The National Inpatient Sample (NIS) database, maintained by the Healthcare Cost and Utilization Project (HCUP), is one of the largest publicly accessible databases in the United States. The database is designed to approximate a nationally representative sample of hospitalizations by inclusion of data from hospitals across the nation. The database allows for generation of national estimates of inpatient hospitalizations by utilization of the “DISCWT (discharge weight)” variable within the NIS.

A strength of the database is its inclusion of hospitalizations across all payer types, thereby enhancing the generalizability of findings across diverse settings. The database contains a wide range of variables of patient demographics, hospital characteristics, diagnoses, procedures, length of stay (LOS), and discharge disposition. The NIS database has been extensively studied within the current state of the neurosurgical and vascular literature.

Following the transition from International Classification of Diseases (ICD)-9 to ICD-10 coding in October 2015, the present study included only hospitalizations coded using ICD-10 diagnosis and procedure codes to ensure consistency and analytic validity.

The NIS database from 2019–2022 was utilized to identify patients undergoing MT for AIS by using ICD-10-Procedure Coding System (ICD-10-PCS) codes. The following procedural codes for MT were used: 03CG3Z7, 03CG3ZZ, 03CG4ZZ. To identify patients specifically with renal failure and AIS, ICD-10-Clinical Modification (ICD-10-CM) diagnosis codes were queried. Patients with renal failure (ICD-10-CM: N17, N18, N19) and AIS (ICD-10-CM: I63.0, I63.1, I63.2, I63.3, I63.4, I63.5, I63.6, I63.8, I63.9, I67.82, I67.89) were included, while patients without renal failure and AIS were excluded. Patients were separated into 2 respective cohorts based on sex (male and female) through use of the “FEMALE” variable embedded within the database [11]. Appropriate ICD-10-CM and ICD-10- PCS codes were gathered through retrospectively analyzing the literature on previous NIS studies, as well as through clinical verification. All analyses used HCUP-recommended discharge weights to produce nationally representative estimates. Trends in procedure volume and discharge disposition prior to applying case weighing are provided in Figs. 1 and 2.

Primary outcomes of interest included in-hospital mortality and non-routine discharge. Demographic, socioeconomic, clinical, and hospital-level covariates included age, race, LOS, total charges, comorbidities, hospital region, primary expected payer, median household income quartile, teaching status of hospital, disposition, and hospital bed size. Non-routine discharge was defined as a discharge other than routine through use of the “DISPUNIFORM” variable available within the NIS database [12]. In the context of the U.S. healthcare system, non-routine discharge generally reflects the need for continued medical care, rehabilitation, or support services and is commonly used as a surrogate marker for increased functional impairment or post-acute care needs.

Categorical variables were compared using Pearson’s chi-square test while continuous variables were compared using t-test. Multivariable logistic regression analyses were performed to identify factors associated with in-hospital mortality and non-routine discharge while adjusting for sex, age, race, primary expected payer, ZIP income quartile, calendar year, weekend admission status, APR-DRG (All Patient Refined Diagnosis Related Groups) Severity of Illness, hospital region, hospital bed size, hospital teaching status, congestive heart failure (CHF), arrhythmia, peripheral vascular disease (PVD), hypertension (HTN), paralysis, chronic obstructive pulmonary disease (COPD), diabetes, liver disease, obesity, weight loss, fluid and electrolyte disorders, deficiency anemia, coagulopathy, dementia, and depression. For hospital teaching status, rural hospitals were selected as the reference category to provide a consistent baseline representing lower-resource care settings, thereby facilitating comparison with more specialized environments such as urban teaching hospitals. Although the proportion of rural cases was small, this approach is consistent with prior health services research examining gradients in hospital resources and care delivery. Appropriate odds ratios (ORs), 95% confidence intervals (CIs), and p-values were reported for the logistic regression analyses. Statistical significance was determined with P-values<0.05.

All analyses were conducted using IBM SPSS Statistics (version 31.0.0.0 [117]; IBM Co.) and RStudio (version 2025.09.2+418). The research procedures herein adhered to ethical principles and STROBE guidelines.

The NIS provides both unweighted discharge-level data and HCUP-provided discharge weights that allow for generation of nationally representative estimates [12]. In the present study, unweighted counts are reported to reflect the actual number of sampled hospitalizations, while weighted values are used for descriptive statistics and inferential analyses to approximate national estimates. Accordingly, values presented in Table 1 reflect weighted estimates and may exceed the unweighted sample size reported in the text.

From the 2019–2022 NIS, a total of 3,280 unweighted hospitalizations involving patients with renal failure undergoing MT for AIS were identified (1,990 males; 2,057 females). After application of HCUP discharge weights, these cases corresponded to a substantially larger nationally representative cohort found in Table 1. All descriptive statistics and regression analyses were performed using weighted data, whereas unweighted counts are provided for transparency. Female patients were older than male patients (mean age 77.56 vs. 72.23 years; P<0.001). Statistically significant sex-based differences were observed across multiple demographic, clinical, and hospital-level variables (Table 1). Female patients had shorter mean LOS (9.41 vs. 10.63 days; P<0.001) and lower mean total hospitalization charges ($216,151.96 vs. $243,450.88; P<0.001).

Comorbidity profiles differed by sex. Female patients had higher prevalence of arrhythmia, valvular disease, pulmonary HTN, paralysis, COPD, hypothyroidism, obesity, weight loss, anemia, dementia, and depression (all P<0.05). Male patients more frequently had liver disease, human immunodeficiency virus, alcohol abuse, drug use, metastatic cancer, and solid tumors without metastasis (all P<0.05). Rates of CHF, PVD, HTN, peptic ulcer disease, lymphoma, fluid and electrolyte disorders, and coagulopathy did not differ significantly between groups.

Discharge disposition differed significantly by sex (P<0.001). Among patients discharged routinely, 63.8% were male and 36.2% were female. Female patients were more commonly discharged to another facility or with HHC. In-hospital mortality prior to adjustment was higher among male patients (51.1% vs. 48.9%; P=0.018).

On multivariable logistic regression adjusting for demographic, socioeconomic, clinical, and hospital-level covariates, female sex was independently associated with increased odds of non-routine discharge (adjusted OR 1.536, 95% CI 1.391–1.696; P<0.001) (Table 2).

Increasing age was associated with higher odds of nonroutine discharge (OR 1.053 per year, 95% CI 1.048–1.058; P<0.001). Compared with White patients, Black race (OR 1.181, 95% CI 1.042–1.339) and Native American race (OR 4.114, 95% CI 1.600–10.578) were associated with increased odds of non-routine discharge (P=0.003 overall).

Relative to Medicare, Medicaid (OR 0.667, 95% CI 0.557–0.799), private insurance (OR 0.786, 95% CI 0.684–0.904), and self-pay status (OR 0.622, 95% CI 0.455–0.851) were associated with lower odds of non-routine discharge (P<0.001). Patients in the 51st–75th income quartile had reduced odds of non-routine discharge compared with the lowest quartile (OR 0.850, 95% CI 0.747–0.969; P=0.015).

Calendar year was significantly associated with discharge disposition, with lower odds of non-routine discharge in 2021 (OR 0.857, 95% CI 0.745–0.984) and 2022 (OR 0.801, 95% CI 0.701–0.916) compared with 2019 (P=0.002). Weekend admission was associated with slightly reduced odds of non-routine discharge (OR 0.895, 95% CI 0.806–0.995; P=0.040).

Several comorbidities were independently associated with increased odds of non-routine discharge, including paralysis (OR 1.654, 95% CI 1.483–1.845), diabetes (OR 1.259, 95% CI 1.143–1.387), weight loss (OR 2.742, 95% CI 1.065–7.055), fluid and electrolyte disorders (OR 1.515, 95% CI 1.358–1.691), coagulopathy (OR 1.557, 95% CI 1.246–1.944), and dementia (OR 1.832, 95% CI 1.436–2.336) (all P<0.05).

On multivariable logistic regression, female sex was not significantly associated with in-hospital mortality (adjusted OR 0.953, 95% CI 0.874–1.038; P=0.266) (Table 3).

Increasing age was associated with higher odds of mortality (OR 1.018 per year, 95% CI 1.013–1.023, P<0.001). Calendar year was associated with mortality, with lower odds observed in 2020 compared with 2019 (OR 0.831, 95% CI 0.734–0.940; P=0.003).

Hospital region and teaching status were significantly associated with mortality. Compared with the Northeast, hospitalization in the South was associated with lower odds of death (OR 0.801, 95% CI 0.706–0.909; P<0.001). Admission to urban teaching hospitals was associated with reduced odds of mortality compared with rural hospitals (OR 0.517, 95% CI 0.318–0.839; P=0.008). Rural hospitals were used as the reference category to facilitate comparison across gradients of hospital resources and specialization. However, given the relatively small proportion of cases originating from rural centers, these estimates may be less stable and should be interpreted with caution. Several comorbidities were independently associated with increased mortality, including liver disease (OR 1.578, 95% CI 1.112–2.241), fluid and electrolyte disorders (OR 1.091, 95% CI 1.002–1.188), and coagulopathy (OR 1.387, 95% CI 1.193–1.612) (all P<0.05). Arrhythmia was associated with slightly reduced odds of mortality (OR 0.911, 95% CI 0.831–0.998; P=0.046). Obesity (OR 0.854, 95% CI 0.759–0.962), weight loss (OR 0.453, 95% CI 0.231–0.891), paralysis (OR 0.718, 95% CI 0.648–0.796), and depression (OR 0.504, 95% CI 0.423–0.600) were also significantly associated with lower odds of in-hospital mortality.

In contrast to mortality outcomes, female sex was independently associated with higher odds of non-routine discharge following MT. This finding remained robust after adjustment for a wide range of covariates and represents a key contribution of the present study. Discharge disposition is a clinically meaningful outcome that reflects not only neurological recovery but also functional status, social support, caregiver availability, and access to post-acute care resources.

Prior literature has documented sex-based disparities in discharge outcomes following stroke with women more likely to be discharged to skilled nursing facilities or require home health services. The present study extends these observations to a highly specific and vulnerable subgroup (patients with renal failure undergoing MT) suggesting that these disparities persist even after accounting for differences in age, comorbidity burden, and hospital characteristics.

Several factors may contribute to this observed disparity. Female patients in the cohort were older on average and exhibited higher prevalence of conditions associated with frailty and functional impairment, including dementia, depression, weight loss, and anemia. While these variables were adjusted for in multivariable models, residual confounding related to baseline functional status and social determinants of health cannot be fully excluded. In the context of renal failure, these disparities may be further influenced by increased frailty, higher burden of systemic illness, and reduced physiological reserve, which may amplify the impact of social and functional determinants on discharge planning. Additionally, women are more likely to live alone or lack informal caregiver support, which may influence discharge planning decisions and necessitate non-routine discharge even in the absence of severe neurological deficits [6].

Importantly, the increased odds of non-routine discharge among women should not necessarily be interpreted as evidence of inferior procedural success or acute neurological outcomes. Rather, this finding underscores the need for sex-aware discharge planning and early engagement of rehabilitation and social work services, particularly for female patients with renal failure who may be at heightened risk for post-discharge dependency.

The present study also identified temporal trends in discharge disposition, with lower odds of non-routine discharge observed in later calendar years compared with 2019. This finding may reflect improvements in stroke systems of care, evolving discharge practices, or increased utilization of home-based services over time. While the COVID-19 pandemic overlapped with the study period and may have influenced hospital workflows and discharge decisions, the directionality of this association suggests potential adaptation of care pathways to better support post-acute recovery.

Socioeconomic factors also demonstrated associations with discharge outcomes. Patients in higher income quartiles exhibited lower odds of non-routine discharge, consistent with prior studies linking socioeconomic status to access to outpatient services, caregiver support, and rehabilitation resources [13,14]. These findings highlight the multifactorial nature of discharge disposition and emphasize the importance of addressing social determinants of health alongside clinical factors in stroke recovery.

One of the principal findings of this study is the absence of a statistically significant association between sex and in-hospital mortality following MT among patients with renal failure. This finding persisted after rigorous multivariable adjustment and contrasts with some prior reports suggesting higher mortality among women following AIS or endovascular intervention [15]. However, it aligns with a growing body of literature indicating that observed sex-based differences in stroke outcomes may be largely attributable to confounding factors such as age, comorbidity burden, baseline functional status, and differences in care delivery rather than sex itself [6].

Patients with renal failure represent a particularly high-risk subgroup, characterized by systemic inflammation, endothelial dysfunction, altered platelet activity, and metabolic derangements that may amplify peri-procedural risk [5]. Within this context, the lack of a sex-specific mortality signal suggests that once patients with renal failure are selected for MT, short-term survival is driven predominantly by age, systemic illness severity, and acute medical complications rather than sex. This finding has direct clinical relevance, as it supports equitable application of endovascular therapy without sex-based bias in patient selection, even among medically complex populations.

Several non-sex-related factors were independently associated with in-hospital mortality in the present study, including increasing age, liver disease, fluid and electrolyte disorders, and coagulopathy. These findings are consistent with prior studies demonstrating that systemic organ dysfunction and metabolic instability play a central role in determining early outcomes after stroke and endovascular intervention [16,17]. Conversely, admission to urban teaching hospitals was associated with lower odds of in-hospital mortality, underscoring the potential impact of institutional experience, multidisciplinary care models, and resource availability on outcomes in high-acuity stroke patients.

The existing literature on sex differences following MT has produced heterogeneous findings. Some prior registry-based analyses have reported worse functional outcomes among women despite similar rates of recanalization, while others have found no significant differences after adjustment [6,8,18]. Few studies, however, have specifically examined patients with renal failure.

By focusing on renal failure patients and leveraging a large, nationally representative database, the present study fills an important gap in the literature. Our findings suggest that while sex does not independently influence short-term survival after thrombectomy in this population, it does play a role in post-acute care needs as reflected by discharge disposition. This distinction is critical for clinicians, as it separates procedural safety and effectiveness from broader recovery and care planning considerations.

Although the present analysis does not include procedural-level data, the findings remain clinically relevant for peri-procedural and post-acute care planning. Specifically, the identification of sex-based differences in discharge disposition within a high-risk renal failure population may inform early multidisciplinary planning, including rehabilitation needs, social support assessment, and resource allocation. These findings may complement procedural decision-making by providing additional context regarding expected recovery trajectories in medically complex patients undergoing thrombectomy. Importantly, outcomes following MT extend beyond technical success and include the broader trajectory of recovery and healthcare utilization. Understanding factors that influence discharge disposition in high-risk populations is increasingly relevant in modern neurointerventional practice, where multidisciplinary care models and post-acute resource planning are integral components of comprehensive stroke care.

This study has several limitations inherent to its retrospective design and use of administrative data. A major limitation of this study is the lack of granular stroke-specific and procedural variables within the NIS, including baseline stroke severity (e.g., National Institutes of Health Stroke Scale), occlusion location, time to reperfusion, and angiographic outcomes such as Thrombolysis in Cerebral Infarction revascularization grade. These variables are critical determinants of both procedural success and clinical outcomes following MT and may differ by sex. An additional limitation relates to the definition of renal failure within the NIS. The use of ICD- 10 codes (N17, N18, N19) encompasses a heterogeneous population including acute kidney injury, chronic kidney disease, and unspecified renal failure. These conditions differ significantly in chronicity, pathophysiology, and associated risk profiles, which may influence outcomes following MT. The inability to distinguish between these entities introduces heterogeneity that may affect interpretation of the observed associations. Additionally, the relatively small proportion of hospitalizations occurring in rural settings may result in less stable estimates when used as a reference group in regression analyses.

Although propensity score matching was considered, multivariable regressions were selected to help preserve sample size and to allow adjustment for relevant covariates given similar cohort sizes found at baseline prior to case weighing.

Coding inaccuracies and misclassification bias are possible, although the use of ICD-10 codes and HCUP-recommended weighting methods is consistent with prior high-quality NIS studies. Finally, the analysis focuses on in-hospital outcomes and does not capture long-term mortality or functional recovery after discharge.