We utilized the STROBE checklist for case-control studies for manuscript development (STROBE, 2023) and adhered to observation case-control study guidelines. We used data from the Veterans Health Administration (VA) Corporate Data Warehouse (CDW). The CDW contains information on VA inpatient and outpatient visits and associated clinical diagnoses and procedures, information on non-VA visits reimbursed by VA, vital signs, VA pharmacy records, laboratory data, and other patient-level variables. We also used Medicare data from the Centers for Medicare and Medicaid Services (CMS) for all Veterans enrolled in or eligible for VHA healthcare and linked these data to VA data.

We examined records from 2000 to 2019 to identify patients eligible for VA care who were enrolled in VHA or Medicare for at least 60 months and had at least 1 VA visit during that time period (Figure 1). Patients enrolled in Medicare Part C at any time during the study period were excluded from the analysis because we only had data available for Medicare Parts A and B. We used ICD and CPT codes (Supplementary Table 1) to identify two case groups: (1) Brain aneurysm only and (2) Brain aneurysm plus non-traumatic SAH. Patients needed 1 ICD code from inpatient visits, 1 procedure code, or 2 outpatient codes qualify. For aneurysm we used codes ICD-9 437.3, ICD-10 I67.1, and CPT codes 61697, 61698, 61700, 61702, 61624. For non-traumatic SAH we used codes ICD-9 430 and ICD-10 I60. We considered the index date to be the date of the first qualifying aneurysm CPT or ICD code. We then randomly selected 10 controls per case from remaining patients who did not have any aneurysm or SAH codes but had a clinical visit on that index date. Cases and controls were matched by age, date of diagnosis (case diagnosis matched to a clinical visit date for controls), and site of index visit (VA vs. CMS). We required that the mental health diagnoses occur in the 5-year period before the index date, i.e., as it relates to the medical record, the diagnosis of a mental health condition preceded the diagnosis of aneurysm without or aneurysm with SAH. After applying selection criteria, 6,320,789 patients were eligible for inclusion. The case and control selection and matching process yielded 35,094 aneurysm-only and 5,749 aSAH patients and 10x matched control patients for each of these groups (Figure 1).

This study was approved by the University of California, San Francisco Human Research Protection Program and the San Francisco Veterans Affairs Healthcare System Research and Development Committee. As the study involved secondary analysis of healthcare records, no participant contact was involved, and informed consent was not required.

We evaluated the 5-year period prior to the index date to establish diagnoses of several mental health conditions using ICD codes: PTSD, major depressive disorder (MDD), generalized anxiety disorder (GAD), and other mood disorders (Supplementary Table 1). We collected additional demographic information (age and gender) and used the same prior 5-year period to assess diagnoses of common comorbidities and stroke risk factors (hypertension, diabetes, and tobacco use).

Differences in patient characteristics between cases and controls were compared using the Chi-square test for categorical variables and Wilcoxon rank-sum test for continuous variables. Univariable logistic regression models were constructed for each mental health predictor with the outcome of brain aneurysm. Analogous models were constructed with the outcome of aSAH. Subsequent models included all psychiatric factors simultaneously to examine the independent associations of the mental health variables with the outcomes. Multivariable models adjusted for sex, hypertension, diabetes, and tobacco use. All analyses were performed with SAS Enterprise Guide version 8.3 (SAS Institute, Cary, NC).

Several factors have been proposed as mechanisms through which mental health disorders could affect aneurysm growth and rupture, including hypertension, inflammation, oxidative stress, and immunoregulatory dysfunction. These mechanisms, whether independently or synergistically, can have a negative effect on the molecular-cellular environment of aneurysms, eventually leading to rupture. Hypertension, a consistent risk factor for aneurysms may result from stress-induced hypothalamic-pituitary axis dysfunction. This increases the secretion of glucocorticoids and circulating catecholamines, which over time leads to vascular smooth muscle cell modulation and endothelial damage (Everson-Rose et al., 2014).

Neuropsychiatric disorders have shown association with elevated inflammation, as demonstrated by increases in C-reactive protein, interleukins, and fibrinogen (Everson-Rose et al., 2014; Beurel et al., 2020). Aneurysmal pathophysiological modeling suggests following an initial insult to the endothelium due to hemodynamic wall stress, upregulation of the inflammatory pathways may initiate a feed-forward injury cycle. The chronic translocation of leukocytes through the endothelium can accelerate degradation of the vessel wall, making the aneurysms more prone to rupture. Levels of proteins such as vascular adhesion molecule 1 and von Willebrand factor, which are important in the leukocyte translocation pathway, have been shown to be higher in individuals with depression (Lopez-Vilchez et al., 2016). Additionally, TNF-a, a cytokine implicated in aneurysm pathogenesis (Starke et al., 2014), has been shown to be elevated in patients with PTSD and MDD. TNF-a upregulates signaling pathways that lead to apoptosis and cellular degradation, such as the expression of calcium channels, Toll-like receptors, and matrix metalloproteinases. TNF-a and interleukin 1 signal downstream pro-inflammatory cytokines and interleukins while MMPs increase vascular permeability, leading to hemorrhage (Liu et al., 2022). Such neurovascular immunological dysfunction may be mitigated in part through flow- and pressure-dependent triggers, though the upstream genetics and may offer opportunities to modify risk either through medical and/or lifestyle changes.

Beyond these inflammatory and immunoregulatory alterations, environmental, occupational and psychosocial stress can increase reactive oxygen species (ROS) and cause vascular dysfunction (Loria et al., 2011). In mouse models, cigarette smoke exposure is increases ROS and NOX1 which precipitates upregulation of proinflammatory/matrix remodeling genes, leading to vascular smooth muscle cell modulation and aneurysm formation and rupture (Laaksamo et al., 2013). Increased oxidative stress and ROS production also affects the renin-angiotensin system and glucocorticoid pathways with downstream effects on endothelial cell function (Griessenauer et al., 2018). Moreover, the upregulation of free radicals, superoxide dismutase and glutathione in the setting of occupational and domestic stress is also associated with MDD (van Sloten et al., 2014). Though no studies have specifically addressed these mechanisms in patients with aneurysms, studies of patients with cardiovascular risk factors have demonstrated stress reduction can reduce oxidative stress and improve markers of endothelial function (Said and El-Gohary, 2016).

Lastly, patients with psychiatric disorders may disproportionately use and/or abuse substances modifying stroke risk. Chronic alcohol use has been implicated as a risk factor for hypertension and stroke (Smyth et al., 2022) and intensity of alcohol use has also been shown to demonstrate a dose-dependent relationship to aneurysm rupture (Can et al., 2018a). Heroin and methamphetamine use have also been implicated in case series of aSAH patients (Can et al., 2018b; Swor et al., 2019). It is possible that pharmacological interventions for psychiatric disorders may also affect stroke risk. There is controversy about how selective serotonin reuptake inhibitors may affect stroke with or without other drug modifiers (Trajkova et al., 2019; Gaffey et al., 2021). More comprehensive, prospective studies will be needed to elucidate these potentially confounding relationships.

As this is an observational study, we cannot prove that psychiatric disorders caused aneurysms or aSAH. We relied on medical records to establish diagnoses of variables and outcomes, and this could be subject to misclassification bias. Our findings may also be influenced by ascertainment bias if patients diagnosed with mental health disorders were more likely to seek care (e.g., undergo brain imaging) and have aneurysms diagnosed. However, aneurysm rupture would likely lead to symptoms that prompted treatment-seeking independent of prior mental health and treatment status. It is encouraging that established risk factors such as hypertension, female sex, and tobacco use were also associated with aneurysmal diagnoses suggesting the results might more accurately reflect reality. Finally, though we attempted to match or control for important factors that are associated with mental health disorders and could impact aneurysm and aSAH risk, residual confounding remains a concern.