates that are based on autopsies. Therefore, population-based cohorts are primarily more representative leading to more generalizable results in comparison with hospital-based cohorts. Additional strengths of well-conducted population-based cohort studies include their prospective data collection of risk factors. On the other hand, most prospective population-based cohort studies have relatively small sample sizes gathered in the past, and they represent small geographical regions. According to the recent systematic reviews on prospective population-based studies focusing on the incidence¹ and case fatality⁶ of SAH, population-based epidemiologic estimates have been reported in 34 and 24 countries worldwide, respectively (Figure). It is most important that apart from one short-term incidence study from Iceland, none of these studies are nationwide; estimates are extrapolated based on smaller geographical regions. Recently, nationwide and externally validated register-based studies have reported up to 50% within-country variations in the incidence and case fatality rates of SAH.^3,5 This means that a population-based study rarely represents the nationwide population due to within-country variances in genetics, environmental risk factors, and health care accessibility. For instance, historical population-based studies from small geographic areas were cited as proof of an exceptionally high SAH incidence in Finland for decades. However, the first nationwide studies on SAH epidemiology have shown that SAH incidence in Finland does not differ from other countries with reliable SAH incidence estimates and similar health care structures (e.g., similar admission policy and case ascertainment).^1,5 Taken together, prospective population-based cohort studies can perhaps still be considered as a gold-standard design, especially when investigating the risk factors of SAH, but other study designs with more comprehensive geographical and temporal coverage, such as administrative register-based studies, may be better justified to determine many nationwide epidemiologic measures of SAH.

Figure. Previous Prospective and Population-Based Cohort Studies Reporting (A) Incidence and (B) Case-Fatality Rates of Subarachnoid Hemorrhage.

Exact locations presented by points and labels. Pink color represents countries where regional findings are commonly extrapolated. Study labels include (A) midyear and incidence¹ (per 100,000 person-years) or (B) case fatality rates⁶ (%) of subarachnoid hemorrhage. ARG = Argentina; AUS = Australia; BAR = Barbados; BRA = Brazil; BRB = Barbados; CHL = Chile; CHN = China; DEU = Germany; DNK = Denmark; ESP = Spain; EST = Estonia; FIN = Finland; FRA = France; GEO = Georgia; GBR = Great Britain; GRC = Greece; HRV = Croatia; IND = India; IRL = Ireland; IRN = Iran; ISL = Iceland; ISR = Israel; ITA = Italy; JPN = Japan; KWT = Kuwait; MEX = Mexico; MTQ = Martinique; NGA = Nigeria; NLD = the Netherlands; NOR = Norway; NZL = New Zealand; PRT = Portugal; RUS = Russia; SWE = Sweden; USA = the United States.

Administrative Register-Based Studies

Retrospective analyses of administrative data, such as nationwide hospital discharge registers and cause of death registers, are also frequently used in the epidemiologic research of SAH.^3,5,54,55 Although such study design offers cost-effective, continuously updated, and nationwide data over several decades, its main pitfall relates to incorrectly registered SAHs. These misclassifications occur particularly in emergency clinics among patients suffering from a severe headache, and when patients with a previously diagnosed and treated SAH seek medical attention again for unrelated reasons. Moreover, misclassifications of traumatic, nonaneurysmal, and intracerebral bleedings are common in administrative data. For example, a nationwide register-based study showed that following a thorough case ascertainment, nearly 40% of the aneurysmal SAH cases in the nationwide administrative register were not aneurysmal SAHs.⁵ After validation and stepwise removal of these false-positive cases representing untypical characteristics for true acute SAHs (e.g., having no registration of emergency admission with SAH as a primary diagnosis, having SAH registration with intracranial injury or arteriovenous malformation as a secondary diagnosis, or having no admission to neurosurgery department but being alive at least a year after first SAH registration), the curated data had a 99.8% positive predictive value for patients admitted to hospitals.⁵ As such characteristics can often be extracted from administrative data sets, register-based studies undergoing similar modifications that lead to high case accuracy in comparison with reference cohorts with unambiguous case identification can be considered even superior to hospital-based and small population-based cohort studies, particularly for investigating the incidence and case-fatality rates of SAH. In addition, such validated administrative data are well-suited for assessing differences by age, sex, within-country region, and by period. Given the almost complete nationwide coverage of SAH cases, properly validated register-based studies may also be useful in risk factor investigations if the identified SAH cases can be linked to persons participating in large representative health studies that collect comprehensive data on population demographics, chronic diseases, lifestyle, anthropometric/blood pressure measurements, and laboratory samples.^8,10,26 Taken together, high-quality administrative data may provide the most comprehensive and reliable information about the epidemiology of SAH but require proper validation and modification against the information of other design types, such as hospital-based or population-based cohort studies. Without such actions, administrative data sets likely overestimate the number of SAH events due to the large number of false-positive registrations.

Future Perspectives

The number of studies on SAH risk factors and other epidemiologic estimates can be expected to grow in the future. Because of the various challenges discussed, it is also likely that many of these studies will report conflicting findings. For this reason, a critical evaluation of the quality of SAH-specific epidemiologic research is of utmost importance. In this context, even artificial intelligence may help to assess the validity of epidemiologic research in the future. For example, it would already be possible to train large language models to recognize the topic of a scientific article, compile the epidemiologic pitfalls of the topic from a research perspective, and even summarize what is still unclear about the topic. Therefore, an artificial intelligence-assisted critical appraisal of epidemiologic research articles could increase the quality of scientific publications and reduce the publication of conflicting and confusing results. Regarding optimal study designs, the utilization of various overlapping design types consisting of hospital-based, population-based, and administrative registration-based cohorts—especially from countries with a limited amount of presently available data—is likewise warranted. Besides clarifying the current controversies and discovering novel causes of SAH among persons without currently known risks, future studies should further investigate why the established causes and risk factors play a role in SAH only in some people. As there is also growing evidence that the effect and prevalence of modifiable risk factors for stroke types including SAH differ by nonmodifiable features, such as age, sex, and race, consideration of such potential effect modificators is also important in future studies. If a group of people with exceptional SAH risk can be identified, this could also guide preventive screening of UIAs. This approach has already been piloted among middle-aged female smokers, with promising results.⁴¹ However, screening of UIAs in relatively healthy and asymptomatic people can lead to overdiagnosis and overtreatment, which both can add costs, impair quality of life, and even increase morbidity. Therefore, screening of UIAs should be targeted to a carefully selected group of people with a very high risk of a future SAH, such as heavily smoking women with high blood pressure values.⁸ Even in this setting, screening would perhaps yield population-level and cost-effective health benefits, if the program succeeds in introducing lifestyle changes to both screen positives and negatives without potentially harmful invasive interventions. In other words, if a targeted screening leads to smoking cessation and optimal blood pressure control in hypertensive heavy smokers, overall health benefits are highly likely, whatever the probability of a future SAH is. Finally, considering the constantly decreasing prevalence of smoking and improved management of hypertension, the underlying causes and risk factors of SAH will likely change over the next decades, especially in high-income countries. Therefore, active, continuous, and updated epidemiologic investigations focusing also on the emerging contributors of the SAH burden are likely informative and necessary.

Conclusions

Smoking and high blood pressure, possibly together with low physical activity, represent currently the only causal risk factors for SAH, whereas evidence supporting the independent role of many other risk factors has remained limited. Considering the inherent limitations of various study approaches, there is no single ideal study design to produce robust epidemiologic estimates of SAH, but these estimates should rely on various overlapping and high-quality sources of information.

Glossary

BMI

body mass index

HRT

hormone replacement therapy

IA

intracranial aneurysm

LDL

low-density lipoprotein

MR

Mendelian randomization

SAH

subarachnoid hemorrhage

UIA

unruptured intracranial aneurysm

Appendix. Authors

Name	Location	Contribution
Ilari Matias Rautalin, MD, PhD	Department of Neurosurgery, University of Helsinki and Helsinki University Hospital, Finland; The National Institute for Stroke and Applied Neurosciences, Auckland University of Technology, New Zealand	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data
Aleksanteri Asikainen, BM	Department of Neurosurgery, University of Helsinki and Helsinki University Hospital, Finland; The National Institute for Stroke and Applied Neurosciences, Auckland University of Technology, New Zealand	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data
Miikka Korja, MD, PhD	Department of Neurosurgery, University of Helsinki and Helsinki University Hospital, Finland	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data

Study Funding

I.M. Rautalin received personal research grants from the Sigrid Juselius Foundation, the Finnish Medical Foundation, the Sakari Alhopuro Foundation, the Finnish Foundation for Cardiovascular Research, and the Maud Kuistila Foundation. A. Asikainen has been supported by the Juho Vainio Foundation, the Päivikki and Sakari Sohlberg Foundation, the Maire Taponen Foundation, The Orion Research Foundation, the Paavo Nurmi Foundation, and the Paulo Foundation.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.