Article Text

Download PDFPDF
Original research
Abdominal aortic aneurysm is associated with subarachnoid hemorrhage
  1. D Andrew Wilkinson1,
  2. Badih J Daou1,
  3. Jeffrey L Nadel1,
  4. Neeraj Chaudhary2,
  5. Joseph J Gemmete1,2,
  6. Byron Gregory Thompson1,
  7. Aditya S Pandey1
  1. 1 Department of Neurosurgery, University of Michigan, Ann Arbor, Michigan, USA
  2. 2 Department of Radiology, University of Michigan Health System, Ann Arbor, Michigan, USA
  1. Correspondence to Dr Aditya S Pandey, Department of Neurosurgery, University of Michigan, Ann Arbor, Michigan, USA; adityap{at}


Background Although intracranial aneurysms (IA) and abdominal aortic aneurysms (AAA) share similar risk factors, little is known about the relationship between them. Previous studies have shown an increased incidence of IA in patients with AAA, though the rate of subarachnoid hemorrhage (SAH) in patients with AAA has not been described.

Objective To use claims data with longitudinal follow-up, to evaluate the incidence of aneurysmal SAH in patients diagnosed with AAA.

Methods We examined longitudinally linked medical claims data from a large private insurer to determine rates of aneurysmal SAH (aSAH) and secured aSAH (saSAH) in 2004–2014 among patients with previously diagnosed AAA.

Results We identified 62 910 patients diagnosed with AAA and compared them 5:1 with age- and sex-matched controls. Both populations were predominantly male (70.9%), with an average age of 70.8 years. Rates of hypertension (69.7% vs 50.6%) and smoking (12.8% vs 4.1%) were higher in the AAA group (p<0.0001) than in controls. Fifty admissions for aSAH were identified in patients with AAA (26/100 000 patient-years, 95% CI 19 to 44) and 115 admissions for aSAH in controls (7/100 000 years, 95% CI 6 to 9), giving an incidence rate ratio (IRR) of 3.6 (95% CI 2.6 to 5.0, p<0.0001) and a comorbidity-adjusted incidence rate ratio (IRR) of 2.8 (95% CI 1.9 to 3.9) for patients with AAA. The incidence of secured aneurysmal SAH was proportionally even higher in patients with AAA, 7 vs 2/100 000 years, IRR 4.5 (95% CI 3.2 to 6.3, p<0.0001).

Conclusion SAH rate was elevated in patients with AAA, even after adjustment for comorbidities. Among risk factors evaluated, AAA was the strongest predictor for SAH. The relative contributions of common genetic and environmental risk factors to both diseases should be investigated.

  • subarachnoid
  • aneurysm
  • hemorrhage

Data availability statement

Data are available upon reasonable request.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Aneurysmal subarachnoid hemorrhage (aSAH) is a devastating condition associated with high rates of morbidity and mortality.1 SAH incidence within the United States is reported to be 10–15 cases per 100 000 patients per year.2 3 Each year, 30 000–40 000 Americans are reported to have SAH and at least 10–15% of patients die before reaching hospital.4 Aortic aneurysms also constitute a major public health problem as rupture of these aneurysms could be fatal.5 Ruptured aortic aneurysms are estimated to kill 13 000 Americans each year. The prevalence of abdominal aortic aneurysms (AAA) within men and women older than 50 years has been reported to be 4–8%, with a higher prevalence in men and an expected increase in incidence associated with an aging population.6–8 Intracranial aneurysms (IA) have a prevalence of 1–5%.9 Both diseases are thought to share similar risk factors with a mix of environmental and genetic causes. Previous studies have described an increased occurrence of IA in patients with AAA,10–14 but the rate of aSAH in patients with AAA has not been described. The goal of this study is to use claims data with longitudinal follow-up, to evaluate the incidence of aSAH in patients diagnosed with AAA.


Data were obtained though review of medical claims in the Optum (Eden Prairie, Minnesota, USA) Clinformatics Data Mart Database, which contains longitudinally linked claims data from patients in a large managed-care network with enrollees throughout the United States. Data comprised all professional claims from January 2001 to December 2014, and all facility claims from January 2004 to December 2014 (figure 1). Because the data were deidentified before we used them, our institutional review board considered this a non-regulated study.

Figure 1

Identification of patients with AAA-SAH and controls. AAA, abdominal aortic aneurysms; LOS, length of stay; SAH, subarachnoid hemorrhage.

Identification of patients with AAA, patients with SAH, and controls

We used the International Classification of Diseases, ninth revision–clinical modification (ICD9-CM) code of 441.4 to identify patients with the diagnosis of AAA, as has been performed previously.15 Admissions for aSAH were identified using ICD-9 code 430 as the primary diagnosis code, and excluding SAH related to traumatic etiology, or SAH resulting from a vascular malformation. Length of stay, discharge location (home vs non-home), and complications of treatment were identified using facility ICD-CM diagnosis codes.

Coverage periods prior to the index AAA diagnosis were excluded from the 'at-risk' period for calculating the incidence of SAH, and aneurysm treatment. We identified 62 910 patients with a diagnosis of AAA and a total of 193 009 years of at-risk coverage, and compared them with age- and sex-matched controls, matched in a 5:1 fashion. For all patients with at least 1 year of enrollment, the first 365 day period of eligibility was used as a lookback for comorbidity ascertainment prior to the at-risk period. For patients with AAA, only coverage after the index AAA diagnosis was included in the at-risk period to ensure that only a population with an established diagnosis of AAA was evaluated.

Identification of aneurysm treatment in patients with SAH

Given limitations in identifying SAH in claims data, as a sensitivity analysis, we further identified a group of patients with admissions for secured aneurysmal SAH (saSAH), which represents patients with an ICD code of aSAH and a concomitant Current Procedural Terminology code for surgical or endovascular treatment, as described previously.16 This group could exclude some perimesencephalic patients or patients with poor-grade SAH whose aneurysms were not treated, but it would have a higher specificity for identifying true aSAH.


We evaluated information pertaining to patient characteristics and comorbidities, including year of birth, sex, race, hypertension, smoking history, and the composite relevant Charlson comorbidities, which include acute myocardial infarction, history of myocardial infarction, congestive heart failure, cardiovascular disease, chronic obstructive pulmonary disease, dementia, paralysis, diabetes mellitus, kidney disease, liver disease, peptic ulcers, rheumatologic diseases, and AIDS. Liver disease, AIDS, and paralysis were removed from the model owing to their low prevalence and the absence of associated cases with the primary outcome. The primary independent outcome variable was the development of aSAH in the AAA and control populations.

Statistical analysis

Chi-square and Fisher’s exact analyses were used for categorical bivariate analysis, and the two-sided Student’s t-test was used for continuous variables, with Wilcoxon analysis for non-normally distributed values. Means with SD, range, median, and IQR were reported for all continuous variables, with counts and percentages for categorical variables. We calculated 95% confidence intervals (CIs) for incidence rates and incident rate ratios (IRRs) using the normal approximation of binomial distribution. A negative binomial multivariate model was applied to estimate adjusted SAH incidence rates and IRRs, given likely overdispersion of the dataset. Marginal means analysis was used to adjust for demographic factors and comorbidities found to be significant (<0.10) predictors of SAH on univariate analysis. Given its low occurrence rate, aSAH was modeled as a non-recurring event, and comorbidity-adjusted IRRs were calculated. Compilation of claims data and analysis was performed using SAS software, version 9.4 (SAS Institute, Cary, North Carolina, USA) and Microsoft Excel 2011 (Microsoft Corp, Redmond, Washington, USA).


Baseline population characteristics

We identified 62 910 patients meeting criteria for the diagnosis of AAA, with a total of 193 009 years of at-risk coverage. There were 314 085 controls matched for age and sex in a 5:1 fashion, with a total of 1 601 640 years of at-risk coverage. Both populations were predominantly male (70.9%). The average age at the onset of the coverage period was 70.8±9.4 years.

Patients with AAA had significantly higher rates of comorbidities than controls, as listed in table 1. Notable differences included higher rates of hypertension (69.7% vs 50.6%, p<0.0001) and smoking (12.8% vs 4.1%, p<0.0001) in the AAA group compared with controls.

Table 1

Characteristics of patients with AAA and control populations

Incidence of aSAH/saSAH

Fifty admissions for aSAH were identified in patients with AAA (26/100 000 patient-years, 95% CI 19 to 44) and 115 admissions for aSAH in controls (7/100 000 patient-years, 95% CI 6 to 9), giving an IRR of 3.6 (95% CI 2.6 to 5.0, p<0.0001) and a comorbidity-adjusted IRR of 2.8 (95% CI 1.9 to 3.9) for patients with AAA (table 2). Though 71% of the AAA and control groups were male (probably due to increased AAA screening in this population), only 54% of those with SAH were male. When only the subset of patients with AAA with a diagnosis of ruptured AAA were analyzed, the IRR compared with controls was similar to patients with AAA overall (3.22 (95% CI 1.02 to 10.03), p=0.046), although with larger confidence intervals due to the smaller number of patients with ruptured AAA.

Table 2

Incidence of aSAH and saSAH in patients with AAA versus controls

The incidence of saSAH was proportionally even higher in patients with AAA than in controls—7/100 000 patient-years versus 2/100 000 patient-years, respectively, with an IRR of 4.5 (95% CI 3.2 to 6.3, p<0.0001). The median time between AAA diagnosis and SAH occurrence was 17 months, with a maximum interval of 126 months.

Factors associated with aSAH

In univariate analysis, factors associated with aSAH included AAA diagnosis, female gender, cardiovascular disease, and diagnosis of dementia, as shown in table 3.

Table 3

Univariate and multivariate analyses for aSAH

Additionally, hypertension, chronic obstructive pulmonary disease (COPD), diabetes, and peptic ulcers were found to approach significance with p values <0.10, and consequently were included in the multivariate model. On multivariate analysis, only a diagnosis of AAA and female gender remained significant, with adjusted IRRs of 2.75 (1.94 to 3.89, p<0.0001) for AAA and 1.85 (1.36 to 2.50, p<0.0001) for female gender.

Management of aSAH

Patients with AAA and aSAH had a higher rate of open surgical treatment of the aneurysm as compared with controls (14% vs 4%, respectively, p=0.028). The rate of endovascular treatment was non-significantly higher in the control population, and the rate of treatment with either endovascular or open techniques was not significantly different between the two groups (table 4).

Table 4

Comparison of aneurysm treatment and outcomes between patients with AAA and control populations.

Outcomes of aSAH

There were no significant differences in length of stay, need for mechanical ventilation, hydrocephalus, procedural complications, and discharge to home rates between the AAA group and controls.


The relationship between the brain and aortic aneurysms has long been recognized10 12 13 but poorly understood, with suspicion of a common underlying pathogenesis and common genetic pathways.17 The prevalence of IA in any group is dependent on the population screened and the method and frequency of screening. True comparisons between patients with and without AAA are difficult as there is substantial uncertainty regarding the true prevalence of IA in a given population. Unlike AAA, which can be detected relatively inexpensively with ultrasound, detection of IA requires CT angiography or MR angiography, and most are not screened. Previous studies have reported an increased incidence of IA in patients with AAA, but it was unclear whether these IA were clinically relevant, or were simply quiescent aneurysms with little chance of future harm to the patient. Any association between aneurysmal SAH with AAA has not been previously evaluated, in part because of the large number of patients needed for evaluation due to the relatively low combined frequency of SAH and AAA.

Through claims data review for patients with a diagnosis of AAA, we report a strong association between AAA and aSAH and propose that a diagnosis of AAA is a risk factor for aSAH. Whether this increased rate of SAH represents a higher underlying incidence of aneurysms in patients with AAA as has been shown previously, an increased risk of rupture of these aneurysms, or a combination of both, remains to be seen. Similarly, whether the increased incidence of SAH in patients with AAA is attributable to a common underlying genetic risk, or simply shared environmental risk factors common to both AAA and IA, is unknown. Regardless of the cause, the increased incidence of IA in AAA is clinically relevant and warrants continued investigation.

Incidence of aSAH

The incidences of aSAH and saSAH were significantly higher in patients with AAA than in controls, with IRRs of 3.6 and 4.5, respectively. Current efforts to find the true prevalence of IA in patients with AAA (and vice versa) are under way,18 and preliminary data from prospective screening suggest that 21% of patients with AAA may have IA.19 Miyazawa et al 20 reported an incidence rate of 7.2% for AAA in patients with IA. Another study by Rouchaud et al, 12 which evaluated 1081 patients with abdominal and thoracic aortic aneurysms, reported an overall prevalence of associated IA of 11.8%, with the rate going up to 12.7% in patients with AAA. Shin et al 21 reported an 11.6% prevalence of IA in 611 patients with aortic aneurysms. A recent study by Kurtelius et al 22 compared 4253 saccular brain aneurysms and 125 fusiform aneurysms with 13 009 matched controls and found that 14.4% of patients with fusiform aneurysms and 1.2% of patients with saccular aneurysms had a diagnosis of aortic aneurysms. The prevalence of aortic aneurysms was higher in patients with brain aneurysms than in controls (1.2% and 0.5%).22 Our study suggests that patients with AAA appear to be at increased risk of IA, and also SAH. We are unable to determine whether the rate of rupture for a given IA in a patient with AAA has a higher risk of rupture than IA in the general population.

Factors associated with SAH

There were inherent differences in baseline characteristics between the AAA group and the control group, with patients with AAA having significantly more comorbidities. The only factors that remained significant predictors of aSAH in multivariate analysis were gender and the presence of AAA. It is not common practice to screen for IA in patients with AAA. However, screening high-risk individuals could identify those at greatest risk and prevent the devastating effects of aSAH. Given the increased incidence of aSAH in patients with AAA, and the possibility of AAA being a risk factor for brain aneurysms, future studies are needed to establish this relation and to evaluate the cost-effectiveness of screening for brain aneurysms in patients with AAA and vice versa.

Gender and AAA/IA

The AAA group was predominantly male due to the increased incidence and screening for AAA in men, for whom ultrasound screening is recommended for all men aged 65–75 years who have ever smoked.23 The incidence of SAH however was proportionally higher in women in our study, with women representing only 29% of the AAA and control groups (which were matched for gender), but 46% and 44% of patients with SAH in, respectively, the AAA and control groups. Male gender is a known risk factor for AAA,24 while female gender is a known risk factor for IA as well as SAH,25 though only in the perimenopausal and postmenopausal periods.26

Age and SAH

Current guidelines for AAA screening by the United State Preventive Services Task Force (USPSTF) recommend a one-time ultrasound screening for all men ages 65–75 who have ever smoked or have a family history of AAA, while professional societies are more liberal, urging screening also in women meeting these criteria.27 28 Consequently, the majority of patients with AAA discovered through screening are older and male, as was the case in our cohort. This may limit the generalizability of our findings, particularly since postmenopausal women are at higher risk for aSAH than men. One implication though is that postmenopausal women with AAA may be at particularly high risk for development of IA or SAH.

Smoking and aneurysms

Smoking is a known risk factor common to both IA and AAA.29 Smoking history is a strong risk factor for AAA, and any history of smoking is included as a factor in the USPSTF recommendation to screen men. Active smoking is a known, dose-dependent, potentially modifiable risk factor for SAH,30 31 and was second only to cocaine use as the strongest predictor of SAH in young patients.32 Studies have found a correlation between decreased smoking rates and population-adjusted incidence of SAH with time,33 although SAH rates have been shown to be decreasing even in populations with stable smoking rates, a possible indication that prophylactic treatment of unruptured aneurysms may be influencing SAH rates.34 Furthermore, previous studies suggest a genetic–environmental interaction between smoking and family history of brain aneurysms.35

In this study, the lack of sensitivity for ascertaining active smoking or smoking history in claims data36 and a lack of granularity with regards to smoking history limit our ability to control for this important risk factor to both AAA and SAH, and the results need to be interpreted in that context. Positive current smoking status was identified for 12.8% of the AAA population in our study, but we suspect that the actual rate may be higher, although this is remarkably close to the currently reported average smoking rate in the USA of 13.7%.37 It is notable that our analysis did not find the presence of a smoking diagnosis code to be a predictor of SAH, despite smoking being an overwhelmingly proven strong predictor of SAH in studies with better patient-specific data. A diagnosis of COPD probably serves as a surrogate marker of significant smoking history, and in our study was moderately associated with SAH on univariate, but not multivariate analyses. These limitations notwithstanding, our finding linking AAA with SAH warrants further investigation to ascertain how much is due to shared environmental risk factors such as smoking. It is notable that previous studies have shown active smoking to be a strong risk factor for SAH (OR=5.0, 95% CI 3.1 to 8.1 compared with matched controls), but that the risk of SAH seems to drop off almost completely on smoking cessation (OR=1.2, 95% CI 0.8 to 2.0).38 Thus, AAA appears to be a stronger predictor of SAH than smoking history, but not active smoking.

Genetic factors: AAA and cerebral aneurysms

Intracranial and aortic aneurysms are thought to share common genetic factors that possibly explain a higher prevalence of SAH in patients with AAA through shared mechanisms of abnormal vascular remodeling.17 39 Genome-wide association studies have identified single nucleotide polymorphisms on several chromosomes to be associated with IA. Specifically, chromosome 9 p21 locus is associated with a higher risk of intracranial aneurysms and other arterial diseases, notably AAA.17 39 Whether or not these genetic commonalities also depict expansion and rupture of IA remains to be further investigated,40 but our study supports the epidemiological linkage between AAA and SAH. As a claims study, we were unable to specifically control for family history of either brain or aortic aneurysms, and further studies will be needed to investigate the interaction of family history and common or independent genetic risk factors. In particular, the presence in both AAA and SAH populations of shared genetic factors that may act synergistically with environmental factors such as smoking may account for a large proportion of aneurysm development. The presence of AAA may serve as a proxy for patients who are genetically predisposed to develop and rupture aneurysms (either AAA or IA) in the presence of environmental risk factors such as smoking. In considering screening, low-cost ultrasound for AAA may serve as a cost-effective method to identify those at higher risk for SAH.

Management of aSAH

IA were more likely to be treated through an open surgical approach in patients with AAA than in the age-matched controls. Although this seems somewhat counterintuitive given that patients with AAA have more comorbidities that may make them less suitable candidates for an open surgical approach, this finding could be related to the aneurysm characteristics rather than patient characteristics. It is feasible that patients with AAA presenting with SAH harbor smaller aneurysms within the anterior circulation, which are more amenable to microsurgical treatment. Regardless of the treatment, patients with AAA do not seem to be at higher risk for SAH-related or treatment-related complications and their length of stay and ability to be discharged to home is not significantly increased.


Although this study relies on previously validated administrative coding algorithms identifying AAA and aSAH, the limitations of claims-based analyses have been well-described previously.41 We performed a sensitivity analysis to account for the lack of specificity of claims data in identifying aSAH by also examining only those who were treated to secure an aneurysm (saSAH), finding an even greater association between AAA and saSAH. As mentioned previously, smoking is under-reported in claims data, and probably accounts for some of the strong association between SAH and AAA. Nevertheless, even if expected due to shared risk factors that we may not be able to control for, the finding of an association between AAA and SAH is novel. In addition, other variables, including family history of IA, which might have been of value, were not available in the dataset. Furthermore, the generalizability of the study is limited as the patient cohort is restricted to a privately insured patient population. As the study includes mostly older patients due to increased screening for AAA in this population, its generalizability to younger patients also remains to be seen.


The rate of SAH is elevated in patients with AAA compared with age- and sex-matched controls, even after adjustment for comorbidities. AAA was the strongest risk factor for aSAH in our analysis, and consideration should be given to the relative contributions of common genetic and environmental risk factors to both diseases. Use of AAA as a screening criterion for IA should be considered.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Ethics approval

Study approved by the University of Michigan IRB (HUM00181249).



  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles