WHAT IS ALREADY KNOWN ON THIS TOPICWHAT THIS STUDY ADDS

A complex interplay of demographic, socioeconomic, and geographic factors influence treatment patterns of high-grade aSAH in the United States.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYIntroduction

Aneurysmal subarachnoid hemorrhage (aSAH) is an emergent neurological disease process which portends significant morbidity and mortality. Approximately 30 000 new cases of aSAH occur in the USA annually, among which nearly half die during hospitalization and fewer than 60% regain functional independence at 1 year.1–3 The prognosis is especially dire for high-grade aSAH (HGSAH) patients, those presenting severely obtunded or comatose following aneurysm rupture, and classified by Hunt and Hess grades 4 and 5. In fact, the initial 1968 description of the Hunt and Hess scale for prognostication in aSAH identified a 100% mortality rate in the highest grade (grade 5) patients and 71% in grade 4 patients.4

Owing to advancements in early surgical and endovascular treatment5–7 for ruptured aneurysms and rigorous pre- and postoperative neurocritical care, data have demonstrated that survival and favorable outcomes may in fact be achieved in approximately 50% of these high-severity patients.8 9 Moreover, a recent study by Goldberg and colleagues assessed the prospective quality of life (QoL) of survivors of HGSAH and found that as many as 80% of survivors were able to achieve good QoL (measured using the EuroQoL-5 Dimension-3 Level questionnaire).10 Thus, increased knowledge of those who would benefit most from intervention is of vital importance in improving outcomes for HGSAH patients.

Despite increasingly favorable outcomes achieved by HGSAH patients over time, there remains a paucity of literature evaluating who receives treatment and at what rate. This study evaluates a large-scale administrative registry and presents a population-level assessment of the geographic, demographic, and socioeconomic characteristics associated with HGSAH treatment in the USA.

MethodsData source

The National Inpatient Sample (NIS), developed and maintained by the Healthcare Cost and Utilization Project (HCUP), is among the largest publicly accessible inpatient care databases in the USA. Annual unweighted data approximate 7 million hospitalizations, reflecting a 20% stratified sample of all HCUP-participating institutions nationally. More information regarding the NIS and data access can be found at: www.hcup-us.ahrq.gov. Given the public accessibility and de-identified nature of the information in this database, this study was not considered human subjects research and thus did not meet the requirements for institutional review board approval. For the same reason, patient consent was neither sought nor required. This manuscript was composed in accordance with Reporting of Studies Conducted Using Observational Routinely-Collected Data (RECORD) guidelines.

Patient selection and cohort development

Weighted adult hospitalizations (>17 years of age) with primary admission diagnoses of aSAH were identified in the NIS during the period of 2015 (fourth quarter, October through December) to 2019 using the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes (ICD-10-CM I60, various) in accordance with established algorithms. Patients with diagnoses related to trauma or congenital cerebrovascular malformations as well as elective admissions were excluded from the analysis to eliminate non-aneurysmal causes of SAH, in accordance with previous literature.11 Thresholds for Hunt and Hess grade were determined from the previously validated NIS Subarachnoid Hemorrhage Severity Score (NIS-SSS)11 optimized for utilization in this dataset and shown to have strong correlation and concordance with Hunt and Hess grade. The NIS-SSS is an integerized composite score determined by the weighting of parameters associated with clinical severity (coma, stupor, cranial nerve palsy, mechanical ventilation, cerebrospinal fluid diversion, hemiparesis, etc), which was then externally validated against Hunt and Hess grade, such that NIS-SSS thresholds were determined for each Hunt and Hess grade (NIS-SSS >9 corresponds to Hunt and Hess grades 4 and 5). Normal grades (Hunt and Hess 1–3) were excluded from the analysis to isolate high-grade severity hospitalizations (Hunt and Hess 4 and 5) (figure 1).

Figure 1

Patient selection and cohort development for patients with confirmed aneurysmal subarachnoid hemorrhage (aSAH) who received either endovascular or surgical treatment or no intervention. H-H, Hunt and Hess; ICD-10-CM, International Classification of Diseases, 10th Revision, Clinical Modification; SAH, subarachnoid hemorrhage.

Baseline demographic and clinical characteristics

The primary exposure of this analysis was geographic area, defined by nine United States Census Bureau Census Divisions: New England (Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut), Mid-Atlantic (New York, Pennsylvania, New Jersey), East North Central (Wisconsin, Michigan, Illinois, Indiana, Ohio), West North Central (Missouri, North Dakota, South Dakota, Nebraska, Kansas, Minnesota, Iowa), South Atlantic (Delaware, Maryland, District of Columbia, Virginia, West Virginia, North Carolina, South Carolina, Georgia, Florida), East South Central (Kentucky, Tennessee, Mississippi, Alabama), West South Central (Oklahoma, Texas, Arkansas, Louisiana), Mountain (Idaho, Montana, Wyoming, Nevada, Utah, Colorado, Arizona, New Mexico), and Pacific (Alaska, Washington, Oregon, California, Hawaii).12 Other demographic and socioeconomic covariates included age (categorized by decade), female sex, race/ethnicity, primary payer status, hospital location and teaching status, hospital bed size (defined by the HCUP based on region of the country, urban-rural designation of the hospital, in addition to teaching status), patient location (a six-category urban-rural classification scheme for counties developed by the National Center for Health Statistics including central and fringe counties of large metropolitan areas, smaller metropolitan counties (subdivided by population), and non-metropolitan counties (divided into micropolitan and non-core categories)), hospital control/ownership, and median household income (by patient ZIP code and stratified by quartile). Other clinical covariates included Charlson Comorbidity Index (CCI) score13 and aneurysm location (anterior circulation, posterior circulation, or unknown location). Mortality risk was derived from the integerized All Patient Refined Diagnosis Related Groups (APR-DRG) mortality risk subclass score, an embedded parameter in the NIS (and other HCUP registries) which rates mortality risk on a 1–4 scale according to admission diagnosis, comorbidity burden, and presentation illness severity. Missing data were identified in three parameters (race/ethnicity, county population size, and income level), and because less than 5% of hospitalizations for these parameters were missing, data were imputed to the dominant category per HCUP statistical and methodological recommendations.14

Clinical endpoints and statistical analysis

Aneurysm treatment by microsurgical clipping or endovascular embolization was evaluated as the primary endpoint of this analysis, and identified by appropriate ICD Procedural Coding Systems codes (03VG0, 03VG3, 03VG4) in accordance with established algorithms.15 Patients not treated by either of these modalities were deemed to have been treated by conservative management, inclusive of medical therapy, withdrawal of care, or other modalities not captured by the aforementioned procedure codes. Favorable clinical outcome was determined by the NIS Subarachnoid Hemorrhage Outcome Measure (NIS-SOM)—the dichotomous NIS-SOM was previously validated in a national cohort of aSAH hospitalizations and defines poor outcome as discharge disposition to intermediate, long-term care, or skilled nursing facility, in-hospital mortality, or placement of tracheostomy or gastrostomy tube, and was shown to have strong concordance with modified Rankin Scale (mRS) scores >2 (suggestive of high levels of disability at discharge).11 Mortality was evaluated secondarily as a form of sensitivity testing.

Statistical analyses incorporated annual hospitalization weights and hospital clustering to account for NIS sampling design and to yield accurate national population estimates. Baseline demographic and clinical characteristics were compared between patients receiving microsurgical or endovascular intervention and those treated with conservative management using Pearson’s χ2 test for binary categorical parameters or by linear-by-linear association for ordinal variables or those with multiple categories. P values <0.10 generated by these univariable tests were determined to detect an association between a given parameter and aneurysm treatment and were subsequently chosen as candidate covariates for multivariable logistic regression analysis. Effect sizes of parameters for treatment with microsurgical clipping or endovascular embolization were presented as adjusted odds ratios (aOR) with 95% confidence intervals (95% CI). Statistical significance was determined following a Bonferroni correction for multiple comparisons. A variance inflation factor (VIF) (with a threshold of 10) was used to evaluate model multicollinearity and parameter redundancy. Sensitivity testing was performed in the multivariable analysis by evaluating the same model in only Hunt and Hess grade 5 patients. All analyses were performed using IBM SPSS (version 29) statistical software (Armonk, NY).

Results

Among 99 460 aSAH hospitalizations identified, 36 795 (37.0%) were HGSAH and included in the primary analysis; 9210 (25.0%) of HGSAH hospitalizations were associated with treatment with microsurgical clipping or endovascular embolization (figure 1). Baseline demographic, socioeconomic, and clinical characteristics which varied significantly between treated and conservatively managed aSAH included age, female sex, APR-DRG mortality risk subclass, CCI score, aneurysm location, primary payer status, hospital location and teaching status, hospital bed size, census division, and hospital ownership/control (table 1). Race/ethnicity, patient location, and median household income did not differ significantly between treated and conservatively managed aSAH, nor did these parameters meet a modest association threshold (P<0.10) for inclusion in multivariable analysis.

Table 1

Comparison of baseline demographic and clinical characteristics between high-grade aneurysmal subarachnoid hemorrhage patients treated with microsurgical clipping/endovascular intervention or conservative therapy

Following multivariable logistic regression analysis, determinants of treatment with surgery or endovascular embolization included female sex (aOR 1.42, 95% CI 1.35 to 1.51), transfer admission (aOR 1.18, 95% CI 1.12 to 1.25), private insurance (ref: government-sponsored insurance) (aOR 1.21, 95% CI 1.14 to 1.28), and government hospital ownership (ref: private ownership) (aOR 1.17, 95% CI 1.09 to 1.25), while increasing age (by decade) (aOR 0.93, 95% CI 0.91 to 0.95), increasing mortality risk (aOR 0.60, 95% CI 0.57 to 0.63), urban non-teaching status (aOR 0.66, 95% CI 0.59 to 0.73), rural hospital location (aOR 0.13, 95% CI 0.7 to 0.25), small hospital bedsize (aOR 0.68, 95% CI 0.60 to 0.76), and hospital geographic region (census division) (ref: New England) (South Atlantic (aOR 0.72, 95% CI 0.63 to 0.83), East South Central (aOR 0.75, 95% CI 0.64 to 0.88), and Mountain (aOR 0.72, 95% CI 0.61 to 0.85)) were characteristics associated with a lower likelihood of treatment (table 2). All parameters included in the model had a VIF <10. Sensitivity testing evaluating the treatment model in only Hunt and Hess grade 5 patients confirmed the findings of the primary analysis.

Table 2

Multivariable logistic regression analysis. Determination of significant demographic, clinical, socioeconomic, and geographic associations with high grade aneurysmal subarachnoid hemorrhage treatment by microsurgical clipping or endovascular intervention

Treated high-grade aSAH patients experienced significantly greater rates of favorable functional outcomes (20.1% vs 17.3%; aOR 1.20, 95% CI 1.13 to 1.28, P<0.001) and lower rates of mortality (25.8% vs 49.1%; aOR 0.36, 95% CI 0.34 to 0.38, P<0.001) in comparison with those not treated (table 3, figure 2). Rates of favorable outcomes (New England 19.2%, Mid-Atlantic 15.8%, East North Central 13.9%, West North Central 13.3%, South Atlantic 20.2%, East South Central 12.8%, West South Central 20.7%, Mountain 15.2%, and Pacific 22.1%; P<0.001) and mortality (New England 41.8%, Mid-Atlantic 42.2%, East North Central 42.0%, West North Central 41.3%, South Atlantic 40.4%, East South Central 50.4%, West South Central 43.5%, Mountain 42.8%, and Pacific 46.5%; P<0.001) differed significantly by census region (table 3, figure 2).

Figure 2

Geographic patterns of treatment (panel A), mortality (panel B), and favorable outcome (panel C) of high-grade aneurysmal subarachnoid hemorrhage (aSAH) in the USA. Shading represents parameter frequency (%). Associated data are also shown in tables 1 and 2.

Table 3

Secondary endpoints. Rates of favorable outcome and in-hospital mortality stratified by census division

Discussion

HGSAH, classified as Hunt and Hess grade 4 or 5, poses significant challenges in terms of treatment and intervention. To our knowledge, this is the first large-scale study exploring national data to determine differences in access to treatment in patients with HGSAH. In this retrospective observational study, we found that female sex, transfer admission, patients with private insurance (compared with government-sponsored insurance), and treatment at a government-owned hospital (compared with private hospitals) were associated with an increased likelihood of receiving treatment for HGSAH. Conversely, increasing age, higher baseline mortality risk, urban non-teaching hospital status, rural hospital location, small hospital size, and certain geographic regions were associated with a lower likelihood of receiving treatment. Although our analysis offers a national perspective on an important clinical entity, it is limited by the shortcomings of retrospective administrative registry research (which are more substantively delineated at the conclusion of the discussion).

The highest percentage of treated patients were located in the Central Midwest, followed by New England regions, whereas the lowest percentage of treatment was observed in the geographic South and the Rocky Mountain states. The highest mortality patterns were seen in the Southern states, specifically Alabama, Mississippi, Tennessee, and Kentucky, followed by the states on the Pacific Coast. Mortality was lowest in Southeastern states on the Atlantic Coast. The most favorable outcomes were seen on the Pacific Coast, followed by the Southern Midwest and Southeast. The lowest favorable outcomes were observed in the Midwest as well as Alabama, Mississippi, Tennessee, and Kentucky, which were also the four states with the highest mortality. Additionally, we found that treated HGSAH patients had significantly higher rates of favorable functional outcomes and lower rates of mortality compared with those who were not treated. Generally, regions with the highest treatment patterns also experienced the lowest mortality rates. There were, however, outliers to the expected trends, as states on the Atlantic Coast such as Florida, Georgia, North Carolina, South Carolina, Virginia, West Virginia, and Maryland all had lower treatment rates but lower mortality and higher favorable outcomes when compared with other regions. It is speculated that socioeconomic and demographic factors may account for this discrepancy.

Previous research has explored the demographic factors influencing the incidence and outcomes of aSAH. Studies have shown higher age-adjusted mortality rates of aSAH in females and Black patients.16 Among HGSAH in our analysis, however, race appeared to have no relationship with interventional treatment as univariable analysis did not demonstrate a strong enough association even for inclusion of the race/ethnicity parameter in multivariable modeling (whereas other demographic and socioeconomic variables demonstrated robust associations with treatment). Recent studies in countries with universal health systems found that female patients, self-employed individuals, and those with high income and employee insurance had a higher likelihood of receiving treatment for aSAH.17 In contrast, our analysis in the multi-payer healthcare system of the USA found that access to private insurance predicted treatment for HGSAH. These patients have access to specialized care, comprehensive stroke centers, and a multidisciplinary team, as well as the financial resources to support treatment.

Previous studies have examined the impact of socioeconomic factors on aSAH incidence, mortality, and outcomes. Rumalla et al found that aSAH patients who received treatment at a major urban stroke center experienced improved outcomes if they were white, had access to primary care providers, and received caregiver support.18 Low socioeconomic status (SES) patients have shown a higher average incidence of aSAH, with a higher likelihood of being active smokers, residing in non-urban areas, and experiencing poorer discharge outcomes overall.19 However, these studies were limited in sample size and geographic coverage and were not applied specifically to HGSAH patients. Most recently, however, Kabangu et al also performed an analysis of NIS data between 2016 and 2020 and discovered that minority patients were significantly more likely to receive surgical intervention and less likely to die than their white counterparts, and they were also significantly more likely to receive life-sustaining interventions including tracheostomy and gastrostomy. This trend persisted even at higher SES classes.20 Another study by Kandregula et al found similar disparities regarding patients with unruptured intracranial aneurysms (UIA) over the period of 2000–2019, with Black patients experiencing slightly improved rates of treatment but Hispanics and other minority groups facing consistent undertreatment.21

Our analysis found 37% of the sampled population presented with HGSAH, and one-quarter received SEI. Procedural intervention has previously been demonstrated to show improved long-term outcomes despite these patients presenting with more acute critical symptoms.5 22 Both treatment modalities were associated with significantly more favorable functional outcomes and lower mortality rates compared with the non-treatment group. Access to treatment varied based on geographic location, with lower treatment rates observed in rural, urban non-teaching, and small hospitals. This suggests that in remote areas, patients may experience disparities in accessing specialized care, and hospitals may lack the necessary neurointervention or interventional radiology equipment to handle aSAH cases. These disparities can arise from the confluence of many different, interconnected factors, such as state budgets, payer mix, and political and financial interests, all of which may contribute to and accentuate differences in the quality of healthcare and outcomes among different regions. Guha et al similarly described a multinational cohort of aSAH in which higher GDP and an increased number of neurosurgeons per capita were associated with improved treatment outcomes.23 Increased age was linked to a lower likelihood of treatment for aSAH, which may indicate the presence of comorbid conditions or age-related treatment preferences, further evidenced by lower treatment rates in those with higher baseline mortality risk.

It is important to draw attention to the number of patients receiving SEI in our cohort (25%) which on surface examination may seem low, while 75% were treated conservatively (as stipulated in the methods section, this includes medical therapy, withdrawal of care, and other treatment modalities otherwise not specified by the procedural codes used to identify surgical and endovascular treatment). Based on trends examined in the literature, it is reasonable to speculate that a large portion of this cohort may indeed have received medical management alone and/or withdrawal of care. A two-center, multinational, retrospective study conducted by Hoogmoed et al determined that 43% of poor grade aSAH patients died during hospitalization, and the major cause of death in this cohort was withdrawal of care (documented in 71% of cases).24 Nearly all patients who did not undergo any intervention (98%) died during hospitalization. Moreover, when evaluating the proportion of Hunt and Hess grades 4 and 5 patients in our study cohort, it was observed that the majority of patients were classified as Hunt and Hess 5 and that these comprised the majority of the treatment group. In principle this would impact the likelihood of treatment as well as clinical outcome; however, sensitivity testing in multivariable analysis determined that the same regional variation in treatment patterns was observed in the cohort of Hunt and Hess 5 patients. Additionally this provides further insight into the unadjusted regional analysis which demonstrates lower mortality and higher rates of favorable outcomes in SEI treated patients despite the high frequency of Hunt and Hess grade 5 patients in this group.

Beyond its retrospective design, the limitations of this study include lack of specific geographic data beyond census region and more granular clinical and radiographic information that may impact treatment decisions (such as aneurysm morphology and specific location). Additionally, we are unable to examine the long-term effectiveness of the treatment method or the real-time factors influencing the decision to pursue intervention or refrain from treatment. In future studies, it is important to explore the concentration of hospitals capable of neurointervention in regions with poor outcomes. Regarding clinical data in this analysis, all information (other than embedded NIS parameters) was determined by ICD codes primarily used for administrative and billing purposes, with a variety of documented errors and inaccuracies.25 Severity of presentation was determined by the NIS-SSS and outcome by NIS-SOM, as described in the methods section of this manuscript. Although the NIS-SSS is a validated severity stratification instrument shown to have high concordance with Hunt and Hess grade, it must be acknowledged that high-grade status was determined indirectly from this score which was derived from retrospective data based on billing codes without knowledge of when during the hospitalization neurological status was determined, thus contributing to potential inaccuracies in the identification of high-grade patients. Similarly, the NIS-SOM was shown to have strong concordance with mRS score; however, there are inherent inaccuracies adapting discharge disposition to functional status, especially since it does not account for clinical events beyond the index hospitalization. It should also be noted that a large portion of patients had an unknown aneurysm location (23% in the SEI group and 62% in the conservative treatment group had a code of I60.9 corresponding to ‘nontraumatic subarachnoid hemorrhage, location unspecified’, as aneurysms of the posterior circulation have been shown to portend poorer outcome in comparison to lesions of the anterior circulation.26

Conclusion

Our analysis of a population-level administrative registry highlights the socioeconomic, demographic, and geographical disparities in accessing life-saving treatment for HGSAH, a diagnosis that is frequently fatal in the absence of appropriate medical intervention. Female sex, transfer admission, private insurance, and treatment in a government-owned hospital were associated with an increased likelihood of receiving treatment for HGSAH, while increasing age, higher mortality risk, urban non-teaching hospital status, rural hospital location, smaller hospital size, and certain geographic regions were associated with a lower likelihood of receiving interventional treatment.

Data availability statement

The data used in this analysis as well as a comprehensive list of billing codes used to define the clinical variables in this study (other than those explicitly denoted in the body of the methods section of this manuscript) are available upon reasonable request of the corresponding author following completion of onboarding and verification procedures as specified by the Healthcare Cost and Utilization Project ( HCUP).

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Not applicable.